Increasing specialist intensity at weekends to improve outcomes for patients undergoing emergency hospital admission: the HiSLAC two-phase mixed-methods study

Article history

The research reported in this issue of the journal was funded by the HS&DR programme or one of its preceding programmes as project number 12/128/17. The contractual start date was in February 2014. The final report began editorial review in February 2020 and was accepted for publication in October 2020. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HS&DR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

© Queen’s Printer and Controller of HMSO 2021. This work was produced by Bion et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2021 Queen’s Printer and Controller of HMSO

Background

The stimulus for the NIHR HSDR programme’s call was the increasing difficulty of providing satisfactory emergency health care in the NHS. The background to this problem was multifaceted. The European Working Time Directive in 2003 limited hours of work;2 the NHS responded by moving junior doctors to shift-working, introducing non-physician clinicians, and increasing the numbers of consultants on new contracts that explicitly identify and reimburse out-of-hours work. This increased staffing and costs, but did not improve team working, training or continuity of care. 3 At the same time, a new contract for family doctors [general practitioners (GPs)] allowed them to withdraw from providing out-of-hours services. 4 Emergency admissions (EAs) continued to rise every year at a rate similar to that of the increase in consultant staff and faster than that of the background population increase, particularly those involving frail elderly people,5 while policies to deliver more health care in the community reduced hospital beds6 without a sufficient compensatory expansion in social care funding. 7 This increased the number of delayed hospital discharges, which adversely affected care quality8 and blocked beds, making it even more difficult to accommodate the growing numbers of EAs. 9 Temporal changes in medical staffing, emergency department (ED) attendances, EAs and hospital beds are shown in Figure 1. We discuss the apparent improvement in reducing delayed discharges of care in Chapter 4.

FIGURE 1.

Trends in NHS England emergency care workload, staffing and beds: 2011/12–2018/19. NHS England overall monthly average number of delayed days; number of overnight beds; annual average number of consultants and specialists, and other doctors; ED total attendance rate (% population); EA rate (% population); and percentage of ED attendances admitted. DTOC, delayed transfer of care; FTE, full time equivalent.

Public concern about the deteriorating quality of hospital care was exacerbated by failings in professional standards identified by the Mid-Staffordshire NHS Foundation Trust Public Inquiry. 10,11 There was also a longstanding perception that hospital care out of hours (at nights and weekends) was unreliable and potentially unsafe because of evidence showing the ‘weekend effect’. The Royal College of Physicians and the Society of Acute Medicine’s survey12 found that only 20% of hospital specialists were available at weekends for periods exceeding 8 hours, 18% reported never attending hospitals at weekends and 73% of acute physicians were not contractually obliged to provide medical care at weekends; only 19% of responding hospitals reported having a formalised rapid response team for acutely ill patients. Only 39% of specialists working in acute medical units (AMUs) reported having protected time for this work free of other duties, and providing care for blocks of time greater than a single day. Suboptimal specialist input had also been identified in the National Confidential Enquiry into Patient Outcome and Death (NCEPOD) audit of deaths following emergency hospital admission:13 at 12 hours following admission, 40% of patients had not been seen by a consultant, and in 12.4% of cases there was no documented evidence of consultant review. Of the 95 cases in which the assessors considered the delay in consultant review to have been unacceptable, the delay was considered to have adversely affected the accuracy or timeliness of diagnosis in 32.6% of cases, and may have contributed to adverse outcomes [intensive care unit (ICU) admission, worsening prognosis or death] in 49.5% of cases.

The profession responded to these findings with proposals for the reconfiguration of hospital services,14,15 and the development of enhanced standards of consultant practice. The Academy of Medical Royal Colleges convened a working group to develop standards for daily consultant review. 15 In the absence of strong direct evidence, the working group arrived at a consensus recommendation, namely that all hospitalised patients should be reviewed by a consultant every day unless prior review had determined that this was not necessary. The Academy of Medical Royal Colleges’ report concluded that systematic evaluation of the standards was required, within a research framework.

Seven-day services

Initiatives to improve access to emergency care were brought together under the UK government’s policy of ‘7-day services’,16 with the development of 10 clinical standards for emergency care,17 six of which directly or indirectly required front-line consultant involvement for delivery. Initially, NHS England required these standards to be introduced at scale and pace within 3–4 years, backed by incentives, rewards and sanctions. This was subsequently diluted to achieving four ‘priority standards’ across the NHS by 2020 (quotations in this paragraph contain public sector information licensed under the Open Government Licence v3.0, URL: http://nationalarchives.gov.uk/doc/non-commercial-government-licence/version/2/). 16 These are:

• standard 2 – EAs assessed by a consultant within 14 hours

• standard 5 – timely access to diagnostic services according to urgency

• standard 6 – access to consultant-directed interventions 24/7

• standard 8 – consultant review of patients in acute medical, surgical and intensive care units twice daily by consultants working blocks of days; daily consultant review of ward patients.

Whether or not these standards were the correct solution to problems of access, and how they were to be measured, monitored and funded, became a source of tension between the professions on one side18 and the government and Department of Health and Social Care19 on the other, with NHS England in the middle. A publication by Freemantle et al. 20 in 2015 presented a surplus weekend admission mortality of 16% as justification for 7-day services as the ‘solution’. During negotiations to change NHS doctors’ contracts to facilitate the introduction of 7-day services,21 the then Secretary of State for Health and Social Care also linked weekend deaths to doctor numbers,22 stating that ‘. . . someone is 15% more likely to die if admitted on a Sunday than on a Wednesday because we do not have as many doctors in our hospitals at the weekends as we have mid-week’ (contains Parliamentary information licensed under the Open Parliament Licence v3.0, URL: https://parliament.uk/site-information/copyright-parliament/open-parliament-licence/) and referring to the weekend effect as a ‘global scandal’. This precipitated a vigorous antipathetic response to 7-day services on social media (Figure 2) and contributed to a threat of strike action by junior doctors. 23 The ‘weekend effect’ seemed to have evolved from a problem requiring scientific exploration into a political tool for implementing health policy. It was against this background that the High-intensity Specialist-Led Acute Care (HiSLAC) project was established.

FIGURE 2.

Frequency of citations of the ‘weekend effect’, 7-day services and integrated care. a, Twitter, Inc., San Francisco, CA, USA (www.twitter.com); b, up to 31 October 2019.

The ‘weekend effect’

The weekend effect was first reported by Bell and Redelmeier in 2001,24 who reported a significantly higher mortality rate associated with EA to hospital at weekends for 23 of the 100 leading causes of death in Canada. The authors were unable to exclude the possibility that patients admitted at the weekend were sicker, but hypothesised that the cause for the surplus mortality was the reduction in medical staffing, particularly senior doctors, in hospital at weekends. In a perceptive accompanying editorial to that paper, Halm and Chassin25 stated that:

Disentangling the potential causal pathways would require painstaking detective work . . . of first accounting for the biologic and social determinants of risk and then identifying the precise differences in processes of clinical care that explain the differences in risk-adjusted outcomes.

Halm and Chassin25

However, few investigators attempted this detective work in the intervening years. Although the literature quantifying the magnitude of the weekend effect continued to accumulate, only one paper had attempted to investigate the cause: a single-centre 8-year study from Dublin that found that patients admitted at weekends were sicker,26 based on adjustment using abnormal biochemistry results. The HiSLAC project therefore chose to focus on Halm and Chassin’s challenge, as the weekend effect was of importance to patients, health professionals, policy-makers and politicians.

How is the weekend effect calculated?

The weekend effect is a complex metric, a ratio of two ratios: the weekend mortality rate divided by the weekday mortality rate. This means that there are four primary mechanisms by which an excess mortality might be generated, as shown in Figure 3.

FIGURE 3.

Potential primary mechanisms of the weekend effect.

The definition of ‘weekend’ requires some elaboration. In the Christian tradition, Sunday was a day for worship and rest from labour; the weekend now usually incorporates Saturday and Sunday in the western world. The Jewish Sabbath is from sunset on Friday to sunset on Saturday; the Muslim days of rest in some countries was from Thursday to Friday, but increasingly now spans Friday to Saturday. The most frequent definition of ‘weekend’ in the scientific literature is that period of the week from midnight on Friday until midnight on Sunday. This is the definition we use in the HiSLAC project. In our literature reviews we have used the data as published in the source documents.

The potential causal pathway for the weekend effect

The causal pathway could include a number of independent or interlinked underlying causes that could be community or hospital based, and could be related to case mix, clinical care quality or measurement artefact (Figure 4).

FIGURE 4.

The acutely ill patient pathway and factors potentially influencing weekend–weekday admission mortality differences. Reproduced with permission from Chen et al. 27 © Author(s) (or their employer(s)) 2019. Re-use permitted under CC BY. Published by BMJ. This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/. This figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Chen et al. 27. © Author(s) (or their employer(s)) 2019. Re-use permitted under CC BY. Published by BMJ.This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/. This figure includes minor additions and formatting changes to the original figure.

Rationale for and design of the HiSLAC project

When the Academy of Medical Royal Colleges’ working party produced its recommendations for daily consultant review,28 it did so recognising that this was a consensus opinion across all royal colleges, and that the supporting scientific evidence was largely indirect and observational in nature. Given the estimated cost of implementing 7-day services (including increased consultant presence at weekends) of between £1.07B and £1.43B,29 research using empirical data incorporating health economics modelling was needed. It was self-evidently impossible to conduct a conventional prospective randomised controlled trial of high- compared with low-intensity consultant staffing, but the planned roll-out of 7-day services provided an unique opportunity for a ‘natural experiment’ conducted under real-world conditions to test the hypothesis that increasing the numbers of (senior) doctors would result in better quality of care for patients following EA to hospital, with weekend admission mortality rates as one of the key indicators.

Doing this required the development of measures of specialist (consultant grade) doctors’ input across the health service, and metrics for care quality that would allow us to map 7-day service penetration and impact over time. At the time of implementation of 7-day services, it was not clear how the uptake of the standards would be measured and monitored. In the event, NHS Improvement required trusts to perform board-assured audits based on local case record reviews. However, the methodology changed in subsequent years, making analysis of secular trends difficult, and the data did not allow for the calculation of the ‘dose’ of specialist input. The 7-day service standard of greatest relevance to potential consultant impact on the weekend effect is standard 2, ‘emergency admissions assessed by a consultant within 14 hours’. For 2017 and 2018, the methodology for standard 2 was based on local case record review of 20 weekday and 10 weekend admissions; the results were presented as a single figure of percentage compliance across the 7 days of the week. 30 In 2019, the methodology was changed to a binary ‘yes’ or ‘no’ response based on each trust’s self-assessment of whether or not 90% of their audited patients had been reviewed by a consultant on weekends and weekdays combined. 31 Neither method permitted the separation of weekend from weekday performance without access to trust-specific data, which we were unable to obtain from NHS Improvement. Moreover, the definition of consultant review is not specified; it could mean direct bedside review by the consultant, remote review (e.g. a ‘board round’), or that a patient had been reviewed by a junior doctor working under the supervision of a consultant present in the hospital but not at the patient’s bedside.

Consultant numbers were available by trust, but trusts provided no information on how those consultants were deployed during the week, in hours or out of hours, or in caring for EAs.

The first phase of the HiSLAC project, therefore, required us to develop measures of the ‘dose’ of specialist input. We also had to acquire sufficient preliminary data (phase 1) to demonstrate that a project extending over several years was achievable (phase 2). Although HiSLAC was initially conceived as a 3-year project, the HSDR programme board proposed that we extend this to 5 years to allow sufficient duration to determine secular change. Appropriate stewardship of public funds was assured by establishing an independent Governance and Oversight Board to control ‘decision gates’ based on meeting specific progress goals.

An overview of the study is presented in Figure 5 and the initial research plan design is shown in Figure 6. We chose two key methods to strengthen the observational nature of the proposed research: triangulation and difference-in-difference analyses. Triangulation employs concurrent use of different (and preferably independent) measures of a particular phenomenon; if similar observations are made using different methods, this increases the confidence in the credibility of those observations. We did this by using quantitative and qualitative (mixed-methods) approaches, and by combining whole-system measurements across NHS England with in-depth examinations of a subgroup of 20 hospitals. Difference-in-difference analyses strengthen causal inferences by reducing the impact of confounding variables (which indirectly influence associations between dependent and independent variables) in studies where random assignment to intervention or control is infeasible. 32 We used this approach in comparing trust-level differences between weekends and weekdays (specialist intensity, admission mortality, error and adverse event rates, and judgements of care quality) and then comparing those differences with change over time (a ‘triple’ difference).

FIGURE 5.

Overview of the HiSLAC project.

FIGURE 6.

The initial research plan flowsheet for the HiSLAC project. CPR, cardiopulmonary resuscitation; LoSLAC, Low-intensity Specialist-Led Acute Care; ICNARC-CMP, Intensive care national audit & research centre – case mix programme; NICE, National Institute for Health and Care Excellence; NHS(E), NHS England; NHS-IQ, NHS Improving Quality; non-op, non-operative; ONS, Office for National Statistics; PAS, Patient Administration System; QA, quality assurance.

The research plan was broadly retained over the 5 years of the project, with two modifications. The first modification was that, despite repeated efforts, we were unable to realise the proposed collaboration with NHS Improving Quality [renamed NHS Improvement] when it assumed the regulatory roles previously held by the Trust Development Authority (for NHS trusts) and Monitor (for NHS foundation trusts) in 2016. Now merged with NHS England, NHS Improvement holds the annual self-reported audit data returns from each trust for implementing 7-day services. These data are reported by individual trusts and are aggregated by NHS Digital for each standard in binary form as having been attained or not, but the original raw data were not made available to the HiSLAC project; therefore, we were unable to compare HiSLAC data on specialist intensity with each trust’s self-reported 7-day service standards. In addition, NHS Improvement’s methodology for collecting 7-day service data and the definitions employed have changed over time, making secular changes difficult to assess.

The second modification was the decision to perform an additional study to test the hypothesis that patients admitted to hospital at weekends were sicker. As this required detailed information, including physiology (vital signs measurements), that is not collected nationally in a standardised format, we performed a single-centre study using the clinical information system available at a large urban university teaching hospital.

Aims and objectives

The aim of the HiSLAC project was to determine whether or not increasing the intensity of specialist-led care at weekends improves outcomes for patients admitted to hospital as emergencies at weekends.

The objectives of the HiSLAC project were to:

• quantify specialist input into the care of EAs and map changes in provision over time

• compare the quality of care in hospitals with high levels of weekend specialist cover with that of hospitals with lower levels, using mixed methods

• determine whether or not the case mix of patients admitted at weekends differs from that of patients admitted on weekdays

• develop a health economics model to estimate the costs and health outcomes [quality-adjusted life-years (QALYs)] associated with increased intensity of specialist provision.

Establishment of the HiSLAC national collaboration

We wished to engage the participation of as many adult acute non-specialist hospital trusts in England as possible. We therefore sought and obtained endorsement of the HiSLAC project from NHS England, the NHS Confederation and the Academy of Medical Royal Colleges in the form of a joint letter to trust chief executives and medical directors asking them to support the project. In the absence of a centralised and accessible list of names and contact details, the project team telephoned the personal assistant to the chief executive of every acute trust in England to obtain this information, which was provided with varying degrees of confidence and alacrity. The supporting letters and a description of the project were then sent by e-mail in February 2014, with follow-up e-mails and telephone calls to non-responders. We asked each chief executive to appoint a local HiSLAC project lead for the trust. Of 141 acute hospital trusts in England, 127 agreed to participate and 115 trusts contributed data to subsequent surveys.

Acquiring data on adult emergency admissions across acute non-specialist trusts in England

We applied to NHS Digital for all acute admission data for English hospitals from 1 April 2007 until 31 March 2018. The data set, Hospital Episode Statistics (HES), is a mandated commissioning minimum data set in which all admitted patient care activity is captured in all English hospitals. The data set is applied consistently across the service, backed by appropriate training in the various coding systems it uses. With the exception of minor modifications, the data set is constant over time. The units of data are ‘episodes’ of care spent by one patient under the care of one consultant. Typically, one spell in hospital consists of just one episode of care, although complex cases can involve several sequential episodes. HES is highly comprehensive, capturing administrative, demographic and clinical information, most of which is coded. There are comprehensive coding and classification systems for diagnoses, procedures, units of administrative geography, specialties, NHS organisations and case-mix categories, among others. Although largely designed for administrative purposes, given the large number of clinical data it contains, HES is commonly used in observational studies of clinical outcomes such as mortality and emergency re-admission.

The HES data sets were stored on a dedicated secure server in the University of Birmingham and processed using Microsoft Structured Query Language (SQL) Server 2008 (Microsoft Corporation, Redmond, WA, USA), which is database software capable of handling large numbers of data. From an initial data set of 220 million patient records, we extracted those records relating to EAs over the decade.

Once in our data warehouse, we added a number of derived variables to the data set. We linked the primary diagnosis on discharge (captured using the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision33) to the Clinical Classifications Software category to obtain a case-mix variable. We mapped the secondary diagnoses to the Charlson Comorbidity Index as a further measure of case severity. We also attached income deprivation scores to the data. This is a domain of The English Indices of Deprivation 2010,34 which is the standard measure of area deprivation used by the UK government. Data on variables such as income, unemployment education and crime are collected at a small level of census geography called a lower-level super output area. Income deprivation, the domain used in this study to adjust for socioeconomic status, is a score based on the proportion of residents of a lower-level super output area whose income is ≤ 60% of the national median income. As this domain uses more up-to-date data than other domains of the score, we chose this as our socioeconomic status indicator.

Hospital Episode Statistics were used in several phases of the study and several iterations of analysis were completed, each differing slightly in objectives, methods and inclusion criteria, depending on the requirement of the study. First, we performed a detailed analysis of in-hospital mortality in financial year 2013/14. In this part of the study (subsequently published),35 we examined the discharge episodes of all patients admitted as emergencies to a list of hospitals that we had previously identified as belonging to non-specialist acute trusts. It was important to understand mortality trends over time in this study, as the introduction of the 7-day services policy was implemented over time and in the context of changes in the underlying epidemiology of needs for emergency care. For this reason, we needed to examine secular trends in hospital mortality. To do this, we analysed mortality data for EAs between financial years 2007/8 and 2017/18. For the longitudinal analyses, EA data were analysed for those trusts that participated in the point prevalence survey (PPS).

For some components of the study, we would have preferred to use HES, but the data were not available. HES are typically made available to the research community (in a cleaned format) approximately 6 months in arrears. To select cases for our case note review, we needed data more quickly than this, so we arranged for participating hospitals to send us pseudonymised extracts from their patient administration system. These data are effectively the raw material of HES, as they are uploaded to NHS Digital, where they are cleaned and a number of derived variables are added. Using data in this ‘raw’ format was problematic as extra work was required to clean the data, including eliminating duplicate records and, in some cases, censoring data that were insufficiently complete. This had to be done for 20 participating hospitals, all of which had slight differences in formatting and data quality. The data were managed similarly to HES in a standalone secure server, where they were uploaded to the Microsoft SQL Server platform. These data were then used to randomly select cases of adult non-surgical EAs to be included in the case note review phase of the study.

Developing and piloting a specialist intensity metric

Ethics

HiSLAC was approved by the Health Research Authority [Integrated Research Application System (IRAS) project ID 139089] and by the Welsh Research Ethics Committee (reference 13/WA/0372) as service evaluation of an existing form of health-care delivery without collecting patient-identifiable data. The committee noted that informed consent was not required for accessing anonymised patient records and only the qualitative research observations and interviews needed consent. Consent for this component (involving focus groups, interviews, and observations employing elements of an ethnographic approach) was obtained from participants after they received the participant information sheet (see Report Supplementary Material 1).

There were no variations in the ethics status of this research during the study.

Introduction

The aim of this mixed-methods review was to quantify the magnitude of the weekend effect on hospital mortality and, concurrently, to conduct a qualitative narrative synthesis of the literature on the mechanisms of the weekend effect. In practice, however, we were not able to identify literature that provided sound evidence on the mechanisms that underpinned the weekend effect. We therefore conducted an exploratory qualitative study consisting of focus groups to explore staff and patient perceptions of how differences in structures and processes of care between weekdays and weekends might contribute to the weekend effect. The methodology was planned and published as a unified parallel process42 and the results were published as two separate papers. 27,43 We provide a summary here.

Methods

This review was registered with PROSPERO (registration number CRD42016036487). MEDLINE, CINAHL (Cumulative Index to Nursing and Allied Health Literature), HMIC (Health Management Information Consortium), EMBASE™ (Elsevier, Amsterdam, the Netherlands), EThOS (Electronic Theses Online Service), CPCI (Conference Proceedings Citation Index; Clarivate Analytics, Philadelphia, PA, USA) and the Cochrane Library were searched from January 2000 to April 2015, with the MEDLINE search updated to November 2017, using terms related to ‘weekends or out-of-hours’ and ‘hospital admissions’. 42 Records were screened by the review team for two component reviews: a systematic review of the magnitude of the weekend effect and the proposed qualitative analysis of the mechanisms of the weekend effect.

Studies reporting predominantly unselected hospital admissions were included in the reviews, covering emergency and elective adult, paediatric, medical, surgical and obstetric admissions. For studies that reported both aggregated and condition-specific weekend effects, only the aggregated data were used in the quantitative analyses of the systematic review. We chose to focus on unselected, rather than condition-specific, admissions to avoid duplicating meta-analyses focusing on condition-specific admissions. 42,44–46 We did not include studies that compared night-time with daytime admissions only. Independent duplicate coding of potentially relevant studies was performed for the first 450 (40%) potentially relevant records to maximise consistency of approach; the remaining studies were then assessed by single reviewers. Final study selection was determined by two reviewers. Any discrepancies in study coding and selection were resolved by discussion or seeking further opinion from other review team members. Risk-of-bias assessment performed independently by two reviewers focused on the level of statistical adjustment categorised as ‘comprehensive’, ‘adequate’ [(1) adjustment for acute physiology; and (2) adjustment for context using other measures of urgency or severity], ‘partial’ and ‘inadequate’.

For the quantitative review, the primary outcome was the weekend effect on mortality. Secondary outcomes included adverse events, length of hospital stay and patient experience. The data were meta-analysed using a Bayesian random-effects model allowing within-study variation and between-study heterogeneity. Analyses were undertaken using (log)-adjusted odds ratios (ORs) (or hazard ratios or rate ratios if ORs were not reported) and the reported standard errors or equivalent. Studies were therefore implicitly weighted by the estimated variance of individual effect estimates. Where multiple estimates based on different reference day(s) were reported, we used the estimate based on or including Wednesday as the reference group. The primary meta-analysis included all types of admissions. Exploratory subgroup analyses were performed for mixed, emergency, elective and maternity admissions. We calculated the I² statistic to quantify statistical heterogeneity between studies (with I² > 50% indicating a substantial degree of heterogeneity). 47 The overall quality of evidence for the primary and secondary outcomes was assessed using the GRADE (Grading of Recommendations, Assessment, Development and Evaluations) framework. 48

Screening of papers for the review of mechanisms of the weekend effect did not identify any high-quality studies with robust evidence of the mechanisms of the weekend effect. As a result, we undertook a qualitative study to explore the views of health-care staff and patients on how the quality of hospital care differed between weekends and weekdays, and how this could contribute to the weekend effect. Quality of care was conceptualised using Darzi’s three elements of effectiveness, safety and patient experience. 49,50 Patients and staff were recruited from three hospital trusts in the Midlands, the north-east and the south-west of England. Patients were recruited through two existing acute-care patient and public involvement groups and through face-to-face recruitment of patients and their relatives during the patient’s stay in the acute medical wards. Focus group moderators and scribes, and interviewers, were trained qualitative researchers. Focus groups were audio-recorded and transcribed verbatim. Data analysis employed thematic analysis,51 supported by NVivo (QSR International, Warrington, UK) qualitative data analysis software, to explore perceptions of quality and safety of care at weekends and the impact on patient outcomes.

Results

Between January 2000 and November 2017, we identified 6441 records, 613 of which passed initial screening. Of these, 319 were subsequently excluded, 224 were considered for qualitative analysis and 68 met the criteria for the quantitative review. None of the papers met inclusion/quality criteria for the qualitative review, but the most relevant were used to inform the topic guides for the qualitative study.

Quantitative review and Bayesian meta-analysis results

Parts of this section have been reproduced with permission from Chen et al. 27 © Author(s) (or their employer(s)) 2019. Re-use permitted under CC BY. Published by BMJ. This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/ The text below includes minor additions and formatting changes to the original text.

The 68 selected studies (see Appendix 3) encompassed more than 640 million admissions (range 824 admissions from a single hospital to 351,170,803 admissions from a national database). 52,53 Patient populations included unselected general admissions (n = 11), medical admissions (n = 1), surgical admissions (n = 3), EAs (n = 34), elective surgical admission (n = 5) and maternity (n = 13).

Definitions of ‘weekend’ varied. The majority (n = 28) of studies defined the weekend as including Saturday and Sunday, whereas 14 provided no definition. Nineteen variably included Friday evening, Monday morning or both, and seven included Friday daytime. Mortality timing included death in hospital, and at 7 and 30 days post admission. These differences may contribute to statistical heterogeneity, although bias is limited by the metric that employs the difference in weekend admission with the difference in weekday admission (as described in Chapter 1).

For the primary outcome of mortality, we found only one study54 that had adjusted comprehensively for potential sources of bias, including acute laboratory tests and source of admission. Three other studies had adjusted for acute physiology in regression models (to which we can now add the HiSLAC study, published subsequently and described in this report; see Chapter 5). 26,55–57 Adjustment using other non-physiological measures related to patient acuity or dependence (e.g. route of admission) had been used in 17 studies. Of the remaining studies, adjustment was partially achieved in 20 and inadequate in 27.

Bayesian meta-analysis of all types of admissions provided a pooled estimate for the odds of death for weekend admissions of 1.16% [95% credible interval (CrI) 1.10% to 1.23%] (Figure 7). There was wide variation in the estimated weekend effect between studies and subpopulations; although the value for I² (measuring between-study variance relative to total variance) is low, there is substantial uncertainty (16%, 95% CrI 0% to 62%). The posterior predictive interval suggests that a new study would have an estimated OR of between 1 (no weekend effect) and 1.34 (odds of death 34% higher).

FIGURE 7.

Bayesian meta-analysis of the weekend effect on mortality following hospital admissions. Mohammed et al. 75 and Ruiz et al. 59 contributed to two estimates for each country, as the weekend effect was estimated separately for different subpopulations (e.g. emergency and elective admissions). ‘Posterior predictive’ indicates the predictive interval (see Quantitative review and Bayesian meta-analysis results) obtained from the Bayesian meta-analysis. I² = 16% (95% CrI for I² 0% to 62%). The I² represents the ratio of between-study variance to total variance in this three-level model. The apparently low I² could be attributed to the between-study variance being relatively small compared with the between-estimate variance within individual studies. As the wide CrI indicates, the I² statistic was estimated with substantial uncertainty. Several studies included in the review were not included in this meta-analysis because of substantial overlap of data between studies, in which case studies judged to have adopted the most comprehensive statistical adjustment were selected.

The multivariate metaregression shows that including measures of acute physiology in statistical adjustment produced estimates of the weekend effect close to null, with ORs for mortality approximately 15% lower than those of other studies (see Appendix 4). The weekend effect is greater for studies including elective admissions (ratio of ORs 1.31, 95% CrI 1.23 to 1.38) and is lower for studies including maternity admissions (ratio of ORs 0.84, 95% CrI 0.77 to 0.90). Finally, across all studies, there did not appear to be a secular trend: the magnitude of the weekend effect was similar over the 17 years. However, a study from the USA based on the National Inpatient Sample reported a reduction in the weekend effect on mortality between 2003 and 2013. 85

Nineteen studies examined the impact of weekend admission on the risk of adverse events. 52,53,65,76,81,84,86–97 There was heterogeneity in populations and types of adverse events, and no consistent direction of effect or association with the magnitude of the weekend effect. No study adjusted for physiological severity of illness, even though sicker patients and non-survivors are more susceptible to adverse events. 98 Only one study examined patient perceptions in relation to EA timing, and found greater satisfaction with information provided in the ED for weekend admissions. 99

Discussion

This mixed-methods examination of the weekend effect is novel in that it combines what is, to our knowledge, the largest quantitative analysis to date with a detailed exploration of care processes at weekends. The use of Bayesian methods for the quantitative review provided a point estimate for the weekend effect (16%), which is in line with other studies, but a wider CrI, which reflects appropriately the variations reported both within and between studies. This higher mortality associated with weekend admission to hospital is a consistent finding across health systems and over time. The qualitative study identified four key deficiencies in processes of care at weekends that are relevant to the health service in England. However, convincing evidence that hospitals are responsible for a causal relationship between processes of care and outcome is elusive for unselected hospital admissions.

Moreover, there is evidence that patients presenting at weekends are sicker than those on weekdays, and both staff and patients observe that the initial care of more severely ill patients following weekend admission is at least as good as, and possibly better than, the care provided on weekdays. Sicker patients might be explained by fewer patients presenting to the ED or being admitted (i.e. a change in the denominator), but could also be a consequence of delays in care of patients in the community pre admission. Reduced community services at weekends were identified by clinical staff as impeding patient discharge, with secondary consequences for timely admission.

We do not imply that the absence of a proven relationship between hospital care processes and weekend admission outcomes means that there is no relationship. The process deficiencies identified by the qualitative study have at least the potential for patient harm, and certainly add to workplace stresses for staff. However, it seems likely that the relationship is complex. For example, standardisation of stroke care has improved very considerably over the last decade, with the development of a centralised hyperacute stroke unit model in London and an integrated model in Greater Manchester;100 however, despite evidence of substantial improvement in reliability of best practice in both centres,101 a secular trend for a reduction in stroke-associated weekend effects between 2008 and 2014 occurred nationally and did not particularly favour London’s centralised model. 102 We conclude, therefore, that insufficient progress has been made since Halm and Chassin called for ‘painstaking detective work’25 to unravel the causal relationships, and that this research needs to examine the whole of the causal pathway, from community to hospital discharge.

Introduction

As described in Chapter 1, during 2012 and 2013, several initiatives were broadly focused on improving equity of access to quality health care in the NHS across all days of the week, particularly (but not exclusively) for emergency care. There was a degree of consensus between the professions and health policy-makers that weekend services were suboptimal, and that front-line medical leadership (GPs as well as hospital consultants) was central to achieving improvements. There were lower levels of agreement about how this could be funded and delivered, and, in 2015, relationships between politicians and health-care professionals deteriorated when, in the midst of contract negotiations, the Department of Health and Social Care19 and the Secretary of State for Health and Social Care22 made public statements linking the weekend effect to a reduction in specialist availability in hospitals at weekends. There were two problems with implying a causative relationship. The first was that there was no evidence to support the assertion. The second was the absence of any objective measure of specialist availability. The HiSLAC project’s 5-year longitudinal study was under way by this stage, and, given the need for objective evidence, we decided to publish the first year’s data in a cross-sectional study35 while continuing to collect data through to 2019. We describe both the published study and the longitudinal study here.

Methods

We described earlier (see Chapter 2) the establishment of the HiSLAC collaboration, the acquisition of HES and the development of the specialist intensity metric based on specialist hours per 10 EAs obtained from a web-based annual PPS on a Sunday and the following Wednesday in June each year. The parallel directorate-level survey used to provide assurance for the PPS is described in Chapter 2; it also offered the opportunity to obtain information on consultant rotas and vacant posts, which we report below.

Over 5 years, 13 hospital sites were subject to trust mergers. In eight cases, the data collected after a merger refer to a combination of hospitals that had previously provided separate data. In the case of three hospitals subject to trust mergers, it was possible to preserve the integrity of the data by treating these as coming from separate sites, whereas, for the remaining two hospitals, the post-merger data were treated as being derived from a separate independent organisation.

Results

The 2013/14 cross-sectional study35

Specialist intensity

Of 141 acute hospital trusts in England receiving unselected emergency medical admissions, 127 agreed to participate and 115 (81.6% of total eligible trusts) contributed data to the survey, providing 15,537 responses. Of these responses, 1003 (6.5%) were from doctors who did not hold a specialist accreditation, and, therefore, these were excluded, leaving 14,534 eligible responses for analysis. The mean response rate was 45% (range 16–79%) and exceeded 40% in two-thirds of trusts.

There were substantially fewer specialists present and providing care to EAs on Sunday (n = 1667, 11.5%) than on Wednesday (n = 6105, 42.0%). This difference was partly offset by the greater average time spent caring for acutely admitted patients on Sunday (5.74 hours) than on Wednesday (3.97 hours). These patterns were consistent across the quintiles of trust size.

For both Sunday and Wednesday, there was a clear relationship between the sum of specialist hours delivered by each trust and the numbers of EAs on that day averaged across the year: as expected, larger hospitals had more specialists and more admissions (Figures 8a and c).

FIGURE 8.

(a) Total number of Wednesday specialist hours and average number of Wednesday EAs; (b) Wednesday specialist hours per 10 EAs and average Wednesday EAs; (c) total number of Sunday specialist hours and average Sunday EAs, 2013/14; and (d) Sunday specialist hours per 10 EAs and average Sunday EAs, 2013/14.

There was substantial variation between trusts when specialist hours were expressed per 10 EAs (see Figures 8b and d), suggesting that factors other than emergency workload influence the amount of specialist time delivered to EAs in any given trust. The median intensity ratio (Sunday divided by Wednesday) was 48% across all trusts, with similar results across quintiles of trust size, as measured using bed numbers for 2013/14. There was no trust for which the Sunday-to-Wednesday ratio was > 1, and in 90% of trusts the ratio was < 0.7 (Figure 9).

FIGURE 9.

Sunday-to-Wednesday specialist hours and ratio per 10 EAs by trust.

Mortality

Using 2013/14 HES data, the logistic model for all 141 acute hospital trusts provided a surplus weekend admission mortality risk of 9%. Trust-specific weekend odds ratios (WEORs) for the 115 trusts contributing to the survey ranged from 0.82 to 1.35, with 96 (83%) trusts recording an OR > 1 (Figure 10). There was no systematic relationship between weekend mortality effect and trust size.

FIGURE 10.

Trust-specific weekend effect ORs with confidence intervals from the logistic regression model, 2013/14.

Specialist intensity and admission day mortality

The correlation between the weekend mortality ORs and the Sunday-to-Wednesday specialist intensity ratio was very low (r = –0.015) and is not suggestive of a clear relationship (Figure 11).

FIGURE 11.

Weekend-to-weekday EA mortality OR and Sunday-to-Wednesday specialist intensity ratio by trust, 2013/14.

Directorate-level questionnaire

Clinical directors reported that the daily review of all patients was the norm on Sundays (Wednesdays) for 50% (86%) of AMUs, 100% (100%) of ICUs and 27% (58%) of acute wards. Consultant vacancies with gaps in cover were reported by 52% of responding EDs, 71% of AMUs, 65% of acute wards and 30% of ICUs.

Conclusion

This cross-sectional study showed a substantial weekend–weekday difference in specialist involvement in the care of patients admitted as emergencies to acute hospitals in England. There was no sign that the variation in the weekend-to-weekday specialist hours ratio was associated with the variation in the weekend-to-weekday admission mortality ratio. This suggests the need for caution in attributing the weekend effect primarily to a lack of consultants at weekends.

Longitudinal study results

Point prevalence survey responses

A total of 116 participating sites contributed 548 sets of survey results during the 5-year period, of which 96 sites contributed data in every year (thus five times in all). The primary analysis was conducted using all available data from the 116 trusts. A total of 66,425 analysable responses were received. The response rate by trust varied from 7% to 93% and declined by 9.4% during the study period, from 45.2% in 2014 to 35.8% in 2018. The average number of specialist responses from which relevant data could be drawn was 121.2 per trust (Table 2).

TABLE 2 - Survey responses over 5 years

The final column shows data from 548 samples (from 116 trusts) over 5 years.

The response rate is the number of clean responses as a proportion of the total surveyed.

	Year					Total
	2014	2015	2016	2017	2018
Participating trusts	112	108	110	106	112	116a
Total number surveyed	34,374	32,978	36,462	35,573	39,357	178,744
Average number per trust	306.9	305.4	331.5	335.6	351.4	326.2
Total number of clean responses	15,535	13,369	13,492	13,841	14,097	70,334
Average number per trust	138.7	123.8	122.7	130.6	125.9	128.3
Response rate (f)b	0.452	0.405	0.370	0.389	0.358	0.393
Total number of specialist responders	14,532	12,657	12,846	12,974	13,416	66,425
Average number per trust	129.8	117.2	116.8	122.4	119.8	121.2

Emergency admissions

Between 2013/14 and 2017/18, EA rates increased by 14.7% (all acute trusts in England), continuing the background trend (see Appendix 6). The increase in EAs per day was greater on weekdays (15.7%) than at weekends (11.3%) (mean weekend-to-weekday ratio 0.78).

Specialist intensity

Between 2013/14 and 2017/18, there was an increase in whole-time equivalent consultant staff of 14.6% across the NHS in England and of 4% among non-consultant doctors (see Figure 1).

The PPS data from 116 trusts are summarised in Table 3, alongside bed numbers and EAs. The results for specialist hours are shown both with and without correction for response rate. Importantly, no adjustment is necessary for the Sunday-to-Wednesday intensity ratio, as the numerator and denominator are affected equally by any such correction. Therefore, any change in the intensity ratio over time is likely to represent a true effect.

TABLE 3 - Specialist intensity over time

Computed from the preceding row using the response rates from Table 2.

Computed as a ratio of hours per trust to EAs per trust (i.e. by pooling across trusts before forming the ratio).

	Year					All
	2014	2015	2016	2017	2018
Number of specialists present on Sunday per trust	14.9	12.8	13.3	12.8	12.3	13.2
Number of specialists present on Wednesday per trust	54.5	44.7	45.0	42.8	39.2	45.3
Total number of hours on Sunday per trust	85.4	79.7	85.6	83.2	80.0	82.8
Corrected for overall response ratea	188.9	196.8	231.4	213.9	223.5	210.7
Total number of hours on Wednesday per trust	216.6	201.8	220.3	205.8	189.9	206.9
Corrected for overall response ratea	479.2	498.3	595.4	529.0	530.4	526.5
Average number of hospital bed numbers per trust	774.9	758.9	792.2	788.5	777.5	778.4
Mean number of EAs on Sunday per trust	87.1	89.5	95.7	94.8	97.7	93.0
Mean number of EAs on Wednesday per trust	106.9	110.8	119.1	122.4	128.0	117.4
Average number of hours per 10 EAs: Sundayb	9.81	8.90	8.95	8.78	8.19	8.91
Corrected for overall response ratea	21.70	21.98	24.19	22.57	22.88	22.67
Average number of hours per 10 EAs: Wednesdayb	20.26	18.22	18.50	16.82	14.84	17.62
Corrected for overall response ratea	44.82	44.99	50.00	43.24	41.45	44.83
Sunday-to-Wednesday intensity ratio (hours per 10 EAs)	0.484	0.489	0.484	0.522	0.552	0.506

There was a significant change in the intensity ratio over time (p = 0.0081). The response curve is flat to begin with (2014–16) and then rises over the next 2 years (2017 and 2018), as shown in Figure 12 (data are provided in Appendix 7). The variation over time was equivalent to an average annual increase in the intensity ratio of approximately 1.5% (standard error 0.46; p = 0.0011) per year, or about six percentage points during the period of the study. There is a significant difference (p < 0.0001) between trusts in the (time-averaged) intensity ratio (Figure 13). Furthermore, there is some evidence (p = 0.0035) of differences in rates of improvement over time. However, it is not clear how, if at all, these differences are related to the size of the trusts and we were unable to identify any other potential causative mechanisms.

FIGURE 12.

Sunday-to-Wednesday intensity ratios and adjusted weekend-to-weekday mortality ORs, 2013/14 to 2017/18.

FIGURE 13.

Sunday-to-Wednesday intensity ratios by trust (mean and 95% confidence interval).

How was the ‘improvement’ in the Sunday-to-Wednesday specialist intensity ratio achieved? If the 7-day services policy initiative were to have had its desired effect, it would have stimulated trusts to increase specialist numbers and, hence, specialist hours on Sundays, either through employing more specialists committed to service delivery at weekends or by redirecting specialist activities from weekdays to weekends. Specialist hours per 10 EAs would increase if more specialists were delivering care to the same number of patients, remain stable if the increase in specialist hours were accompanied by a parallel increase in EAs, or diminish if the increase in EAs outstripped the increase in specialist input or if specialist input actually fell.

Figure 14 shows that, over the 5 years, there was only a very modest overall increase in specialist hours per 10 EAs on Sundays, and a marked reduction on Wednesdays. As shown above, between 2013/14 and 2017/18, EAs increased by 14.7% overall, by 11.3% at weekends and by 15.7% on weekdays (see Appendix 6), and the increase in consultant staff during the same period was 14.6% (see Figure 1). If additional consultant input had been distributed uniformly across all days of the week, it would, therefore, seem likely that, although the increase in consultant staffing would have exceeded the increase in EAs at weekends, it would have been outstripped by the larger increase in EAs on weekdays.

FIGURE 14.

Specialist hours per 10 EAs adjusted for response rates and Sunday-to-Wednesday specialist intensity ratio. Specialist hours are corrected for response rate using the formula n × (1/response rate).

To explore this relationship between specialist hours and EAs in more detail, we present each trust’s mean annual percentage change in specialist intensity for Sundays against Wednesdays in Figure 15. The majority of trusts (n = 71, 62%) achieved an increase in the Sunday-to-Wednesday intensity ratio (i.e. a positive correlation between intensity ratio and year). In this group, 23 trusts increased specialist intensity for both Sundays and Wednesdays, 22 increased Sunday intensity while reducing intensity on Wednesdays, and a further 26 showed a reduction in both, but which was more marked for Wednesdays than for Sundays. Forty-four trusts showed a reduction in specialist intensity ratios over time; of these, 10 had marginally increased Sunday and Wednesday intensity, 15 had increased Wednesday intensity only, and 19 showed a reduction in both Sunday and Wednesday intensity. However, it would be misleading to conclude that actual specialist hours declined during the period. Indeed, there is evidence of a general increase in hours in the majority of trusts for both Wednesdays (63%) and Sundays (61%) (see Figure 15).

FIGURE 15.

Trends in specialist involvement in EAs, 2014–18. (a) Sunday and Wednesday specialist intensity (specialist hours per 10 EAs); and (b) Sunday and Wednesday specialist hours. The average proportional changes per year are computed from trust-level regression analyses of (log-transformed) intensity values. In (a), the 71 (62%) trusts above the diagonal line showed an improvement in the Sunday-to-Wednesday intensity ratio (i.e. positive correlation between intensity ratio and year). Data are from 115 trusts, one of which was excluded because it contributed PPS data for 1 year only.

The final step in elucidating the causes for the change in specialist intensity ratios is to examine the ratio of supply (specialist hours) to demand (EAs) across the 115 trusts. Here, we use the alternative formulation of the weekend-to-weekday specialist intensity ratio:

⁽³⁾ Sunday specialist hours Wednesday specialist hours × Wednesday EAs Sunday EAs .

During the study, each component of this expression increased by about 7%. Using data from Table 3, the ratio of Sunday to Wednesday specialist hours increased from 0.39 (= 85.4 ÷ 216.6) in 2013/14 to 0.42 (= 80.0 ÷ 189.9) in 2017/18, and the ratio of Wednesday to Sunday EAs increased from 1.23 (= 106.9 ÷ 87.1) to 1.31 (= 128.0 ÷ 97.7) during the same period. Together, these changes produced an overall increase of 14% in the intensity ratio, from 0.484 (2013/14) to 0.552 (2017/18). The general increase in EAs was more pronounced for Wednesdays (19.7%) than for Sundays (12.1%), and this change in the distribution of EAs across the week accounts for half of the improvement in the intensity ratio. It reflects a change in the pattern of demand on the service rather than a redirection of specialist resources.

In summary, the increase in the specialist intensity ratio appears to have been achieved by a general increase in specialist hours, which has been more pronounced at weekends, and a general rise in EAs, which has been more pronounced during the week.

Mortality and associations with specialist intensity

Hospital and 30-day post-admission EA mortality rates for all acute trusts in England reduced each year (30-day post-admission mortality reduced from 6.43% in 2007/8 to 5.47% in 2013/14), with weekend and weekday admission mortality rates falling in parallel. From 2014, overall mortality rates stabilised, but, in 2017/18, the weekend 30-day post-admission mortality rate increased from 5.8 to 6.06 and the unadjusted mortality ratio increased from 1.08 to 1.15. Adjustment for baseline characteristics abolishes this apparent increase (see Appendix 8), indicating that the rise in mortality rates is attributable to case-mix differences. The difference between hospital and 30-day post-admission mortality rates increased over these 11 years (Figure 16) for both weekend and weekday admissions, so that the ratio of hospital to 30-day mortality rates diminished. Figure 1 shows that, in the last 2 years, the progressive rise in delayed transfers of care has been reversed, indicating that patients are being discharged more rapidly back to the community. Taken together, these data suggest that an increasing proportion of the mortality risk is being transferred to the community post discharge (see Figure 16 and Appendix 8).

FIGURE 16.

Ratio of in-hospital to 30-day post-admission mortality and overall 30-day post-admission mortality.

The weekend mortality effects (i.e. the adjusted weekend-to-weekday mortality ORs) are shown in Appendix 9 and have been plotted alongside the specialist intensity ratios for each year of the study in Figure 12.

There is no evidence for any change in the weekend admission mortality ratios during this period (p = 0.1971). The weekend effect may differ between trusts (p = 0.0034), but these differences do not appear to relate to hospital size. Details are given in Appendix 10. We found no apparent correlation (r = 0.077) between the weekend mortality effect and the Sunday-to-Wednesday intensity ratio.

Discussion

We have shown that patients admitted as emergencies to acute hospital trusts in England on a Sunday collectively receive, on average, half as much specialist time (hours per 10 EAs) as patients admitted on a Wednesday. There is considerable variation between trusts in specialist intensity, which is independent of hospital size. Over the 5 years of the study, the Sunday-to-Wednesday specialist intensity ratio increased, and this appears to have occurred through several mechanisms: a modest increase in specialist input on Sundays; some redistribution of specialist hours in favour of Sundays; and an increase in the number of EAs, which was more marked for Wednesdays than for Sundays. It seems likely that the 7-day service policy had some effect by transferring specialist time from weekdays to weekends, and increasing consultant numbers, but the impact has been obscured by the increase in EAs during this period.

We have been unable to identify a relationship between weekend-to-weekday specialist intensity ratios and trust-level weekend admission mortality effects in either the cross-sectional analysis or the longitudinal study. Trusts with a narrower Sunday-to-Wednesday specialist intensity ratio do not seem to have a correspondingly smaller weekend effect.

Hospital admissions have continued to increase during the past 11 years. However, there was a progressive reduction in mortality rates until 2013/14, when the improvement in survival appeared to have plateaued and the 30-day mortality rate for weekend admissions increased (see Figure 16). Adjustment for baseline characteristics shows that this increase in mortality is attributable to case-mix differences (see Appendix 8). However, the increasing gap between hospital and 30-day mortality rates suggests that there has been a progressive transition in the place of death from hospital to the community. The reduction in length of hospital stay during the same period would be consistent with this interpretation. Whether this represents a desirable effort to allow those destined to die to do so in the peace of their own homes, or a failure to prevent avoidable deaths through, for example, effective rehabilitation of frail elderly patients who are admitted to hospital at weekends, requires further research at the intersection of community and hospital care at discharge, not just at admission.

The conventional narrative about the perceived – and now measured – gap in specialist input between weekends and weekdays is that patients admitted at weekends are being ‘short-changed’ in some way: that half as much specialist input means half as good care. An alternative interpretation might be that patients newly admitted at weekends and on weekdays are receiving the urgent care they need for this acute phase of their illness, while on weekdays specialists are additionally providing the less time-critical components of acute care targeted at refining treatments and promoting recovery. The decrement in specialist input at weekends might result in delays in progressing patients along the therapeutic pathway, but without contributing additional mortality risk.

Limitations of our study relate to the inherent weaknesses of surveys, namely variable response rates and subjective data recall. We mitigated the response rate issue by comparing weekends with weekdays in each trust. Other factors unrelated to specialist intensity might include unmeasured case-mix differences;26 variation in support services in hospital, including ‘failure to rescue’ (the inability of the system to respond promptly to patient deterioration), which has been shown to explain the difference in surgical outcomes between high- and low-volume centres;89 or a difference in outcomes from patient safety incidents at weekends. 105 We explore these factors in the following chapters, starting with an examination of patient case mix (see Chapter 5) and following with an evaluation of hospital care quality (see Chapter 6).

Introduction

The HiSLAC systematic review (see Chapter 3) identified three studies showing that patients admitted as emergencies to hospitals at weekends tended to be more severely ill, on average, than those admitted on weekdays. 26,55,56 However, it was not clear whether this was actually attributable to greater severity of illness, or the effect was caused by a reduction in the proportion of less severely ill patients presenting to hospitals. 106 Making this judgement required access to certain specific indicators: the ED attendance rate; the ‘conversion rate’ of attendances to admissions; a dynamic measure of physiological severity of illness, not just ‘static’ measures such as age and comorbidities; and the ability to track the initial stages of patients’ journeys in hospital as an indicator of acuity and dependence. The last two indicators can be obtained at trust level from those trusts that have a sufficiently sophisticated electronic patient record only. As there is no standardised national system for electronic patient records, this work can be performed at the local trust level only. We therefore obtained permission to access the database of one of the largest urban tertiary referral centres in the UK, which cares for > 800,000 patients every year and serves a large and diverse local patient population. The results of this study have been published. 57

Methods

This was a retrospective analysis of data that had been prospectively collected between January 2012 and December 2015 from the trust’s Microsoft SQL electronic clinical databases. We identified all adult EAs during this 4-year period, extracting demographic information [age, sex, ethnicity, deprivation, principal diagnosis (International Classification of Diseases and Related Health Problems, Tenth Revision33), comorbidities] and outcome (survival at 30 days post admission and hospital discharge). Re-admissions were classified as new hospital spells. Charlson Comorbidity Index score was calculated and categorised according to the SHMI. 104 Principal diagnosis was grouped using SHMI groupings. Income deprivation index was based on the 2015 Index of Multiple Deprivation,107 which was based on the patient’s residential postcode.

We supplemented these standard measures of case mix with National Early Warning Score (NEWS) data calculated from physiological vital signs (respiration rate, oxygen saturation, temperature, systolic blood pressure, pulse rate, level of consciousness and supplemental oxygen). 108 NEWS was calculated using primary variables for the first full set of vital signs available within the 24 hours following admission; we have termed this ‘NEWS24’. Post-admission severity was also derived from whether or not patients had been transferred to the ICU. At the time of this study (2012–15), vital signs of admitted patients were recorded electronically, except in the ICU, where the complex electronic ICU chart was still in development. We were therefore unable to calculate NEWS for the majority of patients transferred directly to the ICU from the ED. Therefore, the patients were classified into six risk categories using ICU status and NEWS24 values. The (putative) highest-risk category consisted of patients admitted to an intensive care unit within 24 hours (ICU24). The remaining (non-ICU24) patients were assigned, where possible, to one of four NEWS severity bands given by NEWS24 (≥ 7, 5–6, 1–4 and 0). The sixth category consisted of those non-ICU24 patients for whom data for calculating NEWS24 were missing.

Weekend admissions were defined as those admitted on or after midnight on Friday until midnight on Sunday. Admission time was the point of admission into the hospital, not the time when the patient first presented to the ED. Length of stay was calculated by subtracting the day of admission from the day of discharge. Admitted patients who either were discharged or died before midnight on the day of admission are classified as having zero length of stay (Z-LOS).

Results

Population and pathway characteristics are shown in Table 4.

National Early Warning Scores

It was possible to calculate NEWS within the first 24 hours for 90% of those admitted. The documentation of vital signs improved from 86% of admissions in 2012 to 92% of admissions in 2015, and was more complete for weekend admissions (91.7%) than for weekday admissions (89.5%) (Figure 17). NEWS could be calculated within 4 hours of admission for 72.5% of patients and within 8 hours of admission for 83% of patients; completion rates were higher at weekends. As expected, the vital signs required for calculation of NEWS were more likely to be missing for patients transferred to ICU within 24 hours of hospital admission (63.4%) than for patients not transferred to ICU within 24 hours (8.2%), particularly for those transferred directly to ICU from the ED (83.5% missing NEWS) rather than indirectly from another hospital location (19.8% missing NEWS).

FIGURE 17.

The NEWS distribution in first 24 hours of admission and the ratio of the distribution of each NEWS band weekend to weekday, 2012–15.

Weekend admissions had a higher NEWS than those admitted on weekdays (see Table 4) (mean NEWS 1.8 weekend, 1.7 weekday; p = 0.008), and a greater proportion of weekend admissions than weekday admissions were in the higher NEWS bands (see Figure 17). Among patients for whom it was recorded, ICU admission within 24 hours was associated with a higher mean NEWS (2.98) than non-ICU admission (1.70). In the non-ICU population, increasing NEWS values were associated positively with increasing age and length of hospital stay (see Appendix 12).

Mortality rates and mortality adjustment

Crude 30-day post-admission mortality rates are summarised in Table 5. The overall mortality rate was 5.1%, and it was higher for weekend admissions (5.6%) than weekday admissions (5.0%) [crude, unadjusted WEOR 1.13, 95% confidence interval (CI) 1.08 to 1.19]. As expected, mortality was much higher among ICU24 admissions (17.7%) than among non-ICU24 admissions (4.7%; p < 0.001), and there was a significant upward trend in mortality with increasing NEWS band (p < 0.001) among non-ICU24 admissions for whom NEWS was available, ranging from 1.5% (NEWS 0) to 29.1% (NEWS ≥ 7). The mortality rate for non-ICU24 admissions who had missing NEWS was 2.7%, which falls within the range of the two lowest NEWS bands.

TABLE 5 - Rates of 30-day crude mortality by weekend/weekday admission, stratified by ICU transfer and NEWS

Admission group	Crude mortality, n/N (%)			WEOR (95% CI)
	Weekend	Weekday	Total
All admissions	2140/37,979 (5.6)	6257/125,149 (5.0)	8397/163,128 (5.1)	1.13 (1.08 to 1.19)
ICU24	275/1579 (17.4)	673/3781 (17.8)	948/5360 (17.7)	0.97 (0.83 to 1.14)
No ICU24
NEWS ≥ 7	390/1257 (31.0)	1033/3627 (28.5)	1423/4884 (29.1)	1.13 (0.98 to 1.30)
NEWS 5–6	325/2091 (15.5)	862/6008 (14.3)	1187/8099 (14.7)	1.10 (0.96 to 1.26)
NEWS 1–4	895/20,535 (4.4)	2895/66,297 (4.4)	3790/86,832 (4.4)	1.00 (0.92 to 1.08)
NEWS 0	164/10,467 (1.6)	532/34,578 (1.5)	696/45,045 (1.5)	1.02 (0.85 to 1.22)
NEWS missing	91/2050 (4.4)	262/10,858 (2.4)	353/12,908 (2.7)	1.88 (1.47 to 2.40)

Crude WEORs within strata are shown in the final column of Table 5 and in Figure 18. For five of the six risk strata (i.e. ICU24 and the four non-ICU bands with NEWS available), the WEOR CI includes the value 1, which is consistent with the hypothesis that risk stratification has eliminated the weekend effect. The size of the WEOR (1.88; p < 0.001) in the non-ICU24 plus missing NEWS group presents an anomaly. Despite the low average mortality rate (2.7%), it appears that the risk is significantly higher for weekend admissions than for weekday admissions in this group.

FIGURE 18.

Rates of 30-day crude mortality (%) and adjusted WEOR (95% CI): effect of non-availability of NEWS, increasing NEWS band, and ICU transfer within 24 hours.

Among the 146,822 (90%) admissions for whom NEWS could be calculated, the 30-day mortality rate was 1825/34,833 = 5.2% at weekends and 5506/111,989 = 4.9% on weekdays (unadjusted WEOR 1.07, 95% CI 1.01 to 1.13). Adjustment using the standard model had little impact (Table 6 and Figure 19). Adjustment using the NEWS alone annulled the weekend effect (WEOR 1.02, 95% CI 0.96 to 1.08), with no additional contribution when standard adjustment variables were included (see Table 6).

TABLE 6 - The 30-day unadjusted and adjusted WEOR

NEWS included as an uncategorised covariate.

Values from Table 5.

	Unadjusted WEOR (95% CI)	With standard adjustment (95% CI)	Adjustment by NEWSa only (95% CI)	Standard adjustment plus NEWSa (95% CI)
Admissions with NEWS available (n = 146,822)	1.07 (1.01 to 1.13)	1.07 (1.01 to 1.14)	1.02 (0.96 to 1.08)	1.04 (0.98 to 1.11)
All admissions (n = 163,128)	1.13 (1.08 to 1.19)b	1.11 (1.05 to 1.17)	(Standard adjustment plus ICU24/NEWS strata)	1.08 (1.02 to 1.14)
ICU24	0.97 (0.83 to 1.14)b	1.09 (0.92 to 1.30)
No ICU24
NEWS ≥ 7	1.13 (0.98 to 1.30)b	1.14 (0.97 to 1.33)
NEWS 5–6	1.10 (0.96 to 1.26)b	1.07 (0.92 to 1.25)
NEWS 1–4	1.00 (0.92 to 1.08)b	1.03 (0.95 to 1.11)
NEWS 0	1.02 (0.85 to 1.22)b	1.03 (0.86 to 1.24)
NEWS missing	1.88 (1.47 to 2.40)b	1.73 (1.33 to 2.25)

FIGURE 19.

The 30-day WEOR (95% CI) for all patients and those with NEWS available: impact of adjustment variables. a, Risk stratas: (1) ICU24; (2) no ICU24 and NEWS ≥ 7; (3) no ICU and NEWS 5–6; (4) no ICU24 and NEWS 1–4; (5) no ICU24 and NEWS 0; and (6) No ICU24 and NEWS missing.

For the full sample of admissions (including those with missing NEWS), stratified estimates of the WEOR can be calculated using the ICU24/NEWS categories, with allowance for standard adjustment variables as well. The results are presented in Table 6. This approach leads to a modest attenuation of the WEOR from 1.13 to 1.08 (95% CI 1.02 to 1.14), but it does not appear to eliminate the weekend effect altogether, a consequence of the very high WEOR value (1.88) in the non-ICU, missing-NEWS stratum.

Discussion

We have shown that weekend admissions have a higher 30-day mortality rate that attenuates only modestly with standard case-mix adjustment, but effectively disappears when adjusted for physiological severity (NEWS) in the 90% of patients for whom complete NEWS variables could be calculated within 24 hours of admission. This shows that patients admitted to hospital at weekends are sicker – more acutely ill – than those admitted on weekdays. This is not just a matter of proportions, that is that the weekend cohort appears sicker because fewer patients who are less ill are admitted; a substantially larger number and larger proportion of weekend admissions required transfer to intensive care in the first 24 hours following admission, and more of these were transferred directly from the ED at weekends, again indicating that weekend admissions are more severely ill at hospital presentation. Patients admitted at weekends remain in hospital longer than those admitted on weekdays, which is another potential indicator of severity. Importantly, we find no clinically important differences in age, gender or comorbid disease between weekends and weekday admissions; patients admitted at the weekend are more acutely ill, not more chronically ill.

At weekends, more patients are transferred to the ICU directly from the ED, whereas on weekdays those referred to the ICU are more likely to follow an indirect route. This suggests that weekend care is more efficient for sicker patients. Delays in patient flow on weekdays may be a consequence of competition with elective admissions. Indirect (and potentially delayed) admission to intensive care is associated with worse outcomes. 109 Rates of NEWS completion rates for admissions at weekends are slightly better than those for admissions on weekdays. Taken together, these findings suggest that weekend emergency care is more efficient and more reliable than weekday emergency care.

We have also shown that, despite a constant daily ED attendance rate, fewer patients are admitted to hospital at weekends, mainly because of the reduction in community referrals from GPs (family doctors). This is consistent with the finding from England-wide data by Meacock et al. 106 that ‘direct’ hospital admission rates are lower at weekends. The smaller numbers of Z-LOS patients, combined with the higher mortality rate of these patients, suggest that this group contains frail or elderly patients who are more difficult to discharge back to the community at weekends. Taken together, these findings suggest that the search for the causes of the weekend effect needs to include the whole patient pathway, particularly referral patterns, and the organisation of care in the community, not just care in hospital.

Limitations of this study include the constraint that data with this amount of detail could be obtained from a single trust only. Concerns about generalisability are mitigated by the trust’s case mix (large and diverse), the fact that measurement was over several years, and the fact that the effects identified are in the same direction as those in other studies. The ‘missing NEWS’ values could have been addressed using imputation, but this requires that data be missing at random, and, as we have shown, there was a systematic reason for the non-availability of NEWS. Had we used imputation, we would not have identified the link with ICU admission.

Although the sample is large and covers 4 years, the finding that in-hospital care processes at weekends may be more reliable, not less reliable, requires confirmation across a larger sample of hospitals before the weekend effect can be attributed to pre-hospital factors. We therefore undertook a series of mixed-methods studies across 20 acute hospital trusts in England to investigate the quality of care of patients admitted as emergencies at weekends and on weekdays.

Introduction

The preliminary theory underpinning the 7-day services policy initiative was that the higher hospital mortality risk associated with weekend admission could be mitigated by increasing specialist input into the care of patients admitted at weekends. In the preceding chapters, we have confirmed the magnitude of the weekend effect in the worldwide literature, characterised patients’ and staff’s views about deficiencies in the quality of hospital care at weekends, and shown that consultant input into the care of EAs at weekends across acute trusts in England is half of that on weekdays. However, we have been unable to show that this decrement in specialist input is linked to weekend admission mortality risk, or that a modest shift in consultant input over 5 years from weekdays to weekends has reduced that risk. In a single-centre study, we have shown that patients admitted at weekends are sicker than those admitted on weekdays, and that those care processes available for study were performed at least as reliably and efficiently at weekends as on weekdays. Moreover, we found some preliminary evidence of deficiencies in pre-hospital community care at weekends that could contribute to the weekend effect. Our logic model is, therefore, becoming more nuanced and more complex.

To explore these findings in more detail, we undertook a mixed-methods evaluation of the safety and quality of care of patients admitted at weekends to 20 acute hospital trusts. Two rounds of qualitative research were conducted, the first involving all 20 trusts (round 1) and the second involving a subset of eight trusts (round 2). The qualitative research included observations of acute-care delivery up to, including, and following a weekend in each participating hospital, and interviews with front-line staff and senior executives in each site. In parallel, we evaluated error and adverse event rates, and global care quality from the case records of 4000 EAs at weekends and on weekdays to these 20 trusts in two epochs, 2012–13 and 2016–17, representing a period before and a period during the implementation of 7-day services.

Methods

The protocols of these studies have been published. 110,111

Results

Qualitative research: views of front-line staff, patients, relatives and executives on barriers to, facilitators of, and potential impact of 7-day service standards

In this section, we present three key sets of findings: (1) the role of consultants in the organisation and delivery of care at weekends (qualitative research data round 1); (2) the quality and outcomes of care at weekends (case note review and qualitative research data round 1); and (3) the challenges and facilitators of implementation of 7-day standards (qualitative research data round 2).

The role of consultants in the organisation and delivery of care at weekends

In round 1, six qualitative researchers undertook 826 hours of observations of care delivery, spanning weekends, at all 20 sites. A total of 93 interviews were conducted with staff and patients; the professionals interviewed at each site are listed in Table 7.

TABLE 7 - Interviews conducted by site: round 1

Site	Consultants	Registrars	Junior doctors	Nursing	Other	Patients	Total
1	1	1	1				3
2	1			2			3
3	3		2	1			6
4	2		1	1			4
5	2	1	2	2			7
6	1		1		1		3
7	1			1			2
8	1		1	1			3
9	1	1	1	1			4
10	1		2				3
11	4			1		4	9
12	1		2	1		3	7
13	1		1	1			3
14	1		1	1			3
15	1	1		1			3
16	2	1	1	1			5
17	2		2			3	7
18	2	1	3	1		3	10
19	2			3			5
20	1		1	1			3
Total	31	6	22	20	1	13	93

Decision-making: authority and expertise

Consultants are the key decision-makers, both at the individual level and organisationally, from admission to discharge. They categorise and allocate patients to the most appropriate specialty, and ensure that diagnosis and treatment can begin, be amended where necessary and be completed satisfactorily. Consultants use their seniority and expertise to prioritise and advance diagnosis and treatment decisions.

All interviewed staff believed that consultants were crucial in getting decisions made, and that this was one important means through which having more consultant availability could improve weekend care delivery. Observations revealed the extent and breadth of the decisions in which consultants were involved. They included decisions made at the individual patient level and those at an organisational level, particularly important in ensuring patient flow throughout the hospital. At the patient level, consultants were characterised as responsible for determining a diagnosis, deciding which tests are required and determining how urgent these were (and hence making prioritisation decisions). Their presence at weekends and their authority helped expedite decision-making and action on the next steps in a patient’s journey:

And the consultant said that definitely try and get him endoscoped today because they really need to find out what’s happening . . . So he explained to the registrar actually to try and organise this straight away with the gastro[enterology] team. He said there’s a gastro[enterology] list on a Saturday morning, let’s see if we can add this particular patient onto the end of this list.

07OBS

Consultants acted as overseers and co-ordinators of patient pathways, and held responsibility over the weekend for decisions that would alter the pathway of treatment or progress through the hospital system for a patient. Although there were many cases where the specialty of the consultant was not critical to their decision-making, in some cases, access to specialist expertise from someone with the authority of a consultant who could make high-stakes decisions was seen as critical to ensuring optimal patient care:

[Consultant] says to [researcher] ‘So you’ve witnessed a lovely example of why it is good to have consultants in the weekend. My colleague from the ICU needed a specialist input, and my input did make a change for this patient, it did change the patient plan. They were not sure of what to do from a respiratory point of view and I offered them my advice . . . The consultant-to-consultant discussion was very useful in this case’.

08IV03

We observed nurse-led discharge in one of the sites at weekends, but, in the rest, consultants had overall responsibility for making discharge decisions. Consultants were seen as responsible for determining discharge planning and this was facilitated by the availability of consultants at weekends.

The involvement of a consultant could expedite decision-making and patient progress because of the expertise and authority they could bring to bear. Consultant input was seen as particularly important in ‘high-stakes’ decisions, for example decisions to escalate or withdraw treatment, and when patients were acutely unwell.

Consultants used their higher-level overview of patient management plans, along with their authority, to ensure that the steps required in patient care took place and to push for actions to be taken. This included using their authority to access tests that they felt were essential, or to negotiate for patient transfers or access to equipment. There were examples of consultants making assertive orders, making it clear that certain things had to happen, and suggesting that staff could call on them to use their authority to overcome any obstacles:

The consultant had asked one of the junior doctors, a FY1 [Foundation Year 1], to get a patient . . . a MRI [magnetic resonance imaging] so that they could get her discharged. But the junior doctor came back and said we might not get a MRI today, because neurology wants to come and see this patient. But the consultant on the AMU said ring them back and tell them this MRI is the only thing that’s holding up discharge, so we need to get it today. And he said get the name of the neurology consultant who wants to see the patient, with an implication that if there was any hold-up, the consultant would step in and talk to the other consultant and make sure that they didn’t hold up this patient’s discharge.

06OBS

Support for junior staff

Along with patient-level and organisational decision-making, consultants played a key role in monitoring, supporting and teaching junior colleagues, and performing ‘scaffolding’ roles, such as providing supervision, guidance and support to junior doctors. Managing the flow of work included managing juniors, making sure that they completed their tasks in a timely fashion and ensuring that all patient reviews were conducted in a timely manner. Without consultant input, momentum could be lost over the weekend:

If there’s no reason for them to be reviewed [by a consultant at the weekend], the jobs that have been written down aren’t necessarily chased appropriately by the relevant doctors. So things can sometimes get missed . . . I don’t think you get that kind of consistency that you get during the week.

19IV04

Role of consultant limited by infrastructure and resources at weekends

Our analysis identified that the extent to which increased specialist input can improve the quality of service delivery is constrained by broader factors related to the resources available to support them both within the hospital and in the community. The value of having consultants available was limited if junior doctors were not available to action decisions; when other specialist and allied services within the hospital, such as physiotherapy or speech and language therapy services, were in short supply; and where organisational infrastructure was not supportive (e.g. when hospital pharmacies had limited opening hours at weekends). Perhaps most importantly, the ability of consultants to progress patient care through to discharge was limited by the availability of support services in the community:

Thinking about what the problems are at the weekend in hospitals, [consultant] said it’s not the lack of consultants or junior doctors that’s the problem at the weekend. It’s a whole list of things which he reeled off. Inappropriate admissions [. . .]. Lack of social services at the weekend to provide the services that patients need to be discharged.

02OBS

You can’t have that weekend service, that weekend review, unless you put everything else in place, you know we can’t get patient flow without pharmacists. We can’t get people out of hospital without physiotherapists, occupational therapists, social workers. Improving doctors on their own won’t improve flow.

04IV04

Quality and outcomes of care at weekends

We next outline our findings on quality of care and outcomes at the weekend, bringing together case note review and analysis of qualitative research data. We start by describing variation between hospitals in quality of care at the weekend. We then report the findings from the case note review, which triangulates the quantitative findings on quality of care with the findings from the qualitative data.

Observations and views of front-line staff, patients and relatives

In round 1, along with the 826 hours of observations of care delivery spanning weekends in 20 sites, a total of 93 interviews were conducted with staff and patients. The spread of interviewees by site is shown in Table 7. An additional set of telephone interviews was conducted with local project leads in 19 sites between the two rounds of fieldwork. In one site, the project lead failed to respond, despite repeated requests.

An overview of how each site organised and delivered acute medical care at the weekend was produced in the form of a case study (see Appendix 14 for examples). Analysis was conducted by exploring case by case which quality domains were frequently identified as important within each site or reported/observed to be lacking, with reference to the 7-day service standards17 and guidance on acute care from the Royal College of Physicians. 112 A framework analysis was then produced that enabled a cross-case comparison of each domain. The analysis was drawn from both observation field notes and interviews with front-line staff and patients. The key domains identified were:

1. staffing levels for consultants, junior doctors and nurses

2. the extent of senior oversight on medical wards

3. the access to senior expertise

4. continuity of care on AMU over the weekend and into the week

5. handover and communication

6. patient flow (at admission, general bed capacity, and at discharge).

There was substantial variation between trusts in staff perceptions and qualitative researcher observations of the quality of weekend care. Examples are provided in Table 8.

TABLE 8 - Examples of good and limited or poor service quality at weekends

A&E, accident and emergency; COPD, chronic obstructive pulmonary disease; FY1, Foundation Year 1; GI, gastrointestinal; HCA, health-care assistant; ITU, intensive therapy unit; SHO, senior house officer.

Staffing levels: consultants
Good:We do OK for consultant cover over the weekends, I think we’re well staffed. They have several people providing the acute medicine cover over the weekend, endoscopy cover for GI bleeds, a cardiologist who comes in to do a ward round on Sunday. So a lot of issues can be dealt with then02OBS	Limited:We need 16 or 18 consultants in A&E. But we’re about 50% short of what we need03INT01
Junior doctor staffing
Good:I don’t know if you’ve been in to other hospitals, but [our] Trust is completely out of the ordinary, it’s unlike any other hospital at the weekend: there are a ridiculous number of junior doctors here in at the weekend06OBS	Limited:You have got access to the doctors who are on the on-call rota but they are very, very stretched. The registrar and the SHO are mainly clerking in new patients in the AMU, [but there are only] two FY1 doctors running around the whole hospital trying to deal with the sick patients and you know, they’re just so stretched11OBS
Nurse staffing
Good:I’ve got four staff nurses on today [weekend day] so they’ve each got a bay and a side room [. . .] So this morning, we’ve got a staff nurse in each bay and then a HCA taking a bay and a half each16IV02	Limited:It’s not very good, from a nursing perspective, on the ward during the weekend: mainly agency staff and that wasn’t safe and even with the agency staff [the ward sister] still has to see if she can get somebody else to come and help on a weekend07OBS
Senior oversight on wards
Good:At the weekend each medical speciality [sic] has one consultant to go round in the morning seeing the patients that were flagged up for review over the weekend, just in their speciality [sic]. They don’t look after all the medical wards, just their own speciality [sic]16OBS	Limited:There are no consultants in the ward at a weekend. The only doctor is the on-call junior covering the [whole of the] third floor08OBS
Access to specialist expertise
Good:Across the wards, so for every speciality [sic], so elderly care or renal or respiratory, so on there is a consultant that comes in and [. . .] see[s] anybody who is unwell, anybody who is new to their ward, anybody who they are concerned about, anybody who is going to be discharged and do a sort of selective ward round with some of the patients16IV04	Limited:The neurologists . . . don’t work over the weekend. And we don’t necessarily have neurology back-up, in terms of AED[s], antiepileptic drugs05IV01
Continuity of care on AMU
Good:The AMU keeps [junior] doctors on the same area as much as possible, during the week and at the weekend, so they have some continuity. And the consultants for the week try to keep that continuity, so they are following the same cohort of patients06OBS	Limited:So you start on the Saturday not knowing them, and you may see them on Sunday and overnight, but then come Monday they’re no longer your patient, they’re back to another team. So there’s a break in continuity from Friday to Saturday and Saturday to Sunday . . . which can be risky for the patient . . . because things are missed01INT03
Handover and communication
Good:The AMU has the same pattern of handover 7 days a week. The 8 a.m. morning handover from night to day is quite long; they review the AMU patients, sick patients on base wards, patients on ITU that they need to be aware of, and this all gets handed over electronically. 3.30 p.m. on Friday, they go through all the patients quickly and put patients and tasks on the review list and a junior doctor would be on a computer updating it. So, their pattern is 8 a.m. big handover, 3.30 p.m. catch-up, 8.30 p.m. big handover again to the night. And that’s the pattern 7 days a week02OBS	Limited:Sometimes it’s very hard to interpret the jobs that are handed over by the week team that are basically a job list that they can access through their computer system. ‘Patient X review and manage if needed.’ But manage what? So they lose a lot of time because the handovers from the week team are not clear08OBS
Flow: front door
Good:They have systems and structures to avoid admission. The care of the elderly clinic [allows] older patients who come into A&E to go for short-term management to avoid admission. In some cases, the acute medical consultant on call will see patients who are still in the emergency department, so if a patient’s been allocated to the AMU, but they haven’t actually been transferred yet, the acute medical consultant will come back and see that patient as part of their ward round06OBS	Limited:Actually the difference at the weekend is that they also don’t have that short-stay unit open, where the patients who have been referred to the hospital via the GP route were going on during the weekday. That isn’t open on a weekend17OBS
Flow: beds
Good:They have non-clinical administrators who’ve been employed in the Trust for quite a long time with really good tacit knowledge about what can be done and how to push for things to happen. They know the workarounds and who to ask about things; they provide a lot of continuity, because they’re on 7 days a week. They would sit in on handovers, and come on ward rounds, and chase things, particularly patients’ care packages and facilitating discharges to happen06OBS	Limited:I just watched a game of musical beds. There were no spare beds, so patients were stacking up in the corridors, which isn’t good for dignity when you do examinations. Some patients were in chairs. Patients were being swapped between the majors section, which is where I was, and resus[citation]. There were no beds on AMU04OBS
Flow: discharge
Good:The consultant in charge was the discharge doctor of the day [on Saturday], to go round and to review all the patients that had been flagged for discharge on a whiteboard that was in the doctors’ office on the ward. There was a star against each of the patients who had been considered potentially suitable for discharge today, and the consultant’s job was to review them and make decisions for discharge06OBS	Limited:The emphasis on [Sunday] was not to discharge people, whereas on the Friday, [it] was looking through the list, can this patient go home or does this patient have to stay17OBS
Services to support flow: in-house
Good:Therapists, we have on-call therapists rather than the sort of the full complement, which does limit the ability, have to then be a bit more focused on those things, but it’s still there01IV01	Limited:The consultant said, from his point of view, that the biggest issue at the weekend is the fact that patient flow is obstructed because of the lack of allied diagnostic and therapeutic services at the weekend. So he said, again, that patient who needs a [speech and language assessment], Friday it’s difficult, Saturday no, Sunday no. So nothing is going to happen02OBS
Services to support flow: social and community services
Good/limited:One of the good things about the trust is that they don’t have major problems with flow over the weekend because some of the community services, for example the district nurses, the COPD nurses, are employed by the trust, not by the council, and this means that they keep working, they keep functioning over the weekend as well05OBS	Poor:Also outside of hospital, that’s where a lot of the blockage comes from. So a lot of patients actually stuck in hospital and cannot go out because the services aren’t there to assist them outside, there is no social services for instance, there is nobody to receive them out in the community or to organise stuff014IV02

Qualitative researcher ratings of weekend care quality

A RAG rating was produced that identified the key features of the quality of service delivery at weekends. This was conducted by mapping which domains (outlined earlier) appeared to be operating well or were limited in each site. Sites were rated as green (good) on the domain if an element of the service or care was identified as happening as intended, or as not presenting a threat to quality at the weekend. Sites were rated as amber when there was reduced capacity within the domains. Sites were rated as red when there was clear evidence of a problem of quality of service delivery within a domain or where it was clear that the service or care was not happening as intended. To enable triangulation of the findings from the qualitative research with the case record review, colours were weighted for subsequent quantitative analysis (red, 2; amber, 1; and green, 0) to produce a RAG score (Table 9).

TABLE 9 - Qualitative research weekend care quality RAG scores across the 20 trusts

Trust	Staffing			Senior oversight on wards	Access to expertise: specialist availability	Continuity of care in the AMU (nights and weekends)	Handover and communication	Flow: evidence of action taken			Services to support flow		Total weighted score for each trust
	Consultants	Junior doctors	Nurses					To reduce admissions	To manage beds	To facilitate hospital discharge	In house facilities, diagnostics, pharmacy	In the community
1	0	1	0	1	0	1	1	0	0	0	0	2	6
2	0	1	1	0	1	0	0	1	2	2	2	1	11
3	1	1	1	1	0	0	1	0	2	1	0	2	10
4	0	2	0	0	1	2	1	1	2	2	2	2	15
5	0	1	0	2	1	0	0	0	0	0	0	1	5
6	0	0	0	1	1	0	0	0	0	0	0	1	3
7	0	1	2	2	2	1	1	0	1	2	1	1	14
8	1	2	2	2	1	0	2	2	2	2	2	2	20
9	0	2	1	0	0	1	1	1	1	1	2	1	11
10	1	2	1	2	2	1	2	2	2	1	2	1	19
11	1	2	0	1	1	1	1	1	2	2	1	2	15
12	2	1	1	1	1	1	1	1	2	1	1	2	15
13	2	2	1	2	1	1	1	1	1	1	2	2	17
14	1	2	0	1	1	1	0	1	1	1	1	2	12
15	0	2	1	1	1	0	1	1	1	1	1	2	12
16	0	0	0	0	0	0	1	0	0	0	1	2	4
17	2	2	2	2	1	1	1	2	2	1	2	2	20
18	0	1	0	0	0	1	1	0	1	0	1	2	7
19	0	1	0	0	0	1	0	2	2	1	2	2	11
20	1	2	0	2	1	1	1	0	0	2	2	1	13
Sum of domain scores	11	28	13	20	15	14	17	16	23	20	24	33

Case record review: results

Four thousand case records were retrieved. The characteristics of the randomly selected study population were representative of the total hospital admitted population in England (see Appendix 17). Seventy-nine reviewers participated; the mean number of reviews per reviewer was 61 (20–69 reviews), with 800 records reviewed in duplicate. Of 4800 case reviews, in 37 the reviewer found that the case record contained insufficient documentation. This left 4763 case reviews available for analysis, of which 1584 could be used for inter-reviewer reliability assessment (Figure 20). There were also 37 reviews where no global care assessment was offered. Unless otherwise stated, analyses are based on available reviews, not on available case records.

FIGURE 20.

Case record review data acquisition and processing. PAS, Patient Administration System; UoB, University of Birmingham.

Errors, error-related adverse events and global quality of care

Of the 4763 case reviews (50% weekend admissions, 50% weekday admissions), 1178 (24.7%) showed one or more errors in care: 1909 errors were identified in total. Single errors were identified for 15.8% of reviews, and two or more errors were identified for 8.7% (Table 10). The most frequently identified category of error was ‘clinical assessment, investigation or diagnosis’ (31.7%), followed by ‘treatment and management’ (28.9%), ‘communication’ (15.5%) and ‘medication’ (13.3%). In total, 149 error-related adverse events were identified in 120 case reviews (2.52% of reviews, 7.8% of errors), of which 110 (73.8%) were judged to have a > 50% chance of being preventable. Reviewers judged that best practice care had been provided completely in 1891 (39.7%) reviews, substantially in 2002 (42.0%) reviews, partially in 740 (15.4%) reviews, very little in 99 (2.1%) reviews and not at all in 31 (0.6%) reviews.

TABLE 10 - Number of errors by weekend–weekday admission and epoch

Calculated using ‘total number of errors/all reviews excluding errors missing’.

	Both epochs, n (%)			Epoch 1, n (%)			Epoch 2, n (%)
	Total	Weekend	Weekday	Total	Weekend	Weekday	Total	Weekend	Weekday
Total (N)	4800	2395	2405	2400	1199	1201	2400	1196	1204
Number of errors
No error	3585 (74.7)	1807 (75.4)	1778 (73.9)	1752 (73)	880 (73.4)	872 (72.6)	1833 (76.4)	927 (77.5)	906 (75.2)
1	760 (15.8)	375 (15.7)	385 (16)	400 (16.7)	207 (17.3)	193 (16.1)	360 (15)	168 (14)	192 (15.9)
2	256 (5.3)	120 (5)	136 (5.7)	144 (6)	62 (5.2)	82 (6.8)	112 (4.7)	58 (4.8)	54 (4.5)
3	91 (1.9)	40 (1.7)	51 (2.1)	46 (1.9)	20 (1.7)	26 (2.2)	45 (1.9)	20 (1.7)	25 (2.1)
4	39 (0.8)	20 (0.8)	19 (0.8)	26 (1.1)	14 (1.2)	12 (1)	13 (0.5)	6 (0.5)	7 (0.6)
5	14 (0.3)	9 (0.4)	5 (0.2)	7 (0.3)	5 (0.4)	2 (0.2)	7 (0.3)	4 (0.3)	3 (0.2)
6	10 (0.2)	4 (0.2)	6 (0.2)	5 (0.2)	1 (0.1)	4 (0.3)	5 (0.2)	3 (0.3)	2 (0.2)
7	3 (0.1)	1 (0)	2 (0.1)	2 (0.1)	0 (0)	2 (0.2)	1 (0)	1 (0.1)	0 (0)
8	2 (0.0)	1 (0)	1 (0)	2 (0.1)	1 (0.1)	1 (0.1)	0 (0)	0 (0)	0 (0)
13	2 (0.0)	0 (0)	2 (0.1)	2 (0.1)	0 (0)	2 (0.2)	0 (0)	0 (0)	0 (0)
15	1 (0.0)	1 (0)	0 (0)	1 (0)	1 (0.1)	0 (0)	0 (0)	0 (0)	0 (0)
Missing	37 (0.8)	17 (0.7)	20 (0.8)	13 (0.5)	8 (0.7)	5 (0.4)	24 (1)	9 (0.8)	15 (1.2)
Error in care
Yes	1178 (24.5)	571 (23.8)	607 (25.2)	635 (26.5)	311 (25.9)	324 (27.0)	543 (22.6)	260 (21.7)	283 (23.5)
No	3585 (74.7)	1807 (75.4)	1778 (73.9)	1752 (73.0)	880 (73.4)	872 (72.6)	1833 (76.4)	927 (77.5)	906 (75.2)
Missing	37 (0.8)	17 (0.7)	20 (0.8)	13 (0.5)	8 (0.7)	5 (0.4)	24 (1.0)	9 (0.8)	15 (1.2)
Total number of errors
	1909	914	995	1066	501	565	843	413	430
Mean number of errors per patient admissiona
	0.4	0.38	0.42	0.45	0.42	0.47	0.35	0.35	0.36

Weekend–weekday differences

Of the 1909 identified errors, 914 were in weekend admissions and 995 were in weekday admissions (see Table 10). The mean rate of errors per case review was numerically smaller for weekend admissions (0.384) than for weekday admissions (0.417), but this did not achieve statistical significance (rate ratio 0.92, 95% CI 0.84 to 1.02; p = 0.1029) (Table 11). Similarly, there was no difference in adverse event rates, or in global care quality assessments, between weekend and weekday admissions. Nor was there any evidence for temporal change in weekend–weekday differences over time (see Table 11).

TABLE 11 - Results from mixed-effects regression models for error rates, adverse events and quality of care

RR, rate ratio.

RRs for the first nine categories of error; ORs for any error, adverse events and global quality of care; for quality of care an OR < 1 signifies an improvement.

Errors, adverse events and global quality	Epochs (epoch 2 : epoch 1), RR/OR (95% CI)	Day of admission (weekend : weekday), RR/OR (95% CI)	Weekend : weekday (between epochs), RR/OR (95% CI)
Errors
Assessment, investigation or diagnosis	0.75 (0.66 to 0.86)	0.88 (0.78 to 1.00)	1.03 (0.72 to 1.48)
Treatment and management	0.75 (0.66 to 0.85)	0.93 (0.80 to 1.09)	1.07 (0.79 to 1.44)
Communication	0.82 (0.67 to 0.99)	1.02 (0.85 to 1.24)	1.42 (0.99 to 2.04)
Medication	0.76 (0.63 to 0.94)	0.93 (0.74 to 1.17)	0.73 (0.44 to 1.24)
Monitoring	0.81 (0.58 to 1.14)	0.88 (0.68 to 1.14)	0.75 (0.35 to 1.63)
Resuscitation	0.72 (0.42 to 1.23)	1.74 (0.81 to 3.73)	1.28 (0.36 to 4.64)
Infection	2.16 (0.93 to 5.02)	0.70 (0.27 to 1.79)	1.26 (0.19 to 8.25)
Invasive procedures	0.46 (0.20 to 1.07)	1.62 (0.70 to 3.74)	0.99 (0.19 to 5.2)
Other	0.54 (0.32 to 0.89)	1.11 (0.72 to 1.71)	0.96 (0.40 to 2.29)
Any errora	0.80 (0.73 to 0.87); p < 0.0001	0.92 (0.84 to 1.02); p = 0.1029	1.03 (0.83 to 1.29); p = 0.7762
Adverse eventa	0.55 (0.39 to 0.78); p = 0.0008	0.99 (0.69 to 1.41); p = 0.9417	1.61 (0.77 to 3.39); p = 0.2063
Global quality of carea	0.82 (0.74 to 0.91); p = 0.0001	0.96 (0.86 to 1.08); p = 0.4786	0.96 (0.77 to 1.2); p = 0.7307

Temporal trends between epochs (2012–13 and 2016–17)

Error rates between epochs diminished significantly overall (rate ratio 0.80, 95% CI 0.73 to 0.87; p < 0.0001), and also for the four most frequently identified error categories (see Table 11); there was also a significant reduction in error-related adverse events (100 in epoch 1, 49 in epoch 2, OR 0.55, 95% CI 0.39 to 0.78; p = 0.0008) and in care quality (p = 0.0001) (see Table 11): the proportion of reviews attracting the two highest care quality assessments (‘completely’ and ‘substantially’) rose from 79.8% in epoch 1 to 82.4% in epoch 2.

Inter-reviewer differences

Error identification rates

Reviewers varied in their rates of error identification (see Appendix 18) and in their judgements of errors and global care quality (Table 12). Reviewer reliability coefficients were low for the assessment of any error (0.026), but higher for treatment (0.131), communication (0.058) and medication errors (0.072), and for global quality (0.142) (see Table 12). However, the study does not aim to establish the quality of care for any particular patient. Rather, it is concerned with aggregate data in individual trusts and the ratio of weekend to weekday error rates. The lowest meaningful level of aggregation is based on the (approximately) 50 case notes representing the condition of a particular trust in a particular epoch for weekend or weekday admissions. The reliability of such aggregates estimated using the Spearman–Brown formula is much higher, as shown in Table 12.

TABLE 12 - Inter-reviewer reliability

NA, not appropriate.

Computed from the Spearman–Brown formula with 50 case notes per trust.

Omitting categories with fewer than 50 errors reported among the repeat reviews.

Error category and global quality	All reviews (N = 4763)	Errors per review			Repeat reviews (N = 1584 = 2 × 792 reviews in total)
	Errors (n)	Mean (n)	SD	Maximum (n)	Errors (n)	Individual-level reliability	Trust-level reliabilitya
Assessment	903	0.19	0.58	12	272	0.003	0.138
Treatment	824	0.17	0.54	10	265	0.131	0.883
Communication	442	0.09	0.44	14	140	0.058	0.753
Medication	380	0.08	0.34	8	129	0.072	0.794
Monitoring	138	0.03	0.19	4	47	b	b
Resuscitation	38	0.01	0.10	2	12	b	b
Infection	26	0.01	0.08	2	11	b	b
Invasive	25	0.01	0.08	2	5	b	b
Other	75	0.02	0.13	2	25	b	b
Any error	1909	0.40	0.92	15	606	0.026	0.568
Global quality of care	NA	NA	NA	NA	NA	0.142	0.892

Trust-level aggregate quality measure

Error, global quality and qualitative research

The relationship between error rates and global quality of care identified from the case record review at trust level is represented in Figure 21, with trust-specific ORs (for ‘any error’ and ‘global quality’ of care) obtained from the models reported in Table 11. The overall correlation is close to zero (r = 0.06), but this masks an evident – and plausible – relationship between higher error rates and lower global quality of care assessments for 18 of the 20 trusts. Two trusts (labelled 10 and 15 in Figure 21a) did not follow this pattern, recording the lowest error rates alongside relatively poor global care assessments.

FIGURE 21.

Trust-level aggregate analysis. (a) Error and global care quality; and (b) global care quality and qualitative research quality score. The ORs for the presence of (any) error and the ORs for (worse) quality of care are plotted on a logarithmic scale, with a value of 1 corresponding to the average trust.

We explored this further using the qualitative research RAG-rated evaluations of quality of weekend care. The RAG score ranged from 3 (good) to 20 (poor) over the 20 trusts, with a mean value of 12.0. Trust 10 had a high (19, adverse) weighted RAG score, and trusts 14 and 15 had an intermediate score (12) (see Table 9). The RAG score is moderately correlated (r = 0.49; p = 0.0273) with the reviewers’ trust-specific global assessment scores (see Figure 21b). This suggests that the case record reviewers and the qualitative researchers in the field are detecting quality signals that may have something in common, but that less clearly correlate with clinical errors (see Figure 21a).

Of the 12 qualitative research weekend care quality criteria, the one that attracted adverse scores across all 20 trusts was support services in the community for patient flow (impacting on admission and discharge) (see Table 9). Junior doctor staffing and workload were also identified as particular sources of concern at weekends.

Between epochs and specialist intensity

When weekend–weekday differences are aggregated across the entire study period, there is little evident correlation between trust-level weekend effects for quality of care (i.e. weekend-to-weekday ratios pooled over both epochs) and the corresponding ratios for the measure of specialist intensity, as computed over the 5 years of the PPS from 2013/14 to 2017/18 (r = –0.23, p = 0.3341 in Figure 22a). Nevertheless, this does not rule out the possibility that changes in staffing policy over time in an individual trust might have an impact on the weekend-to-weekday ratios of care quality in that trust. To examine this further, the PPS data from 2013/14 were matched with the case-note reviews from epoch 1, and the PPS data from 2016/17 were matched with the reviews from epoch 2. In this way, a ratio of weekend effects between epochs could be calculated for each trust both for global quality of care and for specialist intensity. These ratios (corresponding to differences in differences on the log scale) are shown in Figure 22b. The negative correlation here (r = –0.53) indicates that improved compliance with 7-day consultant care in individual trusts was accompanied by a corresponding equalisation of standards of care across the week, although it does not establish a causal connection. A similar analysis using trust-level weekend effects for the presence of error in place of poor care quality (figure not shown) produced a somewhat smaller negative correlation (r = –0.34).

FIGURE 22.

Trust-level weekend effects for quality of care and specialist intensity. (a) Pooled over both epochs and all PPS time points; and (b) relative changes between the two epochs. In (a) values above (below) 1 on the vertical axis represent care that is worse (better) than average. In (b), ratios greater (less) than 1 indicate that care had worsened (improved) from epoch 1 to epoch 2.

Patient admission pathways

Data on the pre- and post-admission phases are summarised in Table 13.

TABLE 13 - Reviewers’ assessment on pre- and post-admission patient pathways

ADL, activities of daily living; AMU, acute medical unit; DNACPR, do not attempt cardiopulmonary resuscitation; MAU, medical assessment unit.

Including surgical assessment/operating theatre.

For example, coronary care unit, renal unit, respiratory haematology, oncology.

Including high dependency.

The sum of ‘Yes’ and ‘No, but would definitely have been appropriate to limit treatment’ and ‘DNACPR already in place prior to admission’.

	Both epochs, n (%)			Epoch 1, n (%)			Epoch 2, n (%)
	Total	Weekend	Weekday	Total	Weekend	Weekday	Total	Weekend	Weekday
Total (N)	4800	2395	2405	2400	1199	1201	2400	1196	1204
Source of admission
Own home	4059 (84.6)	2052 (85.7)	2007 (83.5)	2052 (85.5)	1038 (86.6)	1014 (84.4)	2007 (83.6)	1014 (84.8)	993 (82.5)
Nursing or residential care home	314 (6.5)	149 (6.2)	165 (6.9)	160 (6.7)	73 (6.1)	87 (7.2)	154 (6.4)	76 (6.4)	78 (6.5)
No information available	291 (6.1)	127 (5.3)	164 (6.8)	119 (5)	52 (4.3)	67 (5.6)	172 (7.2)	75 (6.3)	97 (8.1)
Another hospital	40 (0.8)	20 (0.8)	20 (0.8)	15 (0.6)	8 (0.7)	7 (0.6)	25 (1.0)	12 (1.0)	13 (1.1)
No fixed abode	88 (1.8)	43 (1.8)	45 (1.9)	51 (2.1)	26 (2.2)	25 (2.1)	37 (1.5)	17 (1.4)	20 (1.7)
Missing	8 (0.2)	4 (0.2)	4 (0.2)	3 (0.1)	2 (0.2)	1 (0.1)	5 (0.2)	2 (0.2)	3 (0.2)
Patient condition before admission
Independent	3142 (65.5)	1552 (64.8)	1590 (66.1)	1605 (66.9)	804 (67.1)	801 (66.7)	1537 (64)	748 (62.5)	789 (65.5)
Needing help with some ADL	789 (16.4)	394 (16.5)	395 (16.4)	404 (16.8)	194 (16.2)	210 (17.5)	385 (16)	200 (16.7)	185 (15.4)
Dependent on others for most/all ADL, including personal hygiene	476 (9.9)	270 (11.3)	206 (8.6)	226 (9.4)	128 (10.7)	98 (8.2)	250 (10.4)	142 (11.9)	108 (9.0)
Unable to determine; no relevant information in notes	384 (8.0)	174 (7.3)	210 (8.7)	162 (6.8)	71 (5.9)	91 (7.6)	222 (9.3)	103 (8.6)	119 (9.9)
Missing	9 (0.2)	5 (0.2)	4 (0.2)	3 (0.1)	2 (0.2)	1 (0.1)	6 (0.3)	3 (0.3)	3 (0.2)
Referral to hospital mechanism
999/ambulance transfer to ED	2261 (47.1)	1253 (52.3)	1008 (41.9)	1197 (49.9)	662 (55.2)	535 (44.5)	1064 (44.3)	591 (49.4)	473 (39.3)
Self-presentation to ED (walk in/own transport)	1022 (21.3)	543 (22.7)	479 (19.9)	489 (20.4)	263 (21.9)	226 (18.8)	533 (22.2)	280 (23.4)	253 (21.0)
GP or deputising service referral (documented letter or telephone call)	667 (13.9)	195 (8.1)	472 (19.6)	344 (14.3)	102 (8.5)	242 (20.1)	323 (13.5)	93 (7.8)	230 (19.1)
Unable to determine	644 (13.4)	322 (13.4)	322 (13.4)	271 (11.3)	141 (11.8)	130 (10.8)	373 (15.5)	181 (15.1)	192 (15.9)
Patient instructed by GP to attend ED, no formal evidence of referral	116 (2.4)	34 (1.4)	82 (3.4)	64 (2.7)	18 (1.5)	46 (3.8)	52 (2.2)	16 (1.3)	36 (3)
Urgent care centre or walk-in centre referral	55 (1.1)	33 (1.4)	22 (0.9)	20 (0.8)	7 (0.6)	13 (1.1)	35 (1.5)	26 (2.2)	9 (0.7)
Self-presentation, patient instructed to attend by 111 service	23 (0.5)	9 (0.4)	14 (0.6)	9 (0.4)	3 (0.3)	6 (0.5)	14 (0.6)	6 (0.5)	8 (0.7)
Missing	12 (0.3)	6 (0.3)	6 (0.2)	6 (0.3)	3 (0.3)	3 (0.2)	6 (0.3)	3 (0.3)	3 (0.2)
Admission pathway
ED/pre-admission area	4016 (83.7)	2110 (88.1)	1906 (79.3)	1990 (82.9)	1048 (87.4)	942 (78.4)	2026 (84.4)	1062 (88.8)	964 (80.1)
Direct admission to acute ward	555 (11.6)	189 (7.9)	366 (15.2)	304 (12.7)	104 (8.7)	200 (16.7)	251 (10.5)	85 (7.1)	166 (13.8)
Unable to determine	220 (4.6)	91 (3.8)	129 (5.4)	103 (4.3)	45 (3.8)	58 (4.8)	117 (4.9)	46 (3.8)	71 (5.9)
Missing	9 (0.2)	5 (0.2)	4 (0.2)	3 (0.1)	2 (0.2)	1 (0.1)	6 (0.3)	3 (0.3)	3 (0.2)
Pre-admission vital signs documentation
Yes, full set, and a NEWS	2113 (44.0)	1098 (45.8)	1015 (42.2)	918 (38.3)	474 (39.5)	444 (37.0)	1195 (49.8)	624 (52.2)	571 (47.4)
Full set but no NEWS calculated	1546 (32.2)	796 (33.2)	750 (31.2)	935 (39)	475 (39.6)	460 (38.3)	611 (25.5)	321 (26.8)	290 (24.1)
Incomplete vital signs or time missing	377 (7.9)	180 (7.5)	197 (8.2)	223 (9.3)	107 (8.9)	116 (9.7)	154 (6.4)	73 (6.1)	81 (6.7)
No vital signs documented	436 (9.1)	212 (8.9)	224 (9.3)	159 (6.6)	84 (7.0)	75 (6.2)	277 (11.5)	128 (10.7)	149 (12.4)
Not applicable: patient not admitted via ED: direct admission to ward	313 (6.5)	99 (4.1)	214 (8.9)	160 (6.7)	55 (4.6)	105 (8.7)	153 (6.4)	44 (3.7)	109 (9.1)
Missing	15 (0.3)	10 (0.4)	5 (0.2)	5 (0.2)	4 (0.3)	1 (0.1)	10 (0.4)	6 (0.5)	4 (0.3)
First location post admission
Acute medical unit (AMU/MAU)	1676 (34.9)	822 (34.3)	854 (35.5)	842 (35.1)	392 (32.7)	450 (37.5)	834 (34.8)	430 (36.0)	404 (33.6)
Unable to determine ward type	804 (16.8)	404 (16.9)	400 (16.6)	395 (16.5)	212 (17.7)	183 (15.2)	409 (17.0)	192 (16.1)	217 (18.0)
Clinical decision unit or short-stay ward	782 (16.3)	420 (17.5)	362 (15.1)	419 (17.5)	220 (18.3)	199 (16.6)	363 (15.1)	200 (16.7)	163 (13.5)
Other (please specify)	513 (10.7)	234 (9.8)	279 (11.6)	244 (10.2)	120 (10)	124 (10.3)	269 (11.2)	114 (9.5)	155 (12.9)
General surgerya	452 (9.4)	221 (9.2)	231 (9.6)	214 (8.9)	117 (9.8)	97 (8.1)	238 (9.9)	104 (8.7)	134 (11.1)
Medical subspecialties including high careb	257 (5.4)	128 (5.3)	129 (5.4)	136 (5.7)	54 (4.5)	82 (6.8)	121 (5.0)	74 (6.2)	47 (3.9)
General medical ward	199 (4.1)	97 (4.1)	102 (4.2)	103 (4.3)	55 (4.6)	48 (4.0)	96 (4)	42 (3.5)	54 (4.5)
Older people’s medicine/elderly care unit	56 (1.2)	34 (1.4)	22 (0.9)	22 (0.9)	14 (1.2)	8 (0.7)	34 (1.4)	20 (1.7)	14 (1.2)
Critical care unit/ICUc	36 (0.8)	23 (1.0)	13 (0.5)	18 (0.8)	11 (0.9)	7 (0.6)	18 (0.8)	12 (1)	6 (0.5)
Missing	20 (0.4)	10 (0.4)	10 (0.4)	5 (0.2)	3 (0.3)	2 (0.2)	15 (0.6)	7 (0.6)	8 (0.7)
Rehabilitation	5 (0.1)	2 (0.1)	3 (0.1)	2 (0.1)	1 (0.1)	1 (0.1)	3 (0.1)	1 (0.1)	2 (0.2)
Location appropriateness
Yes, definitely appropriate	2952 (61.5)	1452 (60.6)	1500 (62.4)	1483 (61.8)	723 (60.3)	760 (63.3)	1469 (61.2)	729 (61.0)	740 (61.5)
Probably appropriate	1041 (21.7)	532 (22.2)	509 (21.2)	539 (22.5)	274 (22.9)	265 (22.1)	502 (20.9)	258 (21.6)	244 (20.3)
Unable to determine	643 (13.4)	332 (13.9)	311 (12.9)	295 (12.3)	163 (13.6)	132 (11)	348 (14.5)	169 (14.1)	179 (14.9)
No	137 (2.9)	66 (2.8)	71 (3)	73 (3)	33 (2.8)	40 (3.3)	64 (2.7)	33 (2.8)	31 (2.6)
Unknown	27 (0.6)	13 (0.5)	14 (0.6)	10 (0.4)	6 (0.5)	4 (0.3)	17 (0.7)	7 (0.6)	10 (0.8)
Post-admission vital signs documentation
Yes, full set, and a NEWS	2324 (48.4)	1203 (50.2)	1121 (46.6)	1108 (46.2)	573 (47.8)	535 (44.5)	1216 (50.7)	630 (52.7)	586 (48.7)
No vital signs documented	1151 (24.0)	538 (22.5)	613 (25.5)	501 (20.9)	242 (20.2)	259 (21.6)	650 (27.1)	296 (24.7)	354 (29.4)
Full set but no NEWS calculated	932 (19.4)	459 (19.2)	473 (19.7)	586 (24.4)	281 (23.4)	305 (25.4)	346 (14.4)	178 (14.9)	168 (14.0)
Incomplete vital signs	353 (7.4)	175 (7.3)	178 (7.4)	190 (7.9)	93 (7.8)	97 (8.1)	163 (6.8)	82 (6.9)	81 (6.7)
Missing	40 (0.8)	20 (0.8)	20 (0.8)	15 (0.6)	10 (0.8)	5 (0.4)	25 (1)	10 (0.8)	15 (1.2)
Initial consultant review
Consultant review documented	2259 (47.1)	1107 (46.2)	1152 (47.9)	1088 (45.3)	512 (42.7)	576 (48)	1171 (48.8)	595 (49.7)	576 (47.8)
No evidence for consultant review in first 24 hours	1512 (31.5)	776 (32.4)	736 (30.6)	708 (29.5)	395 (32.9)	313 (26.1)	804 (33.5)	381 (31.9)	423 (35.1)
Consultant review, time not documented, but case record suggests < 14 hours after admission	373 (7.8)	170 (7.1)	203 (8.4)	232 (9.7)	103 (8.6)	129 (10.7)	141 (5.9)	67 (5.6)	74 (6.1)
Probable consultant review but status of doctor uncertain	342 (7.1)	187 (7.8)	155 (6.4)	201 (8.4)	106 (8.8)	95 (7.9)	141 (5.9)	81 (6.8)	60 (5.0)
Unlikely that consultant review occurred during first 24 hours	196 (4.1)	107 (4.5)	89 (3.7)	103 (4.3)	55 (4.6)	48 (4.0)	93 (3.9)	52 (4.3)	41 (3.4)
Consultant review, time not documented, but case record suggests > 14 hours after admission	75 (1.6)	27 (1.1)	48 (2)	49 (2)	16 (1.3)	33 (2.7)	26 (1.1)	11 (0.9)	15 (1.2)
Missing	43 (0.9)	21 (0.9)	22 (0.9)	19 (0.8)	12 (1)	7 (0.6)	24 (1.0)	9 (0.8)	15 (1.2)
Palliative care discussed
No, not required, patient appropriate for full treatment	3476 (72.4)	1702 (71.1)	1774 (73.8)	1765 (73.5)	871 (72.6)	894 (74.4)	1711 (71.3)	831 (69.5)	880 (73.1)
No, but would probably have been appropriate to consider some form of treatment limitation	526 (11.0)	255 (10.6)	271 (11.3)	291 (12.1)	144 (12.0)	147 (12.2)	235 (9.8)	111 (9.3)	124 (10.3)
Yes	321 (6.7)	190 (7.9)	131 (5.4)	136 (5.7)	81 (6.8)	55 (4.6)	185 (7.7)	109 (9.1)	76 (6.3)
No, but would definitely have been appropriate to limit treatment	300 (6.3)	158 (6.6)	142 (5.9)	164 (6.8)	79 (6.6)	85 (7.1)	136 (5.7)	79 (6.6)	57 (4.7)
DNACPR already in place prior to admission	142 (3.0)	73 (3.0)	69 (2.9)	29 (1.2)	15 (1.3)	14 (1.2)	113 (4.7)	58 (4.8)	55 (4.6)
(Sub)total of above three categoriesd	763 (16.0)	421 (17.5)	342 (14.2)	329 (13.7)	175 (14.6)	154 (12.8)	434 (18.1)	246 (20.6)	188 (15.6)
Missing	35 (0.7)	17 (0.7)	18 (0.7)	15 (0.6)	9 (0.8)	6 (0.5)	20 (0.8)	8 (0.7)	12 (1.0)
Palliative care decision appropriateness (of those discussed)
Yes, appropriate decision	318 (99.1)	189 (99.5)	129 (98.5)	135 (99.3)	81 (100)	54 (98.2)	183 (98.9)	108 (99.1)	75 (98.7)
Yes, but patient might have benefited from escalation	2 (0.6)	0 (0)	2 (1.5)	1 (0.7)	0 (0)	1 (1.8)	1 (0.5)	0 (0)	1 (1.3)
Missing	1 (0.3)	1 (0.5)	0 (0)	0 (0)	0 (0)	0 (0)	1 (0.5)	1 (0.9)	0 (0)
Palliative care referral
No	4613 (96.1)	2289 (95.6)	2324 (96.6)	2317 (96.5)	1149 (95.8)	1168 (97.3)	2296 (95.7)	1140 (95.3)	1156 (96)
Yes	104 (2.2)	62 (2.6)	42 (1.7)	46 (1.9)	29 (2.4)	17 (1.4)	58 (2.4)	33 (2.8)	25 (2.1)
Missing	83 (1.7)	44 (1.8)	39 (1.6)	37 (1.5)	21 (1.8)	16 (1.3)	46 (1.9)	23 (1.9)	23 (1.9)
Admission avoidable
No	3493 (72.8)	1787 (74.6)	1706 (70.9)	1769 (73.7)	896 (74.7)	873 (72.7)	1724 (71.8)	891 (74.5)	833 (69.2)
Possibly	939 (19.6)	434 (18.1)	505 (21.0)	463 (19.3)	218 (18.2)	245 (20.4)	476 (19.8)	216 (18.1)	260 (21.6)
Yes	335 (7.0)	156 (6.5)	179 (7.4)	156 (6.5)	76 (6.3)	80 (6.7)	179 (7.5)	80 (6.7)	99 (8.2)
Missing	33 (0.7)	18 (0.8)	15 (0.6)	12 (0.5)	9 (0.8)	3 (0.2)	21 (0.9)	9 (0.8)	12 (1)

Pre admission

The majority of patients were admitted from home. Weekend admissions were substantially more likely to be classed as dependent on others for activities of daily living (weekend 11.3% vs. weekday 8.6%) and to have been brought to hospital by ambulance following an emergency (999) call (52.3% vs. 41.9%), and were less likely to have been referred by a GP (8.1% vs. 19.6%) or to have been admitted directly to an acute ward, bypassing the ED (7.9% vs. 15.2%). These weekend–weekday differences were more marked for epoch 2 than for epoch 1. Weekend admissions were more likely to include patients for whom a palliative care decision was already in place, had been applied at the time of admission, or, in the opinion of the reviewer, should definitely have been made (17.6% vs. 14.2%), with a marked increase between epochs (13.7% vs. 18.1%). The reviewers considered a smaller proportion of weekend than weekday admissions to be definitely avoidable (6.5% vs. 7.4%).

Post admission

The initial location of admission (usually the AMU) was considered by the reviewers to have been appropriate for the majority of admissions, regardless of the day of admission. The initial and subsequent documentation of vital signs and NEWS was more complete at weekends, and better in epoch 2 than in epoch 1. Documentation of an initial review of the patient by a consultant was marginally less likely for weekend (46.2%) than for weekday (47.9%) admissions, but increased during epoch 2 for weekend admissions (49.7%) to a level above that of weekday admissions, which remained constant. Consultant review within 14 hours of admission was identified in 22.9% (weekend) and 24.9% (weekday) of case reviews for epoch 1, increasing in epoch 2 more markedly for weekend than for weekday admissions (29.5% vs. 27.2%, respectively). Combining ‘documented’ with ‘probable’ consultant reviews within 24 hours of admission indicated that this had occurred in 54.2% of cases overall.

Implementing 7-day standards: challenges and facilitators

Round 2 of the qualitative research involved exploring board-level and front-line staff’s views on implementing 7-day standards and services, and conducting focused observations. Here we report the findings from interviews with 43 front-line and executive-level staff in eight sites (Table 14).

TABLE 14 - Professional groups interviewed, by site

Site	Board level (n)	Consultants (n)	Senior nursing (n)	Rota co-ordinators (n)	Total (n)
3	2	3		1	6
7	2	2	1	1	6
10	1	3			4
11	2	2	1	1	6
12	1	2	1		4
16	1	3			4
17	1	2	1	2	6
18	2	2	1	2	7
Total	12	19	5	7	43

Progress in implementing 7-day standards

Interviewees in each site concentrated on discussing the extent to which they believed they were meeting the four priority standards. These were:

• standard 2 – EAs assessed by a consultant within 14 hours

• standard 5 – timely access to diagnostic services according to urgency

• standard 6 – access to consultant-directed interventions 24/7

• standard 8 – consultant review of patients in acute medical, surgical and intensive care units twice daily by consultants working blocks of days; daily consultant review of ward patients.

Trusts’ performances on meeting the four quality standards (based on national audit data)16 are shown in Table 15.

TABLE 15 - Trusts’ performance on meeting 7-day standards 2, 5, 6 and 8 at weekends

N/A, not applicable.

NHS England, July 2019. www.england.nhs.uk/seven-day-hospital-services/progress/

Site	Weekend results (March 2017) (%)				Weekend results (March 2018) (%)
	Standard 2	Standard 5	Standard 6	Standard 8	Standard 2	Standard 5	Standard 6	Standard 8
3	79	100	89	62	86	87	89	N/A
7	64	82	89	60	94	87	89	79
10	49	100	89	53	78	95	100	77
11	87	87	100	92	93	100	100	94
12	55	100	100	63	83	100	100	79
16	81	100	100	97	89	100	100	85
17	74	68	78	69	83	73	100	77
18	68	100	100	83	66	100	100	83

Although most participants believed that they had made some progress in delivering 7-day standards, many recognised that they still had much more that they wanted to achieve and acknowledged the difficulty of delivering the standards with their existing resources. Implementing the standards was challenged by questions about the legitimacy of the standards. Contextual and resource issues affected the extent to which sites could deliver the standards. These included organisational instability such as mergers or changes of leadership. Where sites were struggling with financial and quality pressures, the need to address weekend working remained secondary to improving their overall quality indicators.

Challenges in implementation

Staff engagement with 7-day agenda

Implementing standards was challenged by four concerns: the legitimacy of focusing mainly on increasing consultant intensity at weekends; the evidence that this would in fact reduce mortality; the risk of reducing safety during the week by changing working patterns; and the need to improve services across the board as opposed to focusing primarily on the weekend:

We need to rethink what it is that 7-day process is going to bring us, and what is it that we’re trying to achieve . . . If [condition-specific mortality] goes down from 1.2% to 1.1%, is that worth the investment and the issues that would arise? I don’t know the answer, I definitely ask the question.

03INT01

The implementation of 7-day services most often involved changing the working patterns of existing staff groups. Engagement across the staff body within organisations was sometimes seen as variable. Certain staff groups or ‘clans’ had their own ways of working and could be seen as harder to engage, particularly those who were used to working in an individual and autonomous way:

Actually . . . talking particularly about doctors I think there are some clans if you like or big groups of doctors that are much less likely to embrace change and [. . .] are much more likely to want to carry on doing things their own way as they see fit, which is the way that they, they’ve done it.

11IV03

Different staff groups with differing experiences had different values when it came to working weekends. Some consultants, for example, were seen as reluctant to take on weekend working, having already ‘paid their dues’ as a junior.

Lack of resources

The most debilitating constraint on implementing 7-day services was a lack of staff and the ability to pay for them. This was particularly the case for two sites in ‘special measures’, one with a large financial deficit and the other with both financial and quality issues. Both sites had real problems in recruiting and retaining staff and, thus, in delivering a 7-day service. Justifying the costs to commissioners of employing extra consultants to facilitate patient flow was also difficult. This left little room for proactive efforts to implement 7-day services:

I get the impression that it is not so much working towards a real 7-day service as an effort to keep patients safe with the bare minimum staffing levels because, as I am sure you are aware, staffing recruitment, retention is a huge issue with the trust and . . . it is unrealistic to I think that at the current situation, it is unrealistic to expect a genuine 7-day service.

17IV04

One organisation (site 10) had lost middle-grade trainees after they had been reallocated to one of the other hospitals in the trust. This meant that the site had to use its consultants differently to cover the out-of-hours rotas. The lack of middle-grade doctors was seen as particularly problematic for some sites aiming to make optimal use of their consultant hours:

I think the other thing that’s hitting everyone and us included is of course the situation in relation to juniors gaps because so, gaps in middle-grade rota, particularly in middle-grade rotas during out-of-hours periods as a result of some posts being unfilled, again this is a national problem. And of course that has a knock-on effect on consultants . . . If the consultant ends up doing a resident shift then clearly that will compromise their work the next day.

12IV01

Lack of support from community services at the weekend

All of the sites had difficulty in ensuring patient flow, and, for some, this was seen as the main area of weakness in their ability to provide high-quality care at weekends. Community and social work services were typically unavailable over the weekend and, as a result, consultant-led ward rounds were not felt to be needed at the weekend. Indeed, even others who were performing relatively well on meeting the standards commented that much of the inability to deliver 7-day standards and services was owing to the lack of support from social and community services. The problem of providing good-quality 7-day services often lay beyond their control:

Theoretically we could [conduct geriatrician ward rounds at the weekend]. But the reality is that social services don’t work on the weekends. And nursing homes don’t do assessments on the weekends, so actually there’s no point. There are services that we don’t run 24/7 because there’s no point, because you’d just be paying people to come in and say nothing’s changed since Friday.

18IV04

Facilitators of implementation

Sites that considered themselves to have embraced the 7-day standards described strong leadership as playing a key role in overcoming resistance to changing ways of working. The way that staff were communicated with was considered crucial to encouraging their investment in delivering 7-day services. Indeed, an organisation appearing to listen to its front-line staff was fundamental in encouraging a perception of unity and willingness to work together to achieve weekend working. Taking time to encourage and explain helped to gain staff participation in weekend working, particularly among specialisms:

There was quite a big change to get to 7-day working, and the kind of the 12-hour shifts, and actually, it was a strong leader who did bring people together and . . . Because they were tight, quite a tight group of clinicians, brought them on board so that they did it.

07IV01

Management support and investment, whether in time or in additional staff, was critical to creating the conditions for implementing external standards:

They’ve invested in the acute services quite heavily in the past 4 [or] 5 years, because we’ve gone from four to nine consultants, we’ve gone from four- to six-bedded area, to additional space for ambulatory growth and for this assessment area if you like.

03IV01

Ultimately, however, sites that were more receptive to change were those that had the greatest ‘organisational slack’; they were more likely to have a history of higher performance and to be facing fewer resource and infrastructure challenges. 122

Discussion

We have adopted the relatively novel approach in mixed-methods evaluation of presenting qualitative themed assessments (the RAG rating of progress towards 7-day standards; see Table 9) in a form that has allowed comparison with a completely independent set of assessments from the case record review. The RAG rating was developed through facilitated discussion between the qualitative researchers, who were blinded to the outputs of the case record review. This binocular approach strengthens our conclusions.

Our analysis of the role of consultants at weekends highlighted the mechanisms through which senior specialist input to patient care could contribute to optimising the quality of service delivery at weekends. Consultants have overall responsibility for the diagnosis and treatment of individual patients, conducting or organising patient reviews, and organising and allocating tasks to junior doctors and other members of their service’s teams. They have organisational oversight, contributing to maintaining momentum, and facilitating discharge, helping to maintain patient flow. They also play a crucial scaffolding role, supervising and supporting junior staff. Our analysis also identified, however, that the impact of consultant input to care at weekends was limited when there was a lack of a supportive infrastructure to ensure decisions could be actioned, where key in-hospital services were not available, and when lack of services in the community slowed discharge-planning.

Weekends in hospital differ from weekdays in terms of available resources, care processes and access to expertise, but there is also wide variation between hospitals, with some having developed innovative ways of closing the weekend–weekday gap. The gap seems to affect already-hospitalised patients more than those presenting acutely. At weekends, EAs may even receive more timely care than on weekdays. It is of interest that the qualitative researchers’ assessments in the workplace of weekend care quality should be associated with the global quality judgements of the case record reviewers: a shared perspective emerging from two distinct and independent sources of information.

In terms of care quality, we find no evidence that the quality of in-hospital medical care of patients undergoing EA is lower at weekends than on weekdays. However, we do find strong evidence of a temporal trend towards improved hospital care between epochs: there was a statistically significant improvement in error rates, error-related adverse events and global quality of care assessments. The error-related adverse event rate of 2.5% compares favourably with a recent systematic review finding of 6% across 70 studies between 2000 and 2018. 123 Our findings triangulate well with improvements in in-hospital processes of care between epochs, such as documentation of initial consultant review and recording of NEWS. They are also aligned with the ‘rising tide’ phenomenon of secular improvements in care processes, such as those identified by the evaluation of the Safer Patients Initiative in the previous decade. 124,125

Seven-day services is clearly a complex intervention. Trusts varied in their approaches to introducing this change in service provision. Commitment-orientated approaches and a collaborative culture would seem to be advantageous. The case record review shows a reduction in error rates between epochs, and particularly in those forms of error that are most influenced by doctors: assessment, diagnosis, treatment, prescribing and communication. It seems possible that this may be a consequence of more timely consultant reviews across all days of the week, not just at weekends.

The improvement of in-hospital care processes are in marked contrast to the pre-hospital data, which indicate deficiencies in the provision of community services at weekends, services which have deteriorated with time. The qualitative researchers also identified the interface between hospital and community services, and the decrement in provision of community services at weekends, as distinctly problematic. Compared with weekday admissions, patients admitted at weekends were more likely to be dependent, have a palliative care decision in place and arrive by ambulance to the ED; they were much less likely to have been reviewed and referred by a GP in the community, as would usually be the case. All of these indicators had deteriorated further by epoch 2 (2016–17). These findings are consistent with those of other studies showing that weekend admissions from the community are sicker than those admitted on weekdays56,57 and there are fewer GP referrals at weekends;57,106 and with the documented reduction in social care funding despite increasing demand. 126

Case record review lacks precision when using a single review of a single record by an expert reviewer. 127 Consensus-based joint reviews create an illusory improvement in reliability. 128 Averaging across multiple reviews improves reliability;129 the approach we have adopted, with 4000 case records in total and 4763 usable reviews covering 20 trusts in two epochs, has led to one of the largest reviews undertaken, and distinguishes reliably between trusts (Spearman–Brown coefficients 0.8–0.9). Our conclusions are enhanced by triangulation with qualitative research (involving interviews, and observations employing elements of an ethnographic approach) and employing a difference-in-difference analysis that minimises confounding from variation in case mix. Our analysis of the implementation of 7-day standards identifies the challenges to and facilitators of increasing specialist intensity.

In summary, we describe the role of consultants in providing quality of service delivery in hospitals at weekends, and highlight how the impact of consultants on quality of care is moderated by the availability of supportive infrastructure and resources. Essential pre-conditions for improvement include the perceived legitimacy of the standards, leadership and organisational ‘slack’. However, we find no evidence to support the contention that the weekend effect is attributable to worse in-hospital care at weekends. In-hospital care improved between 2012/13 and 2016/17, and, at trust level, qualitative researchers’ findings on weekend care quality concurred with the case record reviewers’ judgements of global quality. By contrast, we find evidence that community care performs less well at weekends, has deteriorated between epochs and could be the source of the weekend effect. Standards to improve weekend care had a mixed reception in practice and there were significant challenges to implementation, including resource constraints and a lack of focus on improving community services provision. Policy-makers should focus their efforts to improve acute and emergency care on a ‘whole-system’ 7-day approach that integrates social, community and secondary health-care organisation and delivery.

The final stage in the HiSLAC programme was to determine whether or not increasing consultant intensity at weekends would be cost-effective. We address this issue in the next chapter.

Introduction

In 2013, NHS England published The NHS Belongs to the People: A Call to Action,130 in which it called for a public debate on how to develop new ways of working in the health service to respond to the financial crisis, given a funding gap of around £30B between projected spending requirements and resources available between 2013/14 and 2020/21. In the event, the government’s 2015 spending review indicated that additional funding for the NHS, part of which was intended to facilitate the introduction of 7-day services, would have to be accompanied by savings of £22B, suggesting that trusts were likely to be expected to find at least some of the funding for 7-day services from existing resources. 131 However, the true cost (including opportunity costs) of implementing 7-day services was unknown, and the potential benefits are uncertain and unquantified. The cost of training a doctor to the level of consultant was estimated to be £510,411 in 2016, with an annual basic salary of £89,046 and cost per hour of £105. 132 Given the emphasis in the 7-day service standards on expanding consultant input at weekends, it was clear that a major cost component for individual trusts to consider would be doctors’ salaries. The HiSLAC study provided an opportunity to examine the cost-effectiveness of this element of 7-day services on the quality of emergency care.

Economic evaluation of medical products and technologies is standard practice in developed health-care systems,133,134 generally extrapolating effectiveness estimates from randomised controlled trials to generic health end points, such as QALYs. Historically, less attention has been paid to service delivery interventions affecting staffing and system organisation, although this is now starting to change. 135 Service delivery interventions are complex, act further upstream of the patient than drugs or devices, and may have diffuse ‘downstream’ effects. They therefore require large sample sizes for accurate estimation and pose potential difficulties for causal interpretation. 136

The level of specialist intensity is a good example of a complex service delivery intervention acting through multiple causal pathways. It has both direct and indirect effects, that is acting through a mediating variable or not, respectively. 137 Specialist presence affects, for example, both the acute phase of stabilisation, diagnosis and treatment at admission, and the subsequent risk of experiencing preventable harm. Indeed, it was this mechanism that motivated the policy. As discussed in Chapter 1, the observed increased risk of mortality among weekend admissions was most frequently assumed to be a consequence of a reduction in care quality. However, the level of specialist presence is also likely to be correlated with the health of the patient population. 106 A single regression model of mortality on day of admission, as much of the weekend effect literature is based on, cannot separate these two effects, a problem discussed widely at the time. 138 However, a structural equation modelling approach can be used to specify models based on causal hypotheses of the relationships between observed and latent variables. 139–141

In this chapter, we present results from a cost–benefit analysis of a policy of changing the specialist-to-patient ratio, based on a structural equation modelling method. Our objective in using this approach was to develop, compare and estimate models that accounted for the different possible pathways through which specialists might affect the relevant outcomes. By jointly estimating models from across the causal pathway we could separate out the effects of unobserved patient health from those needed for cost–benefit analysis, subject to some key assumptions. We took a Bayesian approach to model specification, estimation and interpretation, which permits the incorporation of prior information and knowledge, typically affords the estimation of more complex joint and hierarchical models, and naturally fits into a decision-making framework.

Aim

The aim was to determine whether or not implementing a policy of increasing specialist intensity at weekends to the level observed in ‘high-intensity’ hospitals is cost-effective.

Methods

Model development

To support the development of the model, an expert workshop and focus group was convened initially to discuss the mechanisms by which increasing the intensity of specialist input at weekends might affect patient outcomes. Based on this discussion and on previous literature, a candidate set of three models of increasing complexity was developed (Figure 23). All of the models included the causal effect of specialists on the risk of experiencing each of the nine types of error and a direct effect on length of stay due to early or delayed discharge. Model 1 has these two pathways only, so that the only mechanism by which specialists could influence adverse events and mortality is by affecting the risk of an error. Unobserved patient health was also included, which had an effect on the risk of mortality and overall length of stay only. However, model 1 may be unrealistic, sicker patients may be taken to hospitals with greater specialist presence or, conversely, less severely ill patients may not be admitted when specialist levels are lower. Model 2, therefore, extends model 1 by allowing for a relationship between the specialist-to-patient ratio and patient health. So, for example, an increase in specialist presence could affect length of stay through three mechanisms: changing the risk of errors, changing the risk of delayed discharge or correlating with the health of the patient population. Model 3 adds an additional extension by allowing for a relationship between patient health and the risk of each type of error and the risk of experiencing an adverse event. More severely ill patients may be at higher risk of experiencing an error as they are exposed to a greater number of interventions and have more complex conditions.

FIGURE 23.

Three candidate models.

Results

Summary statistics

Using data from the PPS, Figure 24 shows the levels of specialist intensity at the participating sites on Sundays and Wednesdays in the two epochs. Although the majority of sites were estimated to have increased their level of specialist presence at the weekend, most also reduced their specialist intensity on a weekday. Across the 20 trusts, the mean level of specialist intensity at the weekend increased from 25.1 hours per 10 EAs to 27.7 hours, but on weekdays intensity declined (52.7 and 49.8 hours per 10 EAs, respectively). As we have shown (see Chapter 4), the reduction in weekday intensity is explained partly by a disproportionately greater increase in Wednesday EAs than in Sunday admissions between the two epochs, but it is likely that specialists have transferred a proportion of their weekday work to weekends.

FIGURE 24.

Mean specialist intensity by trust and epoch, hours per 10 EAs.

Summary statistics of the sample of 4000 admissions by epoch and day of admission are shown in Table 17. There were some changes in the patient population over the two epochs. The proportion of patients arriving by ambulance declined at weekends and on weekdays over time, although it remained relatively larger at the weekend. The proportion of patients experiencing almost all types of error also declined over time; there was no clear pattern showing that the number of errors or adverse events were consistently larger for weekend or weekday admissions. Mortality among weekend admissions was higher than that among weekday admissions in both epochs.

TABLE 17 - Summary statistics of the sample by epoch and day of admission

CPR, cardiopulmonary resuscitation.

	Epoch 1		Epoch 2
	Weekend	Weekday	Weekend	Weekday
Hospital level
n	20	20	20	20
Specialists (hours/10 EAs)	25.5 (15.0)	52.7 (22.0)	26.9 (13.2)	48.4 (20.7)
Individual level
n	993	996	992	988
Age (years), mean (SD)	60.5 (22.5)	60.7 (22.2)	62.8 (22.2)	60.4 (22.3)
Female (%)	53.8	51.3	52.4	55.6
Arrival by ambulance (%)	55.7	44.9	49.0	39.8
Communication error (%)	6.8	8.3	6.9	5.7
CPR error (%)	1	0.5	0.9	0.4
Diagnostic error (%)	15.7	16.3	12.2	11.9
Infection error (%)	0.3	0.4	0.8	0.8
Invasive procedure error (%)	0.9	0.6	0.5	0.3
Medication error (%)	8	7.4	5.4	5.9
Monitoring error (%)	2.7	2.9	2.1	2.3
Other error (%)	2	1.7	1.1	1
Treatment or management error (%)	14.1	14.9	11.3	10.6
Adverse event (%)	3.2	3.5	2.2	1.5
In-hospital mortality (%)	4.9	4.2	4.1	3.6
Length of stay, median (IQR)	2 (1–5)	2 (0–5)	2 (0–5)	1 (0–4)

We can examine the relationship between specialist intensity and markers of patient health to support or refute different models. Figure 25 shows the association between specialist intensity and patient age and the proportion of patients arriving by ambulance. There is evidently a relationship, which would initially refute model 1, as it provides evidence that patient health is associated with level of specialist intensity.

FIGURE 25.

(a) Relationship between specialist intensity and patient age (b) and the proportion of patients arriving by ambulance.

Model comparisons

Model 3 was the preferred model under both comparison statistics. Model 2 was a large improvement over model 1, and model 3 was a (smaller) improvement over model 2. Figure 25 shows the estimated adjusted relationships between specialist intensity and the risk of each of the nine types of error from the three models. For models 1 and 2, which do not allow for patient health to affect the risk of error, there is little evidence of a relationship between specialist intensity and the risk of each type of error (see Appendix 19). For model 3, higher levels of specialist intensity are associated with a reduced risk of diagnostic, communication, medication, monitoring and management errors. For example, an increase in specialist intensity from 25 to 50 hours per 10 EAs is associated with a reduction in the odds of a diagnostic error of 26% (95% CrI 2% to 45%). These results suggest that there is a degree of confounding in models 1 and 2 as we expected; higher levels of specialist intensity are associated with both a lower risk of error and poorer health among patients, and poorer health is associated with a higher risk of error.

Little evidence of a relationship was seen between specialist intensity and the risk of experiencing an adverse event, and there was a large degree of uncertainty about any effects. The risk of an adverse event was very low (1.35%; see Table 17). Diagnostic errors were the most likely to lead to harm; the risk of an adverse event was approximately 10% with a diagnostic error, compared with the next most common error, treatment/management errors, which had a risk of < 1%. As a result, the doubling of specialist intensity from 25 to 50 hours per EAs was associated with a reduction in the number of patients experiencing adverse events only by approximately 2 in 1000. For the same reason, there was little evidence of a relationship between specialist intensity and mortality. Although experiencing an adverse event was associated with an increased risk of mortality in all models, given the relatively low risk of both mortality and adverse events, the effects of specialist intensity are very small. For the same doubling of specialists described above, we would expect five fewer deaths per 10,000 EAs on average.

Figure 26 shows the posterior predictive distribution of length of stay compared with specialist intensity. Both models 2 and 3, which allowed for latent health to be correlated with specialist intensity, mortality and length of stay, showed a shorter expected length of stay with higher levels of specialists. Indeed, the effects were, qualitatively, highly similar, although they were associated with wide CrIs. Most of the observed effect was from the direct effect of specialists rather than from a reduced risk of adverse events.

FIGURE 26.

Posterior predictive distribution of length of stay and specialist intensity: all models.

Cost–benefit analysis

Model 3 was selected as the best-fitting model and was therefore used for the cost–benefit analysis. As Figure 24 and Table 17 show, the 7-day services policy had little discernible effect on the level of specialist intensity in these 20 trusts. However, we can examine the expected net benefits of the policy, had it been implemented with greater fidelity, by predicting the outcomes from model 3 at a level of specialist intensity of 26.9 hours per 10 EAs (the weekend average) and 48.4 hours per 10 EAs (the weekday average). The posterior mean expected net benefit was £177, indicating that the policy is cost-effective at a willingness to pay per QALY of £20,000.

On average, the policy would cost £167 per EA, generate £5 per EA in QALYs from the reduced risk of an adverse event, and save £349 as a result of a reduced length of stay. Model 3 predicted that the length of stay would be reduced by an average of 0.9 days under this policy. Given that QALY gains made a relatively small contribution to the overall expected net benefits, the model was not affected by the choice of cost-effectiveness threshold.

To consider more generally the cost–benefit of increasing specialist intensity from this model, we next examined the effects of increasing specialist intensity incrementally from its lowest to highest observed values. Figure 27 shows the expected net benefits of increasing the level of specialists by 10 hours per 10 EAs. The expected net benefits were greater than zero across the range of specialist intensity. However, they were flat and close to zero above approximately 35 hours per 10 EAs, and the 95% CrI was fairly wide, with a substantial area below zero. Below 35 hours per 10 EAs, there is much stronger evidence that increasing specialist intensity is cost-effective.

FIGURE 27.

Posterior predictive distribution of expected net benefits of raising specialist intensity by 10 hours per 10 EAs with 95% CrI.

Discussion

As described above, one of the main aims of the 7-day services policy was to improve care quality for patients admitted to hospital at the weekend. However, the confounding of the relationship between specialist intensity and care quality by patient health status complicates the task of identifying the effect of the policy from observational data. Indeed, this is a general problem in the evaluation of service delivery interventions, which, by definition, have effects that are mediated by other services and have clinical processes that are diffuse but affect a large proportion of the patient population. 136 The policy was not implemented as designed, and so the expected within-hospital variation in specialist intensity that we planned to use to identify the effects of the policy was not present. However, significant variation in specialist intensity levels between hospitals and by day of admission was present, and this enabled us to examine the relationship of this variation and patient outcomes. Implicit, however, in these analyses is the assumption that any other changes in care quality by day of admission were constant across hospitals in each epoch, which may be a strong assumption.

For this study, we developed structural equation models for this type of evaluation, which enabled us to disentangle the different causal pathways. In doing so, our evidence supports the hypothesis that an increase in the level of specialist intensity is associated with reductions in the risk of errors and adverse events, which is confounded by patient health. However, the effect is small and certainly not large enough to justify the potential outlay required to implement the policy. This agrees with a number of previous studies and discussions that have been published based on cruder analyses. 29 A more significant benefit is estimated in the reduced length of stay associated with increases in specialist intensity, which is potentially large enough to make the policy cost-effective, particularly when specialist intensity is low.

We estimated that increasing specialist intensity at the weekend to weekday levels would reduce the average length of stay by almost a day, thereby significantly reducing the cost of patient care. The qualitative analysis (see Chapter 6) also identified delayed discharge as a key issue at the weekend that results from a lack of specialists. However, the system-level implications of reducing length of stay have not been explored in detail. Overall patient volume and bed occupancy would be unlikely to reduce, but throughput would increase. This would imply an improved efficiency of the health service and potentially improved patient experience as a result of less unnecessary time spent in the hospital, although we did not account for patient experience in our analysis here. Any improvement in care quality would also affect the patient experience, although we reiterate that significant adverse events were exceedingly rare. The types of error that the analysis suggested would be impacted by specialist presence also aligned with those identified in our focus group exercise, further validating the evidence.

There has been growing interest in the evaluation of service delivery interventions. 155,156 Watson and Lilford,156 Lilford et al. 136 and Watson et al. 157 have discussed the necessity of capturing and integrating data from across the causal chain to estimate policy relevant effects and conduct economic evaluations. The analyses in this chapter exemplify how this might be done in practice using a structural equation modelling approach. However, we caution that, in the absence of an experimental methodology, the results cannot necessarily be interpreted as causal effects. Even the more complex models are simplifications of the health system and may miss issues of confounding or reverse causality. However, as we discussed, the data become less and less informative about model parameters as complexity increases. A method of principled Bayesian analysis using qualitative model development and comparison aims to deal with this problem. 158 We believe that our analysis provides the most reliable evidence to date of the cost–benefit of increasing specialist intensity.

We reiterate the issues with the data discussed in previous chapters. Response rates to the PPS were variable and often low. The consequences of this could be significant in terms of bias if the response rate were systematically dependent on other key variables, such as patient health or specialist intensity. However, we do not have reason to believe that this is the case and the results have strong face validity. Although the inter-reviewer reliability of subjective case note review was low at the individual level (see Chapter 6, Case record review: results), it was high when aggregated at trust level and by epoch. By the same reasoning, we expect the ‘noise’ to cancel out in the aggregate. Our results here focus on average effects, and the effect of the high variation in case note review is to increase our uncertainty around these effects rather than to introduce any systematic bias.

Had the 7-day services policy been implemented as designed, the evidence presented in this chapter suggests a high probability that it would have been cost-effective. This may seem to contradict previous findings and discussions (e.g. Godlee,138 Meacock et al. 29), which determined that improvements in quality and reductions in mortality would be unlikely to justify the policy on cost-effectiveness grounds. We would agree with this assessment, but disagree with the overall conclusions, as through reductions in length of stay the policy would likely be justifiable. Thus, we would tentatively conclude that the 7-day services policy was right, but for the wrong reasons.

Overview

In this 5-year programme of research, we used quantitative and qualitative methodologies to investigate why the risk of death seems to be higher for patients admitted to hospitals at weekends than for those admitted during the week. The answer turns out to be complex. Contrary to our initial hypothesis, we find no evidence that this situation is related to deficiencies in hospital medical staffing. We do find circumstantial evidence that the causes may lie in the community, upstream of hospital admission. Importantly, we find that, although the overall quality of hospital care in general is not worse at weekends, and improved between 2013/14 and 2017/18, the limited measures available to us suggest that community care is worse at weekends and seems to have deteriorated with time.

The 7-day services policy was – and is – a commendable package of measures intended to improve the overall quality of care for acutely ill patients and EAs to hospital, but the use of the weekend effect as an argument to effect change was unhelpful. As the weekend effect has been reported for 20 years from most health systems worldwide, the cause or causes needed to be identified to inform policy, rather than merely attributing it to deficiencies in hospital medical staffing.

Determining causation in health services usually relies on interpreting associations, strengthened by the use of instrumental variables, where these can be identified. We took the opportunity to use the introduction of 7-day services as an (uncontrolled) intervention to test the hypothesis that increasing the availability of hospital specialists (consultants, attendings) at weekends would mitigate the weekend effect. As this was, in effect, a ‘natural experiment’, we wished to strengthen the interpretation of the findings using triangulation: looking for similar signals from different and independent sources. We also wanted to ensure that we had adequate representation from those on the front line, namely patients and staff. We therefore used qualitative methods in addition to quantitative measures. We also examined changes over time. By comparing secular changes in weekend and weekday differences within individual trusts, we were able to reduce confounding from between-trust comparisons, for example in case mix, or service delivery and organisation.

Summary of main findings

Here we summarise our main findings and then consider their implications.

Implications for policy and practice

The main implication of our research is that efforts to improve the care of patients who require EAs to hospital must be system-wide and occur across all days of the week. The focus on mortality associated with hospital admission has distracted attention from deficits in community health services at weekends; moreover, mortality is not a good indicator of care quality. 129,164 Our findings provide evidence to support many of the recommendations in National Institute for Health and Care Excellence guideline 94 entitled Emergency and Acute Medical Care in Over 16s: Service Delivery and Organisation on emergency and acute medical care135 and are consistent with a recent systematic review of models of integrated care. 165 They also provide further justification for the promotion of the development of integrated care systems in the NHS149 through Sustainability and Transformation Partnerships166 and acute and emergency care collaborations,167 such as the one in London that incorporates the development of seamless partnerships across social and primary and secondary health care. 168 New research being commissioned by the NIHR Applied Research Collaboration West Midlands169 has a focus on integrated care and a subtheme of acute care. The emergency care pathways for frail elderly people and those with multimorbidity are an important target for further research.

Recommendations for future research

The global burden of emergency medical diseases has been estimated to cause 28.3 million deaths and to contribute to 50.7% of all deaths worldwide. 170 The need for a system-wide integrated approach to improving outcomes through research is well recognised in the UK. 169,171 The HiSLAC study demonstrates that signals from performance indicators conventionally attributed to hospital practice may originate in the community, both before and after discharge from hospital. Our research recommendations, therefore, take this wider view of exploring quality of care throughout the patient pathway.

Conclusions and lessons learned

The HiSLAC study has shown that the weekend effect is most likely to be a system-wide pathway problem that has its origins in community services. Contributory factors in the community include budgetary constraints during the period of austerity that followed the financial crisis of 2008, and the progressive increase in health service utilisation, accompanied by persistent demands for greater efficiencies and reductions in hospital length of stay. The improvements in hospital safety and care quality between 2012/13 and 2016/17 contrast with the deterioration in community care metrics. Although direct comparisons of secondary and primary care were not possible in this study, the evidence we have presented here is consistent with the hypothesis that the closer integration of social, primary and secondary care will provide benefits to patients. 178,179

The marked reduction in hospital errors and adverse events, and the improvement in global quality identified between epochs by the HiSLAC case record reviewers has parallels with the MERIT study,180 the Safer Patients’ Initiative116,125 and the Matching Michigan study;181 it has been described as the ‘rising tide’ phenomenon. 124 The reduction in errors and adverse events demonstrated in HiSLAC could be a manifestation of a non-specific long-term secular trend, but it is also consistent with being driven by the salience of the 7-day services policy, in particular the six standards that require increased consultant involvement in the acute-care pathway. If this were the case, then the 7-day services policy should be regarded as having benefited patients across all days of the week.

Oversight and Governance Committee

Michael Rawlins (chairperson), Jennifer Dixon, Peter Lees, Paddy Storrie, Alastair Henderson, Matt Sutton and Naomi Fulop.

Local project leads from participating NHS trusts

Dr Tim Felton and Dr Tim Cooksley, Manchester University NHS Foundation Trust; Dr Victor Lawrence, Isle of Wight NHS Trust; Dr Michelle Gallagher, Royal Surrey NHS Foundation Trust; Dr Rebecca Thorpe, University Hospitals Bristol NHS Foundation Trust; Dr Andrew Griffiths, Torbay and South Devon NHS Foundation Trust; Professor John Kinnear, Southend University Hospital NHS Foundation Trust; Dr Rupert Negus, Royal Free London NHS Foundation Trust; Dr Achim Schwenk and Dr Cathy Cale, North Middlesex University Hospital NHS Trust; Dr Fiona Wisniacki, The Hillingdon Hospitals NHS Foundation Trust; Miss Jane Wilson, Kingston Hospital NHS Foundation Trust; Dr Matthew Hayman, Taunton and Somerset NHS Foundation Trust; Dr Richard Jee, Dorset County Hospital NHS Foundation Trust; Dr Mark Lipton, Wirral University Teaching Hospital NHS Foundation Trust; Dr Shirley Hammersley, Mid Cheshire Hospitals NHS Foundation Trust; Dr Roope Manhas, Northern Devon Healthcare NHS Trust; Dr Shwe Zin Chit Pan, Bedford Hospitals NHS Foundation Trust; Dr Peter Albert, Bedfordshire Hospitals NHS Foundation Trust; Dr Andrew Alldred, Dr Peter Hammond and Dr David Scullion, Harrogate and District NHS Foundation Trust; Dr Franco Guarasci, Airedale NHS Foundation Trust; Dr Mark Blunt, The Queen Elizabeth Hospital King’s Lynn NHS Foundation Trust; Dr Tina Tamm, Heatherwood and Wexham Park Hospitals NHS Foundation Trust; Dr Ian Reckless, Milton Keynes University Hospital NHS Foundation Trust; Dr Marcus Pittman, Basildon and Thurrock University Hospitals NHS Foundation Trust; Dr Allan Harkness, Colchester Hospital University NHS Foundation Trust; Dr Earl Williams, The Royal Bournemouth and Christchurch Hospitals NHS Foundation Trust; Dr Mark Daly and Dr David Friedericksen, Royal Cornwall Hospitals NHS Trust; Dr John Hollingsworth, Aintree University Hospital NHS Foundation Trust; Dr Magda Smith, Barking, Havering and Redbridge University Hospitals NHS Trust; Professor Anil Hormis and Dr David Harling, The Rotherham NHS Foundation Trust; Dr Anitha Mathews, Dr Suzanne Hamilton and Dr Callum Gardner, North West Anglia NHS Foundation Trust; Dr Paul Molyneux, West Suffolk NHS Foundation Trust; Dr Chris Adcock, Cambridge University Hospitals NHS Foundation Trust; Mr James Hobbs and Mr Martin Cooper, Royal Devon and Exeter NHS Foundation Trust; Dr Mike Clancy, University Hospital Southampton NHS Foundation Trust; Dr Andrew Gibson, Sheffield Teaching Hospitals NHS Foundation Trust; Dr Phil Young, Portsmouth Hospitals University NHS Trust; Dr Emma Vaux, Royal Berkshire NHS Foundation Trust; Dr David Wood, Guy’s and St Thomas’ NHS Foundation Trust; Dr Mehool Patel and Dr David Sulch, Lewisham and Greenwich NHS Trust; Dr Nnenna Osuji, Croydon Health Services NHS Trust; Dr Jane Evans, St George’s University Hospitals NHS Foundation Trust; Dr Viviana Elliott and Dr Jyothi Nippani, South Warwickshire NHS Foundation Trust; Dr Nick Coleman, University Hospitals of North Midlands NHS Trust; Mrs Deborah Price and Dr Andrew Willis, Burton Hospitals NHS Foundation Trust; Dr Sanjeev Srivastava, East Cheshire NHS Trust; Dr Frank Joseph, Countess of Chester Hospital NHS Foundation Trust; Dr William Bernal, King’s College Hospital NHS Foundation Trust; Dr Andrew Haynes, Sherwood Forest Hospitals NHS Foundation Trust; Dr Sam Waddy, University Hospitals Plymouth NHS Trust; Dr Richard de Boer and Dr Mike Iredale, University Hospitals Coventry and Warwickshire NHS Trust; Dr Clarissa Murdoch, The Whittington Hospital NHS Trust; Dr Lee Dowson, The Royal Wolverhampton NHS Trust; Dr Martyn Farrer, City Hospitals Sunderland NHS Foundation Trust; Ms Jo Sampson and Dr Rupert Ransford, Wye Valley NHS Trust; Dr Catherine Free and Dr Gordon Wood, George Eliot Hospital NHS Trust; Professor Erika Denton, Dr Neil Burgess and Mr Peter Chapman, Norfolk and Norwich University Hospitals NHS Foundation Trust; Dr Richard Warner, Salford Royal NHS Foundation Trust; Dr Natalie Walker and Dr Leonidas Zaccharides, Bolton NHS Foundation Trust; Dr Martin Pattrick, Tameside and Glossop Integrated Care NHS Foundation Trust; Dr Debesh Mukherjee, Great Western Hospitals NHS Foundation Trust; Mr Ben Cresswell, Hampshire Hospitals NHS Foundation Trust; Dr Aref Rastegar and Dr Lanitha Srikugan, Dartford and Gravesham NHS Trust; Dr Hassan Paraiso, The Dudley Group NHS Foundation Trust; Dr Rod Harpin, North Cumbria University Hospitals NHS Trust; Dr Raja Reddy, Kettering General Hospital NHS Foundation Trust; Dr Raghu Govindan and Dr Amanda Bisset, Northampton General Hospital NHS Trust; Dr Stuart Henderson, Salisbury NHS Foundation Trust; Dr Timothy Noble, Doncaster and Bassetlaw Teaching Hospitals NHS Foundation Trust; Dr Graeme Sanders, Medway NHS Foundation Trust; Dr Peter Burnham and Dr Kathryn Clark, Royal Liverpool and Broadgreen University Hospitals NHS Trust; Dr Jayachandran Radhakrishnan, Mid and South Essex Hospital Services NHS Trust; Dr Emma Rowlandson, Chelsea and Westminster Hospital NHS Foundation Trust; Dr Andy Morris, The Princess Alexandra Hospital NHS Trust; Dr Emma Rowland, Homerton University Hospital NHS Foundation Trust; Dr Rifat Rashid and Professor Matthew Cooke, Heart of England NHS Foundation Trust; Dr Kilimangalam Narayanan and Dr Christopher Scott, Gateshead Health NHS Foundation Trust; Dr Graham Johnson, Leeds Teaching Hospitals NHS Trust; Dr Catherine Snelson and Dr Mike Berry, University Hospitals Birmingham NHS Foundation Trust; Dr Samer Elkhodair, Dr Nicoletta Gaston, Dr Breid O’Brien and Ms Simone Akuffo-Akoto, University College London Hospitals NHS Foundation Trust; Dr Sean Elyan, Gloucestershire Hospitals NHS Foundation Trust; Dr Eliot Sykes, Northumbria Healthcare NHS Foundation Trust; Professor Richard Donnelly, Derby Teaching Hospitals NHS Foundation Trust; Dr James Price, Oxford University Hospitals NHS Foundation Trust; Mr Pasha Nisar, Ashford and St Peter’s Hospitals NHS Foundation Trust; Dr Sath Nag, South Tees Hospitals NHS Foundation Trust; Dr Andrew Higham, University Hospitals of Morecambe Bay NHS Foundation Trust; Dr Kiaran Flanagan, North Bristol NHS Trust; Dr Pauline Swift, Dr Aik Guam Lim and Dr Steve Hyer, Epsom and St Helier University Hospitals NHS Trust; Dr Paul Stevens, East Kent Hospitals University NHS Foundation Trust; Dr Nick Roper, North Tees and Hartlepool Hospitals NHS Foundation Trust; Dr Patrick McDonald, Southport and Ormskirk Hospital NHS Trust; Dr Nick Tierney, Pennine Acute Hospitals NHS Trust; Dr Sanjiv Jain, United Lincolnshire Hospitals NHS Trust; Dr Mark Ardron, University Hospitals of Leicester NHS Trust; Dr Vinay Bhatia and Dr Clive Lawson, Maidstone and Tunbridge Wells NHS Trust; Dr Michael Van Der Watt, West Hertfordshire Hospitals NHS Trust; Professor Phillip Smith, Miss Jane McCue and Mr Neil Harris, East and North Hertfordshire NHS Trust; Dr Colin Wasson and Dr James Catania, Stockport NHS Foundation Trust; Ms Anna Walker, Worcestershire Acute Hospitals NHS Trust; Dr Rehan Naseer, Calderdale and Huddersfield NHS Foundation Trust; Dr Keith Girling, Nottingham University Hospitals NHS Trust; Dr James Wilkinson, East Sussex Healthcare NHS Trust; Mr Ian Carr, The Mid Yorkshire Hospitals NHS Trust; Dr Heather Brown and Dr Wendy Munro, Brighton and Sussex University Hospitals NHS Trust; Dr David Carruthers, Sandwell and West Birmingham Hospitals NHS Trust; Professor Mark O’Donnell, Blackpool Teaching Hospitals NHS Foundation Trust; Dr Paul Peter and Dr Mike Jones, County Durham and Darlington NHS Foundation Trust; Dr Andrew Huang; Buckinghamshire Healthcare NHS Trust; Mr Charles Thomson, East Lancashire Hospitals NHS Trust; Dr Richard Heinink, Shrewsbury and Telford Hospital NHS Trust; Professor Julian Redhead and Dr William Oldfield, Imperial College Healthcare NHS Trust; Dr Tim Taylor and Dr Rob Haigh, Western Sussex Hospitals NHS Foundation Trust; and Dr Faris Al-Sultan, Weston Area Health NHS Trust.

Case note reviewers

Randeep Mullhi, Jonathon Coates, Neil Patel, Emma Carver, Zoe Kimbley, Nicholas Price, Akshay Nair, Sheena Sikka, Navinee Gilliat, Greg Packer, Anna Blackwell, Emma Christmas, Adebola Adebowale, Sebastien Ellis, Rebecca Mallinson, Farhad Peerally, Daniel Forster, Simmi Krishnan, Vishal Patil, Richard Hopkins, Ruth Todd, Sofia Cavill, James Murray, Imran Ghafoor, Katey Flowers, Paul Dean, Christopher Wright, Shreena Shah, Jonarthan Thevanayagam, Richard Hunt, Karwai Tsang, Mohammed Yusuf, Mark Whitsey, Brian Hogan, Amina Jaffer, Emma Bryden, Rebecca Powell, William Wareing, Yasir Elhassan, Parijat De, Manoj Viegas, Christopher Wharton, Michael Duffy, Richard Heinink, Chris Beet, Elin Roddy, Anna Szczerbinska, Annabel Makan, Manohasandra Majuran Manohasandra, Neel Patel, Behrad Baharlo, James Brown, Jun Jack Wong, Salmaan Mughal, Musarat Hussain, Nick Murch, Georgina Barrows, Ryan Boyle, Syed Habib Zaidi, Ivan Collin, Caroline Kibbler, Monica Bowa-Nkhoma, Clare Hughes, Muniswamy Hemavathi, Thomas Cozens, Khin Nyo, Debbie Beck, Claire Peplow, Devasena Subramayam, Tom McLellan, Gautam Bagchi, David Hartin, Stewart McMorran, Chung Thong Lim, Hiba Mohamed, John Ong, Hoi-Ping Mok, Ayesha Khaliq, Abigail Ford and Alison McInerney.

Additional acknowledgements

Felicity Evison for statistical support with Chapter 5.

Professor Nick Black and Dr Helen Hogan for access to their case record review methodology.

Emma Lancashire, Kam Gareja, Claire Price and Olivia Brookes for project support.

Elisa Liberati, David Kocman, Sarah Chew, Jen Jones and Emma Angell for conducting qualitative research visits to hospital sites (particular thanks go to Emma Angell for her role supporting project management).

Liz Shaw for her contribution to the analysis of qualitative data.

Xavier Armoiry, Caroline Higenbottam, Nicholas Cowley, Ranjna Basra, Chia-Wei Wu and Amy Gosling for their contribution to the systematic review.

Sue Pargeter, Manager, NIHR HSDR programme.

In memoriam

Professor Tim Evans: an outstanding researcher, collaborator, leader, clinician and good friend.

Contributions of authors

Julian Bion (https://orcid.org/0000-0003-0344-5403) (Professor of Intensive Care Medicine) was the HiSLAC project chief investigator. He developed the original design and contributed to methodology development, analysis of data and writing of the manuscript.

Cassie Aldridge (https://orcid.org/0000-0002-9162-3478) (HiSLAC Project Manager) contributed to data collection, methodology development, analysis of data and writing of the manuscript.

Chris Beet (https://orcid.org/0000-0002-9259-9852) (Specialty Registrar in Intensive Care Medicine) contributed to methodology development, development of the case record review and reviewer training materials, and the writing of the manuscript.

Amunpreet Boyal (https://orcid.org/0000-0002-5156-2869) (Research Associate) contributed to data collection, methodology development and writing of the manuscript.

Yen-Fu Chen (https://orcid.org/0000-0002-9446-2761) (Associate Professor, Population Evidence and Technologies) contributed to the systematic review, methodology development and the writing of the manuscript.

Michael Clancy (https://orcid.org/0000-0001-8669-5226) (Consultant Emergency Physician) contributed to methodology development and the writing of the manuscript.

Alan Girling (https://orcid.org/0000-0003-2022-045X) (Honorary Reader in Medical Statistics) contributed to data collection, methodology development, analysis of HES data, statistical analysis and the writing of the manuscript.

Timothy Hofer (https://orcid.org/0000-0003-0434-8787) (Professor of General Medicine) contributed to methodology development and the writing of the manuscript.

Joanne Lord (https://orcid.org/0000-0003-1086-1624) (Director and Professorial Fellow in Health Economics) conducted the health economics analysis and contributed to the writing of the manuscript.

Russell Mannion (https://orcid.org/0000-0002-0680-8049) (Professor of Health Systems) contributed to methodology development and the writing of the manuscript.

Peter Rees (https://orcid.org/0000-0001-5654-6254) (Patient and Public Involvement Lead) contributed to methodology development and the writing of the manuscript.

Chris Roseveare (https://orcid.org/0000-0002-9767-7725) [Consultant in Acute and General (Internal) Medicine] contributed to methodology development and the writing of the manuscript.

Louise Rowan (https://orcid.org/0000-0001-6211-3869) (Communications Fellow) contributed to collaborator engagement, dissemination of findings and the writing of the manuscript.

Gavin Rudge (https://orcid.org/0000-0002-9857-6547) (Research Fellow, Applied Health Research) contributed to data collection, methodology development, analysis of HES data, statistical analysis and the writing of the manuscript.

Jianxia Sun (https://orcid.org/0000-0003-4029-6515) (Statistical Intelligence Analyst) contributed to data collection, methodology development, analysis of HES data, statistical analysis and the writing of the manuscript.

Elizabeth Sutton (https://orcid.org/0000-0002-6691-6287) (Research Associate, Health Sciences) was involved in the qualitative research design and contributed to organising and facilitating the focus groups and interviews, analysis of data and the writing of the manuscript.

Carolyn Tarrant (https://orcid.org/0000-0001-7356-5342) [Professor and Research Group Lead of the Social Science Applied to Healthcare Improvement Research (SAPPHIRE) Group] devised and led on the qualitative research design, and contributed to analysis of data and the writing of the manuscript.

Mark Temple (https://orcid.org/0000-0003-1600-851X) (Consultant Nephrologist) contributed to methodology development and the writing of the manuscript.

Sam Watson (https://orcid.org/0000-0002-8972-769X) (Senior Research Fellow, Population Evidence and Technologies) contributed to the statistical methodology development for the cost–benefit analysis, conducted the health economics analysis and contributed to writing of the manuscript.

Janet Willars (https://orcid.org/0000-0002-7886-3223) (Honorary Visiting Fellow, Health Sciences) contributed to organising and facilitating the focus groups and interviews, analysis and writing of the manuscript.

Richard Lilford (https://orcid.org/0000-0002-0634-984X) (Director of NIHR Applied Research Centre West Midlands, and Research Methodology Lead for HiSLAC) contributed to the writing of the application, design of the project, health economics analysis and the writing of the manuscript.

Publications

Aldridge C, Bion J, Boyal A, Chen YF, Clancy M, Evans T, et al. Weekend specialist intensity and admission mortality in acute hospital trusts in England: a cross-sectional study. Lancet 2016;388:178–86.

Chen YF, Boyal A, Sutton E, Armoiry X, Watson S, Bion J, et al. The magnitude and mechanisms of the weekend effect in hospital admissions: a protocol for a mixed methods review incorporating a systematic review and framework synthesis. Syst Rev 2016;5:84.

Bion J, Aldridge CP, Girling A, Rudge G, Beet C, Evans T, et al. Two-epoch cross-sectional case record review protocol comparing quality of care of hospital emergency admissions at weekends versus weekdays. BMJ Open 2017;7:e018747.

Tarrant C, Sutton E, Angell E, Aldridge CP, Boyal A, Bion J, et al. The ‘weekend effect’ in acute medicine: a protocol for a team-based ethnography of weekend care for medical patients in acute hospital settings. BMJ Open 2017;7:e016755.

Sutton E, Bion J, Aldridge C, Boyal A, Willars J, Tarrant C. Quality and safety of in-hospital care for acute medical patients at weekends: a qualitative study. BMC Health Serv Res 2018;18:1015.

Watson SI, Chen YF, Bion JF, Aldridge CP, Girling A, Lilford RJ, et al. Protocol for the health economic evaluation of increasing the weekend specialist to patient ratio in hospitals in England. BMJ Open 2018;8:e015561.

Chen YF, Armoiry X, Higenbottam C, Cowley N, Basra R, Watson SI, et al. Magnitude and modifiers of the weekend effect in hospital admissions: a systematic review and meta-analysis. BMJ Open 2019;9:e025764.

Sun J, Girling AJ, Aldridge C, Evison F, Beet C, Boyal A, et al. Sicker patients account for the weekend mortality effect among adult emergency admissions to a large hospital trust. BMJ Qual Saf 2019;28:223–30.

Bion J, Aldridge C, Girling AJ, Rudge G, Sun J, Tarrant C, et al. Changes in weekend and weekday care quality of emergency medical admissions to 20 hospitals in England during implementation of the 7-day services national health policy [published online ahead of print October 28 2020]. BMJ Qual Saf 2020.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to available anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HS&DR programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HS&DR programme or the Department of Health and Social Care.