Multicomponent hospital-led interventions to reduce hospital stay for older adults following elective surgery: a systematic review

Article history

The research reported here is the product of an HS&DR Evidence Synthesis Centre, contracted to provide rapid evidence syntheses on issues of relevance to the health service, and to inform future HS&DR calls for new research around identified gaps in evidence. Other reviews by the Evidence Synthesis Centres are also available in the HS&DR journal.

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 16/47/22. The contractual start date was in September 2018. The final report began editorial review in October 2018 and was accepted for publication in March 2019. 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.

Declared competing interests of authors

Rob Anderson is a member of the National Institute for Health Research Health Services and Delivery Research (Researcher-Led) Prioritisation Panel.

Copyright statement

© Queen’s Printer and Controller of HMSO 2019. This work was produced by Nunns 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.

2019 Queen’s Printer and Controller of HMSO

Hospital use by older adults in the UK

Data from the World Health Organization1 indicate that most people across the world can now expect to live until they are aged ≥ 60 years; this is attributed to reduced mortality from childbirth and childhood illness and to declining mortality for older individuals in low-, middle- and high-income countries. 2,3 In the UK, the number of people aged ≥ 60 years is expected to increase from 14.9 million in 2014 to 21.9 million in 2039, with the life expectancy for men and women being 79.4 and 83.1 years, respectively, in 2017. 4 In England, the number of treatment episodes associated with inpatient and day-case activity recorded in NHS hospitals increased from 12.7 to 16.3 million between 2005/6 and 2015/16. 5 During the same period, there was a steady increase in the number and age of patients admitted to hospitals, with the number of combined elective and emergency admissions of patients aged 60–65 years increasing by 57% and the mean patient age increasing from 49 to 53 years. 5

In 2015/16, the largest number of elective and emergency procedures were conducted in patients aged between 65 and 69 years, with the most common procedures in this age group involving bones and joints, diagnostic tests, rehabilitation, the upper and lower digestive tract, and the eye. 5 Older adults admitted to hospital for elective procedures may present a different patient profile from that of younger adults. For example, they may be more likely to have transport difficulties,6 they may be in poor physical health or living with frailty,7 or they may be socially isolated8 or have living arrangements that require additional support following discharge. 9 Older adult hospital inpatients are also at increased risk of peri- or postoperative complications, such as delirium, falls, hospital-acquired infection, pressure sores, muscle wasting (sarcopenia), loss of mobility, poor nutrition and dehydration, cognitive decline and reduced psychological well-being. 10–17 Admission to hospital may also result in the recognition of a previously unidentified frailty syndrome, which is then destabilised by an inpatient hospital stay,18 leading to additional assessment and/or organisation of follow-up care before the patient can be discharged. Such complications can impede patients’ recovery, increase their length of stay (LOS) in hospital and influence their discharge destination. 13 As a result, care pathways may need to be specialised to allow appropriate and effective care.

Within the current financial climate, NHS hospitals are under increased pressure to maintain or improve their provision of care, and ensure the cost-effective delivery of services. The increased number of hospital admissions in an increasingly ageing population indicates that there are both patient-care and financial drivers of the need to manage the length of time older adults need to spend in hospital.

Length of hospital stay as a key outcome

Optimising the LOS in hospital does not consist solely of reducing the number of days until discharge. Discharge should occur when the patient is physically prepared19 and is involved in the decision to discharge20 and when appropriate post-discharge support is in place. 21 Thus, the LOS in hospital could be considered a composite outcome, encompassing multiple indicators of readiness for discharge.

Although LOS is a key outcome, and the primary focus of this review, the patient care pathway does not end on leaving the hospital, and nor does LOS encompass all aspects of patient recovery and well-being. Measures of additional health resource use, hospital re-admissions and complications, as well as indicators of patient recovery and well-being, are of interest when judging the success of treatment.

Hospital-based multicomponent interventions to manage length of hospital stay for planned admissions

Although identifying the optimal point of discharge can be a challenge for health-care practitioners to address on an individual basis, opportunities exist to streamline organisational processes to enhance recovery after treatment and reduce the risk of complications during hospital stay. Although LOS for patients undergoing high-volume procedures has been decreasing overall, large variation remains among patients undergoing the same procedure across different organisations. 22 A Nuffield Trust report23 indicated that around half of patients experiencing delayed discharge had done so because of factors under the direct control of the hospital. In 2007, the Royal College of Surgeons of England24 suggested separating the organisational structures and processes used for people admitted for elective surgery from those used for emergency admissions in order to establish a more predictable workflow, improve continuity of care and reduce LOS. Planned stays, in contrast to emergency admissions, offer hospitals and the wider health and social care systems a more predictable opportunity to structure the organisation and delivery of their service in order to reduce expected risks and optimise patient recovery.

Several organisational strategies or interventions have been developed and implemented both in the NHS and worldwide with a view to standardising service delivery and improving patient recovery and LOS. Comprehensive geriatric assessment (CGA)25 and enhanced recovery after surgery (ERAS)/enhanced recovery protocols (ERPs)26 are examples of such interventions. These are considered to be ‘multicomponent’ interventions in that they combine several different aspects of care that could be delivered individually otherwise.

ERAS Society guidelines27 exist for multiple planned procedures, aiming to provide explicit recommendations for health-care professionals and detailing the care that a patient should expect to receive throughout their inpatient journey. These include preoperative, intraoperative and postoperative aspects of care. Evidence from several systematic reviews indicates the efficacy of the ERAS/ERP26,28 or ‘care pathways’29 and CGA30,31 approaches in reducing patient LOS, or improving other patient outcomes, among both patients undergoing elective procedures and those undergoing emergency procedures. 32 However, systematic reviews in this area frequently do not specifically address older adults26,32 or common elective procedures requiring overnight hospital stays in this population,28 or focus on only one specific part of the patient care pathway, for example discharge arrangements. 9,33,34 Therefore, there is currently a lack of synthesised evidence examining the effectiveness of all multicomponent interventions to improve recovery and/or reduce LOS in older adults undergoing elective treatments requiring inpatient admission.

Aims and objectives of the review

This review aims to assess the effectiveness of multicomponent organisational interventions that aim to improve or accelerate the recovery of older adults undergoing planned (i.e. ‘elective’, non-emergency) treatments requiring hospital inpatient admission.

We will address two research questions:

1. What is the effectiveness of hospital-based multicomponent interventions in reducing length of inpatient stay in hospitals for older adults following planned admission?

2. What is the cost-effectiveness of hospital-based multicomponent interventions in reducing length of inpatient stay in hospitals for older adults following planned admission?

Search strategy

We identified effectiveness studies by searching bibliographic databases, conducting forwards and backwards citation searching of studies that met the inclusion criteria for the review, inspecting the reference lists of topically similar systematic reviews, carrying out web searches, consulting stakeholders, and contacting authors of potentially relevant conference abstracts. We also identified effectiveness studies via the cost-effectiveness searches. Cost-effectiveness studies were identified in the same way, but with adapted search terms (see below) and by inspecting studies in the effectiveness arm of the review for cost data.

The bibliographic database search strategy for effectiveness studies was developed using MEDLINE (via Ovid) by an information specialist (SB) in consultation with the review team and the stakeholders. Search terms were derived from the titles and abstracts of relevant studies identified from background searches and supplemented with relevant synonyms. The search strategy used controlled headings (e.g. MeSH in MEDLINE) wherever appropriate and free-text terms (i.e. terminology used in the titles and abstracts of studies). A multistranded approach was used to maximise the sensitivity of the search as a result of uncertainty about the specific names of procedures and interventions of interest, as follows.

1.  Group 1 terms: included terms for older people and elective procedures commonly undergone by older people. Because we could not be certain of specifying all relevant elective procedures, and nor could we rely on relevant studies describing the population group in the title or abstract, these two sets of terms were combined using the ‘OR’ Boolean operator.

2.  Group 2 terms: included generic terms for multimodal interventions for reducing LOS, such as ‘ERAS’ and ‘fast-track’, and common components of these interventions, such as ‘early ambulation’ and ‘nutritional support’. We also included terms that described reducing LOS, using the format ‘length’ adjacent to ‘stay’ adjacent to ‘reducing’. Because we could not be certain of specifying all relevant interventions, we combined the intervention terms and terms that described their intended effect (reducing LOS) using the OR Boolean operator.

The search terms in groups 1 and 2 were combined using the AND Boolean operator and limited using a study type filter. The filter was developed using adapted versions of Royle and Waugh’s38,39 simplified approaches to identifying randomised controlled trials (RCTs), and the Cochrane Effective Practice and Organisation of Care (EPOC) group’s suggested terminology for identifying non-randomised trials, controlled before-after studies and interrupted time series (Paul Miller, EPOC, 23 August 2017, personal communication). We also inspected the titles, abstracts and controlled headings of known relevant studies. No English-language or date filter was used; however, we retrospectively limited the search results to studies published from 2000 to date of searches using the sort by date feature in EndNote (EndNote X7, Thomson Reuters, New York, NY, USA). This date limit was selected because of the increasing prevalence of so-called ‘enhanced recovery pathways’ in the early 2000s, following the work of Kehlet and colleagues. 40–43 In addition, we wanted to limit the extent to which ‘usual care’ could be considered far removed from current day-treatment pathways.

The MEDLINE search strategy was translated for use in a selection of bibliographic databases, chosen because they were felt to be most likely to contain primary studies most relevant to our research questions. The full set of bibliographic databases comprised:

•  MEDLINE (via Ovid)

•  MEDLINE In-Process & Other Non-Indexed Citations (via Ovid)

•  EMBASE (via Ovid)

•  Health Management Information Consortium (HMIC) (via Ovid)

•  Cochrane Central Register of Controlled Trials (CENTRAL) (via The Cochrane Library)

•  Cumulative Index to Nursing and Allied Health Literature (CINAHL) (via EBSCOhost)

•  Allied and Complementary Medicine Database (AMED) (via EBSCOhost).

We also searched for studies using the Google Scholar^TM and Google^TM (Google, Inc., Mountain View, CA, USA) web search engines. The search strategies for MEDLINE, Google Scholar and Google Search are reproduced in Appendix 1, with the search strategies for other bibliographic databases reported in Report Supplementary Material 1.

The bibliographic database search for the cost-effectiveness review used the same search strategy as above, except the effectiveness study filter was replaced with a cost-effectiveness study filter. The cost-effectiveness filter was derived from a published search for cost-effectiveness studies developed by the team’s information specialist (SB) and refined to meet the specific requirements of our review in discussion with an experienced health economist (RA). 44 The search results were date limited from 2000 to date of search. No English-language filter was applied.

The search strategy was translated for use in an appropriate selection of bibliographic databases, including:

• MEDLINE (via Ovid)

• MEDLINE In-Process & Other Non-Indexed Citations (via Ovid)

• HMIC (via Ovid)

• NHS Economic Evaluation Database (NHS EED) (via Ovid) (although this was discontinued in March 2015, it can still be searched as an historical archive).

The results from the bibliographic database searches and Google Scholar were exported to EndNote X7 and deduplicated by manually checking and using the automatic deduplication function. The results from Google Search were copied and pasted into a Microsoft Word (version 14.0; Microsoft Corporation, Redmond, WA, USA) document for screening, as Google Search has no export function.

We carried out forwards and backwards citation searching on all prioritised studies that met the inclusion criteria (as detailed in Study selection). Web of Science (via Clarivate Analytics), Scopus (via Elsevier) and Google Scholar were used for forward citation searching. If a study was not indexed in Web of Science we searched Scopus, and if it was not indexed in Scopus we searched Google Scholar. Backwards citation chasing was conducted manually by inspecting the reference lists of prioritised studies that met the inclusion criteria.

The first authors of relevant conference abstracts from 2014 to date were contacted by e-mail to ascertain whether the study had been subsequently, or was soon to be, published as a journal article. Conference abstracts published before 2014 were not followed up on the assumption that they would have been published as a journal article already. Finally, we screened studies and the publication lists of authors suggested to us by our stakeholders.

Inclusion and exclusion criteria

The following inclusion and exclusion criteria were applied.

Study selection

The inclusion and exclusion criteria were piloted on a sample of 100 records identified by the database searches by four reviewers (MN, LS, SB and RA) independently. Following discussion, the criteria were refined and applied to the title and abstract of each identified citation independently by two reviewers (LS, MN, SB), with disagreements resolved through discussion. The full text of each potentially relevant paper was obtained and assessed independently for inclusion by two reviewers (SB, MN, LS) using the same method. When necessary, the opinion of a third reviewer was sought (MN, LS, RA, JTC). EndNote software was used to support study selection. A Preferred Reporting Items for Systematic reviews and Meta Analyses (PRISMA)-style flow chart was produced, detailing the study selection process.

We took the pragmatic step of prioritising the following categories of includable studies for full data extraction and synthesis: (1) RCTs conducted in any high-income country and (2) studies of any of includable trial design and conducted in the UK. This step was taken to allow us to manage the size of the synthesis while ensuring that it was based on the highest quality of evidence available, and to allow us to focus on the most relevant evidence within the UK setting. Only cost-effectiveness evaluations associated with prioritised studies were of interest. Minimal data extraction (study details, design and location; sample size, age and reason for admission; intervention type and key features; comparator type; setting; stages of care affected by the intervention) was carried out for the studies that were not prioritised, and these features were tabulated (see Report Supplementary Material 2, Table 1). Cost-effectiveness studies that were eligible for inclusion but were not based on prioritised effectiveness studies are summarised in Report Supplementary Material 7.

Data extraction

A standardised, piloted data extraction form in Microsoft Excel^® (Microsoft Corporation, Redmond, WA, USA) was used to collect data from each of the prioritised papers. Piloting the form for effectiveness studies involved each reviewer (MN, LS, SB) extracting an included study, which was then checked by another reviewer. Following this, all reviewers discussed ways to improve the form. This was repeated for the cost-effectiveness form by three reviewers (RA, LS and MN).

Data extraction was performed by one reviewer (effectiveness studies: MN, LS, SB; cost-effectiveness studies: RA, LS, MN) and checked by a second (effectiveness studies: LS, MN, SB; cost-effectiveness studies: RA, LS), with disagreements settled through discussion. The following data were extracted where applicable and reported.

Quality assessment strategy

The quality of all prioritised studies was independently appraised by two reviewers (LS, MN, SB) using the Effective Public Health Practice Project (EPHPP) Quality Assessment Tool for Quantitative Studies, which is suitable for randomised and non-randomised study designs. 46 Disagreements between reviewers were resolved through discussion. An additional item was considered during quality assessment, namely whether or not LOS was clearly defined. This item was not included when scoring the global quality of each study.

Cost-effectiveness studies were subject to additional appraisal using the Consensus Health Economic Criteria (CHEC) list. 47 Each study was independently appraised by two reviewers (LS and MN) and checked by a third reviewer (RA). Quality assessment informed interpretation of findings, for example by qualifying statistically significant findings in cases of poor study quality, and was not used to exclude studies. Non-prioritised studies were not subject to quality assessment.

Synthesis methods

All studies were grouped into categories based on the anatomical location of the procedure. These groupings were informed by consultation with stakeholders and agreed by two researchers/clinicians (MN and AH) and were as follows: cardiac surgery, colorectal surgery, lower limb arthroplasty, pelvic surgery, thoracic surgery, tumour removal (various locations), upper abdominal surgery, vascular surgery and various surgeries. We expected that this approach would lead to interventions within each group sharing common features, aiding comparison between studies.

Interventions and outcomes were categorised by two researchers (MN and LS). Interventions were classified according to the following broad categories, which were based on terms found in the literature.

• ERP: an intervention consisting of components at multiple stages of the care pathway (i.e. pre admission; post admission but preoperative; perioperative; postoperative but prior to discharge; post discharge).

• Prehabilitation: characterised by a focus on preoperative (usually pre-admission) components.

• Preoperative assessment with care plan (PACP): an assessment prior to hospital admission, with a subsequent care plan for the patient.

• Rehabilitation: characterised by a focus on postoperative components to improve or speed up recovery, delivered while the patient was still in hospital or when they had been discharged; usually based on physical exercise.

• Specialist ward: this involved moving the patient to a different location in the hospital, or a ring-fenced ward, with aspects such as restricted opening times, specialist staff or extra infection control measures.

• Staff mix: the main active ingredient was the provision of particular numbers or types of staff, such as a team of geriatricians, or the provision of extra nurses at key time points.

Comparators were grouped in the same way, with the additional category ‘usual care’ available. The following synthesis was only applied to prioritised studies.

Outcomes were grouped into categories for ease of reporting by two researchers (MN and LS) and confirmed following discussion with stakeholders (JM, AH, DT, CL). Broadly, outcomes were considered as either ‘clinical’ or ‘patient-reported’ in nature. ‘Clinical’ outcomes included those outcomes obtained from patient records or accessed from a database, such as LOS, re-admissions, complications, use of additional care, surgical processes, morbidity and mortality. ‘Patient-reported’ outcomes were considered to be those outcomes that a patient might actively report (although some might be assessed by a third party), such as mental health, quality of life, satisfaction and markers of physical recovery. Outcome categories are defined, with examples, in Appendix 2.

After categorisation and collation of data, effectiveness findings were tabulated and summarised by procedural group. Surgical outcomes were not described but are available on request.

Stakeholder and patient and public involvement

Stakeholder and patient and public involvement is described in Chapter 3.

Stakeholder engagement

Stakeholder involvement was incorporated throughout the review, from development of the protocol to making sense of preliminary results, identifying key messages for dissemination and supporting the preparation of the final report and other outputs. Consultation occurred through a series of individual face-to-face meetings, telephone calls and e-mail correspondence with the following clinical expert advisors:

• A consultant geriatrician (AH) with expert knowledge in the management of adults with multiple comorbidities and complex needs, frailty syndromes and polypharmacy, and with expertise in achieving successful discharge planning and supportive home-based post-discharge interventions.

• A consultant urological surgeon in the UK (JM), who was National Clinical Advisor to the UK Department of Health for the Enhanced Recovery Partnership Programme, and chaired the ERAS Guideline Development Group for the British Association of Urological Surgeons and authored the specialty guidelines. Recognised internationally for work in Enhanced Recovery following major urological surgery and has published widely within this field.

• A clinical lead occupational therapist (CL) in neurology and neurorehabilitation in the UK. His clinical interests include Parkinson’s disease, brain injury and cognitive neuroscience.

• A deputy chief nurse at Royal Devon and Exeter hospital (DT) with 29 years’ experience in nursing, almost entirely within the surgical specialties, and with clinical understanding of the ERP processes and patient journey during the hospital stay.

Stakeholder engagement is documented and the impact of this involvement on the review is described in Table 1.

Patient and public involvement

We met with a group of four adults aged > 60 years for three 2-hour meetings during this review. Each individual had experience of being admitted to hospital overnight for a planned procedure. We planned to learn from their knowledge and experiences to help us identify important outcomes and aspects of care, particularly where they may have been overlooked in the included evidence. We elicited feedback from the group at key stages of the review and aimed to co-produce the plain English summary to maximise readability. The impact of the patient and public involvement on this review is also described in full in Table 1.

Study selection

This section presents the findings from the effectiveness searches. The cost-effectiveness search results are presented in Synthesis of cost-effectiveness evidence, Study selection.

The PRISMA flow chart in Figure 1 summarises the study selection process. Bibliographic database searches identified 17,150 records and supplementary search methods identified 3501 records. Following the removal of duplicates, a total of 10,448 unique records were screened at title and abstract level. The full texts of 583 papers was sought for further consideration. Of these, 579 full texts were successfully retrieved (99.3%). Following full-text screening, 361 papers were excluded for the reasons specified in Figure 1.

FIGURE 1.

The PRISMA flow chart. NR, not reported; SR, systematic review.

In total, 208 studies, reported in 218 articles, met the inclusion criteria for inclusion in this review (not prioritised,40,54–190 prioritised31,191–269).

As described in Chapter 2, Synthesis methods, only RCTs and studies of any design from the UK were prioritised for quality assessment and further synthesis. Studies that were not RCTs and that were conducted outside the UK are described in Report Supplementary Material 2, including the stages of care targeted by each intervention, along with a brief description of the intervention and comparator. The descriptions in Sample characteristics, Intervention characteristics: prioritised studies and Quality assessment apply only to prioritised studies.

The results from the RCTs conducted in any high-income country are presented in Synthesis of evidence from randomised controlled trials. The results from the studies of any comparative trial design that were conducted in the UK are then presented in Synthesis of evidence from the UK. Evidence within these sections is further broken down by procedural type. Within each procedural subsection, key findings are summarised at the start, followed by a detailed analysis. Full tables of outcome data are located in Report Supplementary Material 5 (RCTs) and Report Supplementary Material 6 (UK evidence).

Sample characteristics

Of the 73 prioritised studies (reported in 80 articles), 39 (40 articles) were conducted in the UK,31,191,192,194,195,197,201,203,205–208,210,212,213,215,216,218,224,226,228–232,236–238,241,244,245,250,253,256,257,259,260,262,263,269 13 of which were RCTs. Thirty-four (41 articles) were RCTs conducted in one of 15 other countries, the most common of which were six studies from Germany,202,217,227,240,246–249,255 five from the Netherlands,204,219–221,265,266,268 four from Denmark196,225,233,234,267 and four from Canada. 193,198,200,214 The remaining studies consisted of 24 uncontrolled before-and-after (UBA) trials191,194,195,197,201,203,206–208,212,216,218,228,230,232,237,238,241,245,250,253,257,262,263 and three controlled trials224,229,259 from the UK.

All of the prioritised articles were published in peer-reviewed journals, apart from one, which was a PhD thesis. 233 The majority of articles (81.5%) were published from 2008 onwards, with 50 (61.7%) published since 2011. 31,191,195,197,199,201–203,205–209,211,212,214,216,219,220,222,225–230,235–243,249,250,252,253,255,257–260,262–265,267,268 Data were collected from 26,365 patients across 73 studies, with a mean number of 366 patients per study, ranging from 21 within a RCT221 to 5319 within a controlled trial229 utilising database sampling. The mean proportion of female participants in studies was 44.9% across 70 studies. Eight studies had an upper age limit for inclusion as follows: 75 years,254,256 80 years235,265,268 and 85 years. 202,243,264 By contrast, nine studies199,204,207,218,220,221,223,242,267 exclusively recruited patients aged ≥ 60 years. Studies explicitly excluded patients who lived with cognitive impairment (n = 6200,218,229,254,261,262), had ‘psychiatric illness’ (n = 6196,202,210,231,254,266), had a history of stroke (n = 2218,259), had ‘mental disability’ (n = 2234,267), had periods of dizziness/confusion (n = 2210,244) or were unable to consent (n = 431,209,226,236). By contrast, seven studies199,200,207,215,218,221,223 selected individuals who were at elevated risk of postoperative complications or who were likely to have complex needs or multimorbidities.

The reasons for admission, according to our broad procedural categories, were lower-limb arthroplasty (n = 25 studies195,196,200,202,206,216,218,221–224,230,234,237,238,241,244,245,253,254,256,261,262,267,269), colorectal surgery (n = 19 studies192,198,203,204,209,211,213,214,229,231,232,235,236,239,242,243,251,252,265), cardiac surgery (n = 8 studies193,208,210,215,255,258,259,266), upper abdominal surgery (n = 8 studies191,201,205,226,227,257,263,264), pelvic surgery (n = 3 studies217,225,250), thoracic surgery (n = 4 studies 197,212,228,248), vascular surgery (n = 2 studies31,247), solid tumour removal at various anatomical sites (n = 1 study220), abdominal surgery (n = 1 study199) and a mix of various different surgeries (n = 1 study207).

Table 2 provides a broad overview of the sample characteristics of studies prioritised for synthesis.

Intervention characteristics: prioritised studies

Interventions are summarised in Table 3; they are described in detail in subsequent sections of the synthesis and in Report Supplementary Material 3. The key characteristics are summarised here. The most frequently studied intervention category was ERP (n = 49 studies191,192,194,196,197,199,201,203,206,208,209,211–213,216–218,224–232,234–239,241–243,245,247,248,250–252,254,256,257,261–265). Thirteen studies193,198,200,204,205,210,214,215,221,222,244,258,269 evaluated prehab; two studies207,220 evaluated PACPs; four studies202,253,266,267 evaluated rehab; three studies195,255,259 evaluated specialist wards; and one study223 evaluated a staff mix intervention. The most common comparator was standard care or pre-intervention pathway (in UBA designs), although study authors used a range of descriptions (see Table 3).

Primary intervention delivery sites were at hospital, outpatient clinic, rehabilitation centre or the patient’s home, or a combination of these (see Table 3). The patient usually received the intervention alone, except in 14 studies31,193,195,200,206,211,216,221,232,234,235,245,254,258 where a family member or carer was actively involved in the intervention; in two interventions216,267 that included the involvement of friends or a non-specified support person; in two studies31,245 that involved carers; and in one study31 that involved GPs in the intervention. Ten studies197,209,241,244,247,248,251,261,266,269 did not report the intervention recipient, in which case we assumed that this was just the patient.

Quality assessment

Only two studies,193,270 both of which were RCTs, received a global study quality rating of ‘strong’ using the EPHPP tool. 46 Twenty-six studies,194,196,199,202,205,211,213,215,217,220,222,223,225,227,232,234,235,239,242,243,256,258,264,267–269 of which 24 were RCTs, received a ‘moderate’ quality rating, and the remaining 45 studies191,192,195,197,198,200,201,203,204,206–210,212,214,216,218,221,224,226,228–231,236–238,241,244,245,247,248,250–255,257,259,261–263,266 (22 RCTs) were rated as ‘weak’. Quality ratings are displayed in Table 4, with the full breakdown of scores for each item provided in Report Supplementary Material 4.

The component that contributed most towards the large number of ‘weak’ global ratings was the ‘data collection methods’ item, on which only nine studies193,194,215,232,255,264,267,269,270 received a ‘strong’ rating and one study received a ‘moderate’ rating. 257 To receive a rating of ‘strong’ for this domain, studies had to have discussed the reliability and validity of their primary outcome measure, unless this was well known. In addition, while not contributing towards the global rating of each paper, only 31191–195,197,201,202,206,208–210,215,217,223–227,232,234,235,237,239,250,252,255,257,259,264,268 of the 73 studies prioritised for synthesis (43.2%) clearly defined their LOS outcome. Only two studies229,270 achieved a ‘strong’ rating on the selection bias component, with eight studies205,210,215,221,231,236,238,248 scoring ‘weak’ and the rest scoring ‘moderate’.

Forty-two studies192,193,196,199,200,202,204,205,209–211,213–215,217,220–223,225–227,231,234–236,239,242,243,248,251,254–256,258,261,264,266–270 received a ‘strong’ rating on the study design component, with 31 scoring as ‘weak’. 191,194,195,197,198,201,203,206–208,212,216,218,224,228–230,232,237,238,241,244–247,249,250,252,253,257,259,260,262,263 All of the strongly scoring studies for this component were RCTs. All UBA studies prioritised for inclusion were rated as ‘weak’, as were the two controlled trials. 224,259 Six RCTs198,217,242,244,247,252 received a ‘weak’ rating because randomisation was not described. On the component considering the likelihood of possible confounding of results, the majority of studies were rated as ‘strong’, although one study213 received a rating of ‘moderate’ and 20 were rated as ‘weak’. 192,198,200,201,204,207,212,214,226,230,231,237,241,244,245,251,253–255,261,266,267 Regarding the blinding of outcome assessors and study participants, four studies194,197,250,262 were rated as ‘strong’ and five studies192,209,251,255,264 received a ‘weak’ rating, with the remaining studies ‘moderate’. Studies scored most highly in terms of reporting of numbers and reasons for withdrawal and dropouts. All but 11 studies191,192,194,195,208,236,245,252,257,266,269 were rated as ‘strong’ on this component, including 20 of the 23 UBA studies prioritised for synthesis. 194,195,197,201,203,206,207,212,216,218,228,230,232,237,238,241,250,253,262,263 Two studies194,195 were scored as ‘moderate’ on this item, and 10 studies191,192,208,236,242,245,252,257,266,269 obtained a ‘weak’ rating.

Synthesis of evidence from randomised controlled trials

Interventions to improve recovery from colorectal surgery: randomised controlled trials

There were 17 RCTs (published in 18 papers) evaluating interventions intended to improve recovery and/or reduce LOS for older adults following elective colorectal surgery. Of these, 14 studies192,199,209,211,213,231,235,236,239,242,243,251,252,265,268 evaluated ERP interventions and three198,204,214 trialled prehab. All ERP interventions were compared with usual or standard care, while prehab interventions were compared with walking and breathing exercises,198 rehabilitation214 and home-based exercise advice. 204 The study by Vlug and colleagues268 included comparisons of ERP and standard care in both open and laparoscopic surgery as separate intervention arms.

Four studies were conducted in the UK,192,213,231,236 three were conducted in Italy,242,243,252 two were conducted in each of the Netherlands204,268 and Canada,198,214 and single studies originated from Spain,211 France,239 Switzerland,251 Norway,209 Taiwan199 and South Korea. 235 A total of 3079 people commenced trials, with sample sizes ranging from 25192 to 427. 268 Participants were admitted for hemicolectomy (left and right), resection of malignant lesions, colonic reconstruction, colorectal resection, rectal resection, high anterior resection, colonic resection and extraperitoneal rectal surgery. Across the included studies, the mean proportion of female participants was 44%. Ages within included studies were reported as either median or mean and can be viewed in Table 2.

The two studies by Mari and colleagues242,243 utilised the same intervention with different samples, the latter study focusing on patients of ≥ 70 years of age. van Bree and colleagues265 reported on a small sample of the original 427 participants in the study by Vlug and colleagues268 focusing on recovery of gastrointestinal transit. In this case, we report outcome data from the original sample in the study by Vlug and colleagues268 only, except where van Bree and colleagues265 reported additional, unique outcomes. The study by Chen and colleagues199 included patients undergoing procedures that could be categorised as both colorectal and upper abdominal surgery; we report the data for patients undergoing left and right hemicolectomies as distinct groups in this section, with those undergoing gastrectomy or pancreaticoduodenectomy reported in the upper abdominal section (see Interventions to improve recovery from upper abdominal surgery: randomised controlled trials). 199

Overview: colorectal surgery

Thirteen RCTs evaluated the effectiveness of ERP interventions to improve recovery from colorectal surgery, and three evaluated prehab interventions. This body of evidence is summarised in Table 5. Full details of the effectiveness of interventions in this type of surgery are reported in ERP interventions and Prehab interventions.

TABLE 5 - Summary of studies, intervention components and findings for RCTs evaluating interventions to improve recovery from colorectal surgery

AEI, assessment, education, counselling or information; ANA, analgesia protocol; ANE, anaesthesia protocol; AP, antibiotic prophylaxis; CATH, catheter protocol; CHL, carbohydrate loading; COMM, communication protocol; DP, discharge planning; DRA, drain protocol; EMOB, early mobilisation; EON, early oral nutrition; EX, exercise programme; FM, fluid management protocol; LAX, laxative; nFAST, avoidance of prolonged fasting; nMBP, no mechanical bowel preparation; nNGT, nasogastric tube protocol; NUT, nutrition supplementation or diet management; OT, occupational therapy; Pbio, probiotic/prebiotic; PONV, active prevention of nausea and vomiting; PreM, pre-medication protocol; PT, physiotherapy; SoW, social worker; SURG, surgical approach differs between groups; TP, thromboprophylaxis; WARM, intraoperative warming protocol.

▲, standardised mean difference indicates significant large beneficial effect size (Cohen’s d > 0.80); ▲, standardised mean difference indicates significant medium beneficial effect size (Cohen’s d 0.50 to 0.79); △, standardised mean difference indicates significant small beneficial effect size (Cohen’s d 0.20 to 0.49); ▼, standardised mean difference indicates significant medium detrimental effect size; ◁▷, standardised mean difference indicates no significant difference between groups; ⬆, OR indicates significant beneficial effect; ⇦⇨, OR indicates no significant difference between groups; x, data but standardised mean difference or OR could not be calculated; [blank], not reported; (x/y), finding for x of y measures (e.g. where multiple measures per outcome category).

Patient satisfaction was reported in the study.

Study quality is indicated by the colour of the study details cell: dark green, ‘moderate’; light blue, ‘weak’.

Study details			Intervention components						Outcome categories
First author, year, country (quality assessment)	Sample size	Intervention category	Pre hospital admission	Preoperative	Perioperative	Postoperative	Post discharge	Other	LOS	Complications	Mortality	Re-admissions	Markers of recovery	Mental health	Quality of life	Use of additional care
Chen 2017,199 Taiwan	535	ERP			FM	EON; COMM; EMOB		Oral and nutritional assistance protocol	◁▷(2/2)	⇦⇨(2/2)
García-Botello 2011,211 Spain	125	ERP	AEI	PreM; nMBP; nFAST	ANE; FM; WARM PONV	EON; EMOB; nNGT; CATH			▲	◁▷(4/4)		⇦⇨	x(6/6)
Gatt 2005,213 UK	39	ERP	AEI; PBio	nMBP; CHL; nFAST	ANA; DRA; FM; nNGT; SURG	EON; EMOB			x	⇦⇨	⇦⇨	⇦⇨	x			⇦⇨
Lee 2011,235 South Korea	100	ERP			FM	EON; EMOB; LAX; CATH		Pre admission: involvement of patient and family	x(2/2)	⇦⇨			▲(6/10) ▲(1/10) ◁▷(3/10)
Maggiori 2017,239 France	270	ERP	AEI	CHL; nFAST; PreM	PONV; ANE; FM; WARM	EON; EMOB; ANA			◁▷(2/2)	⇦⇨(2/2)	⇦⇨
Mari 2014,243 Italy	52	ERP		CHL; nFAST	ANE; FM; PONV	ANA; EON; EMOB; nNGT			▲	⇦⇨		⇦⇨	▲(5/5)
Mari 2016,242 Italy	83	ERP		CHL; nFAST	ANE; FM; PONV	EON; EMOB; nNGT; ANA			▲	⇦⇨			▲(4/12) ▲(4/12) ◁▷(4/12)
van Bree 2011,265 Vlug 2011,268 Netherlands, laparoscopic surgery	93; 427	ERP	AIE; EX	CHL; nFAST; PreM	ANA; FM; PONV	EON; EMOB; FM; NUT; LAX; CATH		Guided tour of surgical ward	△(2/2)	⇦⇨(2/2)	⇦⇨	⇦⇨	▲(2/9) ▲(4/9) △(1/9) ◁▷ (2/9)
van Bree 2011,265 Vlug 2011,268 Netherlands, open surgery	93; 427	ERP	AIE; EX	CHL; nFAST; PreM	ANA; FM; PONV	EON; EMOB; FM; NUT; LAX; CATH		Guided tour of surgical ward	◁▷(2/2)	⇦⇨(2/2)	⇦⇨	⇦⇨	▲(2/9) ▲(3/9) △(1/9) ◁▷(3/9)
Anderson 2003,192 UK	25	ERP	AEI; NUT	nMBP; CATH; CHL; nFAST	ANE; ANA; nNGT; SURG	EON; EMOB			▲	⇦⇨		x	▲(1/2) ◁▷(1/2)
Forsmo 2016,209 Norway	324	ERP	AEI	CHL; PreM	ANE; FM	LAX; ANA; CATH; EON; EMOB			x(2/2)	⇦⇨(3/3)	⇦⇨	⇦⇨	x(5/5)
Khoo 2007,231 UKa	81	ERP		NUT	ANE; FM; nNGT	EON; EMOB; ANA; CATH			x(2/2)	⬆	⇦⇨	⇦⇨	x(3/3)			⇦⇨(3/3)
Lidder 2013,236 UK	57	ERP		CHL		EON; NUT			◁▷	⬆(1/4) ⇦⇨(3/4)			◁▷
Muller 2009,251 Switzerland	156	ERP			FM; DRA	ANA; EON; FM; NUT			x	⬆(2/3) ⇦⇨(1/3)		⇦⇨
Pappalardo 2016,252 Italy	50	ERP			ANE; AP; TP	EON; EMOB; nNGT; ANA			x	⇦⇨(3/3)	⇦⇨		x(4/4)
Carli 2010,198 Canada	133	Prehab	EX						◁▷	⇦⇨			◁▷(2/2)	◁▷(2/2)	◁▷
Dronkers 2010,204 the Netherlands	42	Prehab	AEI; EX						◁▷	⇦⇨			▼(1/4) ◁▷(3/4)		▼(2/7) ◁▷(5/7)
Gillis 2014,214 Canada	89	Prehab	AEI; EX; NUT; PT					Weekly telephone contact	◁▷(2/2)	⇦⇨		⇦⇨(2/2)	x	◁▷(3/3)	◁▷(8/8)

Meta-analysis suggests that ERP interventions are associated with a statistically significant, medium-sized effect on LOS, reducing LOS by a mean of 1.8 days (range 0.0–5.1 days) compared with usual care (based on data from eight comparisons); however, this effect was weaker when a much older outlier study with a small sample size was removed from the analysis. 192 This change in LOS did not lead to negative consequences for other clinical outcomes, as meta-analyses indicated no statistically significant difference overall in the rates of re-admissions (data from seven comparisons) or complications (data from 15 comparisons). Furthermore, there was evidence from meta-analyses of statistically significant improvements on patient-reported outcomes following ERP interventions, including return of bowel function, meeting mobilisation goals earlier and achieving control of pain sooner, although these data always came from three or four studies and there was usually high statistical heterogeneity.

Three ERP intervention studies199,236,239 reported no significant differences in clinical outcomes compared with usual care. The interventions and comparators within two of these studies were atypical. The intervention evaluated by Chen et al. 199 mainly involved a series of tailored postoperative care protocols, added to existing ERP pathways which implemented few ERAS recommendations. For example, mechanical bowel preparation, overnight fasting, routine use of drains and nasogastric tubes were all implemented, in contradiction to ERAS Society guidelines. 271 The intervention evaluated by Lidder et al. 236 involved adding preoperative carbohydrate loading and additional postoperative nutrition to the existing ERP provided to the comparator group. This suggests that these elements alone are not enough to reduce LOS compared with existing standards of care, although caution is required because the studies were low to moderate quality, respectively.

The ERP intervention evaluated by Vlug and colleagues268 was effective in reducing LOS when a laparoscopic surgical technique was used, but not when an open approach was used, indicating the potential influence of the surgical technique on the effectiveness of an ERP. However, the statistical results for the two studies were marginally different, and the effects for patient-reported outcomes were similar for patients undergoing both types of surgery. It is unclear why the intervention was not more effective than usual care in the study by Maggiori and colleagues,239 except that there were a number of similar pathway elements between the ‘full fast-track’ pathway and the ‘limited fast-track’ pathway that constituted usual care (Table 6).

TABLE 6 - Components of included ERP interventions and comparators, mapped against ERAS Society guidelines for colonic and pelvic/rectal surgery

C, present in comparator arm; E, present in experimental arm; PONV, postoperative nausea and vomiting; R/P, items from guidelines for rectal/pelvic elective surgery; TEA, thoracic epidural analgesia.

Interventions were described as ‘ERAS’ by study authors.

Both studies used the same intervention.

3 hours’ fasting for fluids.

Both laparoscopic and open procedures performed.

Laparoscopy used on average 47% of patients.

Rectal resections received pelvic drain.

Responses relate to mention of bladder catheter or urinary drainage, regardless of site.

ERAS item	Study (first author and year)
	aForsmo 2016209	aLidder 2013236	Anderson 2003192	Chen 2017199	García-Botello 2011211	Gatt 2005213	Khoo 2007231	Lee 2011235	Maggiori 2017239	bMari 2014,243 2016242	Muller 2009251	Pappalardo 2016252	van Bree 2011;265 Vlug 2011268
Preoperative information, education and counselling	E		E		E	E		E, C	E	E, C			E
Preoperative optimisation			E			E				E, C
Preoperative mechanical bowel preparation avoided			E		E	E					E, C
Preoperative fasting: clear fluids allowed up to 2 hours and solids up to 6 hours prior to induction of anaesthesia	E, C		E			Ec	E,c Cc		E	E	E,c Cc		E
Preoperative carbohydrate treatment	E	E	E			E			Ec	E			E
Pre-anaesthetic medication: no routine long- or short-acting sedatives (colonic); no advantages in using long-acting benzodiazepines, short-acting should not be given to those aged > 60 years (R/P)	E				E								E
Prophylaxis against thromboembolism	E, C				E, C						E, C	E
Antimicrobial prophylaxis and skin preparation	E, C		E, C		E, C					E, C	E, C	E
Standard anaesthetic protocol	E	E, C	E		E, C	E, C	E, C		E		E, C	E	E, C
PONV prophylaxis					E				E				E
Laparoscopy and modifications of surgical access: recommended for colonic surgery	E,d Cd	E,d Cd	E, C	E,e Ce	E,d Cd			E, C	E, C	E, C	E, C		E, C
Nasogastric intubation: nasogastric tubes not routinely used	E, C	E, C	E			E		E, C			E, C
Nasogastric tubes removed before extubation	E, C		E		E	E	E		E, C	E			E, C
Preventing intraoperative hypothermia	E, C	E, C			E, C				E, C	E, C			E, C
Perioperative fluid management: intraoperative fluids guided by flow measurements to optimise cardiac output		E, C			E, C		C		E		E		E
Perioperative fluid management: judicious use of vasopressors											E		E
Perioperative fluid management: enteral route for fluids postoperatively should be used as early as possible, and intravenous fluids should be discontinued as soon as is practicable	E	E, C		E		E	E	E			E		E
No drainage of peritoneal cavity after colonic anastomosis	E,f C		E			E					E		E, C
Transurethral catheter (R/P)
Suprapubic catheter (R/P)													E
Urinary drainage: routine transurethral bladder drainage for 1–2 days is recommendedg					E, C		E, C	E, C		E, C			E
Early removal of bladder catheter	E				E		E	E					E
Prevention of postoperative ileus: mid-thoracic epidural + laparoscopic surgery (colonic)	E,d Cd	E,d Cd	E								E, C		E
Prevention of postoperative ileus: fluid overload and nasogastric decompression avoided	E						E		E		E		E
Prevention of postoperative ileus: chewing gum (R/P), oral magnesium, alvimopan													E
Postoperative laxatives and prokinetics (R/P)	E			E, C				E					E
Postoperative analgesia. Open surgery: TEA using low-dose local anaesthetic and opioids; laparoscopic: an alternative to TEA is a carefully administered spinal analgesia with a low-dose, long-acting opioid		E, C	E		E, C	E	E, C	E, C	E	E	E		E, C
Early oral intake	E	E	E	E	E	E	E	E	E	E	E		E
Perioperative nutritional care	E	E	E	E	E	E	E		E	E	E		E
Postoperative glucose control
Early mobilisation	E		E	E	E	E	E	E	E	E	E, C		E

The evidence for prehab interventions did not indicate a benefit for clinical or patient-reported outcomes, and it is notable that only three RCTs198,204,214 of ‘weak’ quality were identified. There were no significant differences in clinical outcomes for any study, and, although all of these studies measured additional outcomes, only Dronkers and colleagues204 identified a statistically significant difference between groups. Results from the study indicate an increase in the time taken to rise from a chair, a reduction in physical work capacity and an increase in fatigue among patients in the intervention group. 204 This outcome was unexpected, as patients in the prehab group appeared to have performed worse on three outcomes. However, baseline scores for these three measures were similar to post-intervention scores, with near-statistically significant differences between groups. 204 A fairer interpretation of the intervention’s effects may be that it had little influence on these outcomes, rather than a negative impact. As such, caution is needed when interpreting this finding, particularly in the light of the fact that confounders were not controlled for in the study, which received a global quality rating of ‘weak’ overall. With this in mind, the evidence suggests that prehab interventions alone may not convey beneficial or detrimental effects to older adults undergoing elective colorectal surgery.

Elements of prehab (exercise, physiotherapy, nutrition all prior to admission) were present in only three of the ERP interventions,192,213,268 with mixed effectiveness reported. Therefore, little evidence is available to determine whether or not the addition of prehabilitation to ERP interventions might convey benefits on patient recovery.

Caution is advised when interpreting the evidence for colorectal surgery. Study effects should be interpreted alongside detail of the intervention evaluated, as there was considerable heterogeneity in both the content of interventions, and the evidence of effects, as highlighted in Table 5. Furthermore, comparators were often poorly described. Study quality was ‘moderate’ or ‘weak’, with no studies globally rated as ‘strong’.

Enhanced recovery protocol interventions

Components of the ERP interventions and comparators are mapped against ERAS Society recommendations for colonic and rectal/pelvic surgery in Table 6. 272 The full details of both ERP and prehab interventions can be found in Report Supplementary Material 3, Tables 1–4. In summary, the most common ERP components were the provision of preoperative information, education and counselling (8/13 studies); avoidance of lengthy preoperative fasting (8/13); a standardised anaesthetic protocol (10/13); laparoscopic surgery used, where available (10/13); early removal of nasogastric tubes (8/13); thoracic epidural analgesia postoperatively (10/13); early oral intake postoperatively (12/13); perioperative nutritional care (11/13); and early mobilisation (11/13). The least common components were preoperative optimisation (3/13 studies); avoidance of pre-anaesthetic medication (3/13); active prophylaxis against postoperative nausea and vomiting (3/13); judicious use of vasopressors (2/13); transurethral (0/13) or suprapubic catheter (1/13) use; prevention of postoperative ileus with chewing gum, oral magnesium or alvimopan (1/13); and postoperative glucose control (0/13).

Excluding the item for glucose control (specific to patients with diabetes), interventions adhered to a mean number of 14.2 (45.7%) ERAS guideline items (range from 3252 to 24268 items), and comparator groups adhered to a mean number of 4.8 (15.6%) items (range from 0252 to 7211,236,251 items). Interventions branded as ERAS adhered to a mean of 15.5 (50%) of ERAS guideline items, compared with a mean of 13.9 (44.8%) of ERP interventions not branded as ERAS.

In the study by Chen and colleagues,199 the intervention (the Modified Hospital Elder Life Program) integrated relatively few ERAS Society-recommended components, but featured an additional focus on oral and nutritional assistance, with a postoperative communication protocol to facilitate orientation within the hospital environment.

Quality assessment

Seven studies211,213,235,239,242,243,268 were rated as ‘moderate’ overall quality; the rest were rated as ‘weak’. All studies received a rating of ‘weak’ for data collection methods because they failed to describe the reliability and validity of outcome measures. Only six studies192,209,235,239,252,268 clearly defined how LOS had been calculated. The study by Anderson and colleagues192 was rated as ‘weak’ on four domains (confounders, blinding, data collection methods, and withdrawals and dropouts), and five studies198,231,236,251,252 were rated as ‘weak’ on three domains. Seven studies209,211,235,239,242,243,268 scored ratings of ‘strong’ on three domains; in each case, these domains were study design, confounders, and withdrawals and dropouts.

Effectiveness of enhanced recovery protocol interventions at improving clinical outcomes

Report Supplementary Material 5, Table 1, displays clinical outcome data from the 14 studies evaluating ERP interventions in older adults undergoing elective colorectal surgery. After imputation, standardised mean differences between groups for LOS were available in 10 group comparisons from eight studies. 192,199,211,236,239,242,243,268 A forest plot displaying the results of meta-analysis of these eight studies is displayed in Figure 2a, showing that ERP interventions are associated with a reduction in LOS when compared with usual care, associated with a medium effect size (d = –0.51, 95% CI –0.78 to –0.24; p < 0.001). Heterogeneity was large and statistically significant for this effect (I² = 78.9%; p < 0.001), reflecting inconsistency in the evidence. The mean reduction in LOS was 1.8 days (SD 1.5, range 0.0–5.1 days). Examination of Figure 2a suggests that the effect size from the study by Anderson and colleagues192 may be an outlier. Because the study is 8 years older than any other in the analysis, and has the smallest sample size (data from 24 participants), a sensitivity analysis was performed by removing the study from the meta-analysis. The results are shown in the forest plot in Figure 2b, and indicate that there is still a statistically significant reduction in LOS with ERP interventions; however, the effect size reduces marginally (d = –0.45, 95% CI –0.71 to –0.19; p < 0.01) and the reduction in absolute terms decreases slightly to 1.6 days. Sensitivity analysis showing the influence of imputed data on LOS meta-analysis is available in Appendix 4.

FIGURE 2.

(a) Forest plot showing the results of meta-analysis of the effect of ERP vs. usual treatment on LOS following colorectal surgery; (b) forest plot of analysis with Anderson et al. 192 removed. Lap, laparoscopic surgery group; LH, left hemicolectomy/high anterior resection; open, open surgery group; RH, right hemicolectomy.

In the studies presenting data that could not be used to calculate standardised mean differences, the median LOS for patients receiving ERP interventions was lower in all cases. 209,213,231,235,251 Pappalardo and colleagues252 reported that patients receiving ERP had been discharged by postoperative day 6, but around half of those receiving usual care had not.

Nine studies, evaluating 10 comparisons, reported readmission rates following ERP interventions. Figure 3 is a forest plot displaying the results of meta-analysis of re-admission data, indicating the odds of re-admission were similar between ERP and usual care (OR 1.19, 95% CI 0.77 to 1.82). Thirteen studies, evaluating 15 comparisons, reported incidence of complications following ERP interventions. Meta-analysis indicates that the odds of experiencing a complication were not statistically significantly different in patients receiving ERP interventions from those receiving usual care (Figure 4), despite the summary OR indicating a trend towards a reduction in odds with ERP (OR 0.82, 95% CI 0.65 to 1.03).

FIGURE 3.

Forest plot showing the results of meta-analysis of the effect of ERP vs. usual treatment on the odds of re-admission following colorectal surgery. Lap, laparoscopic surgery group; open, open surgery group.

FIGURE 4.

Forest plot showing the results of meta-analysis of the effect of ERP vs. usual treatment on odds of complications following colorectal surgery. AR, anterior resection; lap, laparoscopic surgery group; LAR, lower anterior resection; LH, left hemicolectomy/high anterior resection; open, open surgery group; RH, right hemicolectomy.

Effectiveness of enhanced recovery protocol interventions at improving patient-reported outcomes

Report Supplementary Material 5, Table 2, displays data for patient-reported outcomes for RCTs trialling ERP interventions to improve recovery from elective colorectal surgery in older adults. Multiple outcomes were reported for markers of recovery across the included studies, allowing several meta-analyses to be performed. In total, five ERP studies192,235,242,243,268 contributed to meta-analyses of the following markers of recovery from surgery: time to first flatus, time to first stool, time to resumption of diet, time to reach mobilisation goals and time to reach pain control goals. There was insufficient data for meta-analysis of mental health or quality-of-life outcomes.

Figure 5a is a forest plot of the effect of ERP interventions on the time to resumption of diet after colorectal surgery, when compared with usual care. Across five group comparisons from four studies,235,242,243,268 there was a statistically significant reduction in time to resumption of diet, with a large effect size (d = –2.07, 95% CI –3.11 to –1.03; p < 0.001). In absolute terms, the mean reduction in time to resumption of diet after surgery was 1.7 days (SD 0.7, range 0.7–2.0 days). In Figure 5b, meta-analysis indicates that there was a medium-sized effect indicating earlier passage of first flatus in participants undergoing ERP (d = –0.64, 95% CI –1.05 to –0.24; p < 0.001), with a mean absolute difference of –0.7 days (SD 0.4, range –0.15 to –1.2 days). The evidence for passage of first stool was similar, occurring 1.4 days earlier with ERP (SD 1.2 days, range 0.3–1.4 days) with Figure 5c indicating a large pooled effect size (d = –1.12, 95% CI –1.81 to –0.43; p < 0.001).

FIGURE 5.

Forest plots displaying meta-analyses of patient-reported outcomes following colorectal surgery. (a) Time to resumption of diet; (b) time to passage of first flatus; (c) time to passage of first stool; (d) time to reach ambulation goals; and (e) time to reach pain reduction of control goals. Lap, laparoscopic surgery group; open, open surgery group.

Figure 5d is a forest plot of the effect of ERP interventions on the time to reach ambulation goals following surgery. Goals included walking a target distance, walking safely or independently and reaching preoperative levels of mobilisation. Patients receiving ERP met goals 2.0 days earlier (SD 1.8 days, range 1.3–3.4 days), with a large effect size (d = –1.02, 95% CI –1.62 to –0.42; p < 0.001). Figure 5e shows that there was a small effect size indicating that patients randomised to ERP achieved pain control/goals earlier than those receiving usual care (d = –0.47, 95% CI –0.65 to –0.30; p < 0.001). Pain goals were achieved a mean of 0.8 days earlier (SD 0.2 days, range 0.7–1.0 days).

In studies not reporting analysable data, Forsmo and colleagues209 reported that the median time to passage of first stool and pain control with oral medication occurred earlier in patients receiving ERP, but it took longer to tolerate food without nausea; García-Botello and colleagues211 reported favourable median scores for all outcomes, except for pain scores on postoperative days 2 and 3, which were similar for both groups; Khoo and colleagues231 reported that only 9% of patients randomised to the ERP group felt that they would benefit from a longer inpatient stay, compared with 69% in the usual care group; and Pappalardo and colleagues252 reported similar quality-of-life scores between groups.

Heterogeneity

There was often high and statistically significant statistical heterogeneity in meta-analyses of the effects of ERP on LOS and markers of recovery in patients undergoing elective colorectal surgery, indicating inconsistency between studies. Although heterogeneity was not statistically significant for the meta-analysis of achievement of pain control outcomes (see Figure 5e), it was for other analyses, with I² values ranging from 77.6% (LOS) to 96.5% (time to resumption of diet), indicating lack of consistency between outcome data. Furthermore, 95% CIs were wide in all cases, suggesting that any underlying true effect of interventions was uncertain. Meta-analyses of complications and re-admissions were associated with low and non-statistically significant heterogeneity.

Prehab interventions

Two204,214 of the three prehab interventions began with dedicated pre-intervention information and education, with Gillis and colleagues214 also including assessments and a booklet to facilitate adherence. All three prehab interventions incorporated a programme of physiotherapy that included multiple weekly sessions and a range of aerobic and strength training exercises. Gillis and colleagues214 supplemented these activities with a review of patient nutrition and subsequent optimisation, visits with a psychologist that included motivation and training to reduce anxiety, and regular telephone support. Participants in the study by Carli and colleagues198 also received weekly telephone support and a home visit. Perioperative care was described only by Gillis and colleagues,214 with all participants receiving the same ERAS protocol. Comparators in the three studies were a walking and breathing exercise group,198 home-based exercise advice204 and postoperative rehabilitation,214 precluding meta-analysis.

Effectiveness of prehab interventions at improving clinical outcomes

Report Supplementary Material 5, Table 3, displays clinical outcome data reported in all studies. There were no statistically significant differences in LOS or complications across the three studies. Only Gillis and colleagues214 reported re-admissions, which were similar between groups. Meta-analysis was not performed because the three studies had different comparators.

Effectiveness of prehab interventions at improving patient-reported outcomes

Report Supplementary Material 5, Table 4, displays data for patient-reported outcomes from the studies trialling prehab interventions to improve recovery following colorectal surgery. A number of outcomes were related to markers of physical recovery, quality of life and mental health. Participants in the prehab group in the study by Dronkers and colleagues204 took 5.4 seconds longer to perform the chair rise test than comparators, associated with a large effect size but wide CIs, leading to uncertainty about the true effect (d = 0.88, 95% CI 0.22 to 1.54; p < 0.01). Participants in the prehab group also had lower physical work capacity (d = –0.79, 95% CI –1.44 to –0.14; p < 0.05) and reported greater fatigue (d = 0.70, 95% CI 0.06 to 1.35; p < 0.05) than their comparators; these differences also had large but uncertain effect sizes. There were no differences between groups on any other outcome across the included studies.

Interventions to improve recovery from lower limb arthroplasty: randomised controlled trials

Thirteen RCTs trialled interventions intended to improve recovery and/or reduce LOS following lower limb arthroplasty. Of these, five studies196,234,254,256,261 compared ERP interventions with usual care; five200,221,222,244,269 trialled prehab versus either usual care200,221,222,244 or home exercise;269 two202,267 compared rehab programmes with usual care and one223 trialled a staff mix intervention versus usual care.

Three studies were conducted in the UK,244,256,269 three were conducted in Denmark,196,234,267 two were conducted in the USA,223,254 and one was conducted in each of Canada,200 Germany,202 the Netherlands,221 Taiwan222 and Iceland. 261

A total of 1686 people commenced trials, with sample sizes ranging from 21221 to 505. 223 Procedures included primary or revision unilateral or total hip or knee arthroplasty. Across the included studies, 58.5% of participants were female.

Overview: lower limb arthroplasty

This body of evidence is summarised below and in Table 7. Full details of the effectiveness of interventions in this type of surgery are reported in Enhanced recovery protocol interventions, Prehab interventions, Rehab interventions and Staff mix interventions.

TABLE 7 - Summary of studies, intervention components and findings for RCTs evaluating interventions to improve recovery from lower limb arthroplasty

AEI, assessment, education, counselling or information; ANA, analgesia protocol; ANE, anaesthesia protocol; CATH, catheter protocol; CHL, carbohydrate loading; DP, discharge planning; DRA, drain protocol; EMOB, early mobilisation; EON, early oral nutrition; EX, exercise programme; GOAL, goal-setting; LAX, laxative; MT, motivational talks; NUT, nutrition supplementation or diet management; OT, occupational therapy; PONV, active prevention of nausea and vomiting; PT, physiotherapy; SM, staff mix; SoW, social worker; TP, thromboprophylaxis.

▲, standardised mean difference indicates significant large beneficial effect size (Cohen’s d > 0.80); ▲, standardised mean difference indicates significant medium beneficial effect size (Cohen’s d 0.50 to 0.79); △, standardised mean difference indicates significant small beneficial effect size (Cohen’s d 0.20 to 0.49); ▼, standardised mean difference indicates significant medium detrimental effect size; ◁▷, standardised mean difference indicates no significant difference between groups; ⬆, OR indicates significant beneficial effect; ⇦⇨, OR indicates no significant difference between groups; x, data but standardised mean difference or OR could not be calculated; [blank], not reported; (x/y), finding for x of y measures (e.g. where multiple measures per outcome category).

Patient satisfaction was reported in the study.

526 were randomised; 21 from pilot were excluded before starting intervention.

Intervention components are listed when they only appear in the experimental arm of the study. Study quality is indicated by the colour of the border of the study details cell: dark green, ‘moderate’; light blue, ‘weak’.

Study details		Intervention components							Outcome categories
First author, year, country (quality assessment)	Sample size	Intervention category	Pre hospital admission	Preoperative	Perioperative	Postoperative	Post discharge	Other	LOS	Complications	Mortality	Re-admissions	Markers of recovery	Mental health	Quality of life	Use of additional care
Borgwardt 2009,196 Denmarka	50	ERP	AEI; OT		ANE	ANA	TEL		x			⇦⇨	x(5/5)			⇦⇨(2/2)
Larsen 2008, 234 2008, 2008,233 Denmark	90	ERP	AEI; EX; GOAL; NUT; OT; SW	AEI	PONV	EMOB; GOAL; LAX; OT; ANA; EON	PT		▲			⇦⇨			▲
Reilly 2005,256 UK	41	ERP				ANA; EMOB; EON	AEI; TEL		x	⇦⇨(2/2)			▲(1/5) ◁▷(4/5)
Pour 2007,254 USA	100	ERP	AEI; PT; OT	ANA	ANE	ANA; DRA; EMOB; PT			x	⇦⇨			x(9/9)	x	x(5/5)
Siggeirsdottir 2005,261 Iceland	50	ERP	AEI; EX		TP		EX; PT; OT		▲	⬆		◁▷	▲(1/2) x(1/2)
Huang 2012,222 Taiwan	243	Prehab	AEI; EX						▲	⇦⇨(4/4)			◁▷(2/3) x(1/3)
Williamson 2007,269 UK	181	Prehab	EX; PT						◁▷				◁▷(3/3)	▼(1/2) ◁▷(1/2)	◁▷
Crowe 2003,200 Canada	133	Prehab	AEI; OT; PT; SoW					Pre-hospital day therapy	△	⬆			▲(2/7) △(2/7) ◁▷(3/7)
Hoogeboom 2010,221 the Netherlands	21	Prehab	EX					Spouse/family involved in exercises	x	⇦⇨			◁▷(6/6)		◁▷
McGregor 2004,244 UK	39	Prehab	AEI						x				◁▷(2/2)		◁▷(4/4)
den Hertog 2012,202 Germany	160	Rehab				EMOB; MT; PT; SW			x	⇦⇨					▲
Vesterby 2017,267 Denmark	73	Rehab					PT	Tele-medicine support	x			⇦⇨				⇦⇨(2/2)
Huddleston 2004,223 USA	505b	Staff mix		SM	SM	SM			x	⬆(2/4) ⇦⇨(2/4)

Evidence was generally of low quality, and not all studies reported data that could contribute to analyses of effectiveness. Meta-analysis of two studies suggests that ERP interventions were associated with a reduction in LOS of over 3 days compared with usual care, with neither study reporting evidence of detriment to patient recovery or well-being. Similarly, meta-analysis of two studies evaluating prehab suggests that LOS may be reduced by around 2.5 days compared with usual care. There was, however, insufficient evidence to draw further comparisons between interventions in either category.

Complications were reported in eight studies, with evidence of improvement with interventions in three. 200,223,261

There was not enough evidence to allow the evaluation of the effectiveness of rehab or staff mix interventions. In general, lack of useable data for clinical outcomes in eight studies precludes a balanced and critical overview of the evidence.

The prehab intervention evaluated by Crowe and colleagues200 represents the most multidisciplinary intervention, with patients receiving assessment by an occupational therapist, a physiotherapist, a dietitian and a social worker, and with nurse input, within a day-therapy programme. This programme contains both education and physiotherapy components, which is similar to the intervention evaluated by Huang and colleagues,222 although the latter programme was delivered primarily by a physiotherapist. The interventions evaluated by Hoogeboom and colleagues221 and Williamson and colleagues269 constituted a home exercise and a physiotherapy programme, respectively. In the intervention evaluated by McGregor and colleagues,244 although patients were given the opportunity to practise the exercises they were to complete postoperatively, there was no separate physiotherapy or exercise training programme. The effectiveness of prehab in this study may in part be because the comparator group in this study received only minimal information prior to admission, whereas patients in the comparator groups of other studies received more comprehensive education packages. 221,269 A tentative interpretation based on the limited data available may be that an intervention requires both an educational and a physiotherapy component to be effective in reducing LOS. However, the quality of the studies this interpretation is based on was rated as ‘moderate’222,269 and ‘weak’. 200

The four studies199,200,234,261 that provided statistically significant evidence of reduced LOS had different interventions. Intervention components common to these four interventions were preoperative assessment, education or information of differing degrees of intensities; and a pre-admission physical component, either an exercise programme or physiotherapy. However, these components were also present in other interventions that were ineffective at improving outcomes, suggesting that further evidence is required before it is possible to examine components that determine improvements in patient recovery following lower limb arthroplasty.

Enhanced recovery protocol interventions

Components of the ERP interventions and comparators are mapped in Table 8. We used an ERAS-style intervention component mapping approach, despite absence of ERAS Society guidelines for lower limb arthroplasties. The following items were present in at least three of the five studies: pre-admission assessment, education and counselling; preoperative exercises; preoperative assessment by an occupational therapist or a social worker to identify postoperative needs and/or provision of equipment; minimally invasive surgical technique; avoidance of patient-controlled analgesia; avoidance of routine opiate analgesia; early mobilisation protocol; and follow-up support.

TABLE 8 - Map of components found in interventions and comparators for ERP interventions aiming to improve recovery from lower limb arthroplasty

C, comparator group; E, experimental group; OT, occupational therapist; PCA, patient-controlled analgesia; PO, postoperative; PONV, postoperative nausea and vomiting; PT, physiotherapist; SW, social worker.

Experimental group divided into those who received small incisions of 7–10 cm and those who received standard incisions of 11–19 cm.

Intervention component	Study (first author and year)
	Borgwardt 2009196	Larsen 2008234	Pour 2007254	Reilly 2005256	Siggeirsdottir 2005261
Assessment/education/counselling	E	E	E, C		E
Avoidance of preoperative fasting
Preoperative carbohydrate treatment
Preoperative exercise		E	E		E
Optimisation of physical condition prior to surgery		E	E
Discharge planning: prior to admission
Discharge planning after hospital admission
Preoperative OT/SW assessment of PO needs and/or equipment provision	E	E	E
Admission on day of surgery		E
Pre- or postoperative antibiotic prophylaxis			E, C
Anaesthesia protocol: routine use of spinal or epidural anaesthesia	E, C		E, C
Anaesthesia protocol: short-acting, opiate avoidance	E			E, C
Minimally invasive surgical technique	E, C		Ea	E, C
Wound management protocol	E, C
Avoidance of routine urinary catheter	E, C
No routine drains			E
Avoidance of PCA		E, C	E	E
Avoidance of routine opiate analgesia	E	E, C		E
Early mobilisation protocol	E, C	E	E	E
Dedicated post-surgery physiotherapy exercises, supported by PT		E, C	E
Early resumption of oral intake (fluids, nutrition)		E		E
PONV prophylaxis		E
Laxative use		E
Thromboembolic prophylaxis			E, C		E
Goal-setting with patient		E
Patient encouraged to be active participant in own care/disown sick role		E
Provision of equipment/medication on discharge				E, C
Follow-up support: contact telephone number, telephone call, home visit	E			E	E
Post-discharge exercise regimen and/or PT support		E		E	E
Follow-up outpatient/community clinic visit/referrals			E, C

Quality assessment

Five studies were rated globally as ‘weak’ and eight were rated globally as ‘moderate’. No study was rated as ‘strong’ overall. The study by McGregor and colleagues244 was the only one to be rated as ‘weak’ on three domains: study design, confounders and data collection methods. Only three studies202,223,234 provided a clear definition of their LOS outcomes.

Effectiveness of enhanced recovery protocol interventions at improving clinical outcomes

Report Supplementary Material 5, Table 5, displays clinical outcome data for each study trialling an ERP intervention to improve recovery from lower limb arthroplasty. Two studies234,261 provided data from which standardised mean differences could be calculated. Meta-analysis indicates a reduction in LOS of 3.3 days (SD 0.5 days, range 2.9–3.6 days), with a large effect size (d = –1.26, 95% CI –1.62 to –0.89; p < 0.001) (Figure 6). In the three studies196,254,256 reporting LOS variance as range data, thereby precluding secondary analysis, LOS was always shorter in the ERP group. Sensitivity analysis showing the influence of imputed data on LOS meta-analysis is available in Appendix 4.

FIGURE 6.

Forest plot showing the results of meta-analysis of the effect of ERP vs. usual treatment on LOS following lower limb arthroplasty.

Rates of re-admissions and use of additional support were similar between ERP and usual care, where reported. The odds of experiencing a complication were statistically significantly lower in patients receiving ERP in the study by Siggeirsdottir and colleagues261 (OR 0.25, 95% CI 0.07 to 0.88). The odds of experiencing a complication were similar between groups in the study by Reilly and colleagues. 256 No other studies reported complications.

Effectiveness of enhanced recovery protocol interventions at improving patient-reported outcomes

Report Supplementary Material 5, Table 6, displays patient-reported outcome data for each RCT trialling an ERP intervention to improve recovery from lower limb arthroplasty. Assessments of joint function before discharge were performed in three studies,196,254,256 although standardised mean differences could not be calculated for two of these. 196,254 Reilly and colleagues256 reported that patients receiving ERP had improved knee flexion range of motion, with an increase of around 5°. Although this change was associated with a medium effect size, wide CIs indicate uncertainty about the true effect (d = 0.79, 95% CI 0.16 to 1.43; p < 0.05). There were no differences between groups for other assessments of knee function in this study. 256

Longer-term (up to 6-month follow-up after surgery) assessments of hip and knee function were undertaken using standardised outcomes such as the Harris Hip Score, Oxford Hip Score, Oxford Knee Score, American Knee Society Score, and Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC). Figure 7 is a forest plot showing standardised mean differences for these outcomes, as presented in the three studies providing analysable data. 234,256,261

FIGURE 7.

Forest plot (without pooled effects) displaying standardised mean differences for quality-of-life outcomes related to lower limb arthroplasty. AKSS, American Knee Society Score; EQ-5D, EuroQol-5 Dimensions.

Improvements were seen on two measures following ERP interventions, the EuroQol-5 Dimensions (EQ-5D) quality-of-life measure234 and the Oxford Hip Score,261 both of which were associated with medium effect sizes but wide CIs (EQ-5D: d = 0.50, 95% CI 0.07 to 0.93, p < 0.05; Oxford Hip Score: d = –0.66, 95% CI –1.24 to –0.07, p < 0.05). The change in Oxford Hip Scores observed by Siggeirsdottir and colleagues261 was five points, which could be enough to change the classification of hip arthritis severity.

For outcomes with no variance data, there were reports of a benefit with ERP interventions for the number of patients with lower limb weakness, the number of patients who were well at 3 months and the walking status of patients at discharge. There was little difference in the ratings of confidence or satisfaction with care at the end of the study by Borgwardt and colleagues,196 despite median scores for all other outcomes appearing favourable for ERP patients. Pour and colleagues254 reported favourable mean scores on a number of outcomes in ERP patients, including quality of life and markers of recovery; however, presentation of only range data precluded further analysis.

Prehab interventions

Five RCTs trialled prehab intervention in patients undergoing lower limb arthroplasty (knee,222,269 hip,221,244 and hip or knee200). Pre-admission education was delivered in two studies. 222,244 Huang and colleagues222 delivered this in the form of an educational group programme 2–4 weeks prior to admission, supplemented with an educational booklet. McGregor and colleagues244 delivered a class in which they provided information about the procedure and rehabilitation process, ensuring that exercises could be performed. They also made provisions for adapting the home environment if required. In the study by Crowe and colleagues,200 participants were assessed by an occupational therapist, a physiotherapist or a nurse, and were assigned tailored strengthening exercises. They also received a home visit to plan environmental changes, were supported financially or loaned equipment to make adaptations if required, received dietary/nutrition support and counselling, and were provided with a telephone number for support.

Pre-admission exercise programmes featured in four of the interventions. 200,221,222,269 Hoogeboom and colleagues221 delivered a 3- to 6-week programme of strength and endurance exercises at their physiotherapy outpatient department, as well as using pedometry to increase general activity and including family in the programme. Similarly, the patients randomised to the intervention group in the study by Williamson and colleagues269 underwent a 6-week programme of lower limb strength and balance training. In the intervention delivered by Huang and colleagues,222 lower limb strength exercises were performed at home, and a telephone call was made to patients 1 week before surgery to discuss the home exercise programme and answer any questions. Participants in the study by Crowe and colleagues200 received their individualised programme of strength and endurance training either in an outpatient physiotherapy clinic or from a physiotherapist at home. Any patients in the study by Crowe and colleagues200 who required multidisciplinary prehab, as determined by baseline assessment, attended a day-care hospital. Full details of these interventions can be found in Report Supplementary Material 3, Table 7.

Two of the studies were rated globally as being of ‘moderate’ quality,222,269 and three were rated as ‘weak’. 200,221,244 Only the study by McGregor and colleagues244 was rated as ‘weak’ in three domains (study design, confounders and data collection methods). None of the studies described clearly how LOS was defined.

Effectiveness of prehab interventions at improving clinical outcomes

The results for clinical outcomes are shown in Report Supplementary Material 5, Table 7. Meta-analysis of two studies in which effect sizes were calculable200,222 indicates that LOS was around 2.5 days shorter with prehab, associated with a medium effect size (d = –0.55, 95% CI –0.77 to –0.28; p < 0.001; Figure 8). Median LOS was 6 days in both groups in the study by Hoogeboom and colleagues221 and mean LOS was 3 days shorter in the prehab group in the study by McGregor and colleagues,244 although it was not possible to calculate effect sizes for these two studies. In the study by Williamson and colleagues,269 LOS was similar for both groups of patients. Sensitivity analysis showing the influence of imputed data on LOS meta-analysis is available in Appendix 4.

FIGURE 8.

Forest plot showing the results of meta-analysis of the effect of prehab interventions vs. usual treatment on LOS following lower limb arthroplasty.

No data were provided for use of additional care, morbidity, mortality or re-admissions. The odds of experiencing complications was statistically significantly lower in the study by Crowe and colleagues,200 with odds 75% lower in the prehab group (OR 0.25, 95% CI 0.10 to 0.64). Complications were also reported in the studies by Hoogeboom and colleagues221 and Huang and colleagues,222 but did not differ statistically between groups. Williamson and colleagues269 reported three complications but did not identify the study group in which these occurred.

Effectiveness of prehab interventions at improving patient-reported outcomes

Report Supplementary Material 5, Table 8, displays results for patient-reported outcomes from RCTs evaluating prehab interventions to improve recovery from lower limb arthroplasty. Crowe and colleagues200 reported that patients randomised to prehab achieved the following discharge criteria sooner than controls: able to get out of bed independently (d = –0.36, 95% CI –0.71 to –0.02; p < 0.05); getting all equipment ready for discharge (d = –0.85, 95% CI –1.20 to –0.49; p < 0.001); having meals planned for discharge (d = –1.32, 95% CI –1.69 to –0.94; p < 0.001); and having all discharge criteria met (d = –0.43, 95% CI –0.77 to –0.08; p < 0.05).

Anxiety was significantly greater in the prehab group in the study by Williamson and colleagues;269 this change was associated with a medium effect size (d = 0.52, 95% CI 0.21 to 0.94; p < 0.01), although scores for both groups were in the ‘normal’ range. 273 Figure 9 is a forest plot displaying standardised mean differences (Cohen’s d) after prehab for a variety of outcomes relating to functional recovery after surgery, quality of life, pain and mental health across included studies.

FIGURE 9.

Forest plot (without pooled effects) displaying effect sizes (Cohen’s d) and 95% CIs for function, quality of life, pain and mental health outcomes from studies trialling prehab outcomes. 6MWT, Six-Minute Walk Test; ADL, Activities of Daily Living; HADS, Hospital Anxiety and Depression Scale; HGS, hand grip strength; LAPAQ, Longitudinal Ageing Study Amsterdam Physical Activity Questionnaire; PWC, physical work capacity; VAS, visual analogue scale.

Rehab interventions

Two studies202,267 were RCTs of rehab interventions in patients undergoing lower limb arthroplasty. In the study by den Hertog and colleagues,202 patients in the rehab arm received specialist care in a three-bed care unit, where they were given motivational messages and urged to ‘compete’ with others to meet recovery goals. There was a target discharge day of postoperative day 6. Daily mobilisation goals were set, and 2 hours of physiotherapy were delivered each day in a living room environment. The participants in the intervention arm of the study by Vesterby and colleagues267 received a home visit from a physiotherapist 3 days after their operation, with a video conference 2 and 6 days after the operation. The main component of the intervention was a telemedicine rehabilitation programme, working through a box connected to a television, with interactive comprehensive information about osteoarthritis and guidance on completing exercises and using supplementary aids. Full details of interventions can be found in Report Supplementary Material 3, Table 8.

Effectiveness of rehab interventions

Report Supplementary Material 5, Table 9, displays the results for all outcomes reported in the two rehab intervention studies. Standardised mean differences were not calculable because of the absence of variance data or imputable variance data. However, mean LOS was reported to be 6.5 days shorter in patients receiving rehab intervention in the study by den Hertog and colleagues,202 whereas median LOS was 1 day shorter in the experimental group in the study by Vesterby and colleagues. 267 The number of procedure-related complications was similar in both groups in the study by den Hertog and colleagues. 202

Den Hertog and colleagues202 reported lower scores on the WOMAC measure, with a large effect size (d = –1.52 95% CI –1.89 to –1.14; p < 0.001 – per-protocol analysis). Numbers of unplanned patient telephone calls, hospital visits and re-admissions were similar in both groups in the study by Vesterby and colleagues. 267 Meta-analysis was not possible.

Staff-mix interventions

One study223 trialled a staff mix intervention. The intervention was a hospitalist-orthopaedic team using co-management care. The team comprised hospitalist faculty (no residents) and consultative medical specialty teams (faculty and resident) with a mean length of postgraduate clinical experience of > 6 years. Hospitalists provided postoperative medical care after the surgical team completed initial postoperative orders. In the control group, these duties were undertaken by the regular orthopaedic team. The study was of ‘moderate’ quality overall, offering a clear definition of LOS and rated as ‘weak’ only for data collection methods. 223

Huddleston and colleagues223 collected data on LOS and complications in 469 patients; however, no estimates of variance were reported, precluding the calculation of standardised mean differences. Data are reported in Report Supplementary Material 5, Table 10. LOS was similar between trial arms. The odds of patients experiencing complications was 38% lower in patients in the staff mix intervention (OR 0.62, 95% CI 0.43 to 0.89; p < 0.05). This difference can be largely attributed to the 46% lower odds of minor complications in the intervention group (OR 0.54, 95% CI 0.37 to 0.8; p < 0.01).

Interventions to improve recovery from cardiac surgery: randomised controlled trials

Six RCTs trialled interventions to improve recovery following cardiac surgery. Four compared prehab interventions with usual care193,210,215,258 and there were single studies examining the effectiveness of a specialist ward255 and rehab. 266 The evidence came from the UK,210,215 Canada,193 Germany,255 Australia258 and the Netherlands. 266

A total of 1267 people commenced trials, with sample sizes ranging from 117258 to 309. 266 In all studies, patients were scheduled for coronary artery bypass graft (CABG) surgery, plus in two studies valve surgery or combined CABG and valve surgery was also undertaken. 255,258 Across the included studies, the sample was 21.6% female and 63.7 years old (where mean age was reported).

Quality assessment

One study achieved a global rating of ‘strong’,193 with no ‘weak’ domains. Two further studies were rated as ‘moderate’,215,258 with one weak domain. Goodman and colleagues215 retained < 60% of the recruited participants, and Rosenfeldt and colleagues258 did not report validity or reliability of their primary outcome measure. Of the studies rated as ‘weak’ overall, two were ‘weak’ for two domains (selection bias and data collection methods;210 confounders and blinding255) and one was weak in three. 266 The study by van der Peijl and colleagues266 did not adjust for confounding variables, did not comment on the validity or reliability of their primary outcome, and did not clearly report withdrawals and dropouts. Two of the six studies also failed to clearly define their LOS outcome. 258,266

Overview: cardiac surgery

This body of evidence is summarised below and in Table 9. Full details of the effectiveness of interventions are described in Prehab interventions and Specialist ward and rehabilitation interventions.

TABLE 9 - Summary of studies, intervention components and findings for RCTs evaluating interventions to improve recovery from cardiac surgery

AEI, assessment, education, counselling or information; ANA, analgesia protocol; ANE, anaesthesia protocol; EX, exercise programme; GOAL, goal-setting; MT, motivational talks; OT, occupational therapy; PreM, pre-medication protocol; PSYC, psychological support; PT, physiotherapy; REF, referrals; SW, specialist ward; TEL, telephone support; WARM, intraoperative warming protocol.

▲, standardised mean difference indicates significant medium beneficial effect size; △, standardised mean difference indicates significant small beneficial effect size; ◁▷, standardised mean difference indicates no significant difference between groups; ⬆, OR indicates significant beneficial effect; ⇦⇨, OR indicates no significant difference between groups; x, data but standardised mean difference or OR could not be calculated; [blank], Not reported; (x/y), finding for x of y measures (e.g. where multiple measures per outcome category).

Intervention components are listed where they appear in only the experimental arm of the study. Study quality is indicated by the colour of the border of the study details cell: light green, ‘strong’; dark green, ‘moderate’; light blue, ‘weak’.

Study details		Intervention components							Outcome categories
First author, year, country (quality assessment)	Sample size	Intervention category	Pre hospital admission	Preoperative	Perioperative	Postoperative	Post discharge	Other	LOS	Complications	Mortality	Re-admissions	Markers of recovery	Mental health	Quality of life	Use of additional care
Arthur 2000,193 Canada	249	Prehab	AEI; EX; REF; TEL						▲(2/2)					◁▷	△
Goodman 2008,215 UK	188	Prehab	AEI; MT						◁▷
Rosenfeldt 2011,258 Australia	117	Prehab	AEI; EX; REF; OT; PSYC						x					◁▷	◁▷
Furze 2009,210 UK	204	Prehab	AEI; EX; GOAL; REF; TEL; PSYC						◁▷	⇦⇨	⇦⇨	⇦⇨	x	x		⇦⇨
Probst 2014,255 Germany	200	SW		PreM	ANE; ANA; WARM	ANA; SW			◁▷	⬆	⇦⇨
van der Peijl 2004,266 the Netherlands	309	Rehab				PT; EX		Programme frequency: 2 × day	x				◁▷(4/4)

The six RCTs of interventions to improve recovery after elective cardiac surgery in older adults failed to provide conclusive evidence of benefit. Meta-analysis of the four RCTs trialling prehab interventions indicated a small reduction in LOS, compared with usual care; however, two of the three studies included showed no benefit. There were only single studies for rehab and specialist ward interventions. Although there were statistically significant changes on individual measures of quality of life and risk of complications in two studies, more evidence is required to determine whether or not multimodal interventions can convey any benefit in this population.

The only study to indicate a statistically significant reduction in LOS with prehab was conducted by Arthur and colleagues. 193 This Canadian RCT was of ‘strong’ quality with a fairly large sample size (n = 249), and, as such, is the most robust study in the group, although also the oldest. The authors also observed a statistically significant improvement in physical quality of life. 193 The three other prehab programmes210,215,258 had no statistically significant impact on LOS, but did not result in any adverse effects on rates of readmission, use of additional care210 or quality of life. 258

The intervention evaluated by Arthur and colleagues193 was the only one of the four prehab studies to include both an education component and supervised outpatient exercise training as part of its intervention. It was also the only intervention to involve spouses and family members, by informing them of what to expect when they saw the patient after surgery.

Goal-setting with the aim of increasing activity level was incorporated within the education delivered as part of the ‘HeartOp Programme’ evaluated by Furze and colleagues. 210 The studies by both Arthur and colleagues193 and Furze and colleagues210 incorporated telephone follow-up by nurses at regular intervals prior to surgery, which may also have contributed towards improved outcomes.

Perhaps the combination of individual supervision, regular telephone calls and involvement of family members helped motivate patients to engage with the prehab programme in the study by Arthur and colleagues. 193 The other study to integrate face-to-face education and regular patient contact (via motivational interviewing delivered monthly) did not reduce LOS. 215 The absence of ongoing support and/or supervision within this intervention may limit its effectiveness, or motivational interviewing may not be as effective as being motivated by family members.

Prehab interventions

The interventions trialled in the studies by Arthur and colleagues193 and Furze and colleagues210 included extensive preoperative information, education and counselling. Arthur and colleagues193 followed this with individual exercise prescriptions for supervised 90-minute sessions, twice per week in the lead-up to surgery. In the study by Furze and colleagues,210 participants in the prehab arm also received relaxation therapy to reduce stress and advice on postoperative self-management. In the study by Rosenfeldt and colleagues,258 participants in the intervention group received aerobic exercise physiotherapy, aiming to reach four 30-minute sessions per week. Participants also received relaxation therapy from an occupational therapist. The intervention in the study by Goodman and colleagues215 was a series of monthly visits from a cardiac home-care nurse in order to reduce cardiac risks, establish lifestyle changes, provide education and answer questions.

Effectiveness of prehab interventions at improving clinical outcomes

Report Supplementary Material 5, Table 10, displays data for clinical outcomes for each study trialling a prehab intervention. Meta-analysis of the effects of prehab on LOS yielded evidence of a small reduction in LOS with prehab interventions (d = –0.35, 95% CI –0.68 to –0.02; p < 0.05) (Figure 10). In absolute terms, LOS was reduced by a mean of 0.7 days (SD 0.3 days, range 0.5–1 days); however, CIs for the pooled effect approach zero, indicating inconsistency across outcomes. Prehab was associated with a reduction in LOS in the study by Arthur and colleagues193 (d = –0.67, 95% CI –0.92 to –0.41; p < 0.001), but not in any of the other studies. The main influence on LOS in this study came from the time spent in hospital after surgery, which was an average of 1 day shorter in the prehab group. Sensitivity analysis showing the influence of imputed data on LOS meta-analysis is available in Appendix 4.

FIGURE 10.

Forest plot showing the results of meta-analysis of the effect of prehab interventions vs. usual treatment on LOS following cardiac surgery.

Furze and colleagues210 provided information about the use of additional GP and NHS hospital visits in the eight weeks after discharge, with results being similar between groups. Rosenfeldt and colleagues258 reported that median LOS was the same (6 days) in both groups. Only Furze and colleagues210 reported complications or mortality, both of which were similar in both groups.

Effectiveness of prehab interventions at improving patient-reported outcomes

Report Supplementary Material 5, Table 11, displays data for patient-reported outcomes for each study trialling a prehab intervention. The Short Form questionnaire-36 items (SF-36) was utilised by Arthur and colleagues193 and Rosenfeldt and colleagues. 258 Arthur and colleagues193 presented change scores from baseline on various subscales of the SF-36, with significantly greater changes (indicating improvement) in the prehab group than in the usual care group on the physical functioning subscale and the physical composite summary scales; both changes were associated with small effect sizes. There was no statistically significant difference in the change from baseline between groups for any other subscale of the SF-36. In the study by Rosenfeldt and colleagues,258 participants in both groups scored similarly on both physical and mental composite scores. Furze and colleagues210 calculated QALYs as a measure of health-related quality of life and mortality, but, as they evaluated these prior to surgery, we have not reported them here. Goodman and colleagues215 collected data for mental health outcomes but did not report numerical data. The effects for quality-of-life outcomes are displayed in Figure 11.

FIGURE 11.

Forest plot (without pooled effects) displaying the effects of prehab on quality-of-life outcomes in patients receiving cardiac surgery.

Specialist ward and rehabilitation interventions

Specialist ward and rehabilitation interventions were trialled in one RCT each. The specialist ward in the study by Probst and colleagues255 was a three-bed post-anaesthetic care unit, a patient-to-staff ratio of 1 : 3, limited opening times, strict analgesia regime with pain monitoring, extubation as soon as ready and discharge into a step-down unit. This was compared with the larger conventional unit with 21 beds, where the physician to patient ratio was 12 : 1, staff were less specialised, opening hours were longer, analgesia was liberal, and extubation and discharge were typically delayed.

In the study by van der Peijl and colleagues,266 patients were randomised to high- or low-frequency rehabilitation. In the high-frequency intervention, patients began a progressive exercise programme targeting flexibility, strength, co-ordination, walking and stair climbing. Exercises began the day after surgery and continued at weekends. In the comparator group, exercises were not supervised and were interrupted by the weekend.

The results of these two studies are displayed in Report Supplementary Material 5, Table 12. Despite reduced time in the treatment unit and intermediate care in the intervention group, total LOS was similar in both groups in the study by Probst and colleagues. 255 However, the odds of incurring a complication were statistically significantly lower, by 79%, in the patients randomised to the specialist ward (OR 0.21, 95% CI 0.10 to 0.41).

Median LOS was the same in both groups in the rehabilitation trial by van der Peijl and colleagues,266 and all outcomes related to physical activity and functional independence after surgery were similar between groups.

Interventions to improve recovery from upper abdominal surgery: randomised controlled trials

There were five RCTs199,205,226,227,264 of interventions to improve recovery and/or reduce LOS following upper abdominal surgery. Procedures in these studies were liver resection,205,226 hepatectomy or pancreatectomy,227 pancreaticoduodenectomy199 and gastric cancer surgery. 199,264 Two studies were conducted in the UK,205,226 one was conducted in Taiwan,199 one was conducted in Japan264 and one was conducted in Greece. 227 Four of the studies199,226,227,264 evaluated ERP interventions and one205 evaluated prehab. In all cases, the comparator was standard care. Across the five studies, there were 485 participants, of whom approximately 42% were female. In the study by Chen and colleagues,199 participants were undergoing different abdominal surgeries. The results for those receiving gastrectomy and pancreaticoduodenectomy were treated as separate analyses, both compared with groups of patients receiving usual care for the same procure.

Quality assessment

Three of the studies were rated as ‘moderate’ overall199,227,264 and two were rated as ‘weak’. 205,226 The studies rated as ‘weak’ overall scored the lowest rating across two domains. The study by Dunne and colleagues205 was rated as ‘weak’ for data collection methods and selection bias. Selection bias was rated as ‘weak’ owing to the large number of dropouts. 226 Jones and colleagues226 did not control for important between-group differences. Three studies226,227,264 provided a clear definition of LOS.

Overview: upper abdominal surgery

The evidence for upper abdominal surgery is summarised here and in Table 10. Full details of the effectiveness of interventions can be found in Enhanced recovery protocol interventions and Prehab interventions.

TABLE 10 - Summary of studies, intervention components and findings for RCTs evaluating interventions to improve recovery from upper abdominal surgery

AEI, assessment, education, counselling or information; ANA, analgesia protocol; CATH, catheter protocol; CHL, carbohydrate loading; DRA, drain protocol; EMOB, early mobilisation; EON, early oral nutrition; EX, exercise programme; NUT, nutrition supplementation or diet management.

▲, standardised mean difference indicates significant large beneficial effect size (Cohen’s d > 0.80); ▲, standardised mean difference indicates significant medium beneficial effect size (Cohen’s d 0.50 to 0.79); △, standardised mean difference indicates significant small beneficial effect size; ◁▷, standardised mean difference indicates no significant difference between groups; ⬆, OR indicates significant beneficial effect; ⇦⇨, OR indicates no significant difference between groups; x, data but standardised mean difference or OR could not be calculated; [blank], not reported; (x/y), finding for x of y measures (e.g. where multiple measures per outcome category).

Intervention components are listed where they only appear in the experimental arm of the study. Study quality is indicated by the colour of the border of the study details cell: dark green, ‘moderate’; light blue, ‘weak’.

Study details		Intervention components							Outcome categories
First author, year, country (quality assessment)	Sample size	Intervention category	Pre hospital admission	Preoperative	Perioperative	Postoperative	Post discharge	Other	LOS	Complications	Mortality	Re-admissions	Markers of recovery	Mental health	Quality of life	Use of additional care
Dunne 2016,205 UK	38	Prehab	AEI; EX						◁▷	⇦⇨		⇦⇨		◁▷	⇦⇨(5/5)
Chen 2017,199 Taiwan	535	ERP			FM	EON; COMM; EMOB		Oral and nutritional assistance protocol	▲(2/4)	⬆(1/2) ⇦⇨(1/2)
Kapritsou 2017,227 Greece	63	ERP			nNGT	EMOB			▲	⬆			◁▷(5/5)	◁▷(3/3)
Tanaka 2017,264 Japan	148	ERP		CHL; nFAST	DRA	EON; ANA			▲(2/2)	⬆(1/2) ⇦⇨(1/2)		⇦⇨	▲(1/6) △(3/6) ◁▷(2/6)
Jones 2013,226 UK	104	ERP	AEI; NUT		DRA; FM	DRA; CATH; EON; FM; NUT; ANA			▲	⬆(2/4) ⇦⇨(2/4)	⇦⇨	⇦⇨	▲		x

Meta-analysis of the four RCTs trialling ERP interventions aiming to improve recovery after upper abdominal surgery suggests that LOS can be reduced by around 5 days with ERP compared with usual care. Furthermore, meta-analysis of these four studies199,226,227,264 indicated that the odds of sustaining complications were lower with ERP interventions. Individual studies reported markers of recovery that either improved or remained similar with ERP. There was one trial of a prehab intervention, which led to improvements in some markers of physical fitness, but not in LOS. The successful interventions were heterogeneous, combining multiple components, and, as such, it is difficult to further explore reasons for their success.

Enhanced recovery protocol interventions

Components of the ERP interventions and comparators are mapped against ERAS Society guidelines for liver surgery,274 pancreaticoduodenectomy275 and gastrointestinal surgery253,276–278 in Table 11. The most commonly occurring items were preoperative counselling (2/4 studies); minimal fasting and carbohydrate loading preoperatively (2/4); avoidance of long-acting pre-anaesthetic medication (3/4); early oral intake postoperatively (3/4); early mobilisation (4/4) audit; and feedback (2/4). Usual care in the comparator arms was identical with respect to bowel preparation, pre-anaesthetic medication, antithrombotic and antimicrobial prophylaxis, and the choice of incision. The intervention in the study by Tanaka and colleagues264 was described as an ERAS intervention by the authors. In the study by Chen and colleagues,199 the intervention (the Modified Hospital Elder Life Program) integrated relatively few ERAS Society-recommended components, but featured an additional focus on oral and nutritional assistance, with a postoperative communication protocol to facilitate orientation.

TABLE 11 - The ERAS components included in ERP (E) and comparator (C) arms in trials examining the effectiveness of ERP for upper abdominal surgery. Items derive from guidance for liver surgery unless indicated

C, present in comparator arm; E, present in experimental arm; G, item from gastric surgery guidelines; PD, item from pancreaticoduodenectomy guidelines; POD, postoperative day; PONV, postoperative nausea and vomiting; TEA, thoracic epidural analgesia.

Consumed for 3 days prior to surgery.

Laparoscopy used on average 47% of patients.

Patients from both experimental and control groups had either laparoscopic or open gastrectomy.

Generally avoided and placed if only deemed necessary by operating surgeon.

Patients in experimental group received additional supplements.

Experimental, twice per day; control, once per day.

Only one antiemetic provided.

ERAS item and description	Study (first author and year)
	Chen 2017199	Jones 2013226	Kapritsou 2017227	Tanaka 2017264
Preoperative counselling: patients should receive routine dedicated preoperative counselling and education before liver surgery		E	E, C
Preoperative endoscopic biliary drainage should not be undertaken routinely in patients (PD)
Preoperative smoking and alcohol consumption: abstinence should be attempted for 1 month prior to surgery (PD)
Preoperative nutrition: routine use not warranted, but significantly malnourished patients should be optimised (PD)
Perioperative nutrition: patients at risk should receive oral nutritional supplements for 7 days prior to surgery		Ea
Perioperative oral immunonutrition: limited evidence for use (immunonutrition for 5–7 days perioperatively should be considered – PD)
Preoperative fasting and preoperative carbohydrates load: preoperative fasting does not need to exceed 6 hours for solids and 2 hours for liquids. Carbohydrate loading the evening before surgery and 2 hours before the induction of anaesthesia		E		E
No oral mechanical bowel preparation		E, C	E, C
Pre-aesthetic medication: long-acting anxiolytic drugs should be avoided. Short-acting anxiolytics may be used to perform regional analgesia prior to the induction of anaesthesia		E, C	E, C	E, C
Antithrombotic prophylaxis		E, C
Perioperative steroids administration: steroids (methylprednisolone) may be used before hepatectomy in normal liver parenchyma
Antimicrobial prophylaxis and skin preparation		E, C
Incision: the choice of incision is at the surgeon’s discretion		E, C
Minimally invasive approach: laparoscopic liver resections can be performed by hepatobiliary surgeons experienced in laparoscopic surgery	E,b Cb			E, Cc
Avoidance of prophylactic nasogastric intubation			E
Nasogastric tubes removed prior to extubation (PD)
Prophylactic abdominal drainage: no recommendation		E, Cd
Preventing intraoperative hypothermia		E, C
Early oral intake: eat normal food on POD1. Enteral or parenteral feeding reserved for malnourished patients or those with prolonged fasting due to complications		E,e C		E
Normal diet after surgery without restrictions, beginning carefully and increasing intake according to tolerance over 3–4 days (PD)	E
Postoperative glycaemic control (diabetic patients)
Avoidance of perianastomotic drains (G)				E, C
Urinary drainage: suprapubic catheterisation is superior to transurethral catheterisation if used for > 4 days (PD)
Prevention of delayed gastric emptying
Stimulation of bowel movement: not indicated	E, C
Stimulation of bowel movement: a multimodal approach with epidural and near-zero fluid balance is recommended, including oral laxatives and chewing gum given postoperatively (PD)
Early mobilisation	E	E, Cf	E	E, C
Analgesia: routine TEA not recommended in open liver surgery. Wound infusion catheter or intrathecal opiates can be good alternatives combined with multimodal analgesia
Pain management: opioid-sparing analgesic strategies, including regional analgesia techniques, should be implemented in context of a multimodal analgesic regimen (G)				E
Epidural analgesia: mid-thoracic epidurals are recommended, compared with intravenous opioids (PD)
Intravenous analgesia: some evidence supports the use of PCA or intravenous lidocaine analgesic methods (PD)
Multimodal approach to PONV should be used. Patients should receive PONV prophylaxis with two antiemetic drugs		E, Cg
Fluid management: maintenance of low central venous pressure with close monitoring. Balanced crystalloid preferred over saline or colloids		E
Fluid management: perioperative haemodynamic management: maintain fluid homeostasis avoiding fluid excess and organ hypo perfusion (G)				E, C
Audit: systematic audit		E	E, C

The two studies evaluating ERP pathways with patients undergoing liver surgery226,227 adhered to a mean of 10.5 (45.7%) of the 23 relevant ERAS guidelines, while the comparator groups in these studies adhered to a mean of seven (30.4%) items.

The intervention in the study by Kapritsou and colleagues,227 conducted with patients undergoing hepatectomy or pancreatectomy, adhered to three (9.7%) of the 31 ERAS guideline items relevant to individuals undergoing pancreaticoduodenectomy, as did the comparator group.

The two studies evaluating ERP pathways with patients undergoing gastrointestinal surgery adhered to a mean of six (23%) of the 23 ERAS guideline items, with their comparator groups adhering to a mean of 3.5 items (15.2%). 199,264

The intervention evaluated in the study by Chen and colleagues,199 conducted with patients undergoing a mix of pancreaticoduodenectomy and gastric procedures, adhered to five (17.9%) of the 28 relevant ERAS guideline items. The comparator group adhered to two (7.1%) of the relevant items. 199

Effectiveness of enhanced recovery protocol interventions at improving clinical outcomes

Report Supplementary Material 5, Table 13, displays clinical outcome data for RCTs evaluating ERP interventions to improve recovery from upper abdominal surgery. Meta-analysis of all trials indicated that LOS was a mean of 5.1 days shorter in ERP groups (SD 3.2 days, range 1.2–9.5 days), with a large effect size (d = –1.04, 95% CI –1.55 to –0.53; p < 0.001) (Figure 12). Heterogeneity was high and statistically significant, with wide CIs reflecting uncertainty in the true effect (I² = 82.8%; p < 0.001).

FIGURE 12.

Forest plot showing the results of meta-analysis of the effect of ERP vs. usual treatment on LOS following upper abdominal surgery. PD, pancreaticoduodenectomy.

Complications were reported in five groups across four studies,199,226,227,264 with meta-analysis finding that the odds of sustaining complications were 61% lower in patients receiving ERP (OR 0.39, 95% CI 0.24 to 0.64; Figure 13). Although heterogeneity was low and not statistically significant for this effect (I² = 37.4; p = 0.17), there was a mix of positive and null findings across the included studies. Re-admission rates were similar in the two studies that reported this outcome. 226,264

FIGURE 13.

Forest plot showing the results of meta-analysis of the effect of ERP interventions vs. usual treatment on odds of complications following upper abdominal surgery. C-D, Clavien–Dindo; PD, pancreaticoduodenectomy.

Effectiveness of enhanced recovery protocol interventions at improving patient-reported outcomes

Report Supplementary Material 5, Table 14, displays data for patient-reported outcomes from the studies trialling an ERP intervention to improve recovery from upper abdominal surgery. The study by Jones and colleagues226 found that patients in the ERP arm were medically fit for discharge around 4 days sooner than those receiving standard care, with a very large effect size reported (d = –5.23, 95% CI –6.1 to –4.35; p < 0.001). A range of markers of recovery was reported by Kapritsou and colleagues,227 including pain, sadness, optimism and various stress outcomes, but these did not differ between groups. Tanaka and colleagues264 reported earlier passage of first stool in participants in the ERP group (1 day earlier, d = –0.52, 95% CI –0.85 to –0.18; p < 0.01) but not first flatus. Patients in this study also had greater serum concentration of protein and transthyretin on day 7 following surgery, and experienced less weight loss than patients in the comparator arm at 1 week (0.9% less, d = 0.58, 95% CI 0.24 to 0.92; p < 0.01) and 1 month (1.6% less, d = 0.51, 95% CI 0.17 to 0.84; p < 0.01) after surgery. There was no opportunity to perform meta-analysis as outcomes were not similar between studies.

Prehab interventions

One RCT205 trialled a prehab intervention to improve recovery from upper abdominal surgery, with 35 patients included in the study. Patients in the experimental group performed 4 weeks of exercise sessions tailored to their physical capacity. Outcomes are presented in Report Supplementary Material 5, Table 15. There was no difference in LOS, complications or re-admissions between the groups. The authors evaluated markers of physical conditioning for all participants, also performing a subgroup analysis of ‘high-risk’ patients, as evaluated on study entry. Improvements were seen in a number of these outcomes, associated with large effect sizes, with some benefits also observed in high-risk patients.

Interventions to improve recovery from pelvic surgery, vascular surgery, thoracic surgery and surgery to remove tumours: randomised controlled trials

Several procedure categories featured in only one or two RCTs. These were interventions to improve recovery from pelvic,217,225 vascular31,247 and thoracic surgeries,248 as well as a range of surgeries to remove tumours. 220 Evidence from these studies is summarised below and in Table 12, with full details of intervention effectiveness described in Pelvic surgery, Vascular surgery, Thoracic surgery and Surgery to remove a tumour.

TABLE 12 - Summary of studies, intervention components and findings for RCTs evaluating interventions to improve recovery from pelvic surgery, vascular surgery, thoracic surgery and surgery to remove various tumours

AEI, assessment, education, counselling or information; ANA, analgesia protocol; ANE, anaesthesia protocol; CATH, catheter protocol; CHL, carbohydrate loading; DRA, drain protocol; EMOB, early mobilisation; EON, early oral nutrition; EX, exercise programme; LAX, laxative; nFAST, avoidance of prolonged fasting; nMBP, avoidance of mechanical bowel preparation; PACP, preoperative assessment with care plan; PONV, active prevention of nausea and vomiting; PT, physiotherapy; SURG, surgical approach differs between groups.

▲, standardised mean difference indicates significant large beneficial effect size (Cohen’s d > 0.80); ▲, standardised mean difference indicates significant medium beneficial effect size (Cohen’s d 0.50 to 0.79); ⬆, OR indicates significant beneficial effect; ⇦⇨, OR indicates no significant difference between groups; È, OR indicates significant negative effect; x, data but standardised mean difference or OR could not be calculated; [blank], not reported; (x/y), finding for x of y measures (e.g. where multiple measures per outcome category).

Sample size reported in individual paper from same study.

*107 received allocated regime. Intervention components are listed where they only appear in the experimental arm.

^Patient satisfaction also reported.

Study quality indicated by the colour of the study details cell: light green, ‘strong’; dark green, ‘moderate’; light blue, ‘weak’.

Study details		Intervention components							Outcome categories
First author, year, country (quality assessment)	Sample size	Intervention category	Pre hospital admission	Preoperative	Perioperative	Postoperative	Post discharge	Other	LOS	Complications	Mortality	Re-admissions	Markers of recovery	Mental health	Quality of life	Use of additional care
Pelvic surgery
Jensen 2015,225 Denmark	129*	ERP	AEI; PT; EX			EMOB; EX			x	⇦⇨	⇦⇨	⇦⇨	x(4/4)
Gralla 2007;217 Magheli 2001,240 Germany^	50	ERP		nMBP; nFAST	ANE; WARM; PONV	ANA; DRA; EON; EMOB; FM; LAX	AEI		▲	⬆		⇦⇨	▲
Vascular surgery
Partridge 2017,31 UK	209	PACP	AEI				FAPP	Pre-surgery: medications changed and level of care required advised. Post discharge: longer-term GP follow-up	x	⬆(3/8) ⇦⇨(5/8)		⇦⇨	⇦⇨(2/2)
Muehling 2008,247 2009,246 2011,49 Germany	82; 101; 101	ERP		nMBP; nFAST	CATH	CATH; EON; EMOB; ANA			x	⬆(2/3) ⇦⇨(1/3)	⇦⇨	⇦⇨	x(3/3)
Thoracic surgery
Muehling 2008,248 Germany	62	ERP		nFAST	SURG	EON; EMOB; ANA			x	⇦⇨
Surgery to remove tumours
Hempenius 2013,220 2016,219 the Netherlands	297; 260a	PACP	AEI					Screened for delirium 3 × day	x	⇦⇨(2/2)	⇦⇨	⇦⇨		⇦⇨(2/2)	È(1/13) ⇦⇨(4/13) x(9/13)

Overview: interventions to improve recovery from pelvic surgery, vascular surgery, thoracic surgery and surgery to remove tumours

Pelvic surgery

The two RCTs trialling interventions to improve recovery after pelvic surgery investigated the effect of ERPs. Evidence from the two trials was inconclusive regarding the effects of ERP on any of the outcomes presented. The study by Gralla and colleagues217 demonstrated a statistically significant reduction of LOS within the intervention group. There were also statistically significant improvements for complications and markers of recovery in this study. However, the study by Jensen and colleagues,225 which was of higher quality than the German study, did not report any statistically significant differences between groups for any outcomes. 217 One apparent difference between these studies is that the intervention by Gralla and colleagues217 contained 12 active components, compared with the five in the intervention in the study by Jensen and colleagues. 225 However, with only two studies available, the RCT evidence for ERP to improve recovery from pelvic surgery is both limited and inconclusive.

Vascular surgery

Two studies evaluated interventions to improve recovery from vascular surgery. In the study by Muehling and colleagues,247 patients receiving ERP were less likely to experience complications, and there were no re-admissions or deaths in the study. The study was rated as ‘weak’ and the data presented for LOS precluded secondary analysis.

Partridge and colleagues31 evaluated a preoperative assessment that informed a care plan. Although the study was of ‘strong’ quality, and LOS was reported as 2.3 days shorter in the experimental group, a lack of variance data precluded calculations of effect size or statistical difference. The identification of 58 more new comorbidities in the experimental group than in the usual care group may have contributed to the reduced odds of complications and, ultimately, LOS. However, few conclusions about interventions to improve recovery from vascular surgery can be drawn.

Thoracic surgery

Muehling and colleagues248 evaluated an ERP intervention to improve recovery in patients undergoing thoracic surgery. The study was of ‘weak’ quality and outcomes either could not be analysed or did not differ between groups.

Surgery to remove a tumour

One study220 evaluated an intervention to provide supportive care and prevent delirium in patients undergoing various surgeries to remove a tumour. The study was of ‘moderate’ quality and reported on a wide range of outcomes other than LOS. Fewer patients could return to their previous living arrangements, which was an undesirable outcome. No other differences were observed in the study, and LOS could not be analysed.

Interventions to improve recovery from pelvic surgery: randomised controlled trials

There were two RCTs217,225 of ERP interventions to improve recovery following pelvic surgery, reported across three journal articles. One of the studies took place in Germany217 and one took place in Denmark. 225 The study by Gralla and colleagues217 was reported across two papers, with the publication by Magheli and colleagues240 focusing on aspects of recovery, while the core paper focused on LOS and complications. 217 Patients (n = 179) were admitted for radical cystectomy225 or radical prostatectomy;217 they had a mean age of 67.8 years and 81% were male.

The study by Jensen and colleagues225 was rated as ‘moderate’ quality overall, scoring as ‘weak’ only for data collection methods. Across the two papers reporting on the study by Gralla and colleagues,217,240 all domains were rated as ‘strong’ or ‘moderate’ except for data collection methods. LOS was clearly defined in both studies.

Enhanced recovery protocol interventions

Components of the ERP interventions and comparators are mapped against ERAS Society guidelines for pelvic surgery272 and radical cystectomy279 in Table 13. Usual care in the study by Jensen and colleagues225 already included a number of fast-track items; the experimental group underwent additional prehabilitation, rehabilitation and discharge planning in addition to the existing fast-track protocol. In the study by Gralla and colleagues,217 few ERAS items had been implemented into standard care.

TABLE 13 - Components of included ERP interventions and comparators, mapped against ERAS guidelines for rectal/pelvic surgery and radical cystectomy

C, present in comparator arm; E, present in experimental arm; POD, postoperative day; PONV, postoperative nausea and vomiting; RC, radical cystectomy; R/P, items from guidelines for rectal/pelvic elective surgery; TEA, thoracic epidural analgesia.

Additional exercise programme.

Enema only.

Mini-laparotomy or robot-assisted.

ERAS item	Study (first author and year)
	Gralla 2007,217 Magheli 2011240	Jensen 2015225
Preoperative information, education and counselling		E, C
Preoperative optimisation		E,a C
Preoperative mechanical bowel preparation avoided	Eb	E,b Cb
Preoperative fasting: clear fluids allowed up to 2 hours and solids up to 6 hours prior to induction of anaesthesia
Preoperative carbohydrate treatment for patients without diabetes		E, C
Pre-anaesthetic medication: avoidance of long-acting sedatives
Prophylaxis against thromboembolism		E, C
Antimicrobial prophylaxis and skin preparation	E, C	E, C
Standard anaesthetic protocol	E, C	E, C
Epidural analgesia: TEA is superior to systemic opioids. It should be continued for 72 hours	E	E, C
PONV prophylaxis	E	E, C
Laparoscopic resection recommended for benign disease and rectal cancer (R/P). Minimally invasive approach not recommended outside trial setting (RC)	E, C	E,c Cc
Nasogastric intubation: nasogastric tubes not routinely used	E, C
Nasogastric tubes removed before extubation
Preventing intraoperative hypothermia	E, C
Perioperative fluid management: intraoperative fluids guided by flow measurements to optimise cardiac output (RC)
Perioperative fluid management: judicious use of vasopressors (RC)
Transurethral catheter (R/P)	E
Suprapubic catheter (R/P)
Urinary drainage: transurethral catheter removal on POD1 in low risk patients
Early removal of bladder catheter		E, C
Prevention of postoperative ileus: chewing gum (R/P), oral magnesium, alvimopan
Postoperative laxatives and prokinetics (R/P)	E
Early oral intake 4 hours after surgery	E	E, C
Early mobilisation	E	E,a C
Audit: all patients should be audited for protocol compliance and outcomes		E, C

The intervention evaluated by Jensen and colleagues225 adhered to 14 (53.8%) items, the same number as the comparator group, but with additional exercise components. After excluding the two ERAS items exclusive to radical cystectomy, the intervention evaluated by Gralla and colleagues217 adhered to 12 out of the 24 (50%) ERAS items listed in Table 13, and the comparator group adhered to a mean of five (20.8%) items.

Effectiveness of enhanced recovery protocol interventions at improving clinical outcomes

Table 12 displays outcome data for LOS, re-admissions, additional care, mortality, morbidity and complication where reported in both studies. LOS was reduced by 3.1 days in the study by Gralla and colleagues,217 with a large effect size (d = –2.90, 95% CI –3.73 to –2.12; p < .001). In the study by Jensen and colleagues,225 median LOS was the same in both groups.

The odds of experiencing a complication was significantly lower in patients receiving ERP in the study by Gralla and colleagues217 (OR 0.25, 95% CI 0.07 to 0.83), with patients 75% less likely to experience complications. However, the wide CI indicated that this reduction in odds may have been as low as 17% or as high as 93%. ERP and usual care patients experienced similar rates of complications in the study by Jensen and colleagues. 225 Re-admission rates were similar between groups in both studies, and 90-day mortality was similar in the study by Jensen and colleagues. 225

Effectiveness of enhanced recovery protocol interventions at improving patient-reported outcomes

Report Supplementary Material 5, Table 17, displays outcome data for patient-reported outcomes where these were reported. It was possible to calculate standardised mean difference for only one outcome, which was time to first deflation/defecation in the study by Gralla and colleagues. 217 This occurred 0.4 days earlier after ERP than with usual care, associated with a medium effect size (d = –0.61, 95% CI –1.18 to –0.05; p < 0.05).

Descriptive statistics for patient satisfaction were presented by Gralla and colleagues. 217 There were similar distributions of scores for both groups for satisfaction with the LOS, while slightly more patients thought that the perioperative course was better than expected in the ERP group than in the usual care group (17 vs. 13 patients). 217 Jensen and colleagues225 reported postoperative pain scores, suggesting that patients in the ERP group were in more pain than the patients receiving usual care (16% reporting pain as 4 or above vs. 7%). After 90 days, the distribution of scores on the Clavien–Dindo scale was broadly similar between the groups. 225

Interventions to improve recovery from vascular surgery: randomised controlled trials

There were two RCTs31,247 of interventions to improve recovery and/or reduce LOS following vascular surgery. In the German study by Muehling and colleagues,247 82 patients underwent open repair of an infrarenal aortic aneurysm. We report data from the 2009 paper,246 when full recruitment of 101 patients had been completed, and the 2011 paper,249 which contains additional outcomes (incidence of organ failure and development of systemic inflammatory response syndrome). Patients randomised to the experimental arm received an ERP, primarily consisting of no bowel preparation, reduced preoperative fasting, patient-controlled epidural analgesia, enhanced postoperative nutrition and early mobilisation. ERP was compared with usual care.

In the study from the UK by Partridge and colleagues,31 patients were admitted for endovascular/open aortic aneurysm repair or lower limb arterial bypass surgery. Two hundred and nine participants were randomised either to a preoperative CGA leading to an optimised care plan or to standard preoperative assessment. Full details of the interventions trialled in the two vascular surgery studies can be found in Report Supplementary Material 3, Tables 17 and 18.

The study by Partridge and colleagues31 was rated as ‘strong’ overall, with only one rating of ‘moderate’, for blinding of assessors and participants. However, a clear definition of LOS was not provided. The study by Muehling and colleagues,246,247,249 which was reported across three journal articles, achieved a global rating of ‘weak’. LOS was not clearly defined in any of the three papers reporting on the study.

Effectiveness of interventions

Report Supplementary Material 5, Table 18, displays data for all effectiveness outcomes reported in the two studies trialling interventions to improve recovery from vascular surgery. In the study by Muehling and colleagues,246,249 patients receiving ERP were at lower odds of requiring postoperative ventilation (OR 0.11, 95% CI 0.02 to 0.55), experiencing organ failure (OR 0.34, 95% CI 0.13 to 0.88) or developing systemic inflammatory response syndrome (OR 0.39, 95% CI 0.17 to 0.9). There were no re-admissions or deaths in the study. For LOS and markers of recovery, the authors presented only medians and ranges, precluding secondary analysis.

In the study by Partridge and colleagues,31 LOS was reported as 2.3 days shorter in the experimental group; however, lack of variance data precluded calculations of effect size or statistical difference. The CGA identified 58 more new comorbidities than were diagnosed in the group receiving regular preoperative assessment. Postoperatively, odds of delirium (OR 0.39, 95% CI 0.17 to 0.9), cardiac complications (OR 0.24, 95% CI 0.02 to 0.58), and bowel and bladder complications (OR 0.4, 95% CI 0.22 to 0.74) were significantly lower in the experimental group. There were no statistically significant differences in any other outcomes.

Interventions to improve recovery from thoracic surgery: randomised controlled trials

There was one RCT of an intervention aiming to improve recovery following thoracic surgery. 248 The study compared ERP with traditional care among 59 patients undergoing lung resection. Both the ERP and standard care pathways featured preoperative information, education and counselling, preoperative warming and prevention of fluid overload. In addition, patients in the ERP arm received minimal preoperative fasting, a standardised anaesthetic protocol, early oral intake and early mobilisation. Full details of the intervention can be found in Report Supplementary Material 3, Table 19.

Publication bias: randomised controlled trials

Figure 14 is a funnel plot of the data for LOS from 30 comparisons from the 26 RCTs that provided data from which standardised mean differences could be calculated. 192,193,198–200,204,205,210,211,214,215,217,222,226,227,234,236,239,242,243,255,261,264,266,268,269 As the intercept was below zero, this indicates a tendency towards publication of studies showing beneficial effects of the interventions being evaluated. Studies with more robust data were clustered more symmetrically than those with less robust estimates (small sample sizes), which may indicate that there is bias whereby smaller trials with null or negative outcomes are not included in the data.

FIGURE 14.

Funnel plot to assess publication bias in reporting of LOS across all RCTs. Dotted lines indicate pseudo-95% CIs. SMD, standardised mean difference (effect size) for each study; SE of SMD, the standard error of that effect.

Although publication bias may be suspected in the first instance, it should be noted that interventions were often the implementation of evidence-based protocols to improve recovery from surgery and usually involved additional changes to existing ‘best care’. This may make it unlikely that such interventions would have an adverse effect on LOS, and a negative intercept value could therefore be expected. However, it might be expected that, although interventions ought to lead to reductions in LOS, the influence of such an outcome on risk of complications is less predictable. Therefore, we produced Figure 15, a funnel plot of the data for 36 summary complications outcomes from the 28 RCTs providing such data. 192,198,200,204,205,209,211,213,214,220,222,223,226,227,231,235,236,239,242,246,247,251,252,255,256,261,264,268 Inspection of this funnel plot indicates a symmetrical distribution of studies about the intercept, although this is again in favour of a beneficial effect (lower odds of complications in experimental groups).

FIGURE 15.

Funnel plot to assess publication bias in reporting of complications across all RCTs. Dotted lines indicate pseudo-95% CIs. Log(OR), the log of the OR for each study; SE of log(OR), the standard error of that OR.

In summary, although there is a possibility of publication bias in the evidence from RCTs evaluating interventions to reduce LOS in older adults undergoing elective procedures, the context of the evidence should alleviate concerns that possible unpublished trials with negative results could threaten the robustness of the evidence.

Synthesis of evidence from the UK