Enhanced psychological care in cardiac rehabilitation services for patients with new-onset depression: the CADENCE feasibility study and pilot RCT

Cardiac rehabilitation service provision

In the UK, the NHS routinely offers multidisciplinary cardiac rehabilitation to people who experience an acute cardiac event. These are patients with CHD referred following admission with an ACS. The majority will have had coronary revascularisation, usually percutaneous coronary intervention (PCI) for STEMI, or coronary artery bypass graft (CABG) surgery. Many patients with heart failure may also be admitted with an ACS, and are also eligible for cardiac rehabilitation. Some patients have concomitant valve surgery with CABG, and others may have had a pacemaker or defibrillator in the context of CHD. Some patient groups, such as people with heart failure, who might benefit from cardiac rehabilitation have historically been excluded from cardiac rehabilitation. However, recent national audit data suggest that such exclusions are diminishing and that the diversity of case mix is increasing. In the UK in 2014–15, an estimated 82,127 patients commenced cardiac rehabilitation, including individuals with a primary indication of MI (19.6%), MI and PCI (28.1%), PCI (14.1%), CABG (14.8%), heart failure (4.4%), angina (3.6%), valve surgery (5.9%) and other or unknown conditions (e.g. having a pacemaker, implantable cardioverter defibrillator; 9.6%). 26

Guidance on the recommended content of usual cardiac rehabilitation care has been published by the British Association for Cardiovascular Prevention and Rehabilitation (BACPR),3 the contents of which are broadly comparable with that which is recommended across Europe4 and the USA. 5 Core components include education; exercise and physical activity; diet and weight management; medical management and psychological support. Cardiac rehabilitation follows a seven-stage care pathway (stages 0–6), as illustrated in Figure 1 and Box 1. 27

FIGURE 1.

Cardiac rehabilitation standard patient care pathway. CR, cardiac rehabilitation.

The organisation and delivery of cardiac rehabilitation services are currently in a state of flux. In 2011–12, in the UK, The National Audit of Cardiac Rehabilitation: Annual Statistical Report 201328 identified 290 operationally distinct, comprehensive cardiac rehabilitation programmes (stages 2–5); by 2014–15, 308 core programmes were identified. 26 While working towards nationally agreed standards and a standardised patient care pathway, there remains considerable regional variation in the staffing and skill mix of individual services.

Patient attrition across the cardiac rehabilitation pathway also remains an area of significant challenge. It has been estimated that, for people with one of the four most prevalent indicated conditions (MI, MI and PCI, PCI or CABG) referred to cardiac rehabilitation in the UK, uptake is around 47% and varies considerably by condition (38%, 54%, 40% and 59%, respectively). 26 The number of individuals who go on to attend a structured rehabilitation programme is strongly influenced by gender. It has been estimated that around two-thirds of potentially eligible men start structured rehabilitation, compared with less than one-third of women, and the reasons underlying these differences are not well understood.

As part of the initial cardiac rehabilitation assessment with a nurse specialist (stage 2), individuals are usually asked to complete the Hospital Anxiety and Depression Scale (HADS)29 to assess mood. In 2014–15, it was estimated that 12% of patients exhibited borderline depressive symptoms (i.e. a HADS-D score of 8–10) and 7% were depressed (i.e. a HADS-D score of ≥ 11) during this initial assessment. Of the individuals who completed the final assessment of a structured cardiac rehabilitation programme (stage 5), 8% exhibited borderline depressive symptoms and 4% were depressed during this final assessment. There was considerable regional variation observed in the change scores between localities. 26 These findings indicate that there is considerable scope to improve and standardise treatment for depression among people undergoing cardiac rehabilitation.

Despite the burden of depression, in 2013, only 25 out of 260 cardiac rehabilitation services (10%) reported receiving direct psychologist input. 30 However there is recent evidence that psychologist input is increasing, with 48 out of 261 programmes (18%) reporting such input in 2014. 26 Notwithstanding this progress, the evidence still suggests that patients’ depression usually remains untreated. 31 Locally agreed referral protocols providing for access to psychological care at either the tertiary (stage 1) or community (stages 2–5) levels may be agreed. However, the precise content of such protocols, and the consistency of their implementation, is not clear from audit data.

Although it would appear that structured management of depression is not routinely provided within the majority of cardiac rehabilitation programmes, individuals may access mental health care through existing care pathways for depression provided by other mainstream NHS services. 32–34 Notwithstanding this, there is a need to deliver models of care to increase access to evidence-based depression treatments for people undergoing cardiac rehabilitation.

Effectiveness of psychological care for people with depression

Effective care for people with depressive symptoms includes the provision of psychological and/or pharmacological interventions, as well as the effective co-ordination of mental health-care services. 32–34

Clinical effectiveness of psychological care for people with coronary heart disease

In the previous section, Effectiveness of psychological therapies in the general population, we reviewed the evidence of clinical effectiveness of psychological therapies and mental health collaborative care models tested in the general population samples of people with depressive symptoms. There is also a considerable body of evidence regarding the clinical effectiveness of psychological therapies designed specifically for use for people with CHD. Similar psychological techniques (e.g. CBT, BA) may be applied in isolation or combined with other psychological interventions (e.g. stress management or problem-solving skills training) as part of multifaceted treatment packages. The target population and setting can vary. Some interventions target people with coronary disease irrespective of their baseline psychological health, whereas others adopt a more selective approach, offering interventions only to those individuals screening positive for an existing psychological condition.

Two systematic reviews56,57 have used meta-analytic techniques to synthesise evidence on the clinical effectiveness of psychological interventions for people with CHD. Direct comparisons between these reviews are problematic because of the important differences in the application of methods and definitions. Both reviews selected studies in which the direct effects of psychological interventions were compared with a comparator group (mostly usual medical care with or without cardiac rehabilitation) on measures of depressive symptoms. However, Dickens et al. 56 selected studies with any length of follow-up (ranging from 5 days to 12 months), and reported only psychological outcomes. In contrast, Richards et al. 57 selected studies that followed up participants for a minimum of 6 months (range 6 months to 10.7 years) and reported evidence for clinical events (e.g. mortality, cardiac morbidity), as well as the psychological outcomes of depressive symptoms and anxiety and stress levels. Both reviews also undertook metaregression analyses, seeking to explore whether or not intervention effectiveness was mediated by their use in unselected populations, as opposed to targeting people with existing psychological conditions. Although the studies applied different taxonomies, both reviews also sought to identify potential explanatory components of psychological interventions. Notwithstanding differences in study methods, there was some consistency in the findings from these reviews.

Analysing data from 62 studies (17,397 participants), Dickens et al. 56 observed a small but statistically significant improvement in depression for psychological interventions compared with UC (SMD 0.18, 95% CI 0.12 to 0.24). Subgroup analysis identified certain treatment components with small beneficial effects on depressive symptoms, including general education, problem-solving, skills training, exercise, CBT and relaxation. In a subgroup analysis, there was also a small effect observed in favour of psychologically based interventions (SMD 0.21; 12 trial arms), when targeted at people with coronary disease and clinical depression, as compared with untargeted populations.

In the second review, Richards et al. 57 synthesised data from 35 studies (10,703 participants), including 12 studies that recruited patients with an established psychopathology (eight studies selected people with depression), and from 19 studies reporting some levels of psychopathology within their participant samples. The interventions tested were mostly multifactorial in nature, designed to address a number of different treatment goals (e.g. reducing stress levels, improving mood states, improving coping strategies, emotional support), through combining psychological techniques (e.g. stress management, CBT or behavioural therapies).

Richards et al. 57 found no significant differences in the effects of psychological interventions versus the comparator group in terms of total mortality (23 trials, 7776 participants; RR 0.90, 95% CI 0.77 to 1.05) and the risk of subsequent revascularisation (13 trials, 6822 participants; RR 0.94, 95% CI 0.81 to 1.11) or of a non-fatal infarction (13 trials, 7845 participants; 0.82, 0.64 to 1.05). However, there was some evidence that cardiac mortality was reduced (11 trials, 4792 participants; 0.79, 95% CI 0.63 to 0.98). When considering psychological outcomes, interventions based on psychological therapies resulted in significant improvements in the participant-reported levels of depressive symptoms (19 trials, 5825 participants; SMD –0.27, 95% CI –0.39 to –0.15), anxiety levels (12 trials, 3161 participants; SMD –0.24, 95% CI –0.38 to –0.09) and stress levels (eight trials, 1251 participants; SMD –0.56, 95% CI –0.88 to –0.24). Metaregression exploring a limited number of intervention characteristics found no significant predictors of intervention effects for the outcome of cardiac mortality. Unlike Dickens et al. ,56 psychological therapies combined with adjunct pharmacology (when deemed appropriate) prescribed for an underlying psychological disorder were found to be more effective than interventions that did not combine psychological and pharmacological therapies (difference in effect size β = –0.51; p = 0.003). For anxiety, interventions recruiting participants with an underlying psychological disorder appeared to be more effective than those delivered to unselected populations (β = –0.28; p = 0.03), although this finding was not replicated for the outcome of depressive symptoms.

Although somewhat different in focus and methodology, both reviews consistently demonstrated that providing psychological interventions for people with established CHD yields small, but statistically significant, improvements in participant-reported levels of depressive symptoms. However, given the multicomponent nature of the interventions tested, combined with the diverse patient groups and settings in which such interventions were tested, it is not possible to ascertain which psychological treatment components may be most effective and for whom.

Observational study

Qualitative study

Design of the pilot randomised controlled trial

The second phase of research included a pilot RCT, the testing of economic data collection procedures and a nested qualitative study.

An external pilot cluster RCT69,70 was undertaken. We recruited eight comprehensive cardiac rehabilitation teams (clusters) and randomised five of these to EPC plus UC embedded within cardiac rehabilitation (intervention), and the other three to usual cardiac rehabilitation care (control). The decision to allocate more clusters to the intervention arm was informed by data emerging from our feasibility study (see Chapter 4, The transition between feasibility study and pilot study), which suggested that participant recruitment would be slower than anticipated. We therefore sought to ensure that a sufficient number of nurses and patient participants had exposure to EPC to support the aims of the qualitative study interviews (i.e. recruiting enough participants with experience of EPC).

Randomisation and allocation concealment

A cluster randomised design was adopted in preference to individual randomisation to ameliorate concerns around potential contamination between the intervention and control arms (i.e. intervention participants coming into direct contact with control participants and sharing aspects of their treatment). Our EPC intervention included training for cardiac rehabilitation nurses to provide mental health-care co-ordination to participants with depressive symptoms. Once trained, we judged that it would be very difficult operationally for the nurse to apply care co-ordination to only the subset of their patients randomised to receive EPC (as would be required with individual randomisation).

Randomisation was carried out by the trial statistician (FCW), who was independent of the recruitment of rehabilitation teams or patient participants. The allocation sequence was generated using computer-generated random numbers. Cluster randomisation was balanced (to the extent possible) by team type (community, hospital or mixed community and hospital teams) and patient throughput at the initial cardiac rehabilitation assessment (NHS care pathway stage 2/3; categorised as low/high). The latter cut-off point between low/high throughputs was determined post hoc after scrutinising the mean monthly throughput across cardiac rehabilitation teams, with ‘low’ being equivalent to ≤ 22 patients per month, and ‘high’ being equivalent to > 22 patients per month (these cut-off points were based on a natural break in the distribution of team throughput assessed during a preparatory audit collecting data for a 2-month period in January and February 2015). The throughput of new patients assessed by cardiac rehabilitation teams was selected as a stratifying variable, as this appeared to be a better predictor of team workload than staffing levels, and would ensure that a sufficient number of patient participants were recruited into each trial arm. Team setting was selected as a stratifying variable, as our feasibility study suggested that teams working in hospital or community settings, or in a combination of both, may encounter different resource issues that could affect the delivery of EPC.

Randomisation took place after all eight cardiac teams had been recruited. Each team was informed of its allocation by the trial researcher.

Trial interventions

Settings and study population

Participant eligibility criteria

The participant inclusion and exclusion criteria applied in the pilot trial were essentially the same as those described for the observational study within the feasibility phase (see Participant inclusion and exclusion criteria). The only difference related to the underlying cardiac condition of potential participants. Over the last 3 years, UK cardiac rehabilitation services have been offering care to increasingly diverse cardiac populations. 26 Although the vast majority of patients were expected to be referred for rehabilitation as a result of an ACS, any patient referred for rehabilitation was potentially eligible, and the sample for the pilot study could now include, for example, patients with heart failure.

Participant recruitment and consent procedures

The process for identifying and recruiting trial participants followed the same procedure as outlined for the observational study conducted within the feasibility phase (see Participant recruitment procedure). Potential participants from teams allocated to UC or to the intervention group were recruited following the same procedures, with the only difference being that participant information sheets were tailored to reflect the care provided by their rehabilitation team. At the time of recruitment, all patients recruited to the trial were also asked if they would consider being approached for an interview as part of the study.

Data collection procedures

To facilitate data collection in two of our sites, which were remote from the main trial centre, and to respond to the emerging geography of the trial, it was essential that remote data collection procedures were established. We negotiated with a local clinical research network and university-based research team, both of which provided local, trained research staff to collect participant and note review data in its area. Remote staff were carefully briefed in respect of the design of the study and in data-capture procedures and processes.

Participants in both trial arms completed baseline measures before they were scheduled to commence the comprehensive cardiac rehabilitation programme. Participants were then invited to take part in two face-to-face follow-up interviews at 5 months and 8 months post baseline.

The blinding of participants, cardiac nurses or researchers extracting data on study outcomes was not possible as a result of the cluster design. However, the final analysis was carried out by the trial statistician, who was blind to treatment allocation.

Statistical analysis plan

Trial data estimated cardiac rehabilitation team recruitment and patient participant study completion rates. As this was a pilot study, we did not make definitive estimates of clinical effectiveness and costs; our primary aim was to document the adequacy of trial procedures, intervention acceptability and outcome measures. All pilot study data were collected and reported [in accordance with the Consolidated Standards of Reporting Trials (CONSORT) statement]95 using good practice guidelines for cluster trials. A detailed statistical analysis plan, including health economics, was developed prior to undertaking the analysis of trial data (see Appendix 10). In summary, the recruitment rate, uptake in the intervention and control arms, outcome completion rates and attrition rates were reported (with 95% CIs). It is a known feature of cluster randomised trials that when participants are recruited after clusters have been randomised this can lead to selection bias. 96 The recruitment rate and participant characteristics at baseline were compared between the trial arms to assess whether or not there was evidence of selection bias. Participant outcome data (including resource use) were analysed descriptively (e.g. means, SDs) for each arm at each follow-up. As per the commissioning brief, we reported the outcome effect sizes (i.e. between-group mean differences and 95% CIs) based on an intention-to-treat analysis. We did not use analytical methods that allow for clustering because the number of clusters and participants per cluster were too small. For this reason, and the fact that this is a pilot study, the CIs should be interpreted cautiously and no p-value is presented. We carried out a complete-case analysis, including only those participants who provided outcome data. No data were imputed as this is not a definitive trial of intervention effectiveness. Patient treatment preferences for mental health care were documented at baseline.

Economic evaluation

Adopting a societal costing perspective, we piloted the methods required to collect resource and service use data in order to estimate costs to the NHS, costs to social care and personal social services and relevant costs to participants and their carers/families (including NHS and privately funded mental health care). If feasible, we planned to distinguish the use of services for (1) depression or anxiety, (2) other mental health problems and (3) other health reasons or social needs. In addition, we also undertook a preliminary assessment of the cost of providing EPC within cardiac rehabilitation teams. Service use data were collected from both (1) routine/administrative sources (e.g. hospital records, GP records, community mental health team records and, where appropriate, social care records) and (2) via participant self-report using a Service and Resource Use Questionnaire (SRUQ). The SRUQ was adapted from the Client Services Receipt Inventory (CSRI),97 with input from the patient and public involvement (PPI) group, and some testing during the feasibility phase. A detailed plan describing the analysis of health economics data is presented in Appendix 10. In summary, we assessed the completeness, validity and reliability of patient-reported data versus routine administrative data on the types and amounts of health care/service use by participants, and over the follow-up periods planned for a definitive trial. This approach enabled better judgements to be made regarding which types of health service, care professional or other resource use were most likely to be affected by the expected trial outcomes, and to allow us to determine the optimal source of obtaining such data.

Nested qualitative study

Ethics and governance considerations

The Royal Devon and Exeter NHS Foundation Trust acted as the trial sponsor. The CADENCE study protocol was reviewed, and a favourable ethics opinion obtained from the National Research Ethics Service (NRES) Committee South West – Exeter (reference number 14/SW/0139). The relevant NHS research governance approvals were obtained prior to commencing fieldwork for sites participating in the feasibility and pilot phases of work. All protocol amendments were notified through existing pathways in the UK NRES and NHS research governance systems, and the trial team ensured that all participating cardiac rehabilitation teams, staff, patients and trial registries were informed of any changes to the protocol (see Appendix 12).

The CADENCE study was registered with a trials registry [as International Standard Randomised Controlled Trial Number (ISRCTN) 34701576] and has been adopted by the UK NHS Clinical Research Network (UKCRN ID: 17105). All researchers and nurses involved in participant recruitment underwent good clinical practice training.

Data management

All personal information obtained about participants or staff for the purposes of recruitment or data collection (e.g. names, addresses, contact details, personal information) was kept confidential and held in accordance with the requirements of the Data Protection Act 1998. 98 Each participant recruited to the study was assigned a research number, and all outcome data entered were encrypted and stored without the subject’s name or address. Electronic study records were held on a Structured Query Language (SQL) server database managed by the Peninsula Clinical Trial Unit (PenCTU), which was stored on a restricted-access secure server maintained by the University of Plymouth. Data were entered into the database via a bespoke web-based data entry system, encrypted using Secure Sockets Layer (SSL). Access to electronic data was permission based, with access to identifiable information limited to those processing questionnaires and directly involved in data collection from participants. Data entered onto the database were backed up according to the PenCTU standard operating procedures. Copies of the study data, retained at study sites, were securely stored for the duration of the study prior to archiving at the University of Exeter. Electronic copies of correspondence sent to study participants were stored securely on a password-protected computer at the University of Exeter, and paper-based information held in a locked filing cabinet in the research team office at the University of Exeter Medical School. Names and participant details were not passed onto any third parties, and no named individuals were to be included in the write-up of the results. The only occasion in which personal information would be passed onto a third party would be if we considered there to be a risk of serious harm to a research participant, and normally this would occur only after discussion with the person concerned and with their GP (where appropriate).

All study data will be kept for 10 years under secure conditions on the University of Plymouth and the University of Exeter secure servers. We sought consent from participants for core outcome data from the pilot study to be released in an anonymised form to an independent data repository, following publication of the key planned outcome papers. We anticipate that for the data set containing the main study outcome variables, the approximate date of release to the archive would be 1 year following completion of the grant (to allow time for the publication of key outputs). Although anonymised, as a result of the sensitive nature of the data collected, the data will be shared under special licence for educational and research purposes only. The data would be released only to users, approved by the programme management group, following completion of user identification and intended use documentation administered through the data repository. Data would also be subject to standard secure storage and usage policies of the data repository.

Safety of participants and researchers

Day-to-day management

The chief investigator (JLC) and the scientific lead (SHR) supervised and monitored the strategic development and progress of the project, liaising with the wider team and methodological leads on a routine basis. The research staff members directly involved in data collection were based in Exeter. Two research fellows were appointed to manage the quantitative (observational study, RCT) and qualitative components of the study. Both research staff were responsible for managing the day-to-day aspects of the research relevant to their own role, supported by an administrator at peak periods of fieldwork activity. Katrina Turner provided qualitative expertise/supervision, Rob Anderson was the health economics lead, Obioha Ukoumunne was the statistical lead and Andy Gibson led PPI engagement.

Monthly project management meetings were organised to ensure that co-applicants and lay advisors (2–4 advisors) were informed of progress and contributed to the ongoing steering of the research. Face-to-face meetings were held every 3 months, with the remainder being held via teleconference. Co-applicants were expected to attend a minimum of one meeting every 3 months. Lay advisors were asked to attend meetings every 3 months, with their interim specific input sought as required via the PPI lead.

Trial Steering Committee

A TSC with an independent chairperson (Professor Glyn Lewis, triallist and psychiatrist) was appointed in accordance with MRC’s guidelines. Other TSC members were recruited with an appropriate range of perspectives and expertise. Members included two lay advisors (Rev. Paul Lanham and Mr Ron Grant), a cardiac rehabilitation expert (Professor Gill Furze), a qualitative methodology expert (Professor Anne Rogers) and a statistical expert (Dr Sally Kerry). Three TSC meetings were held at critical points in the project timeline (start-up, transition between feasibility and pilot study and study end). The TSC monitored the scientific quality, ethics and general progress of the project, receiving regular reports from the study team. This committee also monitored the serious adverse event and adverse event data emerging from the study. As no definitive data on clinical effectiveness/safety were collected, the TSC agreed that it would absorb the role of a Data Monitoring and Ethics Committee for the duration of the study.

Recruitment

Participant recruitment and retention

Participant recruitment for the feasibility study took place between September 2014 and March 2015. Participant flow from eligibility screening through to study recruitment and completion of the 5-month assessment is outlined in Figure 2.

FIGURE 2.

Feasibility study: participant flow diagram. CR, cardiac rehabilitation; DNA, did not attend.

The nurses screened 203 patients for study eligibility during their initial cardiac rehabilitation assessment, of whom 14 (7%) did not complete a PHQ-9. A total of 39 out of 203 patients (19%) were found to have depressive symptoms (i.e. score of ≥ 10 on the PHQ-9) during their initial assessment. Eight participants were already receiving active treatment for pre-existing depression and one was deemed ineligible for cardiac rehabilitation at the point of assessment, leaving 30 out of 203 participants (15%) meeting the CADENCE study definition of new-onset depression, who were therefore eligible for the trial. Of the 30 eligible patients, nine (30%) were recruited to the trial; this represents 4% of the 203 patients initially screened.

Five of the 30 participants (17%) eligible for the feasibility study were not offered study entry, the reasons for which are presented in Figure 2. Nine of those offered study entry (36%) agreed to take part and provided baseline data. All patients who were due to be followed up for the observational study completed their 5-month assessment. Based on these data, it took approximately 7.1 weeks per team to recruit each participant.

The majority of the 203 patients screened attended a community-based cardiac rehabilitation clinic (164 out of 203; 81%), with many fewer invited to attend a hospital-based service (39 out of 203; 19%). The mean age of patients attending an initial nurse screening clinic was 65.2 years (range 25–87 years), 157 out of 203 (77%) were male and the vast majority (198 out of 203; 97.5%) reported their ethnicity as ‘white’ (see Figure 2).

The sociodemographic characteristics of study participants (n = 9) recorded at baseline are presented in Table 3. The mean age of the sample recruited was 60.4 years, and most were male (7/9). All participants reported their ethnicity as ‘white’, and their preferred language was English. One-third of participants were working full time, with just under half currently in retirement. The majority of the sample were homeowners, and just under half were either married or living with a partner. The nature of the cardiac event that initiated referral to cardiac rehabilitation was recorded for eight participants. Three participants had a MI with PCI (insertion of stent, coronary angiography and left heart catheterisation), one had a MI, one had a MI with atrial flutter, two had a CABG, and one had a PCI (heart catheterisation) and left ventricular dysfunction.

TABLE 3 - Feasibility study: participant sociodemographic characteristics at baseline

max., maximum; min., minimum.

Characteristics	Baseline values (N = 9)
Gender (male), n (%)	7 (78)
Mean age (years) (SD; min., max.)	60.4 (18.2; 26, 85)
Ethnicity (white), n (%)	9 (100)
Preferred language (English), n (%)	9 (100)
Marital status, n (%)
Married/living as married	4 (44)
Single	2 (22)
Divorced	1 (11)
Widowed	2 (22)
Employment status, n (%)
Working full time	3 (33)
Retired	4 (44)
Unemployed jobseeker	1 (11)
Unemployed as a result of ill health	1 (11)
Housing status, n (%)
Homeowner	7 (78)
Tenant	2 (22)

Sample characteristics

Participant-reported outcome measures

At baseline, the CIS-R was used to determine a primary and secondary psychological diagnosis. Only two patients did not receive a diagnosis, with the other participants ranging in their diagnoses from mild depressive episodes to severe depressive episodes, and one participant having agoraphobia (Table 5).

Most participants (eight out of nine) reported that they felt that they had low mood, and when asked what professional help they would consider, around half indicated that a non-pharmacological approach was preferable (four out of eight), although two participants ultimately preferred not to receive any professional help.

The mean BDI-II score at baseline was 21.8 (SD 9.8) and the BAI score was 18.2 (SD 12.4). By the 5-month follow-up, participants reported lower mean levels of depression (BDI-II score of 18.0, SD 13.7) and anxiety (BAI score of 16.1, SD 14.3; Table 6). Four out of nine participants had a 50% reduction in scores from baseline at 5 months, and three out of eight were in remission (BDI-II score of < 14) at 5 months. Although all participants had reported not taking antidepressants at baseline, one participant reported taking antidepressant medication at follow-up for their low mood.

Most participants (eight out of nine) reported being ex-smokers at baseline, and at the 5-month follow-up, one participant had recommenced smoking (see Table 7). Participant-reported health outcomes, including self-reported cardiac diagnoses and other long-term health conditions, at baseline and the 5-month follow-up, are presented in Table 7. The EQ-5D-5L had a mean of 0.707 (SD 0.255) at baseline and 0.631 (SD 0.351) at 5 months; equivalent data for the mean VAS score were 55.7 (15.9) and 67.7 (21.2), respectively. The HeartQoL score had increased from a mean of 16.0 (SD 8.4) at baseline to 25.1 (SD 12.7) at 5 months.

Participants reported a positive experience of their cardiac rehabilitation programme, based on both the CSQ and FFT (Table 8), although one participant reported being mildly dissatisfied and ‘extremely unlikely’ to recommend the service to friends and family. Most participants reported that the quality of the service they received was excellent (six out of nine) and they were very satisfied with the amount of help they received (six out of nine). The majority of the participants reported being very satisfied overall (seven out of nine) and that they were willing to return to the programme in the future (eight out of nine). Six participants were extremely likely to recommend the service to friends and family (six out of seven).

Acceptability of study procedures

At follow-up, eight out of nine participants had completed participant-reported outcome data for all measures. In terms of the acceptability of study procedures, most participants reported that the baseline and follow-up assessments were of ‘the right length’ and had ‘the right number of questions’, although two participants felt that there were slightly more questions in the baseline assessment than they would have liked (Table 9). Most participants did not report noticing any repetition of questions at the baseline (four out of seven) and follow-up assessments (five out of seven).

Early feedback from nurses and the patient and public involvement group

The PPI group and two cardiac rehabilitation nurses attended a preliminary meeting to discuss the intervention design facilitated by the researchers (DAR and SHR). Lay advisors included two people who had used BA and two who had received cardiac rehabilitation. The PPI group offered insights into experiencing cardiac rehabilitation and/or BA. PPI advisors and nurses also discussed the potential, and challenges, of implementing BA and mental health-care co-ordination within cardiac rehabilitation services.

The relevant points arising from this meeting were supplemented with feedback from early meetings with individual cardiac rehabilitation nurse teams. Key areas were the realisation that community teams would be conducting nearly all the initial patient assessments, which is the point at which participants enter the study, and hence the community teams would have a much bigger role in recruiting participants than the hospital team; nurses being concerned about changing to the PHQ-9 and the GAD-7 from their current mood measurement tool (the HADS) and cardiac rehabilitation services differing in how they operated, and the need to accommodate these differences during implementation of the EPC intervention.

Observations of usual care in participating sites

The qualitative researcher conducted seven observational sessions relating to three teams during the early phase of the feasibility study (Table 10). Nurses working in different regions allowed patients a choice of venue for their rehabilitation, which also enabled nurses to steer patients towards the venue that they thought would be most suitable for the individual. In each region, nurses conducted their rehabilitation programmes in different ways; for example, one community nurse tended to run group fitness class sessions, and had less private space, whereas another community nurse had more clinic time and was able to offer more one-to-one clinic appointments. Box 2 details the insights gained from these observations.

Observations of nurse intervention training

The CADENCE training programme was delivered to four nurses from the three teams. Key points from the field notes taken during observations of nurse training were organised into three main categories: (1) general reflections on training (2) practical issues about the training materials and (3) intervention delivery issues. Nurses welcomed new knowledge that would help in managing patients’ mental health needs. Nurses gained new knowledge about the existence of, and how to refer patients to, existing community mental health services (e.g. IAPT) and on how to manage mental health risk issues when they arose. While undergoing the training nurses highlighted practical issues, such as the layout of the patient and nurse handbooks and how to make them more user-friendly. In terms of intervention delivery, nurses expressed concerns over how long BA would take to deliver in a typical cardiac rehabilitation session and the extra paperwork involved. Many of the points were resolved interactively during the training, whereas others (e.g. time concerns around implementation) would require monitoring and feedback once nurses implemented EPC.

Observation of clinical supervision sessions with nurses

Two telephone clinical supervision sessions with one nurse and two supervisors were observed, and six sets of supervision records were reviewed. These observations identified two key practical challenges. First, that the paperwork supporting nurse recording of EPC within the clinical notes required improvement to more accurately document the care received by the patient. For example, it was difficult to identify the number and length of BA sessions offered as part of the cardiac rehabilitation programme, and it was unclear what activities (e.g. symptom monitoring, diary review) constituted an adequate session of BA. Second, there was a need to develop procedures to ensure that nurses informed the clinical supervisors and research team when a participant was discharged from EPC. Embedded within this was ensuring that the nurse had a clear definition of when EPC ceased, as this might be different from when an individual stopped engaging with the wider cardiac rehabilitation programme.

Nurse interviews post training

All four nurses who attended the EPC training were interviewed within 4 weeks of completing their training. The interviews lasted between 25 and 36 minutes. Owing to the small number of nurses interviewed, to protect anonymity, quotations reproduced in this report from these interviews have not been tagged to indicate which interviewee is being quoted.

Nurses commented that the training was well organised, relevant and stimulating. The small group size (5–10 nurses was seen as optimal) provided a supportive and confidence-building environment. Nevertheless, the training was viewed as intense and thought-provoking (‘a good brain-ache’), with particular challenges being the initial understanding of the BA concept and how to explain BA to patients, and, later, how to explain and implement specific components of BA with patients. Role play was used throughout and, although daunting, the nurses commented on the benefits of this technique from observing and learning from each other.

Nurses were generally happy with the contents and completeness of the supporting materials provided. However, they found it difficult to differentiate between the participant handbook and nurse handbook, and requested that these were made distinctive in terms of colour/design. Nurses suggested having a few sentences and possibly some pictures in the nurse handbook that they could use with patients to help them describe/introduce BA to participants. They also requested access to electronic copies of the various materials, so that they could print copies off or e-mail them to participants. There was also some concern that the added BA materials might overload some participants, and nurses would need to make decisions about what to give out.

With regard to implementing EPC, prior to the training, nurses said that they had felt that they could provide only superficial support to patients experiencing low mood because they did not have the tools or resources to deal with it. After training, they were positive about the expected benefits of using BA with their patients:

I do a bit [working with patients with mental health issues] already but it tends not to, currently not to get to the nitty gritty of it but you just touch the surface with it and offer them some support.

Nurses expressed an eagerness to start using their new skills, but also concerns about how they would fit the BA sessions around their current workload. A few of the nurses had made plans about how they should implement the CADENCE programme within their rehabilitation sessions:

. . . we thought we’ll probably do the exercise sessions and then have the time with the client afterwards [yeah] when everybody else has gone [. . .] Rather than at the beginning where you might think, ‘oh god I’ve only got 20 minutes I’ve got to stop, no, you can’t talk anymore’, and that’s negative isn’t it?

Before participants were officially recruited to CADENCE, some of the nurses began using the PHQ-9 and the GAD-7 instead of the HADS. Unlike the HADS, the PHQ-9 included a question about suicide risk. The nurses commented that they preferred these measures and felt that the EPC training had given them the confidence to discuss mental health risk issues and a clear pathway to follow to manage these situations:

I wouldn’t have known what to do or what to say or [yeah] what actually was available for people so, and actually we never asked that question [no], we’ve never asked that question since I’ve been doing rehab[ilitation] so it is something that we should be . . .

. . . having a proper structured risk assessment and a plan of ‘what if’ just means that I wouldn’t have any worries talking about it.

Some nurses had not practised mental health-care co-ordination before, and several had not been aware that they were able to access IAPT services. Information on mental health-care co-ordination had provided nurses with options over referral, and it was evident that they now felt more able to direct patients:

I didn’t know anything about mental health-care co-ordination at all [no] before, not at all. I wouldn’t have known where to go and I wouldn’t have known what to do so it opened up a lot of avenues for us definitely.

. . . so I think in the programme, the training we had has already had an impact on enhancing the psychological services people will get through cardiac rehab[ilitation], which can only be a good thing.

Despite having been trained and being made aware that ongoing clinical supervision would be available, when imagining EPC working in practice, nurses remained concerned that they would not have enough time to deliver the intervention effectively:

. . . we’ve been given this great training and all of us we felt were very positive and we can imagine this really working, really, really well. But I think it would be really awful if we find then, well we can do this, but we can’t do it properly because we don’t have enough time.

Qualitative interviews after experiencing implementation of the intervention

Nurses and participants both commented that they felt that the EPC intervention was timely and appropriate within the context of cardiac rehabilitation. Participants explained how their cardiac event had left them feeling vulnerable and low in mood, and that they were glad to have the opportunity to discuss these feelings. Nurses remarked on how providing the intervention had provided them with new ways of supporting their patients, and the training had given them a pathway to deal more fully with individual’s mental health issues.

Refinements to the enhanced psychological care intervention prior to the pilot study

Several key refinements were made to the EPC intervention during the transition period. The most significant change was the fundamental redesign of the intervention from a nurse-led intervention, with significant emphasis on delivering the BA component within cardiac rehabilitation, to a patient-led, nurse-supported programme. The pilot intervention placed greater emphasis on the mental health-care co-ordination as the core component, rather than BA, and provided nurses with clear guidance on how to structure care. Details of the practical changes to procedures are described in Table 11. The EPC intervention tested within the pilot study is also described in detail in Enhanced psychological care.

Refining pilot study methods

Data from the qualitative study (see Table 11) were also used to refine study procedures prior to piloting. For example, to support the extraction of data relevant to the assessment of intervention fidelity, a structured form was developed (see Appendix 2) to assist nurses in recording key elements of the EPC intervention delivered to participants within each cardiac rehabilitation session. A team profile questionnaire was also developed, in order to capture information on the variability of service provision (e.g. team size, settings) across teams that elected to participate in the pilot trial. It was also apparent that even after a substantive redesign of the EPC intervention designed to reduce nurse workload, implementation issues would need to be revisited in qualitative interviews conducted with nurses and patient participants.

Data from the observational study also informed pilot study procedures in five main ways:

1. The eligibility criteria for participants recruited into the trial were broadened to reflect changing practice, whereby increasingly diverse patients were being referred for cardiac rehabilitation (i.e. patients with heart failure, not just patients with ACS).

2. The content of the case note review paperwork was revised to decrease the time burden on researchers collating detailed information around consultations with health professionals.

3. The information sought regarding participant employment and antidepressant use (via the SRUQ) was amended and simplified.

4. The assumptions underpinning the accrual of participants into the pilot RCT were deemed unrealistic based on new, previously unknown process data obtained on participant recruitment and retention (see Participant recruitment and retention). As a result, a working paper was drafted, outlining the implications regarding the design of the pilot study (see Appendix 9). In summary, we recommended that the 6-month recruitment period was not long enough to recruit sufficient participants to estimate the pilot study primary outcome (follow-up rate) with the required precision. After reviewing the sample size calculations and exploring different scenarios for our assumptions, we therefore recommended that the recruitment period be extended by 3 months (i.e. a total recruitment period of 9 months). Even with this extension, if the conditions of the feasibility study were replicated in the pilot trial, the likely accrual of participants into the pilot study might still result in the recruitment of only 43 participants (compared with the original target of 64). Despite our estimates based on feasibility data, the request to extend participant recruitment in the pilot study was rejected. As a consequence, a priori, we predicted the pilot study would fail to achieve its recruitment targets, anticipating recruitment within the region of 43 participants (half of whom would receive EPC).

5. The lower than expected participant accrual rate also had serious implications for the nested qualitative study. We had originally intended to undertake maximum variation sampling from a pool of EPC participants (see Data collection), to obtain diversity in terms of their sociodemographic characteristics, as well as their engagement (or otherwise) with EPC. In reality, the lower accrual rate meant that we had to invite most of the trial participants to take part in a qualitative interview.

Recruitment

Participant recruitment

Owing to limited numbers of patients within the individual cardiac rehabilitation teams, contrary to the statistical analysis plan (see Appendix 10), we did not attempt to report participant recruitment rates for each team on a weekly basis. We report aggregate data across teams within each trial arm for the number of participants who were eligible for, and subsequently recruited to, the trial. The flow of cardiac nursing teams and patients through the pilot stage of the CADENCE study is summarised in Figure 3. Cardiac nursing teams started patient recruitment on different dates depending on approvals (see Table 12), with individual teams recruiting for between 3 and 6 months during the period of June 2015 to 31 December 2015.

FIGURE 3.

Pilot study: CONSORT flow diagram (with cluster extension). CR, cardiac rehabilitation.

A total of 614 patients (389 in the EPC arm and 225 in the UC arm) were screened for eligibility at hospital and community-based cardiac rehabilitation clinics. The mean age of patients attending an initial nurse screening clinic was 66.2 years (range 32–95 years), with the majority being male (481 out of 614; 70.2%) and reporting their ethnicity as white (579 out of 614; 94.3%).

A total of 29 participants were randomised via their clusters (cardiac rehabilitation teams): 15 patients were randomised to the EPC arm and 14 to UC. The flow of participants at different stages of the trial is shown in Figure 3. Patients were screened at cardiac rehabilitation appointments, with 89 out of 614 (14.5%) scoring ≥ 10 on the PHQ-9 questionnaire, indicating depression. Of these 89 patients, 55 out of 614 were found to be eligible to participate in the trial (9%, 95% CI 7% to 12%); 33 were not eligible, as they were either actively being treated for depression (n = 21; 3%, 95 CI 2% to 5%) or were individuals with depression who were excluded for other reasons (e.g. alcohol and/or drug dependency problems, were currently ineligible for cardiac rehabilitation based on their cardiac condition or had language and communication problems).

Fifty-one of the 55 patients deemed to be eligible by a nurse were invited to participate. Of the four patients not invited, one patient declined any further input from the cardiac rehabilitation service, two patients were not offered EPC as the nurse was at capacity and could not take on more patients and the final patient reported a marked improvement in mood and did not feel that they required EPC. Of the 51 patients offered study entry, 21 (41%) declined participation for a variety of reasons including a preference to see their GP (n = 6); feeling better and would prefer to see how mood goes while taking part in cardiac rehabilitation (n = 3); that their mood was a long-standing problem unrelated to this current hospital stay (n = 2); or they did not want the EPC intervention (n = 2) (nine patients did not give reason). Of the 55 eligible patients, 30 out of 55 (54%, 95% CI 41% to 68%) were initially recruited (one of whom withdrew prior to providing baseline data). Four eligible patients, all in the EPC arm, were not invited to participate in the trial.

Of the denominator of 614 patients who attended an initial nurse assessment, the percentage recruited was 5% (95% CI 3% to 7%). Of the 29 participants who were recruited to the trial, 27 provided data at the 5-month follow-up (93%, 95% CI 77% to 99%). At 8 months, data were not able to be collected for 8 out of 29 participants because of late recruitment to the trial. Of the 21 participants for whom 8-month data could have been obtained, data were collected for 17 (81%, 95% CI 58% to 95%).

Sample characteristics

Sociodemographic characteristics, smoking status and baseline psychological descriptors of participants entered into the study are reported in Table 14. Both trial arms had approximately equal proportions of males and females; all participants were white and all apart from one participant reported that English was their preferred language. The mean age of the sample recruited was 62.7 years in the EPC arm and 68.1 years in the UC arm. There was, however, some imbalance between the arms, with the EPC arm having a higher proportion of participants in the three most deprived deciles (6 out of 15; 40%) than UC (2 out of 14; 14%).

In terms of depressive symptoms, the EPC group reported higher levels of depression at baseline than the UC group, as recorded by both the PHQ-9 (14.2 vs. 11.9) and the BDI-II (18.4 vs. 12.5) scores. At baseline, 10 out of 15 (67%) participants in the EPC arm had a BDI-II score of ≥ 14 compared with 7 out of 14 (50%) in the UC arm. The highest BDI-II score in the UC group at baseline was 20, whereas there were seven participants in the EPC group who had a score of ≥ 20 at baseline. Of note, participants from the more deprived backgrounds also had higher baseline BDI-II scores (see Table 14).

The majority of participants were former smokers (four in the EPC arm and five in the UC arm); none was current smokers. There was a wide range of time since quitting smoking, with some participants having quit several years ago and six participants having quit < 6 months prior to entering into the study.

A post hoc comparison was conducted on the sociodemographic characteristics (age, gender and ethnicity) and PHQ-9 scores of patients who were invited to participate but who declined (where sufficient data were recorded on the screening logs). Of the 51 eligible participants who were offered study entry, 15 out of 29 (52%) male patients and 15 out of 22 (68%) female patients accepted. Of 49 eligible white patients invited to participate, 19 (39%) declined, as did both of the two Asian/Asian-British patients. The mean age among 30 eligible patients who accepted participation was 63.8 years (SD 9.5 years), range 47–78 years; among 21 patients who declined, mean age was 61.1 years (SD 12.3 years; range 36–83 years). For the acceptors, mean PHQ-9 score was 13.4 (SD 4.1; range 10–25); for the decliners, mean PHQ-9 score was 14.4 (SD 4.6; range 10–24).

The indicating cardiac events, as recorded in GP or cardiac nurse records, and participant self-reported cardiac conditions at baseline are recorded in Table 15. Trial participants had experienced a wide range of cardiac events that precipitated their referral to cardiac rehabilitation according to their clinical records, the most frequent of which were MI (STEMI or NSTEMI) and PCI.

Enhanced psychological care intervention delivery

Seven nurses from four teams provided EPC to one or more participants (one team recruited no participants to the study). Three teams had two nurses providing EPC and one team had a single practitioner. Most nurses had their own participants allocated to them, with some shared cases when the index nurse was unavailable.

Thirty supervision calls were made to the teams over a 7-month period (August 2015 to February 2016). Some teams opted for joint supervision (i.e. both nurses listening in to all cases), whereas others had individual calls related to their own cases, and then handed the telephone to the other team nurse.

Participant-reported outcome measures

Psychological outcomes

The participant-reported outcomes relating to their mood at each follow-up point are reported in Tables 20 (5 months) and 21 (8 months), and Figure 4. In terms of depressive symptoms, at the 5-month follow-up, the mean BDI-II score was reduced from baseline in both groups. Three participants in the EPC arm did not report a BDI-II score at the 5-month follow-up (baseline BDI-II scores were 14, 25 and 31). For 26 participants, the mean difference in BDI-II scores at the 5-month follow-up (with adjustment for baseline score), comparing EPC with UC was 1.7 (95% CI –3.8 to 7.3). For the 17 participants available for analysis, by 8 months the mean difference in BDI-II scores (EPC vs. UC) was 4.4 (95% CI –1.4 to 10.2).

TABLE 20 - Pilot study: participant-reported mental and physical health outcomes at baseline and 5 months

IQR, interquartile range; max., maximum; min., minimum; OR, odds ratio.

Between-group comparison. The mean difference is adjusted for baseline score.

At baseline, 7 out of 14 UC participants and 10 out of 15 EPC participants had a BDI-II score of ≥ 14; participants with scores of ≤ 13 at baseline were excluded.

CSRC can be calculated only for participants with a baseline BDI-II score of ≥ 14.

Outcome variable	Follow-up time point in each trial arm				Mean difference or ORa (95% CI)
	Baseline		5 months
	UC (N = 14)	EPC (N = 15)	UC (N = 14)	EPC (N = 13)
BDI-II
Mean score (SD), n; min., max.	12.5 (4.9), 14; 4, 20	18.4 (7.4), 15; 7, 31	7.4 (3.7), 14; 2, 14	10.2 (8.4), 12; 2, 33	1.7 (–3.8 to 7.3)
50% reduction in score from baseline, n/N (%)	–	–	7/14 (50)	5/12 (42)	0.71 (0.16 to 3.25)
Remission (score of < 14) from baseline,b n/N (%)	–	–	6/7 (86)	3/7 (43)	0.13 (0 to 1.37)
MCID in score (17.5% reduction from baseline), n/N (%)	–	–	11/14 (79)	10/12 (83)	1.36 (0.22 to 8.31)
CSRC in score from baseline,c n/N (%)	–	–	3/7 (43)	2/7 (29)	0.53 (0.07 to 4.33)
BAI
Mean score (SD), n; min., max.	12.5 (4.4), 14; 4, 21	19.3 (10.7), 15; 4, 36	9.2 (4.0), 14; 4, 15	14.7 (8.3), 13; 3, 27	4.6 (–0.8 to 10.0)
EQ-5D-5L
Mean score (SD), n; median (IQR)	0.869 (0.085), 14; 0.893 (0.866–0.924)	0.751 (0.182), 15; 0.751 (0.652–0.886)	0.875 (0.115), 13; 0.901 (0.843–0.951)	0.885 (0.065), 13; 0.866 (0.828–0.924)	0.045 (–0.023 to 0.113)
EQ-5D VAS
Mean score (SD), n; min., max.	63 (24), 14; 4, 95	49 (19), 15; 10, 100	72 (21), 14; 30, 100	72 (21), 12; 30, 100	5 (–13 to 22)
HeartQoL
Mean score (SD), n; min., max.	21.3 (9.1), 14; 5, 37	15.2 (9.9), 15; 1, 35	31.5 (4.9), 14; 23, 38	22.9 (10.3), 13; 4, 41	–8.2 (–14.9 to –1.4)

TABLE 21 - Pilot study: participant-reported mental and physical health outcomes at baseline and 8 months

IQR, interquartile range; max., maximum; min., minimum; OR, odds ratio.

Between-group comparison. The mean difference is adjusted for baseline score.

At baseline, 7 out of 14 UC participants and 10 out of 15 EPC participants had a BDI-II score of ≥ 14; participants with a score of ≤ 13 at baseline were excluded.

CSRC can be calculated only for participants with a baseline BDI-II score of ≥ 14.

Outcome variable	Follow-up time point in each trial arm				Mean difference or ORa (95% CI)
	Baseline		8 months
	UC (N = 14)	EPC (N = 15)	UC (N = 8)	EPC (N = 9)
BDI-II
Mean score (SD), n; min., max.	12.5 (4.9), 14; 4, 20	18.4 (7.4), 15; 7, 31	7.0 (3.5), 8; 1, 12	12.6 (6.6), 9; 1, 22	4.4 (–1.4 to 10.2)
50% reduction in score from baseline, n/N (%)	–	–	4/8 (50)	2/9 (22)	0.29 (0.04 to 2.09)
Remission (score of < 14) from baseline,b n/N (%)	–	–	4/4 (100)	2/5 (40)	0 (0 to 1.10)
MCID in score (17.5% reduction from baseline), n/N (%)	–	–	5/8 (63)	6/9 (67)	1.20 (0.18 to 7.99)
CSRC in score from baseline,c n/N (%)	–	–	2/4 (50)	1/5 (20)	0.25 (0 to 3.56)
BAI
Mean score (SD), n; min., max.	12.5 (4.4), 14; 4, 21	19.3 (10.7), 15; 4, 36	6.4 (4.6), 8; 0,12	12.7 (7.1), 9; 4, 24	5.0 (–1.2 to 11.1)
EQ-5D-5L
Mean score (SD), n; median (IQR)	0.869 (0.085), 14; 0.893 (0.866–0.924)	0.751 (0.182), 15; 0.751 (0.652–0.886)	0.876 (0.092), 8; 0.868 (0.830–0.951)	0.827 (0.116), 9; 0.874 (0.697–0.951)	–0.041 (–0.145 to 0.064)
EQ-5D VAS
Mean score (SD), n; min., max.	63 (24), 14; 4, 95	49 (19), 15; 10, 100	64 (18), 7; 40, 90	64 (12), 8; 50, 80	1 (–17 to 19)
HeartQoL
Mean score (SD), n; min., max.	21.3 (9.1), 14; 5, 37	15.2 (9.9), 15; 1, 35	28.3 (6.7), 8; 17, 36	28.6 (7.6), 9; 16, 36	0.1 (–7.9 to 8.0)

FIGURE 4.

Pilot study: BDI-II scores for participants with baseline and at least one follow-up observation. (a) UC; and (b) EPC.

Around half the participants with data available at both baseline and 5 months (n = 26) had a 50% reduction in BDI-II scores at 5 months: 5 out of 12 (42%) in the EPC arm and 7 out of 14 (50%) in the UC arm, yielding an odds ratio of 0.71 (95% CI 0.16 to 3.25). A 17% reduction from baseline symptoms (the MCID) was observed for 10 out of 12 participants in the EPC arm (83%) and 11 out of 14 participants in the UC arm (79%); by 8 months, this was 6 out of 9 (67%) and 5 out of 8 (63%), respectively.

In the EPC arm, seven participants with a BDI-II score of ≥ 14 at baseline also had follow-up data at 5 months, of whom three (43%) were in remission (i.e. with a BDI-II score of < 14). Only seven participants in UC had a baseline BDI-II score of ≥ 14 and follow-up data at 5 months, of whom 6 (86%) were in remission at 5 months. Among these participants, in the EPC arm 2 out of 7 (29%) had a CSRC, compared with 3 out of 7 (43%) in UC.

The maximum BDI-II score at the 5-month follow-up was 14 in the UC arm. In the EPC arm, there were two participants who scored 15 and one participant who scored 33 (this participant scored 25 at baseline). Therefore, overall, in both groups, with the exception of one participant in the EPC arm, all 5-month BDI-II scores were in the moderate depression range or below.

At 8 months, the highest BDI-II score in the UC arm was 12, whereas in the EPC arm there were five participants with a score of ≥ 14. The highest score in this group was 22; this participant scored 31 at baseline and did not provide sufficient BDI-II data at 5 months. The participant who scored 33 at 5 months scored 18 at 8 months.

In terms of anxiety, the mean BAI score was also reduced at 5 months in both groups, with a between-arm mean difference between EPC and UC of 4.6 (95% CI –0.8 to 10.0). By 8 months, the between-group difference was 5.0 (95% CI –1.2 to 11.1).

Roughly equal numbers of participants in each arm believed that they had low mood, and there was a general preference in both groups to receive non-drug support for their low mood.

Cost of delivering the intervention

The estimated cost per participant receiving the EPC intervention, and the main elements of this cost, are presented in Table 23.

The calculation presented in Table 23 makes no adjustment for the small scale of the pilot trial, and so it is inevitably an overestimate of what the actual cost would be if the intervention were to be delivered routinely to larger numbers of patients. For example, in this calculation, the initial training of nurses accounts for a significant proportion of the overall cost of providing EPC, but this would, in reality, be shared across a larger number of patients receiving EPC over a longer period.

The estimated time nurses take to deliver EPC elements within rehabilitation accounts for a relatively small amount of the overall cost (£63 of the estimated £959 per participant). This was estimated by trial researchers from the information that has been collected to describe the delivery of the intervention (see Appendix 3). However, the qualitative interviews with nurses, conducted as part of this study’s process evaluation, showed clearly that the nurses found it extremely difficult to distinguish clearly their EPC-delivering time from other aspects of the care they usually provide. Nevertheless, we believe that the cost breakdown shown may help those developing and evaluating similar nurse-delivered interventions in hospital and community settings to estimate the cost of their intervention. The estimated cost of designing and developing the intervention, including intervention manual development and the creation of related learning materials, has been separately estimated and is available from the authors.

Cost of health service use

Table 24 shows the cost of health service use by participants in the trial, based on self-reported service use in the two time periods (baseline to 5 months, and 5 months to 8 months). This analysis provides an overall indication of the types of health service use that accounted for the majority of health service costs incurred in this group. All participants reported using GP services in both periods, and the majority also had hospital outpatient appointments. Although a minority had hospital inpatient stays, these account for a large proportion of the total costs of health-care use.

Owing to the small sample size, any differences in costs between the trial arms should be ignored, as they could easily be chance findings. For example, the > £700 difference in mean hospital inpatient costs between the EPC group and the UC-only group (baseline to 5 months) is almost wholly attributable to one participant in the EPC group who reported having two 10-day hospital admissions as a result of asthma attacks during that period (total estimated cost: £10,518). This is not likely to be due to the EPC intervention.

However, these costs could provide useful point estimates and SDs for planning future definitive randomised trials of similar interventions in similar patient groups.

Health-related quality-of-life and utility scores

The EQ-5D-5L index scores at baseline, 5 months and 8 months, calculated using the most recent and reliable UK population-derived preference weights for health states,88 are presented in Tables 20 and 21. Apart from the previously noted difference at baseline between the two trial groups, there were no notable differences or changes over time. If a definitive trial were to be conducted, these quality-of-life estimates could be used to determine a trial sample size based on the expected cost-effectiveness of the intervention.

Comparing self-report service use data with data from clinical and service records

As well as participant-reported health service use data, the study also gathered selected health service use data from clinical records held by the participant’s general practice, and those held by the participant’s cardiac rehabilitation nurse. Owing to the curtailed follow-up of some participants at 8 months, combined with study attrition, the self-reported service use data are available for small samples. Although service use data from GP records were available from all but two trial participants, complete self-reported service use data (for both baseline to 5 months and 5 months to 8 months) were available for only 17 of the 29 study participants.

For the 17 participants with data from both sources, Table 25 summarises the level of agreement in the number of services used between participant-reported data and GP records. There was poorest agreement between self-reported service use data and GP records for the number of visits to GPs or practice nurses, with no participants recalling the same number of GP visits as their GP records showed, and only 3 out of 17 participants (18%) recalling the same number of practice nurse visits. This might be expected for these frequent types of health service use. Further inaccuracy is added when asking about the type of health professional seen (whether practice nurse or GP).

For the less frequent, and more significant and costly, occurrences of accident and emergency (A&E) attendances and hospital inpatient stays, the level of crude agreement between the two data sources was much higher, with 11 out of 17 participants (65%) recalling the same number of A&E attendances as their GP records show, and 14 out of 17 participants (82%) recalling the same number of hospital inpatient stays.

For all four types of health service use/episode, there seems to be no systematic tendency to either overestimate or underestimate the number of episodes of service use (with similar numbers and proportions over- and under-estimating). This partly explains the broadly similar mean number of episodes of each type. For example, although none of the participants recalled the same number of GP visits as their GP records showed, the mean number of GP visits was the same whether calculated from self-reported data or GP-recorded data. For the potential purposes of costing within an economic evaluation, this provides some evidence to support the use of self-report service use data in this patient group (and over these recall periods), so long as loss to follow-up and questionnaire non-completion could be minimised.

The cardiac rehabilitation nurse notes only consistently recorded the use of other hospital services and, thus the level of agreement could be estimated only for the number of hospital admissions (inpatient stays) reported by participants and nurse records. Overall, there was a high level of crude agreement, with 15 out of 17 (88%) agreement between participant reports and nurse records of inpatient stays for the 8-month follow-up period. The mean number of self-reported hospital admissions was 0.5 (SD 0.9), compared with 0.3 (SD 0.6) admissions, as recorded in the nurse notes. The intraclass correlation coefficient, reflecting the strength of association, was 0.55. No participants underestimated the number of hospital stays, although two participants (12%) overestimated the number of hospital admissions.

Patient interviews

The purpose of these interviews was to gather qualitative evidence on the content of usual psychological care and to assess the acceptability of EPC and study materials to patients. Interviews were conducted once participants had completed their 5-month assessment for the trial, and most had completed their fitness programme (including EPC where relevant).

Nurse interviews

Feasibility study

We successfully recruited three cardiac rehabilitation teams in the south-west of England, drawing on local expert knowledge of services and the geographical accessibility of those services. Following screening of 203 potential participants, 30 individuals with ACS (15%) met our eligibility criteria for admission to the study and, of these, nine agreed to take part. All nine participants completed a follow-up assessment at 5 months. These participants were somewhat younger than the overall group of patients attending initial screening by cardiac nurses, but all provided routine sociodemographic information, and all undertook the baseline CIS-R assessment used to determine primary and secondary psychological diagnoses.

In the feasibility study, we restricted follow-up to a single end point at 5 months, and all participants provided psychological outcome data and data related to smoking status and a range of health outcomes at baseline and follow-up. All feasibility study participants provided information at follow-up regarding their experience of care. The burden of study procedures appeared to be broadly acceptable to patient participants.

Early direct observations of the cardiac rehabilitation nurses delivering UC indicated variations across cardiac rehabilitation teams in respect of service delivery. Variation was also observed in the facilities available for cardiac rehabilitation, the number of patients participating in group sessions and the availability of private space.

Participating cardiac nurses were successfully trained in delivering the EPC intervention, and in monitoring and managing patients with depression who were attending cardiac rehabilitation. That training was supported by the development of a nurse training manual, the content of which highlighted the key elements of EPC regarding mental health-care co-ordination and BA. Observations of nurse intervention training indicated that nurses welcomed their new knowledge and skills. Issues were identified relating to the layout of the patient and nurse handbooks, which were used to guide the content and delivery of the intervention, and there were early concerns expressed in respect of how long EPC might take to deliver in a typical ‘routine’ cardiac rehabilitation session. Observations of clinical supervision sessions with nurses identified concerns relating to the quality of paperwork initially provided to document the care given to patients, and a need was also identified to develop procedures to alert clinical supervisors and the research team when a patient was discharged from EPC.

Interviews conducted with patients as part of the feasibility study indicated that participants felt comfortable with nurses and valued having dedicated one-to-one time to talk about their mental health. Some identified the rather uncomfortable sense of having been ‘singled out’ for treatment – and, in this context, some patients felt that a group approach to delivering the intervention might have been preferable.

Interviews conducted with nurses following the training sessions highlighted overall satisfaction with the organisation and delivery of the training, and with the content and completeness of the supporting materials provided. Nurses reported that they felt that the training had been effective, resulting in greater confidence in the management of patients with psychological problems. Nurses reported benefit from the newly acquired ability to undertake a structured assessment of the patient’s psychological status, including the assessment of suicide risk. Notwithstanding this, nurses identified significant challenges in integrating the delivery of EPC within routine cardiac rehabilitation because of time pressures and the availability of suitable space to allow effective delivery of the EPC intervention. Thus, nurse feedback clearly articulated that the model of EPC tested within the feasibility study was unsustainable within routine cardiac care.

Transition between feasibility and pilot studies

Findings from the feasibility study directly informed the design and delivery of the subsequent pilot study. In particular, aspects of the EPC intervention were refined and redesigned prior to the pilot study, and a range of practical refinements of the pilot study research procedures were proposed.

It was clear that issues pertaining to nurse workload were a major concern, particularly regarding the ability of nurses delivering routine cardiac care to adopt and implement additional work relating to delivery of the intervention and the research procedures. For this reason, we refocused the EPC intervention. The change in the delivery of EPC between the feasibility and pilot phases involved an increased emphasis on the mental health-care co-ordination component of EPC and a shift from a nurse-led intervention to a patient-led, nurse-supported programme of self-help BA embedded within cardiac rehabilitation, reflecting the need to reduce the impact of EPC on nurse time, albeit with nurses trained to support participant engagement. These changes were expected (and indeed shown) to reduce length of session, although it remains possible that the reduction in nurse input may reduce the effectiveness of the intervention.

In the transition period, the eligibility criteria for participants recruited into the pilot trial were broadened, reflecting changing practice within the wider NHS cardiac rehabilitation services through which individuals with other cardiac conditions are increasingly being managed. In reality, this meant that although participants were recruited if they had experienced some form of ACS, there was an increasing likelihood that they would also have other existing cardiac comorbidities (e.g. atrial fibrillation, heart failure). In addition, data collection procedures were refined, with the case note review paperwork and information on antidepressant use being simplified, and additional information being sought regarding participant employment status. As the nurse handbook and participant handbook developed for the feasibility study were well received, practical suggestions emanating from the feasibility study informed the revision and some redesign of these materials (e.g. using colour coding and improving the scripts provided to introduce concepts).

Of significant importance, the assumptions underpinning the accrual of participants into the planned pilot RCT were deemed unrealistic when considered in the light of the findings of the feasibility study. This was carefully considered and reviewed in a paper prepared for the TSC scheduled during the transition period between feasibility and pilot studies. A proposal was submitted to the funders with a view to securing an extension of funding to allow an additional 3-month period of recruitment in the pilot trial, and an additional 3-month period for consideration of the findings and writing-up of the results. This funding request was not agreed. Before commencing the pilot trial, we therefore anticipated that we would not achieve our original pilot study recruitment target, instead anticipating recruitment of around 43 participants (instead of the original 64 targeted). This revised participant accrual rate had implications for our sampling approach within the qualitative study. Thus, we revised our pilot study protocol accordingly, by altering the randomisation ratio from 1 : 1 to five EPC teams to three UC teams to ensure that a sufficient number of participants would experience the EPC intervention.

External pilot trial

The aim of the pilot trial was to test the methods and procedures required to undertake a fully powered evaluation of the clinical effectiveness and cost-effectiveness of cardiac rehabilitation teams implementing EPC for individuals with new-onset depressive symptoms using cardiac rehabilitation services, compared with a comparator of usual cardiac rehabilitation care. We successfully recruited eight cardiac rehabilitation nursing teams, randomising these to one of two trial arms: UC or EPC plus UC. Teams were drawn from a range of settings in South West England and the East Midlands. Most teams recruited offered centre-based rehabilitation programmes. The time of our first approach to the teams to the point of participant recruitment was lengthy (9 months in one area). In two regions, participant recruitment was delayed by over 2 months from the point that all necessary research training had been completed, as we waited for NHS R&D trust approvals to be issued. This effectively reduced the recruitment window from 6 months to 4 months for this team, placing increasing pressure on our ability to achieve our target sample size.

Of the 614 cardiac patients screened for eligibility by the eight teams within a 4- to 6-month recruitment period, 55 patients were eligible to take part in the trial and, of these participants, 29 provided baseline data. Of these 29 participants, 27 (93%) provided 5-month data. Of the 21 participants eligible to provide 8-month data, 17 (81%) did so. Participants in both trial arms were broadly comparable in respect of the range of cardiac diagnoses/events that had precipitated their referral for cardiac rehabilitation.

Twenty-one patients were depressed on assessment at baseline screening by nurses, but were not eligible for inclusion on account of having pre-existing treatment for depression prior to referral to cardiac rehabilitation. The requirement for inclusion of only patients with ‘new-onset’ depression (see Rationale for current research) was set out in the original commissioning brief, and this is an area that future research commissioners might want to revisit.

Cluster randomisation, balanced by team type and patient throughput at the initial cardiac rehabilitation assessment, was successfully implemented, although one randomised team withdrew immediately following randomisation on account of coincidental service changes (a regional reconfiguration of cardiac rehabilitation service provision), which made participation impractical. At short notice, the withdrawn team was replaced with another team matching closely in terms of patient throughput and population mix.

The two trial arms were not well balanced in certain patient characteristics. Participants in the UC arm were older than those in the EPC intervention group. Participants in the EPC arm were more likely to be from deprived backgrounds, to evidence higher levels of depressive symptoms at baseline, and to report lower levels of HRQoL than participants receiving UC. Given the small number of participants and the fact that randomisation was at the level of the cluster, a balance of participant characteristics between arms would not have necessarily been expected. We anticipate that this would not prove to be an issue of concern in a larger, fully powered study.

Two participants (one in the EPC arm and one in the UC arm) were prescribed antidepressants at a therapeutic dose during the study follow-up period. In addition, three participants were prescribed amitriptyline in small doses, in the range usually prescribed by GPs for pain or sleep disturbance. Amitriptyline prescribing is potentially a cause of concern in this group of patients, as older forms of tricyclic antidepressants such as these are generally considered to be relatively contraindicated on account of the presence of cardiac disease.

We relied on case note review of cardiac nurse and GP notes to obtain certain clinical data that might reasonably be assumed to be comprehensively captured in such records. The availability of physiological and biochemical outcome data varied among participants, generally with greater availability of physiological data in GP or cardiac rehabilitation notes than of biochemical data (lipid profile). The cardiac diagnoses/procedures that precipitated referral to cardiac rehabilitation were also inconsistently reported within nurse and GP notes. A future trial should consider revising procedures to ensure that these data are adequately captured by, for example, seeking further data from cardiology services.

Participants in both trial arms were offered attendance at broadly similar numbers of rehabilitation sessions, and similar proportions of participants in both treatment arms attended these sessions. However, four out of the five participants in the intervention arm who attended one or fewer cardiac rehabilitation sessions also had higher baseline depression scores (BDI-II scores of ≥ 14). This suggests that some patients who might have benefited from EPC did not take up that opportunity, although the reasons for this were not clear. In addition, some patients were not offered EPC in the pilot because of the nurse workload, with the nurse lacking capacity to offer EPC to more than three participants at any given time. Patient factors may also be important. Qualitative interviews with trial decliners found that some patients chose not to take part in the trial as they did not identify themselves as experiencing low mood or were not keen to receive more information relating to their cardiac event. We also noted that 4 out of 10 EPC participants who attended at least six rehabilitation sessions had baseline scores below the threshold BDI-II score of 14, indicative of ‘mild’ depression.

Participants’ use of resources was collected from both self-report questionnaires completed by participants and a structured review of both GP routine records and cardiac rehabilitation records. Good or very good agreement between patient and either type of record was observed regarding A&E attendance and hospital admission, but individual-level agreement was poor in respect of the number of primary care appointments attended (when compared with GP records). Differences in the mean number of GP attendances between GP record and self-report data, however, were more similar. For the potential purposes of costing within an economic evaluation, this provides some evidence to support the use of self-report service use data in this patient group (and over these recall periods) – so long as loss to follow-up and questionnaire non-completion could be minimised.

Across the 5-month follow-up period, the mean BDI-II score improved from baseline in both arms of the trial, with the mean between-group difference in BDI-II score (EPC minus UC, adjusted for baseline scores) of 1.72 units, but with a wide CI, encompassing zero. We were able to estimate the distribution of BDI-II scores in our study population, although, being a single study, future research might want to review relevant available evidence from similar or related populations. 82 We also reported additional data on different ways to calculate and compare depression status at follow-up (e.g. remission, MCIDs). Participants in both trial arms reported improved scores for anxiety and heart-related quality of life at the 5-month follow-up, although generic HRQoL, as measured by the EQ-5D-5L, remained unchanged. However, the pilot study sample size was small and likely underpowered, and this, combined with evidence of baseline imbalance between the two trial arms and the lack of adjustment for clustering in the data, does not allow us to draw any formal inferential conclusions regarding the overall direction or magnitude of outcome changes within or between arms.

Perhaps reflecting the nature of the underlying morbidity, two individuals had recommenced smoking, one in each trial arm, by 8 months. Participants in both trial arms reported improved scores for anxiety and heart-related quality of life at 5 months, although generic HRQoL, as measured by the EQ-5D-5L, remained unchanged. Although participants in both arms reported a positive experience of their cardiac rehabilitation, there was some evidence of a more positive experience among participants of EPC when compared with UC.

From semistructured interviews with trial participants, it was clear that they viewed one-to-one dedicated nurse time for psychological care as important to achieving timely recovery from their depression. Participants saw EPC within cardiac rehabilitation as reflecting a ‘holistic approach’ to care, and the delivery of EPC occurring shortly after the cardiac event as important. Despite this, it was evident that a clearer explanation of BA as a key component of EPC was needed, and participants also made it clear that a flexible approach to delivering the intervention was needed as, in the views of at least some participants, ‘one size does not fit all’. A small number of interviews with patients who declined participation identified that their reasons for declining included not feeling that low in mood (despite having questionnaire scores indicative of depression), not being keen to discuss mood with health-care professionals and having concerns regarding ‘information overload’ at the point of attending cardiac rehabilitation.

Interviews with nurses showed that they felt appropriately equipped to deliver the EPC intervention in spite of only limited experience in delivering EPC to patients over the course of the pilot trial. The training in mental health-care co-ordination (including clinical supervision during implementation) and adoption of the PHQ-9 to measure depressive symptoms (replacing the HADS) was well received by nurses, and was collectively viewed as improving their ability to effectively manage mental health and risk. BA was seen by nurses as a viable intervention, although one that was challenging to deliver to some patients and, in particular, to some individuals who were seen as not being receptive to the approach offered. Although there was a general perception of nurses that many patients seemed to benefit from EPC, this was tempered by a view that more of their dedicated time was needed to allow the effective delivery of EPC.

Methods to support economic evaluation

As for the other components of the pilot trial, a principal aim of the economic component was to establish the data collection methods required to support a definitive economic evaluation. We piloted methods required to collect resource and service use data among trial participants that would be relevant to costs to the NHS, costs to social care and personal social services and costs to participants and their carers/families. Service use data were collected from routine/administrative sources, as well as via participant self-report, using a SRUQ (the latter being adapted from the CSRI, following input from the PPI group). Our study identified the estimated cost per participant receiving the EPC intervention, and identified the main elements contributing to this overall estimate of cost. We identified a total overhead cost of providing EPC of £13,384; a substantial proportion of this cost (93%) was accounted for by nurse training. Given the relatively small number of participants, the estimated cost per participant of £904 is likely to be an overestimate, on account of small numbers. Given the nature of the intervention, it proved difficult to identify time spent by nurses in ‘delivering EPC’, as this time was distributed across all of the nurse interaction with the participant rather than in discrete, quantifiable amounts of consultation time.

We successfully gathered information from participants regarding their use of health services across their period of involvement with the study, with data available for the majority of participants. Therefore, patient self-report service use data may be sufficient for providing mean attendance rates in trial arms. The five main categories of health service use on which we captured administrative record data appeared to be intuitively correct, although data collection was intensive, requiring significant researcher time and effort to conduct manual case note review of, and data extraction from, GP and cardiac nurse records. Although we did observe differences between trial arms in resource use estimates, we interpret this with considerable caution and would make no inference regarding the magnitude or direction of any differences on account of the small participant numbers involved. In addition, it is important to separately identify high-cost care episodes (e.g. inpatient admissions) that are unlikely to be related to the intervention or the targeted conditions, but which could cause spurious cost differences between trial arms; therefore, we captured the reasons for such care episodes. The assessment of HRQoL and utility scores using the EQ-5D-5L index scores was undertaken successfully and would have the basis, in a larger trial, of providing quality-of-life estimates that are potentially useful for determining trial sample size, based on the expected cost-effectiveness of the intervention in this population of participants.

Strengths

The CADENCE study brought together an experienced team of researchers and mental health clinicians who designed and delivered a successful mixed-methods feasibility study and pilot trial in response to a national commissioning brief issued by the NIHR. In general, the aims of the feasibility and pilot study were achieved. We successfully recruited three cardiac rehabilitation centres to the feasibility study and eight centres to the external pilot trial (and an additional replacement pilot trial site). Nurse training was provided in respect of EPC for patients attending cardiac rehabilitation with depression.

Our qualitative data found that the training was welcomed by the cardiac nurses, who saw this as an extension of their clinical skill base and knowledge. It was clear that ‘usual psychological care’ was highly variable between teams. As a consequence of intervention training, the assessment of psychological symptoms and the risk assessment of potentially suicidal patients were welcomed. The intervention was both developed and manualised by drawing on a theoretical base previously acknowledged in NICE’s guidelines33,34 for the management of depression in people with long-term conditions. The intervention was developed by experts with experience of training NHS therapists in BA working closely with cardiac nurse specialists, and with the intervention involving regular review of the patient’s psychological state, risk and regular clinical supervision available to nurses implementing the intervention. The trial was supported by the development of two manuals (targeting nurse and patient participants), both of which were subsequently refined following feedback from patients within the trial and our PPI group, as well as from participating nurses delivering the intervention. Data from qualitative observations and interviews provided a deeper insight and understanding of our quantitative findings at each stage of the study. The manuals were modified in the light of feasibility findings in preparation for the external pilot trial. Each EPC session was structured to include the monitoring of depression and risk by nurses, participant self-monitoring (including patient self-identification of activities and triggers to depressive behaviours and thinking) supporting the replacement of negative behaviours in patients with more positive behaviours and the use of mental health-care co-ordination followed an algorithm developed for the study.

Overall, 29 participants were recruited into the pilot trial, 27 out of 29 participants (93%) were successfully followed up at 5 months and 17 out of 21 participants (81%) provided data at the 8-month follow-up. Although we did not meet our original recruitment target (64 participants, subsequently revised to 43 participants), we do believe that recruiting and retaining sufficient participants is possible within the context of a future, larger, definitive, fully powered RCT. The EPC intervention generally appeared to be acceptable to participants, with some evidence of higher satisfaction among participants receiving EPC than among those receiving UC.

In line with the commissioning brief, our intention was to recruit patients attending cardiac rehabilitation who had evidence of new-onset depression occurring since the cardiac event. Although the overall prevalence of depression in this group of patients was in line with expectations, approximately one-quarter of patients (in both the feasibility and pilot studies) had symptoms of depression when initially attending for cardiac rehabilitation and had received treatment for depression in the 6 months preceding the cardiac event, and thus were deemed ineligible for the study. In addition, a small number of depressed individuals were not offered study entry, as the nurse had effectively ‘reached capacity’ in terms of the number of participants receiving EPC at a particular point in time. We would propose that, in the planning of any future research, consideration be given to extending recruitment to include all patients with depression attending cardiac rehabilitation.

Limitations

Prior to undertaking this work, and during the course of preliminary funding consideration, we agreed a change from the NIHR commissioning brief, which had originally envisaged a 1-year follow-up period. Projected study costs and anticipated attrition among this vulnerable group of patients led to this agreement, and our study design involved primary outcome data collection at 5 months and a second (final) follow-up at 8 months. However, the CADENCE study proved to be a challenging study, with tight timelines making necessary changes to the protocol difficult to manage.

Delays encountered in securing trust R&D approval meant that, on occasion, we had nurses who were trained in delivery of the intervention but who were unable to deliver the study until local trust processes had been addressed (these delays related to the negotiation of excess treatment costs). 103 These delays were, in some settings, substantial, and were a source of frustration to the research team and the nurse participants. More significantly, these delays had the potential to undermine the effective delivery of the intervention. Some nurses experienced delays between training and operationalising their skills, although the impact of this was ameliorated, by and large, through ongoing clinical supervision once recruitment commenced. It also limited the number of patients to whom each nurse was able to offer EPC; thus, they had, by and large, only limited experience of delivering the intervention. Negotiation of excess treatment costs has been identified as an impediment to the effective delivery of nationally funded research104,105 and a source of concern within the UK health research community; resolving the issue should, we believe, be a matter of urgency.

Delays were further exacerbated by challenges in managing the range of geographically dispersed recruitment centres, and in respect of research administration and bureaucracy. Overall, we recruited 29 participants – in comparison with the 64 envisaged in the external pilot trial and the revised target of 43, which we anticipated achieving after a study extension request was declined. In the event, 8 of the 29 participants in the external pilot trial were recruited ‘late’ (i.e. at a point when we could collect only 5-month follow-up data from them before data collection closed). Although we were able to estimate our primary outcome (response rate at the 5-month follow-up), our estimate of attrition at 8 months should be interpreted with caution.

We did not recruit the anticipated numbers of participants from ethnic minorities, despite the fact that some of our recruitment centres (in Bristol and Coventry) were located in areas with an ethnic mix in line with the sociodemographic profile of England. There was also an imbalance in participant characteristics and outcomes at baseline between the pilot trial arms in respect of some key features, which is likely to have arisen by chance as a result of the small numbers of clusters and participants.

Although we achieved our stated aim of testing the adequacy of methods used for collecting economic data, consistently with our pilot study aims, we did not undertake a full economic evaluation of the interventions and, hence, for example, we did not collect baseline data on previous patient resource use. We also did not capture and cost the additional time that patients spent completing their mood–activity diaries as part of the EPC intervention. As part of a definitive trial, these methodological issues would need to be resolved.

Challenges were encountered in embedding nurse-led EPC interventions in routine cardiac rehabilitation care. Although trained in study intervention and research procedures and potentially willing to participate, nurses found it very difficult to ring-fence dedicated time in which to deliver the intervention. Following the feasibility study, we refocused the intervention more towards a patient-led intervention, informed by the principles of BA. We found from our qualitative studies that participants welcomed the intervention, with some describing substantial personal benefits in respect of increased activity and improved confidence regarding their health status. Although the EPC tested during the pilot study was designed to reduce the impact on nurse workload, it was clear that this ‘lighter’ version of EPC still posed significant challenges to the nurses attempting to integrate it into routine practice. Evidence for this can be seen in the low fidelity to care co-ordination in both the feasibility and pilot studies, with fewer than 50% of participants having such activity documented in their clinical records. This was despite much greater emphasis being placed on mental health-care co-ordination in the pilot study EPC intervention, suggesting that intervention fidelity remained problematic.

Summary of recommendations for future study designs

Based on our findings, we propose the following recommendations for the design of future research studies evaluating the impact of psychological interventions within cardiac rehabilitation:

1. Appropriately locating psychological care. The qualitative work identified the importance to both patients and nurses of having a psychological component to rehabilitation that is clearly embedded within rehabilitation service delivery – not, for example, developed as a parallel service offered within psychiatric services.

2. Embedding the CADENCE EPC intervention within current routine cardiac rehabilitation practice is not realistic. Although cardiac rehabilitation nurses can be successfully trained in the delivery of EPC to patients, significant barriers to implementation remained, relating to the lack of capacity within existing cardiac rehabilitation services. As a substantive increase in cardiac rehabilitation nurse capacity is unlikely to be feasible; a new model of care may be needed. Our data suggest that consideration should be given to delivering EPC by dedicated mental health workers, possibly such as IAPT-trained PWPs, working in an integrated manner embedded within routine cardiac rehabilitation services. The issue here is not one of lack of professional willingness by cardiac rehabilitation nurses to engage in extended roles and develop/implement psychological skills for the benefit of their patients, but, rather, the evident and limiting practical constraints rendering their full engagement in EPC impractical. Should PWPs be employed to deliver EPC, it is important that they are presented to patients as a core component of the cardiac rehabilitation programme (i.e. not as another onward referral), and that the acceptability of this approach be tested in future research.

3. Refocusing of the EPC intervention from a nurse-led to a more patient-focused intervention and revision of the intervention materials. As a result of our feasibility study findings, we altered EPC to a patient-led, nurse-supported programme and tested this in the external pilot trial (in response to the issues identified in point 2 above). Some modification of the written material used to support the delivery of EPC during the feasibility study was necessary, but, having made these modifications, the revised material appeared to be acceptable to nurses and patients engaged in the CADENCE study. Notwithstanding this, some further refinement and simplification of patient materials may still be required to aid engagement with written materials. Newer modes of intervention delivery (e.g. online materials) may also prove to be worthy of exploration. Depending on the availability of dedicated therapists within the cardiac rehabilitation services as in point 2 above, the revised, patient-focused intervention appeared to offer potential as a means of delivering EPC to this population of patients, and one which may be worthy of fuller evaluation in a larger study.

4. Need for longer follow-up period. Longer periods of follow-up (of 12 months or longer), as envisaged in the original commissioning brief, are desirable from a scientific point of view (e.g. in respect of monitoring infrequent patient safety-related issues and clinical adverse events), but their adoption would add considerably to the cost, and participant attrition rates may increase if longer follow-ups are adopted in a future trial. Although the CADENCE pilot trial proved to be a logistically challenging study to deliver, we believe that the added costs and risks of attrition are worthwhile when balanced against the need for robust evidence on the longer-term effectiveness of enhancing psychological care within cardiac rehabilitation services.

5. Adoption of the BDI-II as the primary outcome measure. The BDI-II appeared to be suitable for assessing the severity of depression, with options to examine the change in symptoms from baseline as a continuous measure, and also to examine the proportion of individuals entering remission during the course of the trial (going from scores of ≥ 14 to < 14) or those experiencing clinically important changes in symptoms. Several other outcomes we worked with might also be considered as competing for ‘primary outcome’ status in a future trial, such as depression status determined by use of the PHQ-9 measure (as adopted in the recently published ‘COBRA’ trial). However, it is important that the primary outcome measure assessing depression is different from tools used clinically to monitor mood as part of a psychological intervention. The decision on exact primary outcome is not an absolute issue, and researchers undertaking a future larger study would wish to review the issue carefully in the light of other relevant published evidence available at the time.

6. Use of a cluster intervention design. If EPC (as tested in our pilot study) was evaluated in a definitive trial, a cluster randomised design is required to mitigate the risk of contamination within participating study centres (clusters), arising as clinicians delivering the intervention might not be able to fully ‘turn on’ or ‘turn off’ EPC delivery in the management of individual patients. However, should trials embed PWPs within cardiac rehabilitation teams, cluster randomisation may not be necessary, which in turn will reduce the size, logistical challenges and costs of a future trial.

7. Patient inclusion criteria. All depressed patients attending cardiac rehabilitation following an acute cardiac event, not just those with new-onset depression following the recent acute cardiac event, should be considered for study entry.

8. Study size and logistical challenges. When considered in the light of a power calculation (reported in Sample size for a future definitive study) supporting the design of a future definitive study, it seems likely that recruitment of participating teams would need to involve a substantial proportion of all available cardiac rehabilitation teams currently operating in the UK. Given this, it seems likely too that such a study would require the early and full involvement of the NIHR Clinical Research Network, drawing on its existing structures to support the geographically dispersed recruitment process.

9. Economic evaluation methods. To avoid spurious differences in costs between trial arms, for high-cost care episodes (e.g. inpatient admissions), greater efforts should be made to collect and corroborate administrative record data, but also to prespecify research processes for deciding whether or not such episodes are related to the intervention and targeted condition(s). For less costly, but more frequent, care episodes (such as general practice attendances) and where case note review is highly resource intensive, patient self-report service use data may be sufficient for providing mean attendance rates in trial arms, so long as reasonably high levels of participant follow-up can be assured. Future resource and service use data collection for similar interventions should always give scope for patients and carers to report their time and other costs involved in receiving care or rehabilitation.

Reflections on the study findings from a PPI perspective are provided in Appendix 14.

Sample size for a future definitive study

Interview 1 (post training)

• Experiences of the EPC training (how well did it support EPC delivery, teaching style, length).

• Views of the nurse handbook and participant handbook (layout, content, how they will use them, any improvements required).

• Implementing training (how knowledge gained will help to integrate EPC delivery into current practice, any gaps in training, most helpful areas, any changes required in the training).

• Views of EPC (positive aspects, concerns, anything else needed to support delivery).

Interview 2 (after delivering the intervention)

• Experiences of delivering BA (addressing low mood, understanding and explaining BA, areas useful or struggling with; telephone vs. face-to-face delivery).

• Using the materials (e.g. the nurse handbook and participant handbook, mood diaries, other tools).

• Training (extent to which training was tailored to deliver EPC, any areas needing more input).

• Managing mental health risk issues and care co-ordination.

• Support (feedback on supervision sessions).

• Impact on relationship with patients and perceived mental health role.

• Practicalities (finding space, time to deliver EPC, managing nurses’ workloads).

• Impact of integrating EPC on the existing service on wider team(s) (factors that hinder/help/improve smooth delivery).

Reviewing the assumptions underlying patient recruitment

The patient throughput into the feasibility study teams was lower than anticipated. The latest data for the period 23 September to 10 March 2015 identified 159 patients screened by one of the teams participating in the feasibility study. National Audit (2011–12) estimated that approximately 17 patients per month per team would be assessed, i.e. a total of 272 patients during our recruitment phase.

Patient throughput into the feasibility study is lower than national data would suggest, at around 59% (159/272).

National audit data also estimated that around 17% of patients report depressive symptoms during their initial cardiac rehabilitation assessment as assessed using the Hospital Anxiety and Depression Scale. This figure includes patients with existing and new onset depression. Our feasibility data has broadly confirmed this figure.

21% (34/159) of patients assessed during the feasibility study were found to have depressive symptoms (score 10 or more on the PHQ-9) during their initial cardiac rehabilitation assessment, of whom 7 were receiving active treatment for pre-existing depression and 1 was deemed ineligible for cardiac rehabilitation at the point of assessment.

27/159 (17%) met the CADENCE study definition of ‘new onset’ depression.

We assumed that all eligible patients would be offered treatment, and that around 50% of those offered study entry would agree to take part and enter the observational study. In reality, this estimate was too high.

5/26 (19%) of the patients eligible for feasibility study were not offered study entry. Of those offered study entry 9/26 (35%) agreed to take part. When considering only those offered study entry, approximately 43% (9/21) accepted the invitation to take part; this figure is closer to that estimated a priori.

For the purposes of sample size estimation, we assumed a conservative loss to follow-up rate of 50%. In reality, most depression trials (including our own trials e.g. COBRA, COBALT, TREAD) which collect data via researcher interviews report a loss to follow-up of < 20%. The estimation of patient attrition was not a stated aim for the feasibility study, but rather a goal for the pilot study. Notwithstanding this, we will be able to report preliminary data on this parameter arising from the feasibility study:

Due to the slower than anticipate patient recruitment, we will be unable to estimate the loss to follow-up rate at five months until late August 2015. As of 01/04/15, all three patients (100%) who were due to be followed up for the observational study completed their assessments.

Estimating patient recruitment in the pilot study

Analyses of projected CADENCE participant recruitment are based on recruitment data derived from four cardiac rehabilitation (CR) teams (RD&E, Mid Devon, Exeter and Exmouth) as of 11^th March 2015. This translates into a combined recruitment period (i.e. adding the recruitment periods of the four teams together) of 16 months; equivalent to 64 team–weeks, i.e. 1 month = 4 weeks. During this period, there were 159 screenings of potential participants, resulting in recruitment of 9 participants to the trial. Based on these data, it takes approximately 7.1 ‘team–weeks’ to recruit each participant.

The pilot study will recruit eight teams (i.e. an additional four teams). If the eight teams recruit at the same rate as the existing four feasibility teams:

Over a 6-month period with all 8 teams recruiting (26 weeks; 208 team–weeks) we would anticipate recruiting approximately 29 participants.

If the recruitment period were to be extended to 9 months (39 weeks; 312 team–weeks) we would anticipate recruiting approximately 44 participants.

Estimating patient recruitment in the pilot study

The table below summarises the margin of error with which the follow-up rate can be estimated for different hypothetical samples sizes (number of participants). The margin of error (based on the width of the 95% confidence interval for the follow-up rate) is shown under scenarios where the follow-up rate is: (a) 50%, (b) 80%.

Sample size for CADENCE pilot study: margin of error for estimating follow-up rate in different scenarios

The 80% follow-up rate is realistic. With the 64 participants we are targeting, we can estimate a follow-up rate of 80% with margin of error ± 9.8%. If recruitment was lower than expected, say 43 participants, we could still estimate this follow-up rate with margin of error ± 12%. If recruitment was 29 participants, as we would currently project from our feasibility data, we would only be able to estimate the follow-up rate with a margin of error around ± 14.4%.

Ensuring sufficient data are available for the purposes of the qualitative interview study

In addition to ensuring that the pilot study has sufficient power to quantify the primary outcome (follow-up rate), sufficient numbers of patients must also be enrolled in the trial to support sampling for the qualitative study. Here we propose interviewing 10 ‘decliners’, i.e. people who elect not to take part in the trial, and 30 patients who take part. Trial participants from the following groups will be interviewed: 10 participants receiving usual care, 20 receiving the trial intervention (12 completers and 8 non-completers).

The (cluster) randomisation ratio in the pilot trial is 5 intervention teams: 3 control teams, equating to 40 and 24 participants per arm respectively, assuming the original target of 64 participants is achieved, or 27 and 16 participants respectively assuming a lower target of 43 participants is set.

6.1. Service use data

• Routine administrative sources (hospital records, GP records).

• Service and Resource Use Questionnaire – [SRUQ; adapted Client Services Receipt Inventory (CSRI) with input from a Patient and Public Involvement (PPI) group]. The question items will be based on previous versions of the CSRI questionnaire which was developed for people with mental health problems living in the community, and has been adapted and used in a number of studies by health economists at University of Exeter Medical School. It has questions which comprehensively capture information about the use of:

  1. hospital care (inpatient, A&E, outpatient)

  2. primary care (GP, practice nurse, telephone/face-to-face/home visits)

  3. use of other care professionals

  4. receipt of disability or illness-related welfare allowances/payments

  5. receipt of informal unpaid care (from relatives or friends)

  6. receipt of medication for specific conditions.

To reflect the likely primary clinical outcomes, most of these questions will focus on items (i) to (vi) above in relation to mental health problems. It will also aim to link to the planned process evaluation data; for example, where referral data will be recorded about the psychological services to which patients in this group have been referred.

The SRUQ will also include a free-text survey question which asks the participant to directly report any difficulties in understanding or answering the service and resource use questions.

6.2. Cost of providing enhanced psychological care within cardiac rehabilitation

Trial records and dialogue with those directly involved in developing and delivering the intervention. This will particularly cover the amount of additional time for CR nurses to become trained in delivering the intervention, and to deliver the BA and care co-ordination components of the intervention.

6.3. Economic ‘data analysis plan’

Since the primary aim is to establish – that is, to try out and refine – the intended data collection methods required to support a definitive economic evaluation, the ‘data analysis plan’ will comprise the following elements:

• A full estimation of service and resource use of the pilot RCT participants on the basis of available data sources and draft questionnaire items, documenting the estimation process and the difficulties encountered and any unexpected omissions or suspected inaccuracies.

• Descriptive statistics of the completeness of data from the different sources and at different time points (i.e. item response for those who started completing a follow-up questionnaire).

• Assessment of the validity or accuracy of self-reported service use data, where they can be compared with routine administrative sources [using kappa scores, percentage where both sources of data agree (in number of contacts of each type individually), and percentage of participants who overestimated and underestimated the number of each type of contact compared with case records]. In general, routine administrative sources will be presumed to be more accurate than self-reported service use data. However, extreme differences between participant reported and routine administrative data sources will be sampled and investigated to see if delays or errors in clinical record-keeping or data entry may explain some of the difference (for example, if a participant recalls a hospital admission which has not been captured by routine hospital records).

• Qualitative assessment of the difficulty of completing the SRUQ questionnaire. Including assessment of responses to a free-text survey question which asks the participant to directly report any difficulties in understanding or answering the service and resource use questions.

• Description of any difficulties encountered in processing the data.

  • The preliminary assessment of the cost of delivering the EPC intervention within CR will be developed iteratively by the study’s health economist co-investigator, as a spreadsheet with a number of the relevant stakeholders (i.e. trial manager, intervention developers, cardiac nurses) and wherever possible separately identify those costs which are effectively ‘sunk costs’ – and would be unlikely to be incurred again (or to the same extent), for example, development of the intervention manual; and are mainly associated with the running of the RCT rather than the delivery of the intervention per se.

The spreadsheet will have rows for those resources or staff time involved in designing and developing the intervention (meetings, materials/manuals); training cardiac nurses in EPC (nurse time, trainer time, training materials); any additional cardiac nurse time involved in screening for depression and additional nurse time to co-ordinate care or deliver BA. Wherever possible, it will separately show the number of units of resource (e.g. minutes of cardiac nurse time) and the unit price of each type of health-care resource (e.g. £ per minute of nurse time).