Background: Innovative ways of delivering care are needed to improve outcomes for older people with multimorbidity. Health coaching involves ‘a regular series of phone calls between patient and health professional to provide support and encouragement to promote healthy behaviours’. This intervention is promising, but evidence is insufficient to support a wider role in multimorbidity care. We evaluated health coaching in older people with multimorbidity. Methods: We used the innovative ‘Trials within Cohorts’ design. A cohort was recruited, and a trial was conducted using a ‘patient-centred’ consent model. A randomly selected group within the cohort were offered the intervention and were analysed as the intervention group whether they accepted the offer or not. The intervention sought to improve the skills of patients with multimorbidity to deal with a range of long-term conditions, through health coaching, social prescribing and low-intensity support for low mood. Results: We recruited 4377 older people, and 1306 met the eligibility criteria (two or more long-term conditions and moderate ‘patient activation’). We selected 504 for health coaching, and 41% consented. More than 80% of consenters received the defined ‘dose’ of 4+ sessions. In an intention-to-treat analysis, those selected for health coaching did not improve on any outcome (patient activation, quality of life, depression or self-care) compared to usual care. We examined health care utilisation using hospital administrative and self-report data. Patients selected for health coaching demonstrated lower levels of emergency care use, but an increase in the use of planned services and higher overall costs, as well as a quality-adjusted life year (QALY) gain. The incremental cost per QALY was £8049, with a 70–79% probability of being cost-effective at conventional levels of willingness to pay. Conclusions: Health coaching did not lead to significant benefits on the primary measures of patient-reported outcome. This is likely related to relatively low levels of uptake amongst those selected for the intervention. Demonstrating effectiveness in this design is challenging, as it estimates the effect of being selected for treatment, regardless of whether treatment is adopted. We argue that the treatment effect estimated is appropriate for health coaching, a proactive model relevant to many patients in the community, notjustthoseseeking care. (Continued on next page) * Correspondence: firstname.lastname@example.org NIHR School for Primary Care Research, Centre for Primary Care, Manchester Academic Health Science Centre, University of Manchester, Williamson Building, Oxford Road, Manchester M13 9PL, UK Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Panagioti et al. BMC Medicine (2018) 16:80 Page 2 of 15 (Continued from previous page) Trial registration: International Standard RandomisedControlledTrial Number (ISRCTN12286422). Keywords: multimorbidity, older adults, health coaching, depression Background receive — and provide informed consent — or decline. Multimorbidity, defined as ‘the co-existence of two or Whether or not a patient consents to treatment, for the more chronic conditions, where one is not necessarily purposes of this design, they remain part of the interven- more central than the others’ , is highly prevalent . tion arm. All those eligible but not selected are not con- Patients with multimorbidity are a major focus of health tacted for participation and become controls. systems, but they face barriers to accessing high-quality The TWiCs design has two potential advantages. It care [3–5], and they incur high costs . Recently, clinical more closely mimics the process of treatment decision- guidelines for multimorbidity have highlighted the need making in routine care, as patients are offered a treat- for innovative models of care . Successful self- ment (which they can decline) rather than being offered management will be crucial for improving the health out- two treatments, then allocated at chance. The design comes of patients with multimorbidity, but the current also provides a different (and in some contexts more evidence for effectively managing multimorbidity is weak. useful) estimate of the effects of the offer of treatment A recent Cochrane review reported only 18 trials , with amongst all those who are eligible, rather than amongst some evidence for interventions targeted at risk factors a subset who agree to receive the treatment. As such, it such as depression or specific functional difficulties. The may have greater relevance for treatments designed to review concluded that there is an urgent need for inter- have broad ‘reach’ amongst the wider population. Exam- ventions that can help patients with multimorbidity to ples would include diabetes prevention programmes  better self-manage their conditions to prevent exacerba- and self-management programmes for older people with tions and avoid expensive care utilisation . long-term conditions [16, 17]. For self-management to be cost-effective at a population level, interventions must be delivered to a significant pro- Health coaching as a population health intervention portion of the population in need, not just those moti- Self-management is critical for patients with long-term vated to participate. This is described as ‘reach’ . conditions. A model that has received significant atten- Evidence of reach is often lacking in trials of self- tion is health coaching, defined as ‘a regular series of management, because only a proportion of those meeting phone calls between patient and health professional...to the eligibility criteria actually participate . Evidence of provide support and encouragement to the patient, and reach can be particularly problematic amongst people promote healthy behaviours such as treatment control, with multimorbidity because they are often excluded from healthy diet, physical activity and mobility, rehabilitation, trials . This study aimed to evaluate the impact of an and good mental health’ . intervention that can be used with a large number of Various types of health coaching exist that differ in patients, using a trial design that can better assess the content, delivery (face to face, remote), and personnel. likely population benefit of the intervention. An important issue is whom is targeted for health coaching. It can be provided for patients predicted to be The ‘trial within a cohort’ as a test of intervention ‘reach’ high users of services or following events such as hos- In a conventional trial, participants receive information, pital discharge . Although the rationale for such then provide consent to participate and are randomised. targeting is clear, many patients identified as high users Critically, patients are told about the different treat- of care revert to lower patterns over time without inter- ments available, but only half are randomised to each. vention . There may be an argument for broader Patients with preferences for one treatment may be less strategies targeting the wider population of patients who likely to take part . are currently well but whose current self-management is The ‘Trials within Cohorts’ (TWiCs) design more not optimal. These patients can be described as being closely mimics the way treatment decisions are made in less ‘activated’. Patient activation is defined as how well routine care . A cohort of participants are recruited a patient understands his/her own role in personal and followed up systematically. Under the form of health care, reflecting knowledge, skills and confidence TWiCs used here, all eligible participants in the cohort [21, 22]. Activation may be a method of targeting coach- are identified, and a sample is selected at random. ing to maximise benefit. Another important factor may Patients selected for the intervention are contacted and be depression, which is associated with poor outcomes offered the treatment, which they can either decide to in multimorbidity and may be important in self- Panagioti et al. BMC Medicine (2018) 16:80 Page 3 of 15 management . Treatment burden is an additional assistance with self-management, defined via scores on factor of relevance in this patient population. It is the Patient Activation Measure (PAM) . The PAM defined as ‘the impact of the “work of being a patient” allows activation to be categorised into four levels. Level on functioning and well-being’ [24, 25] and occurs when 1 includes passive recipients of care, level 2 includes the tasks of managing multiple conditions become a det- those who lack the basic knowledge and confidence to riment to health and well-being. self-manage, level 3 is those who have the basic know- An increasing number of systematic reviews have been ledge but lack the confidence and skills to engage in published on the effectiveness of health coaching. Most self-management and level 4 is those who have the suggest significant, modest short-term benefits, and knowledge, confidence and skills and may only require some also support longer term gains [26–33]. However, support during times of stress . We included patients it is difficult to generalise these findings to care for in PROTECTS whose scores placed them in level 2 or 3 people with multimorbidity, as many trials are focussed of activation, because these patients showed some evi- on people with only one long-term condition [28, 32]. dence of self-management which could be improved by Further research is indicated to examine the impact of health coaching. health coaching, assessing reach and the cost- effectiveness of this intervention amongst patients with Randomisation and masking multimorbidity. As noted earlier, patients eligible for the trial are identi- fied from the cohort and randomly selected for treat- Methods ment. We piloted these procedures in 50 patients to test Study design and participants the rate of uptake of the new treatment. After assess- The study was embedded in a wider integrated care ment of eligibility, we selected patients to be offered programme to improve care for older people with long- health coaching at random, using appropriate central term conditions in North West England. The CLASSIC randomisation through a clinical trials unit to ensure study is a longitudinal cohort study evaluating this inte- concealment of allocation. In this pragmatic evaluation, grated care programme. Embedded within CLASSIC, the we did not blind either patients or providers. Proactive Telephone Coaching and Tailored Support (PROTECTS) trial used the TWiCs design to assess the Procedures cost-effectiveness of health coaching for patients with The intervention was health coaching, as defined earlier. multimorbidity. PROTECTS is reported as per Consoli- The content of the health coaching was based on three dated Standards of Reporting Trials (CONSORT) guide- core mechanisms: lines (see Additional file 1: CONSORT checklist). The trial protocol is also included as an additional file 1. Telephone health coaching involved support and (Additional file 2). encouragement to the patient to promote healthy The integrated care programme was delivered to pa- behaviours around diet, exercise, smoking and tients over the age of 65 with at least one long-term alcohol, through provision of information and condition, and we recruited these patients to the motivation for long-term conditions. The core CLASSIC cohort . FARSITE is a software package health coaching materials include telephone and as- (http://nweh.co.uk/products/farsite) that enables cen- sociated patient tracking and management software, tralised searching of general practitioner (GP) records. and health coaching scripts for lifestyle support. FARSITE was used to generate a list of eligible 2. Social prescribing involved links to resources in the patients in each practice, and the results were pro- wider community through the community and vided to general practices to allow them to remove voluntary sector [37, 38]. Access to local resources any patients meeting the exclusion criteria (patients was provided through either PLANS in palliative care or with reduced capacity to consent) (http://www.plansforyourhealth.org/, a self- prior to asking them for consent. A total of 12,989 assessment tool for users to assess their health and patients were eligible between November 2014 and social needs, with links to relevant community re- February 2015. If they did not respond, they were sources and local support) or the Ways to Well- sent a reminder 3 weeks later. Participants were of- being site (on-line resources and information, no fered an incentive of a £10 voucher. At baseline, 4377 longer available in the form used in the trial). people (34.2%) returned a questionnaire. We did not 3. Low-intensity support for low mood included have access to data on non-respondents. assessment of common mental health problems, For inclusion in PROTECTS, patients had to have 2 or simple lifestyle advice and behavioural techniques more self-reported long-term conditions from a list of to manage mood, and use of appropriate risk 15 , and must have been assessed as needing some assessment protocols [39, 40]. Panagioti et al. BMC Medicine (2018) 16:80 Page 4 of 15 Six monthly phone calls to participants were planned. - Quality of life. The World Health Organization The receipt of four out of the six planned calls was con- Quality of Life brief measure (WHOQOL-BREF) is a sidered a complete ‘dose’ of the intervention. 26-item measure of global quality of life (QOL), which The PROTECTS intervention was delivered by a has been validated in a large international population ‘health advisor’ (a National Health Service (NHS) with physical and mental long-term conditions. QOL is Agenda for Change Band 4 worker) with skills in infor- measured across four domains: physical, psychological, mation technology and communication, as well as social and environmental, as well as a single-item scale experience in working with the general public. Advisors for QOL . We used the physical domain score as already had experience with coaching for diabetes and the most relevant in relation to the PROTECTS use of social prescribing. The health advisor attended intervention. 3 days of training specific to working with low mood. They were given a manual which outlined the key ele- Secondary outcome measures were: ments of the low-intensity intervention used (behav- ioural activation, cognitive restructuring, problem - Depression. The Mental Health Inventory (MHI-5) is solving). They also received monthly group clinical a 5-item scale which measures general mental health supervision which focussed on working with low mood. . This measure is well validated for identifying de- The health advisor were further supported by a special- pression symptoms, with a higher score indicating bet- ist nurse manager and received additional advice on ter mental health [44, 45]. The recommended cutoff mental health and social prescribing (i.e. referral to rele- score of 60 was used to indicate the presence of ‘prob- vant community resources) from the research team. able depression’ , although we used the continuous Patients routinely had continuity in their coach for the score in the analyses. duration of their treatment. There were no formal links - Self-care. The Summary of Diabetes Self-Care Activ- with primary care as part of the intervention. The health ities (SDSCA) is a 7-item measure assessing the num- coaching was delivered via telephone from a central ber of days per week respondents engage in healthy and NHS facility. Proactive, monthly calls of around 20 min unhealthy behaviours (i.e. eating fruits and vegetables, were made for a period of 6 months, with the option for eating red meat, undertaking exercise, drinking alcohol additional calls to deal with complex patients or issues and smoking) . of risk. Health coaching staff were trained to customize calls to the individual patient. Provision of support for Power and statistical analysis low mood and social prescribing were made where At the time of study development, there were no be- appropriate. spoke methods for powering this TWiCs design, and we The design meant that the comparator for patients used conventional methods . We powered the study meeting the eligibility criteria who were not selected for to have 80% power (alpha 5%) to detect a standardised the intervention was usual NHS care. We collected de- effect size of 0.25 on any continuous outcome measure. tails of that care for the economic evaluation. Allowing for 25% attrition amongst participants — and assuming that outcome measures at baseline correlate 0. 5 with their respective follow-ups — 504 patients were Outcomes indicated, with 252 randomised to treatment. The PROTECTS was nested within the CLASSIC cohort, CLASSIC cohort included 1306 patients eligible for which used a wide range of measures, varying at differ- PROTECTS, and we randomly selected 252 to be offered ent time points. A pre-specified subgroup of primary the intervention. The uptake rate was lower than antici- outcomes were used in PROTECTS. All outcomes were pated, and we therefore offered the intervention to a fur- collected via postal survey at four time points across the ther 252 patients. This resulted in a final intervention study: at baseline, then at 6, 12 and 20 months. The group of 504 of which 207 consented to the interven- protocol was registered and updated in a registry tion, with the remaining 802 as controls. However, under (ISRCTN 12286422). the TWiCs framework, all 504 patients offered treatment The primary outcome measures were: remain in the treatment group in analysis, including those who declined. In consequence, the eventual effect - Self-management. The PAM is a self-report measure size detectable at 80% power was 0.39 amongst the sub- of patient knowledge, skills and confidence in self- sample consenting to treatment. management for long-term conditions [22, 36, 41]. The analysis followed intention-to-treat principles and We used the short 13-item version. The score is a pre-specified analysis plan. In summary, we report the categorised into four levels for eligibility determination, trial and analysis according to updated CONSORT stan- although we used the continuous score in the analyses. dards and utilising the extension for pragmatic trials Panagioti et al. BMC Medicine (2018) 16:80 Page 5 of 15 . The main hypothesis test of the intervention was typically larger, but the power to detect an effect is not that the overall effect of the intervention is zero. The greater, since the variance of the estimate increases pro- primary analysis used complete cases only. Condition portionately . group was used as a binary variable. All outcomes were treated as though continuous and normally distributed (in all cases both skewness and kurtosis were < =1.0) and Cost-effectiveness analysis analysed using linear multiple regression. Baseline values The primary outcome measure for the economic of outcomes and a set of pre-specified covariates consid- evaluation was the EuroQOL 5-Dimension 5-Level ered prognostic of outcome were included in all ana- (EQ-5D-5L) , a generic measure of health-related lyses: gender, age (categorised as 65–69, 0–79, 80–98), QOL covering five domains (mobility, self-care, usual health literacy , social support , patient activa- activities, pain/discomfort, anxiety/depression). This tion, depression and quality of life (physical health new version was developed due to concerns over the domain). Robust estimates of variance were used ac- lack of sensitivity to change of the original scale, and counting for the clustering of patients within practices. consists of five severity levels for each domain. We ran two sensitivity analyses. The first repeated the Published English general population preference primary analyses using multiple imputation to include weightings were used to convert responses to a single cases with missing baseline or follow-up data. Missing utility index . data values were imputed using chained-equation mul- The perspective of the economic analysis was that of tiple imputation and scores on all available outcome the English NHS. Individual patient-level health care measures and patient demographics at baseline and resource utilisation over the trial period was collected follow-up. Twenty multiple imputation sets were used to from two sources. The number of GP contacts in the ensure stability of results. The second sensitivity analysis previous 6 months was collected from self-report data at assessed the robustness of the primary analysis results to 6-monthly intervals. Hospital utilisation was extracted removal of the pre-specified covariates from the model from linked administrative patient records provided by (not including the outcome at baseline). the NHS, divided into emergency admissions (short Health coaching in the trial was delivered by an exist- stays ≤5, long stays > 5 days), elective admissions, elect- ing service managing other patients outside the trial, ive day cases, outpatient attendances and accident and rather than a bespoke service. This, combined with the emergency (A&E) department attendances. time taken to administer and analyse the cohort and The economic analysis assessed the incremental cost- randomly select the groups, meant that no patient was effectiveness of the offer of health coaching compared offered treatment until 6 months after the baseline with usual care from the perspective of the NHS. EQ- assessment for the CLASSIC cohort, and for some the 5D-5L data were combined with in-hospital mortality offer was not made until month 12 or later. This caused information from the secondary care utilisation data, variations in the duration of time before start of the applying a utility value of 0 upon death. Quality- treatment (range 259 to 513 days after baseline assess- adjusted life years (QALYs) were calculated using the ment). Length of follow-up from end of treatment to area under the curve method assuming linear extrapo- 20 months follow-up was similarly variable. Thus, the lation of utility between time points. QALYs in the trial is considered to have run over 20 months, with pa- second year of the trial were discounted at an annual tients receiving treatment at any time after the initial rate of 3.5% as specified by NICE . 6 months. As these implementation delays were not Intervention costs were estimated combining the cost anticipated, the pre-specified analysis plan stated that of training and supervision, written materials and deliv- the primary analysis would assess the change in out- ery of the health coaching sessions. The intervention comes between baseline and 20 months follow-up. was offered to all participants selected, although only The design provides an estimate of the mean effect in 189 received at least one call. Only patients receiving at people offered treatment. Compared to a pragmatic trial, least one call were assigned treatment costs, and the which provides an estimate of the mean effect in people intervention costs were therefore estimated based on agreeing to treatment, the effect is ‘diluted’ by the pro- these 189 participants. portion of patients in the treatment arm who do not Patient-level resource utilisation data were combined consent to treatment. An estimate of the treatment ef- with relevant unit cost data for the price year 2014– fect in those patients consenting to treatment was 2015 to calculate total costs. Unit costs not available for derived through application of a complier average causal this price year were inflated to 2014/2015 prices using effect (CACE) analysis [51, 52]. The CACE estimator the consumer price index . Costs occurring in the was obtained by dividing the mean effect estimate by the second year were discounted at a rate of 3.5% . Unit proportion giving consent . The CACE estimate is cost figures were sourced from the Personal Social Panagioti et al. BMC Medicine (2018) 16:80 Page 6 of 15 Services Research Unit’s unit costs of Health and Social those, 1306 were eligible for PROTECTS. Of the 1306, Care 2015 and national NHS Reference Costs [58, 59]. 504 were randomly selected to the intervention, and the Follow-up questionnaire completion dates were remaining 802 eligible participants acted as controls. missing in a small number of cases (n =2). In these The flow of participants is shown in Fig. 1. The baseline instances, dates were imputed using the mean length characteristics of participants are presented in Table 1. of time between baseline and follow-up for the sam- ple for the purpose of QALY and cost calculations. Treatment uptake and adherence Missing information on age and gender were sourced Signed consent to health coaching amongst those eli- from the linked hospital administrative data, where gible was received from 207/504 (41%) of those selected, available (gender n =6, age n = 35). For the remaining although only 189 actually received calls (38%). The individuals with missing age (n = 30) or missing base- baseline characteristics of consenters and non- line EQ-5D-5L (n = 29), mean imputation was used to consenters are reported in Additional file 3: Table A. A ensure independence from treatment allocation . multivariate logistic regression exploring baseline factors For missing EQ-5D-5L and resource use data, we associated with consent found that only younger age used multiple imputation by chained equations (ICE) (odds ratio (OR) = 1.08, 95% confidence interval (CI) = 1. to generate 50 imputed datasets assuming the data 03–1.14) and higher education (OR = 4.07, 95% CI = 2. were missing at random. The independent variables 08–7.94) predicted consent to health coaching. specified in the imputation models were age, gender, Among those who consented, 167/189 (85%) re- treatment arm and baseline EQ-5D-5L. To account ceived 4+ calls (the predefined ‘dose’). Assessment of for non-normality, predictive mean matching was call content showed that diet and exercise were the used which forces imputations to only take values most common areas dealt with (in 70% and 57% of observed in the original dataset. Multiple imputation patients respectively), whereas 25% of patients re- (MI) was conducted using Stata’s ICE package, and ceived social prescribing and around 23% received analysis using Stata’s MI package. support for low mood. The incremental cost-effectiveness ratio (ICER) was calculated, adjusting for age, gender, and baseline EQ- Outcomes 5D-5L index score . To assess uncertainty surround- Table 2 shows the patient-reported outcomes for pa- ing the estimates and to account for the typically skewed tients selected for the offer of health coaching and those nature of cost data, incremental costs and QALYs were not selected. The adjusted mean differences were small bootstrapped using pairwise bootstrapping with replace- for all of the primary and secondary outcome measures ment using 10,000 replications. Cost-effectiveness planes and did not reach statistical significance (p > 0.05). The plot these 10,000 bootstrap replications of the ICER non-significance of all group differences was confirmed estimates to illustrate the uncertainty around the point in both sensitivity analyses. estimate of the ICER in probabilistic terms. Finally, cost- Using CACE analysis, the estimated treatment effects effectiveness acceptability curves (CEACs) were plotted on participants who took up the intervention were to graphically represent the probability of the interven- higher, but with correspondingly wider non-significant tion being cost-effective across a range of cost- confidence intervals (Table 2). effectiveness thresholds. The primary economic analysis was based on a Economic analysis comparison on the full sample with MI. A sensitivity Complete data necessary for the economic analysis were analysis was performed using only the complete case available for 45% of the sample (584/1306). sample for which there were no missing data. We Table 3 shows EQ-5D-5L utility scores at each time also took advantage of the implementation delays to point and the total QALY gain over 18 months for the perform a further sensitivity analysis separating the complete case sample. Patients selected for the offer of trial period into two parts: baseline to 6 months health coaching reported slightly lower EQ-5D-5L scores follow-up, where no treatment had yet been received; at baseline. This steadily fell at each time point for the and 6 months to 20 months follow-up, where we ex- usual care group (0.664 at 18 months follow-up), whilst pect any treatment effects to occur. Stata version 14 remaining stable for the health coaching group (0.691). was used in the analysis. The mean unadjusted QALYs for usual care were 1.105, and 1.124 for health coaching over the study period. Results The resources required to deliver the health coaching Recruitment, retention and baseline characteristics intervention are presented in Additional file 3: Table B. In total, 12,989 patients were identified as eligible for The average cost per individual receiving the full course the cohort, and at baseline 4377 (33.6%) participated. Of of health coaching (6 calls) was £148.27. In addition to Panagioti et al. BMC Medicine (2018) 16:80 Page 7 of 15 Fig. 1 PROTECTS CONSORT diagram the direct costs, the analysis also considered the wider long stays), day cases, and A&E attendances were higher NHS resource utilisation. Table 4 reports the average in usual care. Overall, mean costs were higher in health utilisation by resource category for the complete case coaching (£4000.88) than usual care (£3424.16). The sample. Overall, there was a pattern of greater use of average intervention costs in health coaching were £79. emergency care amongst the control group, whilst the 29. This is lower than the £148.27 estimated for a course group offered health coaching used more planned of health coaching because not all individuals took up or services. completed the health coaching. Table 5 presents the average costs of the resource util- isation of the complete case sample. The list of unit costs and resources is available in Additional file 3: Table Cost-effectiveness analysis: full sample with imputation C. The most costly category was outpatient appoint- Table 6 presents the adjusted estimates of the effects of ments, followed by elective admissions and GP appoint- the offer of health coaching on the incremental costs ments. These are all planned care services, the costs of and QALYs compared to usual care in the full sample which were higher in the health coaching group. Con- with imputed data, controlling for age, gender and base- versely, the costs of emergency admissions (short and line utility. Panagioti et al. BMC Medicine (2018) 16:80 Page 8 of 15 Table 1 Baseline characteristics of participants Characteristics Not selected (n = 802) Selected (n = 504) Total (n = 1306) Mean (SD) age 74.2 (6.4) 75.4 (6.8) 74.7 (6.6) Age in categories: 65–69 years 216 (26.9) 115 (22.8) 331 (25.3) 70–79 years 385 (48.0) 230 (45.6) 615 (47.1) 80–98 years 155 (19.3) 140 (27.8) 295 (22.6) Sex (%): Female 441 (55.0) 270 (53.6) 711 (54.4) Male 357 (44.5) 232 (46.0) 589 (45.1) Health literacy: Never 536 (66.8) 322 (63.9) 858 (65.7) Rarely 100 (12.5) 57 (11.3) 157 (12.0) Sometimes 87 (10.9) 63 (12.5) 150 (11.5) Often/always 59 (7.4) 44 (8.7) 103 (7.9) Living status (%): Live with partner or others 509 (63.5) 315 (62.5) 824 (63.1) Live alone 288 (35.9) 188 (37.3) 476 (36.5) Education (%): No qualifications 352 (43.9) 221 (43.9) 573 (43.9) School level qualifications 68 (8.5) 56 (11.1) 124 (9.5) College degree or higher 349 (43.5) 191 (37.9) 540 (41.4) Mean (SD) chronic conditions 6.8 (2.6) 6.8 (2.5) 6.8 (2.6) Mean (SD) index of multiple deprivation 31.0 (18.8) 33.0 (18.6) 31.8 (18.7) Employment (%): Retired or not economically active 748 (93.3) 472 (93.7) 1220 (93.4) Working or other 39 (4.7) 23 (4.6) 62 (4.8) Ethnicity (%): White 786 (98.0) 489 (97.0) 1275 (97.6) Non-white 11 (1.37) 12 (2.4) 23 (1.8) Mean (SD) GP visits in past 6 months 3.1 (2.0) 3.0 (1.9) 3.1 (1.9) Mean (SD) patient activation 57.8 (6.0) 57.6 (5.6) 57.8 (5.9) Mean (SD) quality of life (physical health) 55.3 (19.8) 54.0 (18.8) 54.8 (19.4) Mean (SD) depressive symptoms 65.3 (21.3) 65.3 (21.8) 65.3 (21.3) Possible depression diagnosis (%): Depression 371 (46.3) 227 (45.0) 598 (45.8) No depression 426 (53.1) 265 (52.9) 691 (52.9) Mean (SD) self-care activities 3.8 (0.9) 3.8 (0.9) 3.8 (0.9) The offer of health coaching is associated with a mean important to consider the uncertainty surrounding incremental total cost increase of £150.58 (95% CI £– this estimate. The cost-effectiveness plane plots the 470.611, £711.776) and a mean incremental QALY gain 10,000 bootstrap replications of incremental cost and of 0.019 (95% CI –0.006, 0.043). QALY estimates (Fig. 2). The replications are clus- Whilst there are no statistically significant differ- tered in the north-east quadrant in Fig. 2 (positive ences in either costs or QALYs, the point estimate of health gain and increased cost). Health coaching re- the ICER is £8049.96 per QALY. This would repre- sulted in an incremental QALY gain in 94% of boot- sent a cost-effective intervention at the standard cost- strap replications and was higher cost in 69% of per-QALY threshold of £20,000–30,000. However, it is replications. Panagioti et al. BMC Medicine (2018) 16:80 Page 9 of 15 Table 2 Intention-to-treat analyses of primary and secondary outcomes, using complete cases Intervention group Control group Comparison CACE estimates (eligible patients selected (eligible patients not (estimated points change for treatment) selected for treatment) in those consenting to treatment) N Mean (SD) N Mean (SD) Adjusted p value Adjusted difference in difference means (95% CI) in means (95% CI) Primary outcomes Patient Activation Measure (PAM) 326 62.88 (14.39) 577 61.92 (13.24) 1.44 0.133 3.69 (−1.17 to 8.53) (−0.46 to 3.33) WHO Quality of Life —- physical health 327 55.74 (19.15) 577 55.41 (18.72) 1.62 0.099 4.15 (−0.82 to 9.12) (WHOQOL) (−0.32 to 3.56) Secondary outcomes Depression (Mental Health Inventory, 325 75.74 (16.40) 583 74.29 (17.26) 1.00 0.373 2.56 (−3.20 to 8.36) MHI-5) (−1.25 to 3.26) Self-care (SDSCA) 321 3.49 (1.09) 572 3.54 (1.10) −0.04 0.58 −0.10 (−0.49 to 0.28) (−0.19 to 0.11) Adjusted for covariates gender, age, health literacy, social support, patient activation, depression and quality of life The CEAC (Fig. 3) demonstrates how the probabil- Discussion ity that health coaching is cost-effective increases Principal outcomes with the decision-maker’s willingness to pay. At the We evaluated the role of health coaching in the care of lower bound threshold of £20,000 per QALY, there multimorbidity. We showed reasonable levels of interven- is a 70% probability of health coaching being cost- tion uptake amongst older patients with multimorbidity effective. This rises to 79% at the upper bound of who were not actively seeking help with self-management. £30,000. Compared with usual care, health coaching A large proportion of those who accepted the referral to is likely to be cost-effective in 50% or more cases if health coaching received a defined ‘dose’.Assistancewith decision-makers are willing to pay £8180 or more diet and exercise were the most common interventions for a QALY. within health coaching, although support for low mood The results of the cost-effectiveness analyses were and social prescribing were also present for a significant similar when a complete case analysis was undertaken minority. (see Additional file 4). The post hoc sensitivity ana- Analysis of health outcomes demonstrated no signifi- lysis analysing costs and outcomes separately in the cant benefit associated with health coaching. However, first 6 months post baseline (when no health coaching the economic analysis suggested that health coaching was received) confirmed that the period in which resulted in an incremental increase in both costs and participants actually received treatment was driving QALYs. When a QALY was valued at £20,000, there was outcomes, as the effects were restricted to the period a 70% probability that health coaching was cost-effective. in which health coaching was delivered (see Figures C The economic analysis suggested that health coaching to F in Additional file 4). led to higher utilisation of planned services and lower use of emergency hospital services than usual care. Strengths and limitations Table 3 HRQOL outcomes (EQ-5D-5L) amongst the complete In addition to its large size and focus on multimorbidity, case sample this trial employed the novel ‘Trials within Cohorts’ de- Usual care (n = 378) Health coaching (n = 206) sign. This design provides evidence of ‘reach’ because it Mean SD Min Max Mean SD Min Max assesses uptake amongst people not actively seeking treat- Baseline 0.708 0.23 −0.18 1 0.696 0.236 −0.102 1 ment. A major criticism of conventional trials is that they show effectiveness of an innovation in a very selected 6 months 0.691 0.247 −0.185 1 0.709 0.228 0.018 1 group of patients, which then fails to ‘scale’ because of is- 12 months 0.685 0.254 −0.246 1 0.694 0.237 0 1 sues such as low rates of acceptability amongst the wider 18 months 0.664 0.264 −0.18 1 0.691 0.26 0 1 population, and differences between those who take part QALYs 1.105 0.374 −0.29 1.723 1.124 0.355 0.055 1.683 in trials and those eligible for the intervention . Panagioti et al. BMC Medicine (2018) 16:80 Page 10 of 15 Table 4 Resource utilisation amongst the complete case sample Baseline to 6 months Type of service Usual care (n = 378) Health coaching (n = 206) Mean (95% CI) Mean (95% CI) Secondary care contacts Emergency short stay 0.063 (0.039—0.088) 0.058 (0.026–0.091) Emergency long stay 0.026 (0.009–0.044) 0.024 (0.003–0.045) Day case 0.172 (0.104–0.240) 0.112 (0.059–0.165) Elective admission 0.024 (0.008–0.039) 0.029 (0.002–0.056) Outpatient 4.992 (4.162–5.823) 6.553 (4.977–8.130) A&E attendance 0.156 (0.110–0.203) 0.131 (0.083–0.179) GP appointments 3.111 (2.791–3.431) 3.039 (2.641–3.437) 6 months to 12 months Secondary care contacts Mean (95% CI) Mean (95% CI) Emergency short stay 0.050 (0.027–0.074) 0.039 (0.006–0.072) Emergency long stay 0.040 (0.010–0.069) 0.019 (0.000–0.038) Day case 0.127 (0.069–0.185) 0.053 (0.017–0.090) Elective admission 0.029 (0.009–0.049) 0.029 (0.002–0.056) Outpatient 4.595 (3.650–5.540) 6.403 (5.126–7.680) A&E attendance 0.159 (0.108–0.209) 0.097 (0.041–0.153) GP appointments 2.783 (2.527–3.039) 3.058 (2.696–3.421) 12 months to 18 months Secondary care contacts Mean (95% CI) Mean (95% CI) Emergency short stay 0.132 (0.091–0.174) 0.068 (0.028–0.108) Emergency long stay 0.045 (0.022–0.068) 0.034 (0.009–0.059) Day case 0.196 (0.107–0.284) 0.180 (0.105–0.254) Elective admission 0.040 (0.020–0.059) 0.063 (0.027–0.099) Outpatient 7.185 (6.064–8.307) 9.893 (8.570–11.217) A&E attendance 0.275 (0.207–0.343) 0.170 (0.112–0.228) GP appointments 2.865 (2.599–3.131) 2.922 (2.543–3.302) Table 5 Resource use costs amongst the complete case sample Type of service Usual care (n = 378) Health coaching (n = 206) Mean (£) Mean (£) 95% CI 95% CI Secondary care costs Emergency short stay 146.87 (112.25–181.48) 98.95 (64.27–133.63) Emergency long stay 313.76 (190.97–436.54) 219.08 (101.92–336.24) Day case 343.61 (212.29–474.93) 238.36 (166.87–309.86) Elective admission 310.71 (203.04–418.38) 405.96 (201.93–609.99) Outpatient appointment 1851.42 (1605.13–2097.70) 2521.95 (2139.57–2904.32) A&E attendance 76.66 (62.69–90.63) 51.79 (39.33–64.24) Mean total costs of secondary care contacts 3043.02 (2626.02–3460.03) 3536.09 (2979.87–4092.31) GP appointments 381.14 (350.96–411.32) 392.50 (351.72–433.28) Health coaching costs – 79.29 (69.59–88.99) Mean total cost 3424.16 (2999.98–3848.34) 4007.88 (3444.57–4571.18) Panagioti et al. BMC Medicine (2018) 16:80 Page 11 of 15 Table 6 Cost-effectiveness analysis: full sample with imputation Health coaching (n = 504) over usual care (n = 802) Mean Bootstrapped standard error Bootstrapped 95% CI Incremental cost (£) 150.583 316.941 −470.611 771.776 Incremental QALYs 0.019 0.012 −0.006 0.043 ICER £8049.96 However, this trial also has important limitations, the collection of baseline measures — with correspond- some of which are directly associated with the TWiCs ingly wide variation between end of treatment and design. A conventional pragmatic trial assesses inter- 20 months follow-up. The logistics of the research and vention effects on those consenting to treatment, with capacity within the service meant that no participant an assumption that there will be non-adherence was offered the intervention prior to the 6 months amongst consenters which will reduce any interven- follow-up. Changes in health or behaviours over this tion effect (as these are included in any intention-to- period may have an impact on the effectiveness of an treat analysis). The current design estimates the mean intervention, possibly reducing differences between effect of selection for treatment, and again all patients groups. Nevertheless, delays in accessing treatment are selected for treatment must remain in that group in common in routine service delivery. Another ‘trial the intention-to-treat analysis. The proportion of within a cohort’ (the Depression in South Yorkshire selected patients who do not take up the intervention (DEPSY) trial) achieved a somewhat higher consent rate in a ‘trial within a cohort’ will likely always be larger of 51%, but with 19% of those selected uncontactable than the proportion of consenting patients who do . DEPSY experienced a much higher attrition rate in notcomplywithtreatment in aconventionalprag- the treatment arm, 32% compared to 13% of controls, matic trial. In consequence, the inclusion in the PRO- and we found some evidence for differential attrition. TECTS treatment group of 59% of participants These and other TWiCs design-related issues are con- selected for the intervention who did not take it up sidered in a related publication . — including 10% who were uncontactable — greatly The trial cannot answer the question of whether health diluted the overall treatment effect compared to con- coaching is effective and cost-effective for multimorbid- trols, and resulted in a detectable standardised effect ity in the longer term. The health coaching intervention (amongst those consenting to treatment) of 0.39, consisted of three mechanisms, but the design does not rather than the 0.25 initially powered for. We have allow us to estimate their distinct contribution. Nearly since published specific methods for estimating sam- half of the patients reported symptoms of depression, ple sizes for this type of design . and although support for low mood was provided fre- Our ability to detect an effect is likely to have been quently, it may have to be a more significant aspect of further reduced by the use of data collected at fixed time interventions in patients with multimorbidity . The intervals, as start of treatment varied greatly relative to economic analysis was based on 45% of patients who Fig. 2 Cost-effectiveness plane: full sample with imputed data Panagioti et al. BMC Medicine (2018) 16:80 Page 12 of 15 Fig. 3 Cost-effectiveness acceptability curve: full sample with imputed data returned complete data, which may limit the general patients assessed as in need, but who may not necessar- conclusions. Although multiple imputation was used to ily be seeking self-management support. There will nat- impute missing data values, this cannot fully adjust for urally be interest in the effects on those patients who unmeasured factors that may affect both outcomes and engaged. Although per-protocol analyses can be used, questionnaire completion; hence, the cost-effectiveness such an approach is vulnerable to bias. Some published findings may be subject to residual confounding. How- trials have assessed the effects through propensity ever, a sensitivity analysis comparing cost-effectiveness matching of the subset who engaged . The CACE in the 6 months prior to the intervention — in which analysis is the preferred model for assessment of effects time the majority of attrition occurred — with cost- in those who receive the intervention, as under certain, effectiveness under the intervention found the effects though usually reasonable, assumptions it provides an restricted to the latter period. unbiased estimate of effect. Finally, this trial was conducted amongst patients with mul- Further development of the intervention may have to timorbidity in one area in the UK primarily composed of consider different approaches to targeting, or more white patients. Ethnic minority groups report poorer experi- choice around the exact nature of the intervention to ence of care , and we do not know whether the effective- better align with patient preferences. Qualitative re- ness, reach and cost-effectiveness of health coaching are search conducted alongside the trial will be published different in ethnic minority groups with multimorbidity. in the full study report and may provide insights into Although we have described this as a population health these issues . The group entering the trial did approach, we did restrict to certain groups depending on report significant numbers of conditions, and it is pos- baseline activation, so ‘reach’ was somewhat limited by design. sible that they were too ill to benefit from the interven- The response rate of patients to the initial cohort recruitment tion. As noted earlier, existing treatment burden may was in line with previous studies in this area [65, 66], but is be high in these patients, and although the coaching is potentially another source of bias, and with very limited designed to support self-management, it is possible that demographic data on non-responders to the initial cohort, we adding more self-management may exacerbate issues in were unable to assess overall representativeness. Although treatment burden . Our model of using activation patient inclusion in the cohort was based on data within clin- to target the intervention is in line with the suggested ical records, patients self-reported types of long-term condi- uses of the measure  and reflects previous health tions, and these were not validated against clinical diagnosis. coaching studies which have suggested the importance of avoiding patients who are too ill or too well to bene- Interpretation of the results in the context of the wider fit . There is good evidence that activation predicts literature many outcomes, but the evidence that activation can It was felt that this design was a relevant test of health predict differential benefit from interventions is not as coaching as a population health strategy, reaching out to strong . Panagioti et al. BMC Medicine (2018) 16:80 Page 13 of 15 The pattern of health utilisation shown in the different outcomes, the trial demonstrated that health coaching groups is of interest. Many interventions for older led to no changes in activation or quality of life. How- people target those who demonstrate high levels of ever, the economic analyses showed that the intervention health care utilisation, on the basis that this is where was likely to represent a cost-effective use of resources reductions are most likely to be made. Nevertheless, it at conventional levels of willingness to pay. The eco- can be difficult to reduce utilisation in such patients in a nomic analysis examines the effect of health coaching comparative study , as patients identified on the using a generic measure of health-related quality of life, basis of high use may demonstrate regression to the which may detect broader impacts of the intervention mean, may not be particularly amenable to intervention not captured by the primary trial outcomes. It also con- and may be present in small numbers in the population siders the trade-off between differences in costs and ef- . One of the largest trials of health coaching under- fects associated with the intervention. taken used a risk prediction score for inclusion in the Decision-makers may not be convinced of the benefits trial, but it failed to demonstrate overall benefits in of health coaching in the absence of evidence of clinical terms of admission rates . The approach taken in improvement. However, resource utilisation patterns PROTECTS was different, as patients were identified on highlighted interesting results which warrant further in- the basis of showing capacity for improvement in activa- vestigation. Individuals offered health coaching had tion. Such patients are prevalent, and the results sug- higher utilisation of planned services and lower use of gested that the intervention might reduce emergency emergency hospital services. Health coaching may have use of care. However, the positive impacts of such had a positive impact by increasing individuals’ wider change were ameliorated by increases in elective use engagement in the health service. Due to the limited and overall increases in costs. Another very large trial follow-up period of the trial, we are not able to assess of health coaching which showed reductions in costs whether such increased engagement with planned ser- had an additional focus on ‘preference sensitive’ vices is maintained. shared decision-making rather than self-management Health coaching in patients with multimorbidity did alone . not lead to significant benefits on the primary measures As noted earlier, the recent Cochrane review reported of patient-reported outcome. The optimal role of this only limited evidence for patients with multimorbidity model of care within integrated care systems for patients , although there was a suggestion that interventions with multiple long-term conditions remains unclear. targeted at risk factors such as depression or specific functional difficulties might be more effective. Whilst Additional files our intervention had a depression component, it was not the primary focus as in other interventions in multi- Additional file 1: CONSORT 2010 checklist of information to include morbidity , and it is possible that the broad focus on when reporting a randomised trial. (DOC 218 kb) self-management behaviour change is less impactful than Additional file 2: Protocol for the PROTECTS trial. (DOC 356 kb) a specific focus on a single area such as depression, Additional file 3: Table A. Comparison of participants consenting with especially in the context of an intervention of limited those not consenting. Table B. Costs of the health coaching intervention. Table C. Other NHS unit costs (XLSX 17 kb) duration. Alternatively, our focus on depression may Additional file 4: The results of the cost-effectiveness analyses in have paid insufficient attention to other psychosocial is- complete case analysis. (DOCX 2606 kb) sues that might be present in these patients, such as anxiety or functional disorders. It is equally possible that Abbreviations for patients with fairly high levels of multimorbidity, the CACE: Complier Average Causal Effect; CEACs: Cost-effectiveness acceptability dose of the coaching was simply insufficient . A lon- curves; CLASSIC: Comprehensive Longitudinal Assessment of Salford’s Integrated Care; ICER: Incremental cost-effectiveness ratio; MHI-5: Mental ger treatment might have increased effectiveness, Health Inventory; NHS: National Health Service; NICE: National Institute of although with restricted resources, increasing the length Health and Clinical Excellence; PAM: Patient Activation Measure; of treatment will clearly restrict ‘reach’. PROTECTS: Proactive Telephone Coaching and Tailored Support; QALYs: Quality-adjusted life years; QOL: Quality of life; SDSCA: Summary of Self-Care Activities; TwiCs: ‘trial within a cohort’; WHOQoL-BREF: World Health Conclusions Organization Quality of Life brief measure Patients with multimorbidity are a major part of the Acknowledgements workload of health systems, and findings from large We thank North West E Health and the National Institute for Health Research evaluations of new models of care for this patient group (NIHR) Clinical Research Network: Greater Manchester for assistance with the are directly relevant to clinicians and policy decision- recruitment of the CLASSIC cohort, as well as staff at the participating practices. We thank the health advisors and their managers for their makers. The interpretation of the results will depend on assistance in delivering the intervention. For assistance with the CLASSIC the relative weight placed by decision-makers on clinical study, we thank ‘Salford Together’— a partnership of Salford City Council, and economic outcomes. To readers focussed on clinical NHS Salford Clinical Commissioning Group, Salford Royal NHS Foundation Panagioti et al. BMC Medicine (2018) 16:80 Page 14 of 15 Trust, Greater Manchester Mental Health NHS Foundation Trust and Salford 7. Farmer C, Fenu E, O'Flynn N, Guthrie B. Clinical assessment and Primary Care Together. management of multimorbidity: summary of NICE guidance. BMJ. 2016;354: i4843. Funding 8. Smith SM, Wallace E, O'Dowd T, Fortin M. Interventions for improving Funding was provided by the UK NIHR (grant 12/130/33). This paper outcomes in patients with multimorbidity in primary care and community represents independent research funded by the NIHR, project 12/130/33. settings. Cochrane Database Syst Rev. 2016;3 https://doi.org/10.1002/ Views and opinions are those of the authors and do not necessarily reflect 14651858.CD006560.pub3. those of the NHS, NIHR, NIHR Evaluation, Trials and Studies Coordinating 9. Bodenheimer T, Lorig K, Holman H, Grumbach K. Patient self-management Centre (NETSCC), Health Services and Delivery Research (HS&DR) or of chronic disease in primary care. JAMA. 2002;288(19):2469–75. Department of Health. 10. Glasgow R, McKay H, Piette J, Reynolds K. The RE-AIM framework for evaluating interventions: what can it tell us about approaches to Availability of data and materials chronic disease management? Patient Educ Couns. 2001;44:119–27. The datasets used and/or analysed during the current study are available 11. Treweek S, Dryden R, McCowan C, Harrow A, Thompson AM. Do from the corresponding author on reasonable request. participants in adjuvant breast cancer trials reflect the breast cancer patient population? Eur J Cancer. 2015;51(8):907–14. Authors’ contributions 12. Kenning C, Coventry P, Bower P. Self-management interventions in PB, DR, KL, PC and TB were involved in the acquisition of funding and patients with long-term conditions: a structured review of the role of design of the trial. LR, KH, AB, LR and MS were involved in the administration multimorbidity in patient inclusion, assessment and outcome. J of the study and recruitment of the patients. The statistical analysis was Comorb. 2014;4(1):37–45. conducted by MP, MH and DR, with the economic analysis conducted by BP 13. King M, Nazareth I, Lampe F, Bower P, Chandler M, Morou M, Sibbald B, Lai and RM. MP and PB drafted the manuscript, and all authors provided R: Conceptual framework and systematic review of the effects of comments and approved the final manuscript. participants’ and professionals’ preferences in randomised controlled trials. Health Technol Assess 2005, 9(35):1-186. Ethics approval and consent to participate 14. Relton C, Torgerson D, O'Cathain A, Nicholl J. Rethinking pragmatic This study was approved by the National Research Ethics Service Committee randomised controlled trials: introducing the ‘cohort multiple randomised North West–Lancaster (reference number 14/NW/0206). All participants controlled trial’ design. BMJ. 2013;340:c1066. consented to join the cohort study (CLASSIC), with those selected for the 15. Aziz Z, Absetz P, Oldroyd J, Pronk N, Oldenburg B. A systematic review of intervention providing consent for the trial. real-world diabetes prevention programs: learnings from the last 15 years. Implement Sci. 2015;10:172. Competing interests 16. Lorig KR, Sobel D, Ritter P, Laurent D, Hobbs M. Effect of a self-management The authors declare that they have no competing interests. program on patients with chronic disease. Eff Clin Pract. 2001;4(6):256–62. 17. Kennedy A, Reeves D, Bower P, Lee V, Middleton E, Richardson G, Gardner C, Gately C, Rogers A. The effectiveness and cost effectiveness of a national Publisher’sNote lay led self care support programme for patients with long-term conditions: Springer Nature remains neutral with regard to jurisdictional claims in a pragmatic randomised controlled trial. J Epidemiol Community Health. published maps and institutional affiliations. 2007;61:254–61. 18. McLean S, Protti D, Sheikh A. Telehealthcare for long term conditions. BMJ. Author details 1 2011;342:d120. NIHR School for Primary Care Research, Centre for Primary Care, Manchester 19. Steventon A, Tunkel S, Blunt I, Bardsley M. Effects of telephone health Academic Health Science Centre, University of Manchester, Williamson 2 coaching (Birmingham OwnHealth) on hospital use and associated costs: Building, Oxford Road, Manchester M13 9PL, UK. Manchester Centre for cohort study with matched controls. BMJ. 2013;347:f4585. Health Economics, Division of Population Health, Health Services Research & 20. Roland M, Abel G. Reducing emergency admissions: are we on the right Primary Care, Manchester Academic Health Science Centre, University of track? BMJ. 2012;345:e6017. Manchester, Manchester M13 9PL, UK. School of Nursing, Midwifery & Social 21. Hibbard J, Gilburt H. Supporting people to manage their health: an Work, University of Manchester, Manchester M13 9PL, UK. Mental Health introduction to patient activation. London: The King’s Fund; 2014. and Addiction Research Group, Department of Health Sciences and Hull York 22. Hibbard J, Stockard E, Mahoney E, Tusler M. Development of the Patient Medical School, University of York, York YO10 5DD, UK. NIHR Collaboration Activation Measure (PAM): conceptualizing and measuring activation in for Leadership in Applied Health Research and Care – Greater Manchester patients and consumers. Health Serv Res. 2004;39:1005–26. and Manchester Academic Health Science Centre, University of Manchester, 23. Katon WJ, Lin EHB, Von Korff M, Ciechanowski P, Ludman EJ, Young B, Manchester M13 9PL, UK. Salford Royal NHS Foundation Trust, Stott Lane, Peterson D, Rutter CM, McGregor M, McCulloch D. Collaborative care for Salford M6 8HD, UK. patients with depression and chronic illnesses. N Engl J Med. 2010;363(27): 2611–20. Received: 27 November 2017 Accepted: 6 April 2018 24. Eton D, de Oliveira D, Egginton J, Ridgeway J, Odell L, May C, Montori V. Building a measurement framework of burden of treatment in complex patients with chronic conditions: a qualitative study. Patient Relat Outcome References Meas. 2012;3:39–49. 1. Boyd CM, Fortin M. Future of multimorbidity research: how should 25. May C, Montori V, Mair F. We need minimally disruptive medicine. BMJ. understanding of multimorbidity inform health system design? Public 2009;339(aug11_2):b2803. Health Rev. 2010;32(2):451–74. 26. Dejonghe LAL, Becker J, Froboese I, Schaller A. Long-term effectiveness of 2. Salisbury C. Multimorbidity: redesigning health care for people who use it. health coaching in rehabilitation and prevention: a systematic review. Lancet. 2012;380(9836):7–9. Patient Educ Couns. 2017;100(9):1643–53. 3. Sinnott C, McHugh S, Browne J, Bradley C. GPs’ perspectives on the 27. Oliveira JS, Sherrington C, Amorim AB, Dario AB, Tiedemann A. What is the management of patients with multimorbidity: systematic review and effect of health coaching on physical activity participation in people aged synthesis of qualitative research. BMJ Open. 2014;3(9):e003610. 60 years and over? A systematic review of randomised controlled trials. Br J 4. Coventry P, Fisher L, Kenning C, Bee P, Bower P. Capacity, responsibility, and Sports Med. 2017;51(19):1425–32. motivation: a critical qualitative evaluation of patient and practitioner views about barriers to self-management in people with multimorbidity. BMC 28. Boehmer KR, Barakat S, Ahn S, Prokop LJ, Erwin PJ, Murad MH. Health Health Serv Res. 2014;14:536. coaching interventions for persons with chronic conditions: a systematic 5. Paddison C, Roland M. Better management of patients with multimorbidity. review and meta-analysis protocol. Syst Rev. 2016;5(1):146. BMJ. 2013;346:f2510. 29. Hill B, Richardson B, Skouteris H. Do we know how to design effective 6. Yoon J, Zulman D, Scott J, Mciejewski M. Costs associated with multimorbidity health coaching interventions: a systematic review of the state of the among VA patients. Med Care. 2014;52:S31–6. literature. Am J Health Promot. 2015;29(5):e158–68. Panagioti et al. BMC Medicine (2018) 16:80 Page 15 of 15 30. Wolever RQ, Simmons LA, Sforzo GA, Dill D, Kaye M, Bechard EM, Southard 51. Dunn G, Emsley R, Liu H, Landau S, Green J, White I, Pickles A. Evaluation and ME, Kennedy M, Vosloo J, Yang N. A systematic review of the literature on validation of social and psychological markers in randomised trials of complex health and wellness coaching: defining a key behavioral intervention in interventions in mental health. Health Technol Assess. 2015;19:1–116. healthcare. Global Adv Health Med. 2013;2(4):38–57. 52. Dunn G, Maracy M, Dowrick C, Ayuso-Mateos J-L, Dalgard O, Page H, 31. Ammentorp J, Uhrenfeldt L, Angel F, Ehrensvard M, Carlsen EB, Kofoed PE. Lehtinen V, Casey P, Wilkinson C, Vazquez-Barquero J, et al. Estimating Can life coaching improve health outcomes? — A systematic review of psychological treatment effects from a randomised controlled trial with intervention studies. BMC Health Serv Res. 2013;13:428. both non-compliance and loss to follow up. Br J Psychiatry. 2003;183: 323–31. 32. Barakat S, Boehmer K, Abdelrahim M, Ahn S, Al-Khateeb AA, Villalobos NA, 53. Koop JC. On an identity for the variances of a ratio of two random Prokop L, Erwin PJ, Fleming K, Serrano V, et al. Does health coaching grow variables. J R Stat Soc Ser B Methodol. 1964;26(3):484–6. capacity in cancer survivors? A systematic review. Popul Health Manag. 54. Herdman M, Gudex C, Lloyd A, Janssen M, Kind P, Parkin D, Bonsel G, Badia 2018;21:63–81. X. Development and preliminary testing of the new five-level version of EQ- 33. Kivela K, Elo S, Kyngas H, Kaariainen M. The effects of health coaching on 5D (EQ-5D-5L). Qual Life Res. 2011;20(10):1727–36. adult patients with chronic diseases: a systematic review. Patient Educ 55. Devlin N, Shah K, Feng Y, Mulhern B, van Hout B. Valuing health-related Couns. 2014;97(2):147–57. quality of life: an EQ-5D-5L value set for England. London: Office of Health 34. Blakemore A, Hann M, Howells K, Panagioti M, Sidaway M, Reeves D, Bower Economics; 2016. https://www.ohe.org/publications/valuing-health-related- P. Patient activation in older people with long-term conditions and quality-life-eq-5d-5l-value-set-england. Accessed 7 Feb 2017 multimorbidity: correlates and change in a cohort study in the United 56. NICE. Developing NICE guidelines: the manual. Process and methods Kingdom. BMC Health Serv Res. 2016;16(1):582. guides. London: National Institute for Health and Care Excellence; 2014. 35. Bayliss E, Ellis J, Steiner J. Seniors’ self-reported multimorbidity captured 57. NICE. Guide to the methods of technology appraisal 2013. Manchester: biopsychosocial factors not incorporated in two other data-based morbidity National Institute for Health and Care Excellence; 2013. measures. J Clin Epidemiol. 2009;62(5):550–7. 58. Curtis L, Burns A. Unit Costs of Health & Social Care 2015. Canterbury: 36. Hibbard J, Mahoney ER, Stockard J, Tusler M. Development and testing of a Personal Social Services Research Unit; 2015. http://www.pssru.ac.uk/project- short form of the patient activation measure. Health Serv Res. 2005;40(6 Pt pages/unit-costs/2015/. Accessed 7 Feb 2017 1):1918–30. 59. Department of Health. NHS reference costs 2014 to 2015. London: 37. Brandling J, House W. Social prescribing in general practice: adding Department of Health; 2015. https://www.gov.uk/government/publications/ meaning to medicine. Br J Gen Pract. 2014;59(563):454–6. nhs-reference-costs-2014-to-2015. Accessed 7 Feb 2017 38. South J, Higgins T, Woodall J, White P. Can social prescribing provide the 60. Faria R, Gomes M, Epstein D, White I. A guide to handling missing data in missing link? Prim Health Care Res Dev. 2008;9(4):310–8. cost-effectiveness analysis conducted within randomised controlled trials. 39. Lovell K, Bower P, Richards D, Barkham M, Sibbald B, Roberts C, Davies L, Pharmacoeconomics. 2014;32:1157–70. Rogers A, Gellatly J, Hennessey S. Developing guided self-help for 61. Manca A, Hawkins N, Sculpher M. Estimating mean QALYs in trial-based depression using the Medical Research Council complex interventions cost-effectiveness analysis: the importance of controlling for baseline utility. framework: a description of the modelling phase and results of an Health Econ. 2005;14:487–96. exploratory randomised controlled trial. BMC Psychiatry. 2008;8(1):91. 62. Viksveen P, Relton C, Nicholl J. Benefits and challenges of using the cohort 40. Lovell K, Richards D. A recovery programme for depression. London: Rethink multiple randomised controlled trial design for testing an intervention for Mental Illness; 2007. https://www.rethink.org/resources/r/recovery-programme- depression. Trials. 2017;18:308. for-depression-booklet-april-2012. Accessed 7 Feb 2017 63. Coventry P, Lovell K, Dickens C, Bower P, Chew-Graham C, McElvenny D, 41. Schmittdiel J, Mosen D, Glasgow R, Hibbard J, Remmers C, Bellows J. Hann M, Cherrington A, Garrett C, Gibbons CJ, et al. Integrated primary care Patient assessment of chronic illness care (PACIC) and improved for patients with mental and physical multimorbidity: cluster randomised patient-centered outcomes for chronic conditions. J Gen Intern Med. controlled trial of collaborative care for patients with depression comorbid 2011;23(1):77–80. with diabetes or cardiovascular disease. BMJ. 2015;350:h638. 42. Skevington S, Lotfy M, O'Connell K. The World Health Organization's 64. Burt J, Lloyd C, Campbell J, Roland M, Abel G. Variations in GP-patient WHOQOL-BREF quality of life assessment: psychometric properties and communication by ethnicity, age, and gender: evidence from a national results of the international field trial — a report from the WHOQOL group. primary care patient survey. Brit J Gen Pract. 2016;66(642):E47–52. Qual Life Res. 2013;13(2):299–310. 65. Kennedy A, Bower P, Reeves D, Blakeman T, Bowen R, Chew-Graham C, 43. Berwick DM, Murphy JM, Goldman PA, Ware JE Jr, Barsky AJ, Weinstein MC. Eden M, Fullwood C, Gaffney H, Gardner C, et al. Implementation of self Performance of a five-item mental health screening test. Med Care. 1991; management support for long term conditions in routine primary care 29(2):169–76. settings: cluster randomised controlled trial. BMJ. 2013;346:f2882. 44. Yamazaki S, Fukuhara S, Green J. Usefulness of five-item and three-item 66. Reeves D, Hann M, Rick J, Rowe K, Small N, Burt J, Roland M, Protheroe J, Mental Health Inventories to screen for depressive symptoms in the general Blakeman T, Richardson G, et al. Care plans and care planning in the population of Japan. Health Qual Life Outcomes. 2005;3:48. management of long-term conditions in the United Kingdom: a controlled 45. Kelly MJ, Dunstan FD, Lloyd K, Fone DL. Evaluating cutpoints for the MHI-5 prospective cohort study. Br J Gen Pract. 2014;64(626):568–75. and MCS using the GHQ-12: a comparison of five different methods. BMC 67. Härter M, Dirmaier J, Dwinger S, Kriston L, Herbarth L, Siegmund-Schultze E, Psychiatry. 2008;8:10. Bermejo I, Matschinger H, Heider D, König H-H. Effectiveness of telephone- 46. Toobert DJ, Hampson SE, Glasgow RE. The summary of diabetes self-care based health coaching for patients with chronic conditions: a randomised activities measure: results from 7 studies and a revised scale. Diabetes Care. controlled trial. PLoS One. 2016;11(9):e0161269. 2000;23(7):943–50. 68. Bower P, Reeves D, Sutton M, Lovell K, Bakemore A, Hann M, Howells H, 47. Reeves D, Howells K, Sidaway M, Blakemore A, Hann M, Panagioti M, Bower Meacock R, Munford L, Panagioti M, et al. Comprehensive Longitudinal P. The cohort multiple randomized controlled trial design was found to be Assessment of Salford Integrated Care (CLASSIC): a mixed methods study of highly susceptible to low statistical power and internal validity biases. J Clin the implementation and effectiveness of a new model of care for long-term Epidemiol. 2018;95:111–9. conditions. Health Serv Deliv Res. 2018. (in press) 48. Zwarenstein M, Treweek S, Gagnier J, Altman D, Tunis S, Haynes B, Xman A, 69. Lin E, Katon W, Rutter C, Simon G, Ludman E, Von KM, Young B, Oliver M, Oher D, CONSORT and Pragmatic Trials in Healthcare (Practihc) groups. Ciechanowski P, Kinder L, et al. Effects of enhanced depression treatment Improving the reporting of pragmatic trials: an extension of the CONSORT on diabetes self-care. Ann Fam Med. 2006;4(1):46–53. statement. BMJ. 2008;337(nov11_2):a2390. 70. Wennberg D, Marr A, Lang L, O'Malley S, Bennett G. A randomized trial of a 49. Morris NS, MacLean CD, Chew LD, Littenberg B. The Single Item Literacy telephone care-management strategy. N Engl J Med. 2010;363(13):1245–55. Screener: evaluation of a brief instrument to identify limited reading ability. BMC Fam Pract. 2006;7:21. 50. Mitchell PH, Powell L, Blumenthal J, Norten J, Ironson G, Pitula CR, Froelicher ES, Czajkowski S, Youngblood M, Huber M, et al. A short social support measure for patients recovering from myocardial infarction: the ENRICHD Social Support Inventory. J Cardiopulm Rehabil. 2003;23(6):398–403.
BMC Medicine – Springer Journals
Published: May 30, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera