A case management occupational health model to facilitate earlier return to work of NHS staff with common mental health disorders: a feasibility study

Clinical outcomes

Clinical improvement in CMDs, such as a decrease in the number or severity of symptoms, can be captured using a range of well-established assessment tools. In this feasibility study we used the Patient Health Questionnaire-9 items (PHQ-9) and Generalised Anxiety Disorder-7 (GAD-7) because they are brief and designed to be completed by the patient, changes in their score accurately reflect changes in symptoms of depression and anxiety, and they are free to use. 7,34 Furthermore, PHQ-9 is used by GPs and practitioners involved in the Improving Access to Psychological Therapies (IAPT) initiative, providing an opportunity to compare the outcomes of this study directly with routine care in non-NHS workers. In addition, we collected information on adherence to and change in use of medication.

Occupational outcomes

We measured time to full RTW (defined in this study as working the same days or hours per week as before sickness absence in an identical or equivalent role for at least 4 weeks) and time to partial RTW (defined as working any number of hours in any role) through the ESR and self-completion questionnaires. A study examining a multistakeholder perspective on the definition of RTW after sickness absence due to CMD found that definitions of RTW based on working days and hours may not accurately reflect the priorities of all stakeholders. 35 Therefore, we selected the most appropriate questions and discussed these with our PPI representatives, who approved the final version for our study.

A number of instruments have been developed to measure presenteeism and workability, such as the Work Ability Index (WAI)36 (to explore the effect of physical co-morbidity on work ability) and the Work and Social Adjustment Scale37 (to measure global functioning). Initially, it was our intention to use both of these measures of occupational outcomes to allow us to compare the results from the measures and to recommend one or both for use in a full trial. However, we reviewed the use of these in the research team and with our two external, international experts (Professor Carel Hulshof and Dr Karen Nieuwenhuijsen). The Work and Social Adjustment Scale contains only one item on work: ‘Because of my problem, my ability to work is impaired’ (score 1–8). We considered this unsuitable as a robust work ability outcome for this study. We then compared the WAI with the World Health Organization Disability Assessment Schedule 2.0 (WHODAS)38 at the suggestion of our international experts. Although the WAI is widely used in rheumatology medicine in assessing RTW, its validity has not been assessed in mental health. The WHODAS appears to have more face validity in mental health and work research because it has two specific items relating to areas that are important functional modalities and are frequently impaired in workers with CMDs. These are learning a new task and concentrating on something for > 10 minutes. Both of these impairments are known to be associated with CMDs. The WHODAS has been used in mental health studies on RTW. 39

Prognostic indicators

At baseline, we collected demographic data about personal characteristics such as age, sex, job, previous sickness absence and history of CMD and physical ill health. We enquired about expectations of full RTW and self-efficacy with regard to RTW because both of these indicators are strongly associated with RTW outcomes. 40,41 Self-efficacy with regard to RTW was measured by the Return-to-Work Self-Efficacy (RTW-SE) scale,42 a self-report tool that has shown promising reliability and prediction of actual RTW within 3 months.

Cost-effectiveness measures

To assess cost-effectiveness in a future trial, we collected data on the health and social care services that may be used more or less as a result of the intervention.

To this end, we used a short, self-completed version of the Client Service Receipt Inventory (CSRI). 43 The objective was to determine which services were used during study follow-up and how often. Although the questionnaire was relatively short, the list of services was deliberately comprehensive; this list can be reduced as a result of the feasibility study. A full trial will involve linking costs for treatment and control arms with clinical outcomes and also quality-adjusted life-years (QALYs). The most widely used measure for generating QALYs is the EuroQol-5 Dimensions (EQ-5D). 44 The EuroQol-5 Dimensions, five-level version (EQ-5D-5L), was used in the study, and the relationship between the utility weights derived from it and the clinical measures was explored. This indicated the appropriateness of the measure in a full study. Sensitivity of the utility weights to change in other measures was assessed using correlations and standardised mean responses. Costs of delivering the package of interventions, including OH nurse time, participant and manager time, and cost of training OH nurses in case management, were measured.

Other measures

In addition to the above, we collected information about rates of recruitment, adherence to the intervention in those allocated to receive it and the management of those allocated to CAU. In addition, we collected information on rates of follow-up, and participants’ referral to and uptake of IAPT services and the government’s Fit for Work service, in accordance with the funder’s commissioning brief.

Quantitative data

We collected the same data on all participants and participating trusts. Data from all groups (to which participants were allocated) were used to answer questions about rates of recruitment and follow-up, as well as the acceptability and performance of methods to assess possible outcome measures. Data from group A (intervention arm) and group D (intervention arm) provided information about the acceptability and costs of the intervention, as well as rates of adherence to the intervention. Data from group B (CAU arm) and group C (CAU arm) provided information about the distribution of possible outcome measures in the absence of the intervention (which would help in power calculations for a subsequent trial). Comparison of the management of participants in groups B and C gave an indication of the potential for major contamination. The follow-up period was 6 months from recruitment.

Before recruitment started, sites were advised on how to collect sickness absence data from their trust’s ESR. We assisted participating OH departments in liaising with their HR departments to ensure that the HR departments accurately recorded fit notes reporting CMD with the correct code (S10) in the ESR system.

Data on sickness absence for a CMD were collected at 6 months after the participant entered the study using data coded as S10 on the participating centre’s ESR and by self-report questionnaire. Additionally, we collected data about the distribution of possible outcome measures, including change in anxiety/depression, health-related quality of life, change in antidepressant use and RTW during the 6-month study period. Full and partial RTW was measured by total days of sickness absence before RTW (whether partial or full RTW). We collected information about participants’ referral to and uptake of IAPT services and the government’s Fit for Work Service.

Questionnaires were completed by study participants at the time of entry to the study and at 3 and 6 months after entry to the study. All participants in both arms completed the same baseline and intermediate questionnaires; however, responses provided in the intermediate questionnaire determined which version of the final questionnaire was used. Version 1 was used for participants back at work in any capacity for more than 4 weeks at the 3-month time point. Version 2 was administered to participants on sick leave due to a CMD at the intermediate questionnaire (including those who had not returned to sustained work) and those who did not return the intermediate questionnaire.

Participants were invited to complete hard-copy questionnaire booklets, and pre-paid envelopes were provided to facilitate return to the central research team. A series of reminders were used to encourage return of the study questionnaires. Reminders were sent as follows: one reminder via e-mail, one reminder via telephone and, finally, a second copy of the questionnaire booklet sent to the participants’ preferred postal address. This reminder strategy was used successfully in another OH study47 and was adopted for use in this study. Contact details were collected with the baseline questionnaire so that the follow-up questionnaires (and reminders, if needed) could be sent.

We recorded reasons for non-participation, baseline characteristics of those eligible for inclusion, adherence to the intervention in those allocated to receive it and the management of those who were not allocated to receive the intervention. We also recorded any protocol violations, reasons for not completing the intervention and any adverse events. Costs of delivering the package of interventions, including OH nurse time, participant and manager time, and cost of training OH nurses in case management, were also measured.

At the sponsor site (Guy’s and St Thomas’ NHS Foundation Trust), as well as at another non-participating NHS trust where a co-investigator was employed, additional exploratory work was completed to better understand the factors affecting the identification and referral of staff who were on sick leave with a CMD. Where appropriate and with relevant permissions, this included reviewing information on existing management referrals made to the OH department.

Qualitative data

Qualitative data collection was undertaken using semistructured focus groups and one-to-one interviews, once the quantitative data collection was complete. The qualitative work explored views and experiences in relation to:

• strategies for screening and recruiting study participants, and for promoting the study across trusts (including barriers and enablers)

• motivation for taking part in the study

• RTW processes

• communication pathways established during the study

• use of study documentation and resources (usefulness and acceptability)

• case manager training workshop (content, delivery and acceptability)

• delivery of the case management intervention (how it worked in practice)

• recommendations to consider for a future full trial.

Most of the interviews were conducted locally at each of the participating sites. Individual telephone interviews were conducted for those unable to attend in person. Purposive sampling was used to invite participants to take part in the focus group and interviews. Pilot interviews were carried out to assess comprehension, relevance and appropriateness of the interview schedule. We conducted one or two focus group sessions at each participating site, with the exception of Royal Papworth Hospital NHS Foundation Trust, which decided to withdraw from the study soon after participant recruitment commenced and chose not to participate in the follow-up interviews.

The central research team provided the local field workers with wording to promote and advertise upcoming focus group sessions, and requested that they promulgate this information via appropriate communication channels. Those interested in taking part were asked to contact the central research team so that further details (date/time/location) could be provided. Study participants (sick-listed workers participating in the study) were individually invited by e-mail to take part in a one-to-one interview. All participants in the focus groups and individual interviews were sent a participant information sheet at least 1 week before sessions were held and were given a chance to ask any questions that they may have in a telephone call with a member of the central research team. Prior to the commencement of each session, participants were asked to complete a consent form. For telephone interviews, verbal consent was taken and recorded on a paper consent form. Sessions took place in a quiet location and typically lasted 60–90 minutes. Most focus groups were facilitated by two members of the central research team (GG and VP) and one-to-one interviews were facilitated by one member of the central research team. Sessions were recorded on a digital audio-recorder and transcribed verbatim shortly after each session by a fully skilled medical secretary contracted by Guy’s and St Thomas’ NHS Foundation Trust. Audio-recordings were retained after transcription and will be kept for 20 years. No identifiable information was included during the transcription process. Transcripts were stored electronically using password protection. All transcriptions were uploaded onto NVivo version 8 (QSR International, Warrington, UK) to facilitate data management and coding.

Quantitative analysis

In keeping with the objectives of this feasibility study, no formal statistical analysis beyond simple descriptive presentation of results was undertaken.

Rates of recruitment were calculated from the numbers of eligible participants and those who agreed to participate in the study. Characteristics of eligible participants were recorded at baseline and were summarised through means, medians, standard deviations (SDs) and interquartile ranges (IQRs). Outcome variables, including anxiety/depression scores, health-related quality of life scores and use of antidepressants, were summarised separately for baseline and follow-up. RTW time (full, partial or modified) was summarised and we compared data on RTW from participants’ self-report with data provided by case managers.

For each possible outcome measure, the prevalence of missing data was quantified. We calculated the proportion of participants who completed the intervention and the proportion of completed data sets.

Qualitative analysis

The interview and focus group transcripts were analysed using a thematic analysis approach as a guide. 48 This was chosen as the most suitable method to use on the basis that it is not aligned to any pre-existing theoretical or epistemological framework, and is widely considered a flexible and useful research tool. The qualitative components for the study, including the questions used in the interview schedules, were developed with the realist perspective in mind because we were interested in understanding the experiences and views of those who took part in the study (participants, OH case managers, line managers and HR personnel) as described in their own accounts.

The aim of the thematic analysis at this stage was to identify important and meaningful patterns in the data and to provide insights to account for the experiences and views of those involved in the study. As outlined in the study protocol, we initially planned to follow a six-phase, structured approach:

1. transcription of all interviews by a professional transcriber

2. familiarisation with text and creation of an initial list of emerging themes

3. coding of transcriptions uploaded to NVivo, a qualitative software package that allows data to be annotated as codes and cross-referencing of these codes

4. categorisation and interpretation through additional coding phases and development of representative themes and theoretical concepts emerging from analytical induction and cross-checking with an additional researcher on the team

5. identification of thematic frameworks in additional discussions with the team, which focused on further refinement through constant comparison within and between codes to ensure that the framework reflected the data

6. linking of findings with existing or newly generated theoretical concepts and models to provide context and confirm the relevance and robustness of the key findings of the study.

However, for pragmatic reasons we decided to adopt the following approach to the qualitative analysis. We initially created separate thematic coding frames in NVivo for each of the three qualitative data collection groups (focus groups, case managers and study participants). The final coding frame for each group can be found in Appendix 1. Most insights were derived primarily from research areas of specific interest included as topics in the interview guide (deductive analysis). These included site set-up, study promotion, participant recruitment, delivery of the intervention, stakeholder engagement and communication, and case manager training. In this regard, the analysis stopped short of developing overarching themes and subthemes (or a thematic map) because our focus was on addressing each of the specific research questions and objectives. As recommended by Braun and Clarke,48 we used this analytical approach, which did not occur in a linear fashion but rather reflected a process of going back and forth between different phases. A brief description of activities undertaken is given below.

Rigour and trustworthiness

Several strategies were used during the study to demonstrate rigour and trustworthiness that were in keeping with fundamental principles of qualitative methods and analysis, for example collecting data using a variety of methods, constructing an audit trial to describe the approach used (including during the thematic analysis phase) and a peer review process to validate and question the analytic linkage being made between the data, coding framework and emerging insights.

A distress protocol was developed. The research team members who conducted the individual and focus group sessions were qualified health-care professionals and had experience providing counselling support. This allowed us to provide immediate support, if necessary, as well as providing the participant with information about local services and resources. In addition, the interviewers were supported by co-investigators with extensive clinical experience in occupational medicine and occupational psychiatry. We took a number of different measures to prevent and minimise upset; for example, participants were given the option to skip questions if they did not wish to provide an answer. Participants were also informed during the consent stages that they were free to withdraw from the study at any stage without giving a reason.

A further in-depth systematic thematic analysis exploring the broader myriad of issues (e.g. organisational, personal and cultural) relating to the provision of OH support for NHS staff with a CMD is proposed to be undertaken at a later time, with the emerging themes and subthemes published subsequently. At the time of writing this report, we have already progressed with a second independent rater, who was not involved in the formulation of the thematic coding frame and has not carried out any coding using the thematic coding frame, and who is now conducting an independent analysis of three random transcripts, one transcript from each study group. The purpose of this exercise is to compare and contrast relevant patterns in terms of insights that were identified by the two independent assessment approaches.

Economic analysis

Study design

This was a systematic review of the literature.

Key questions:

1. Which workplace-based interventions are effective in improving RTW outcomes for workers with a CMD?

2. What are the key elements of effective interventions?

3. Are any interventions specific to the health-care sector?

4. Are any interventions specific to the UK?

Methods

Outputs

We reported the systematic review based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. 54

Results

A total of 934 articles were retrieved from five databases (Figure 2). In total, 40 articles were included for qualitative synthesis and are listed in Table 2. Our search was sensitive because it included all CMDs. The majority of the 572 records excluded at the title/abstract stage were excluded because of studies not presenting results on CMDs.

FIGURE 2.

The PRISMA flow diagram of included studies. CDSR, Cochrane Database of Systematic Reviews.

The following were key findings and recommendations from the systematic review and informed the development of the case management intervention; they relate to questions 1 (which workplace-based interventions are effective in improving RTW outcomes for workers with a CMD?) and 2 (what are the key elements of effective interventions?):

• We found strong evidence suggesting that workplace-focused interventions significantly reduce the time until partial RTW, and low-quality evidence that they do not significantly reduce time to full RTW compared with no treatment.

• The systematic review included four economic evaluations: two workplace-focused interventions were found to be cost saving from the perspective of the employer60,71 and two studies showed no economic benefit compared with CAU. 21,80

• A review of qualitative studies58,69,78,86 that focused on the RTW process found that, from the employee perspective, the following considerations were prominent: (1) concerns about reduced working capacities, (2) difficulty setting limits in demanding work situations, (3) a sense of responsibility and a fear of being a burden to an employer, (4) recognition of exhaustion and (5) the need to control cognitions and behaviour such as perfectionism. An interesting finding was that most workers were able to describe solutions; however, few workers expressed an intention to implement or utilise the solutions in the workplace, with the exception being structural adaptations of work demands.

• Several studies have demonstrated that a number of baseline characteristics appear predictive of a longer duration of sickness or poor RTW outcomes. These include a low level of education, a history of sickness absence, low self-esteem, low social functioning, older age and negative expectation regarding RTW.

• Evidence suggests that different stakeholders have different perspectives regarding barriers to RTW. For example, employers tend to underestimate the importance of the work environment, whereas health-care professionals emphasise remaining at work, maintaining contact with employers and support in the work environment as important.

• We found strong evidence that interventions should be based in the workplace and should involve problem-solving and an element of work-focused behavioural therapy.

For question 3, we found no interventions specific to the health-care sector.

For question 4, we found no interventions specific to the UK.

Collectively, the literature review suggested that our new intervention should include:

• identifying obstacles to RTW11

• work-focused CBT12

• focused problem-solving13

• focus on participant engagement and MI techniques, optimisation of clinical treatment, promoting sleep hygiene, goal-setting, written RTW plans based on discussion between participants and their manager14

• workplace adjustments including flexible working and graded RTW, regular reviews, and communication of the RTW plan with other health-care professionals, in particular GPs who are treating participants11,16

• maintenance of contact between the line manager and the sick-listed worker. 17

Discussion

The systematic review exercise was undertaken as an extension of the earlier review by Pomaki et al. ,81 and provided an important opportunity to incorporate updated information and learnings from the literature when developing our new case management intervention. We confirmed that interventions to improve RTW following sick leave with a CMD should be based in the workplace and based on work-focused behavioural changes. Although we undertook a comprehensive literature review, a more recent systematic review and meta-analysis52 exploring predictors of RTW for people on sick leave with CMDs was not completed during this present study, and so the findings and recommendations of this could not be taken into consideration when developing our case management intervention. 90–94 Another, concurrent review94 suggests that future interventions designed to facilitate RTW for people with CMD should attempt to ameliorate high workload issues. Our systematic review enabled us to identify key components of our proposed intervention.

Study design

This was a cross-sectional survey of OH departments providing OH services to NHS trusts and health boards.

Methods

We identified OH departments providing OH services to NHS trusts and health boards from two lists. Specifically, 122 providers of NHS OH services were identified from the NHS Health at Work Network and four providers were identified from the Commercial Occupational Health Providers Association. These 126 providers were invited to complete the survey of CAU.

We designed a 12-item electronic questionnaire that enquired about how OH providers currently manage NHS staff who go on sick leave with a CMD. We included questions on the type(s) of trust to which they provide OH services, who delivers the intervention, the nature of the intervention, how the line manager is involved, whether or not the workers have rapid access to mental health assessments or treatment (including CBT), and whether or not a worker’s treating health-care professional is involved.

The questionnaire was piloted among the 35 OH departments taking part in the SCIN (Skin Care Intervention in Nurses) trial. 96 We confirmed that nine OH departments were willing to take part in the feasibility study and use electronic OH data collection systems that would enable us to collect the data required for this study. We included OH departments that provide a service to acute and mental health trusts. NHS staff working in ambulance trusts have the highest sickness absence rates of all health-care workers, and the organisational structure of ambulance trusts is very different from that of acute trusts. In the light of this, we were keen to include an ambulance trust in our feasibility study. Because only one of the 35 centres we contacted provided OH care to an ambulance trust, we contacted one of the largest ambulance trusts in England (West Midlands Ambulance Service University NHS Foundation Trust). This trust completed the questionnaire and confirmed its willingness to participate in the feasibility study. Accordingly, we selected six NHS OH providers to take part in the feasibility study, with one as a reserve. The reserve site was intended to be opened if one of the existing sites withdrew at a very early stage during the study. Although Papworth Hospital NHS Foundation Trust subsequently withdrew from the study, it was not feasible because of time constraints to train an additional case manager from the reserve site.

Data analysis

We used descriptive statistics to summarise the patterns of management of CMDs.

Outputs

The information we collected gave us a picture of usual management of NHS staff who go on sick leave with a CMD.

Results

A total of 49 out of 126 (39%) OH providers who were invited to complete the survey responded. We found that the majority (98%) of respondents reported that they had an organisational policy specifically relating to the management of sickness absence that also included triggers for referral for staff on sick leave with a CMD. In 63% of cases, referrals by line managers of sick-listed staff with a CMD to OH occurred between 8 and 28 days after commencement of sickness absence. A total of 10 respondents (20%) accepted referrals on the first day of absence and one respondent accepted referrals as soon as a fit note was received. Timely access to assessment by a psychiatrist or therapist was available to 82% of respondents, whereas early access to treatment by a mental health professional, through online CBT or via a counselling or employee assistance programme service, was available to 88% of respondents. Table 3 details the usual clinical aspects covered during the first consultation.

A key finding from the survey of CAU is that a large proportion of OH providers either do not deliver or only sometimes deliver a case management approach or regular, timed reviews (38.8% and 36.7%, respectively) for sick-listed workers with a CMD. Moreover, of those who reported that they do provide a form of case management or regular timed reviews, specific details with regard to the nature of both features were unknown. Forty-five OH providers (92% of responders) reported that a qualified OH professional, usually an OH nurse, undertook the first consultation. Seventeen (35%) OH providers always used face-to-face appointments for first consultations, while 22 (45%) OH providers used face-to-face consultations more than half of the time. The remainder predominantly used telephone consultations. All respondents provided contact and support for line managers of staff referred for CMDs. The most frequent advice provided was to maintain contact with sick-listed staff (86%). Most (80%) OH providers provided training for managers to better support staff with CMDs. The survey also found that some level of support or advice is provided by OH providers to line managers of NHS workers on sick leave with a CMD. Contact with the staff member’s GP or mental health professional was less consistent, and when this occurred contact was most commonly made by e-mail (Figure 3).

FIGURE 3.

Methods of contact initiated between OH providers and other stakeholders.

Discussion

This was the first national survey to report on current OH management practices for NHS staff with a CMD. This survey found noteworthy variation in the elements comprising the first OH consultation for NHS workers who are on sick leave with a CMD, particularly with regard to the time between the commencement of sick leave (in response to a CMD) and initial OH consultation. Our survey of CAU capitalised on one of the stated outputs of the SCIN trial: ‘to establish a network of NHS OH departments which would be in a good position to deliver future studies’. 96 Because only 39% of OH providers participated in the survey, an important limitation was the potential impact of non-response bias. Therefore, it is reasonable to argue that OH providers that are able to deliver superior OH services may have been more motivated to respond to the survey and, therefore, the results from this survey may not reflect routine OH management practices for sick-listed staff with a CMD. With regard to the variation in the method of conducting first consultation (i.e. telephone vs. face-to-face), these differences could be due to variation in OH service delivery (e.g. in-house OH service vs. external, contracted OH service). The findings from the survey highlight a need to promote the standardised adoption of early case management practices in the provision of OH services, which may facilitate the early and sustained RTW for NHS staff with a CMD. Embedding a standardised approach to the management of NHS staff with a CMD will also enable future widespread evaluation of OH services.

Study design

This was a two-stage, iterative consultation based on methodology that the chief investigator (Ira Madan) and co-investigator (Julia Smedley) have used in the development of national evidence-based guidelines. 97 This section describes the various stages used during the intervention development phase.

Stage 1: mapping of evidence from the literature and expert feedback

Stage 2: stakeholder workshop

Results

The preliminary case management intervention was developed using the iterative process (i.e. using results from the systematic review of the evidence on RTW and mental health, and mapping of the evidence onto the preliminary intervention) and was subsequently delivered in the feasibility study. As far as practicable, this is presented in keeping with the Template for Intervention Description and Replication (TIDieR) reporting requirements101 (Figure 4).

FIGURE 4.

Manualised model of the intervention (including recruitment and CAU pathways). PIS, participant information sheet.

The intervention was based on a case management approach delivered by an OH nurse (case manager) trained in a CBT-based approach to problem-solving in the context of CMDs. The assessment of the worker on sick leave followed a biopsychosocial approach and the focus of the intervention was on the present (forward looking): the case manager encouraged and facilitated active engagement with the sick-listed workers rather than providing passive instruction. The intervention included identification of barriers to RTW; problem-solving by the participant and manager, as recommended by Arends et al. 13 and Waddell et al. ;11 work-focused CBT [via online programmes, e.g. the HeadGear smartphone application (app) (version 1.1.27, Black Dog Institute, University of New South Wales, Sydney, NSW, Australia), or face-to-face therapy, depending on local access] as recommended by Nieuwenhuijsen et al. ;12 two evidence-informed leaflets on RTW (one for managers and and one for employees); optimisation of clinical treatment, goal-setting, peer-support networking, a written RTW plan based on discussion between the participants and their manager, and workplace adjustments including flexible working and graded RTW in discussion with the participant’s manager, as recommended by Waddell et al. ;11 maintenance of contact between the line manager and the sick-listed worker as recommended by Nieuwenhuijsen et al. ;17 communication of the RTW plan with the treating health-care professionals, including the participant’s GP; regular timed reviews as recommended by Stern and Madan16 and Waddell et al. ;11 and referral to an occupational physician or secondary services, if required. The intervention was delivered on receipt of a fit note (or management/self-referral to OH) or soon after. Consultations between case managers and sick-listed staff occurred in person, although there was the option for case managers to maintain contact by telephone, if necessary.

We used evidence from the systematic review to develop a bespoke information leaflet for managers on how to communicate with the sick-listed employee. The leaflet would be given to the line manager by the case manager. The line manager had the option to ask the case manager for additional practical advice on how to initiate supportive communication with a sick-listed employee who has a CMD. The evidence from the systematic review also informed the design of a bespoke sleep hygiene leaflet, especially for NHS workers with a CMD, as well as a My Action Plan (MAP) template to record actions employees would take to tackle their identified problems.

Discussion

The stakeholder group played a vital role in developing the intervention and ensuring that it would be acceptable to participants (employees), case managers, line managers and employers, which is in keeping with recommendations published in guidelines on developing and evaluating complex interventions. 45 Based on participants’ advice, we refined the pathways and processes of the intervention to maximise the chances of its success.

Based on the findings from this feasibility study, the preliminary manualised case management intervention was modified and finalised (see Figure 4).

Method

We developed a series of data collection tools that included self-report participant questionnaires and site-level case report forms that were completed by field workers (case managers and independent assessors). The study questionnaires were shared and discussed with our PPI representatives to obtain their views on comprehension and acceptability, particularly on the appropriate questions to measure RTW outcomes that are important for workers. We did not undertake reliability analysis on the data collection tools; instead, the research team made a decision about their suitability based on a review of the completeness of the data collected at the end of the study coupled with feedback on acceptability from our PPI representatives, participants, case manager and local field workers. This allowed us to assess whether or not the proposed data collection tools would be of sufficient scope to reliably report on each of the outcome measures.

Output

The output was a series of data collection tools for use in the feasibility study, including suitable questionnaires.

Results

A series of data collection forms (case report forms) were produced to record specific data relating to clinical, occupational, prognostic and cost-effectiveness measures. Data collection forms included participant self-completed questionnaires (i.e. previously validated tools including the PHQ-9, GAD-7, WHODAS, Work and Social Adjustment Scale, RTW-SE scale, CSRI and alcohol-use tools) and tools for the collection of site-level data by local field workers (including tools to record screening and recruitment activity, provision of CAU interactions, delivery of case management interventions, consultation time duration and sickness absence duration). The data collection tools used in this study are given in Table 4.

The site-level data collection tool capturing information on consultation time duration and number of follow-up consultations required was also suitable for use in the study. However, in relation to approaches for measuring time to RTW outcomes (partial or full), the results showed that there was poor agreement between the self-report and case manager feedback, and, unfortunately, this was consistent across the five possible outcomes (i.e. ‘RTW, any’, ‘RTW and stay continuously for > 4 weeks without any further sickness absence’, ‘returned to modified duties’, ‘returned to unmodified duties’, ‘returned to normal hours’). This observation reflects the accumulating evidence showing the inherent difficulty in reliably measuring RTW among sick-listed staff. 102

During the conduct of this study, researchers from the Netherlands undertook a project aimed at developing core outcome measures for sickness absence that can be used in research projects. In relation to measuring absenteeism (periods of sickness absence), the researchers recommended that recall of, at a maximum, 2 months could be collected by self-report from participants. Nevertheless, we found only modest consistency in the capacity of participants to accurately self-report ‘duration’ of recent episodes of sickness absence (i.e. during the study period) when these self-report data were compared with organisational sickness absence data (see Figure 9). Furthermore, we also found that not all participants were able to accurately recall each individual episode of sickness absence during the study period when we compared these self-report data with organisational data (although we also acknowledge that organisational sickness absence data are prone to inaccuracies). Collectively, this suggests that self-reported sickness absence is not a reliable measure to use alone.

A further complication with capturing RTW and sickness absence data from case managers and field workers during delivery of the case management intervention was the lack of access to trusts’ HR systems that record this information. In the light of these challenges, it is proposed that alternative, more reliable methods to capture RTW data are explored. One possible means would be to make use of short messaging service (SMS), sending texts to study participants on a regular basis (weekly/fortnightly) to request details of recent sickness absence periods. A notable advantage of this method is that it would help to minimise recall bias.

We also found that the site-level data collection tools (audit form and checklist form), as well as the participants’ study questionnaires, were reliable tools to measure intervention fidelity (uptake and adherence to individual components of the case management interventions). However, a notable shortcoming with the 6-month study questionnaire for participants in the intervention arm was that it did not capture important information on reasons for non-uptake of and non-adherence to the intervention. This additional source of information would have been helpful to gain a better understanding of the reasons why some participants chose not to engage with specific components of the intervention.

In terms of acceptability and feasibility regarding use of the data collection tools, various approaches were used by case managers and field workers to document and record study-specific data. Some found it more helpful to make initial handwritten notes during consultations rather than using the study’s data collection tools during clinic appointments. The benefits of this approach were that it allowed them to focus on the interaction with staff as this helped them to establish a therapeutic relationship with staff. Moreover, some case managers pointed out that an excessive amount of study documentation requiring completion by them and their participants could be considered an intrusive element in OH consultations, and others felt that the format of some of the data collection tools relating to RTW was not easy to follow or complete and they did not reflect the range of possible scenarios staff may present with. This is exacerbated further when OH clinic sessions are busy and when OH nurses are under pressure to keep within designated appointment times. Others found it difficult to remember to complete the case management form during OH consultations, which meant that important study data were missing at the end of the study. To mitigate some of these logistical issues, some of the field workers and case managers suggested linking the study’s data collection forms, specifically those relating to RTW, to existing electronic OH records to streamline the recording of this information, as this would also help to minimise any disruptions during the consultations. The use of an electronic tablet for field workers and participants to record study data was also proposed as a worthwhile alternative to paper records, although we have found in previous research, and in this present study, that implementation of a reminder strategy helped to optimise the return of paper questionnaires. 47 Moreover, the participants’ questionnaires were also considered by some case managers and participants to be too lengthy and the requirement to complete so many was considered an onerous burden. Furthermore, it was generally felt that clearer information was needed in terms of the purpose and timing of each questionnaire during the study period. It was also recommended that a useful strategy to optimise the response rate of study questionnaires in a future trial would be to administer these at the time that participants attend their follow-up review consultations, rather than relying on the postal delivery-and-return method, or to provide participants with the option of completing questionnaires online.

Discussion

The series of participant questionnaires used in the study were found to be of suitable scope to assess current (and changes in) clinical state relating to mental functioning (anxiety and depression indicators), workability/occupational functioning, RTW self-efficacy and health economic measures.

Agreement between periods of absence with a CMD as reported by participants and as extracted from ESRs was described graphically. Additionally, agreement between RTW patterns as reported by participants and in case-manager forms were described in cross-tabulations and assessed using the kappa statistic (see Figure 9).

Methods

We developed a 2-day bespoke training package and the workshop was held in London, UK, during September 2017. The workshop training comprised the following topic areas:

• Background to the study.

• Use of the study protocol, screening for eligibility and taking consent, completion of data collection tools, flow charts, use of the study pack, and safety issues.

• Clinical assessment of CMD/risk assessment, and identifying and challenging unhelpful beliefs.

• Problem-solving and promotion of engagement, theory and practice. The learning objectives for the problem-solving approach were as follows –

  • recognising engagement and disengagement

  • how to share information to empower the worker to set up their action plan

  • good-quality listening.

• Use of clinical and case pro formas, writing RTW plans and improving RTW self-efficacy.

• Communication – RTW leaflets; communication with GP, line manager and others; and access to CBT.

• How to establish the peer network.

• Suicide risk assessment, including a practical session.

  • Practical session – use of case management pro formas and roleplay.

• Helping managers to stay in touch.

• Seeking support and help – case manager network.

• Knowledge assessment questionnaire – pre and post completion of the training.

• Training evaluation feedback.

The small group size of up to six case managers (a maximum of two per intervention trust) allowed for in-depth training and discussion. We based our training on our experience of running case manager training days with an emphasis on problem-solving and the promotion of engagement (JS and RS),24 and on mental health training days for OH nurses (MH and IM). 103 Previously, we had shown that small-group training on mental health for OH nurses results in good knowledge transfer and use of knowledge in practice, as well as increased confidence and levels of satisfaction with the training. 104 The essential skill required of case managers is the capacity to create a supportive, empathic atmosphere in which they explore and make salient the client’s own reasons for returning to work, discuss their barriers to returning to work and support the client (when ready) to identify an acceptable pathway to change by facilitating structured personal goal-setting. This goal-setting element was orientated to increasing activities at home, gradually introducing and scaling up engagement with the employer, and phased RTW.

Therefore, training for case managers comprised a number of related elements that were integrated into the manual and training:

• Clinical assessment of CMDs in the OH context, delivered by a consultant occupational physician and consultant psychiatrist. This component of the training covered an overview of CMDs (aetiology and prevalence); symptom and functional enquiry; mental state assessment; use of standardised questionnaires (e.g. GAD-7 and PHQ-9); principles of the optimal treatment of CMDs (counselling, talking therapies, drug treatment, and online CBT in conjunction with input from a therapist), and availability of treatments locally; addressing and overcoming barriers to work; problem-solving using a specified framework; RTW planning, including indications for phased hours, minimising shift work, exposure to stressors at work and work volume; and the management of relationships at work.

• Problem-solving and MI. Case managers were taught to deliver problem-solving in the spirit of MI to promote participant engagement by Robert Shannon, who is experienced in teaching and delivering MI training in a variety of settings including chronic disease and OH case management.

• A combination of workshop input and follow-up individualised supervision by way of regular teleconferences with case managers. The follow-up supervision sessions took place on three occasions, were between 30–60 minutes in duration and were largely unstructured in nature. These sessions provided an important opportunity for the newly trained case managers to report back on delivery of the intervention and any unforeseen issues that they may have encountered. In particular, these sessions were an opportunity for case managers to describe how they were approaching the participant screening and recruitment activities, the clinical assessments, and problem identification and problem-solving components of the intervention, and how they felt participants were engaging with the intervention. It was not deemed necessary to undertake additional roleplay exercises during the follow-up supervision sessions. We also incorporated a practical patient-actor/case manager roleplay session into the training workshop so that problem-identification, problem-solving and action-planning skills could be practised by all trained case managers.

• CBT skills. Case managers were coached in basic CBT techniques. This was provided by a psychiatrist, but stopped short of conventional clinical CBT training, focusing instead on the use of CBT approaches (e.g. positive thought cycling) with a workplace focus, for example defusing negative thought cycling or rumination on problems in the workplace. We trained the case managers in risk assessment of serious self-harm and suicide in the study participants. We built an escalation procedure into the training, which took into account the case managers’ current local procedures.

• Use of the study protocol and communications with line managers and GPs; setting up a peer-support network.

• Safety issues – procedures for safeguarding vulnerable individuals, which were dovetailed with the local procedures at the case manager’s trust.

Evaluation

• We evaluated the training by using the first three levels of the Kirkpatrick model of educational evaluation (learner satisfaction, learning outcomes and performance improvement),104 adapting the questionnaire we used in previous evaluation of mental health training for OH nurses. 103

• We monitored the integrity of the case management delivery by conducting a case file-review audit exercise at the end of the study period. The purpose of this exercise was to assess the extent to which all components of the case management intervention were delivered. The case file review audit was undertaken at the end of the follow-up period by an independent non-case manager employed at the participating sites (see Table 9 for results).

Results

Overall, the results showed a sustained, gradual increase in knowledge across the subject area following participation in the training, although it was also noted that (‘correct’) knowledge in the pre-training assessment had diminished (for several questions) at the post-training evaluation for up to two case managers. Table 6 presents the overall increase/decrease in assessment scores for the group.

Accordingly, future training of case managers will need to ensure that additional information is covered in the training with respect to the basic aetiology and epidemiology of CMDs and approaches to problem-solving in the context of CMD.

Overall, participants rated the training workshop very favourably in terms of acceptability, with most responding with ‘strongly agree’ to the following seven statements:

1. The objectives of the training were clearly defined.

2. Participation and interaction were encouraged.

3. The content was organised and easy to follow.

4. The materials distributed were helpful.

5. The trainers were knowledgeable about the training topics.

6. The trainers were well prepared.

7. The time allotted for the training was sufficient.

Additional feedback from case managers was captured on the post-training evaluation form and is summarised below (Table 7).

Table 8 provides an overview of the number of participants who contributed to the qualitative data collection. Table 9 provides an overview of adherence to the intervention.

The following analysis is based on qualitative data and represents the views of and feedback from case managers.

Overall, the case managers felt that the 2-day training workshop was well developed, covered a diverse range of relevant subject areas and was delivered by facilitators from key specialities (occupational medicine, occupational psychiatry and organisational psychology). Importantly, the case managers felt that the workshop supported them in developing new skills and competencies that were required for them to take on the new case manager role:

But it was good and the information that they gave you to be able to do the motivational interviewing and go through the case management and the problem-solving, I mean it was all well delivered and able to be lifted and taken back to my practice with no difficulties.

Case managers expressed the view that the skills they developed in the training enabled them to deliver an enhanced level of OH support during the study period; more specifically, it allowed them to shift the focus of their OH consultations away from information gathering only to an approach that actively sought to engage staff in a more participatory and therapeutic relationship. Participation in the training workshop was also recognised as a valuable opportunity for them to build on their own professional skills and capabilities, and allowed the case managers to use their new knowledge and skills in other areas of clinical practice:

We’re constantly short of time, short of clinic slots, short of staff, and whilst there is a place for telephone, I think we’ve gone totally virtually to telephone and I think the balance is not quite there and I don’t think I realised that until actually I came out of the lovely long face to face, where I had really good interaction with somebody. I felt as if I’d moved them along during the process of that assessment rather than just getting information from them, so it wasn’t just about gaining information, it was about gaining that information, putting some things into place, helping them to identify what they needed to do next, you know, identify some of their problems, talk to them about their action plan, and that they were leaving with something to go away with and do, and the first one just felt fabulous, it really did feel that I’ve made a real difference.

Prior clinical experience working in OH was considered a valuable quality to support case managers in taking on the new role. Those without significant OH clinical experience to draw on were therefore considered less likely to feel confident in facilitating discussions with clients in terms of problem-solving and action-planning. Moreover, incorporating their new skills and knowledge when delivering the case management intervention meant that some case managers felt suddenly less capable in their OH role than they had felt beforehand. Adopting a new style of working had initially slowed down their delivery of OH clinical practice and there was a consensus that proficiency in the application of their new case manager skills takes time to master. It was generally agreed that ongoing opportunities for case managers to use their new skills would help them to build the required competencies and confidence to deliver the case management intervention in the future. Several case managers reported accessing collegial support from other local case managers during the study, and these interactions were found to provide a useful opportunity for case managers to mentor each other as skills were developing:

[. . .] probably a little bit wooden, you have your own style of doing things and then when you’ve got to change that, I think it feels unnatural, probably not to the person that you are actually seeing, they probably don’t notice any difference but I guess I went from being able to do consultations almost on, not quite on autopilot because you are thinking about it, but in terms of the way you ask things to actually having to think about how I was phrasing every question, whether it was an open question or a reflected statement back and things like that.

We did support each other, yes, we talked about the cases afterwards, we’ve got quite a supportive team anyway and we’ve had to be over the last sort of few years, so we do discuss difficult clients and difficult cases, there’s always time at the end of the day to be able to do that and I’m [case manager’s] line manager so we often get together and have chats about things, but yes we certainly did support each other on this.

In terms of the timing for delivering the training, some case managers expressed concern that knowledge and skill development were likely to diminish if they were not used soon after training, and others felt that an opportunity for refresher training to underpin skills development would be practically beneficial and would help to prevent a decay in skills:

I think the actual sessions themselves were really good. I seem to remember there was a bit of a gap between doing the case management training and actually starting the recruitment onto the study, so actually by the time we started the recruitment process I probably did not feel as fresh as I could have done and I did have to go back several times through some of the slides and bits and pieces to re-get my head round some of the processes that we were supposed to be doing.

In particular, the assessment of CMDs and the suicide risk assessment components of the training were positively received and provided case managers with a useful opportunity to enhance their existing clinical knowledge and skills in this area or, for more novice OH case managers, acquire important foundation knowledge and new practical skills that they could then use when delivering the case management intervention. Most case managers were found to adopt a flexible yet holistic approach when assessing risk-related concerns (e.g. exploring suicide/self-harm and work-related issues). However, several case managers felt that the ‘assessment of risk’ predominantly related to a person’s capacity to perform their job role instead of suicide or self-harm:

I probably did feel more comfortable with asking the questions around suicide and deliberate self-harm and things like that than I have done previously.

The roleplay exercise was considered by some case managers to be a useful component in which to practice MI techniques during training; however, some felt that more time should be given for experiential learning so that new skills and techniques could be rehearsed in different scenarios. It was also noted that these practical roleplay sessions could make case managers feel anxious; they felt genuinely concerned that peers may negatively assess their capability to deliver the case management intervention effectively, preferring instead to incorporate new knowledge and to practise new skills and techniques in real-life situations:

There was one particular session, it was all really useful, it was one particular session where we did, normally I hate roleplay but we did a roleplay session and I found that really, I think that was towards the end of the [inaudible] once we’d had the basis of it, I found that really helpful in terms of applying it and that really made me sort of think about how you have to alter your perception of supporting people, so rather than advising people to do certain things, it was about turning it around and allowing people to come to their own decisions and conclusions, so I found that particular session very very useful from it all.

Generally, case managers had a preference for making use of the Microsoft PowerPoint^® 15.0 (Microsoft Corporation, Redmond, WA, USA) slides that had been used during the training, as well as referring back to personal notes to familiarise themselves with the content post training, particularly with regard to the action planning and problem-solving components, and only a few case managers found it necessary to access the video-recordings of the workshop to support them in their new role:

[T]he slides were what was presented so that tended to be what I then reflected on, mentally relating back to the session.

Independent assessors at intervention/mixed intervention sites completed an audit form for 11 participants in the study. The purpose of this exercise was to assess delivery of each component of the case management intervention.

The OH case managers described the training workshop in positive terms and expressed the view that they had acquired the necessary new skills and competencies to deliver the case management intervention, particularly in relation to conducting a comprehensive clinical occupational history and mental state assessment, engaging and guiding participants in problem identification, problem-solving and action planning, as well as risk assessment. Post training, case managers provided useful feedback in terms of where refinements could be made to the content and method of delivery. This included the need to ensure that training provided additional information with respect to the basic aetiology and epidemiology of CMDs, and approaches to problem-solving that are specific to mental health symptomology. Moreover, the case management training workshop provided a valuable opportunity for the case managers to build on and enhance their existing clinical and interpersonal skills.

The independent audit of case notes found evidence that all participants in the intervention arm had engaged in the interactive problem identification and problem-solving components of the intervention and there was evidence that a formal RTW plan had been devised for all participants. All participants were provided with the sleep hygiene leaflet and the majority were signposted to follow-up counselling support provided either in the workplace or by an external provider. All line managers were given a copy of the formal OH report following delivery of the case management intervention. No participants were found to have engaged in local peer support, nor was there evidence that case conferencing (with line manager/sick-listed employee and case manager) or workplace visits had been required.

Discussion

Overall, we found that it was both feasible and acceptable to train experienced OH nurses as case managers. However, based on the workshop evaluation scores and qualitative feedback, it is important to carefully consider the prerequisite criteria when selecting OH nurses to be trained in delivering the intervention.

Prior OH experience was found to be a key attribute, since this provided a foundation on which newly trained case managers could learn and grow, coupled with benefits of having access to local collegial support from other trained case manager colleagues. A strength was the professionalism, enthusiasm and commitment of the case managers during their involvement in the study. Nevertheless, it is important to recognise that skill decay may occur relatively quickly if the new clinical skills and knowledge required to deliver the OH case management intervention are not applied in clinical practice soon after the training. This has important implications in terms of the timing of case manager training sessions, and the potential need to offer refresher training and to facilitate access to ongoing mentoring support as new skills are consolidated.

Study design

We hypothesised that the impact of an intervention to improve RTW in NHS staff who go on sick leave with a CMD would best be assessed through a cluster RCT in a variety of NHS trusts, provided that contamination could be avoided (i.e. delivery of the intervention did not inadvertently modify the management of controls). In population, intervention, comparator, outcomes (PICO) format, the design of the RCT that was envisaged can be summarised as follows:

• Population – NHS staff in a variety of NHS trusts who have been on sick leave with a CMD for 7 or more consecutive days and less than 90 consecutive days.

• Intervention – a complex intervention comprising various clinical and workplace provisions that, individually or in combination, are thought likely to be effective in this population (either from direct evidence or because they are known to be effective in other working populations or conditions).

• Comparator – CAU (i.e. the provision of existing OH care as usual and support provided to NHS staff on sick leave with a CMD).

• Outcomes – reduction in anxiety/depression; change in use of medication for CMDs; early, part, full and sustained RTW; change in health-related quality of life and well-being; relapse rates; and adverse events. We assessed the cost-effectiveness of the intervention both from a NHS and a societal perspective. In addition, the financial implications for employers were investigated.

To answer research questions 6 and 7, we recruited six OH departments from those that confirmed that they wished to participate in the study in response to an expression of interest correspondence sent to NHS trusts who participated in the SCIN trial, with one of the site’s being an ambulance trust. We selected NHS OH departments representative of the range of NHS trusts in the UK, providing OH services to a mixture of acute, mental health and ambulance trusts. All included trusts had electronic OH data capture systems, did not have case management in use at the time of the study and employed more than two OH nurses who advised on RTW in workers with CMD. One trust (Royal Papworth Hospital NHS Foundation Trust) withdrew from the study prior to the commencement of participant recruitment, citing OH staff storages.

We allocated the participating trusts to mixed intervention/CAU (groups A and B), CAU only (group C) and intervention only (group D) based on sampling groups, with the aim of achieving broad comparability between the types of trust (Table 10). Sites were not randomly allocated to a particular treatment arm during the study design; instead, for pragmatic reasons, this decision was made by the research team based on feedback from individual sites, such as whether or not they had the capacity to deliver the case management intervention. For instance, OH services at our NHS ambulance trust were delivered by an external OH provider and it was not deemed feasible to implement a new case management intervention or train external OH staff within the existing contractual arrangement.

Methods

Outputs

• An assessment of the feasibility and acceptability of the intervention to workers who go on sick leave with a CMD, line managers, OH departments and HR managers.

• A manual for training case managers.

• An assessment of the feasibility, acceptability and impact fidelity of the training in strengthening motivation and CBT-directed goal-setting skills.

Results

During the recruitment phase, 42 NHS workers were screened for eligibility. A total of 24 participants were consented to take part in the study and 18 NHS workers were not consented for various reasons (e.g. ineligible, declined). It is noteworthy that a proportion of sick-listed staff were already on a RTW plan at the time of their first OH appointment, although we did not collect any data on the duration of their sickness absence prior to their first OH appointment. A Consolidated Standards of Reporting Trials (CONSORT) flow diagram of the screening, recruitment and response rate to the study questionnaires is provided (Figure 5).

FIGURE 5.

The CONSORT flow diagram. a, Total number of staff on sick leave with a CMD during the study at participating sites (does not reflect the number of staff invited to take part or referred to their local OH department). One participant returned a questionnaire at 3 and 6 months but not at baseline.

Table 11 presents the total workforce population size, (estimated) number of staff on sick leave with a CMD, number of sick-listed staff who were screened for eligibility and number of staff who were recruited to take part during the 7-month recruitment period, as well as the follow-up response rates to the study questionnaires, by site. We chose to collate the total size of workforce populations and the total number of staff sick-listed with a CMD during the study period because of the smaller than expected number of participants who were screened for eligibility during the study period and to ascertain how many potentially eligible participants were not referred to OH by their line managers during the study period.

In this study, we found that decisions on why referrals to OH were and were not made by line managers when staff were on sick leave with a CMD varied considerably. For example, we found that there was significant variation across participating sites in the organisational requirements (commonly referred to as ‘trigger points’), as outlined in sick leave policies, for when to initiate referrals to OH for staff who are on sick leave with a CMD. At one participating site (University Hospitals of Leicester NHS Trust), the policy recommends an early referral to OH for any staff who are on sick leave with stress- or anxiety-related conditions, and offers further guidance on what specific line management actions (e.g. contact employee in first week of absence, refer to OH, conduct stress risk assessment) should be taken in these situations. Similarly, at Guy’s and St Thomas’ NHS Foundation Trust, managers are encouraged to manage sickness absence at a local level in the first instance, with OH referrals to be made at 4 weeks of sickness absence. However, the OH service’s policy at Guy’s and St Thomas’ NHS Foundation Trust105 recommends that early OH referrals should be considered in situations where it is unclear if an employee is currently under the care of a health-care professional for their CMD. It is a requirement at West Midlands Ambulance Service University NHS Foundation Trust that staff who go on sick leave with a CMD are referred to OH on the first day of sickness absence. By contrast, other sites provide no policy recommendations specific to the management of CMD presentations or recommend that referrals to OH for any reason should be initiated only when staff have been absent from work for > 4 weeks or if managers observe a regular pattern in periods of sickness absence. With this backdrop, it is likely that requiring line managers to adhere to a different OH referral requirement for the purpose of this research (as stipulated in the study protocol), which is fundamentally incongruent with existing organisational policy requirements, may result in cognitive dissonance for them.

In spite of trigger points outlined in these policies for initiating referrals to OH, this study found that a substantial number of staff who were on sick leave with a CMD during the study period were not referred to OH by their line managers. This reflected similar observations in the previous CAMEOS (CAse Management to Enhance Occupational Support) trial. 23 Furthermore, we found that the proportion of sick-listed staff with a CMD who were referred to OH within 8–28 days was smaller than the proportion reported by the OH providers who participated in the survey of care as usual.

The research team deliberated on the lower than expected recruitment response rate in the context of the overall number of staff on sick leave with a CMD during the study period and concluded that there are several plausible explanations to account for this anomaly:

• Sick-listed staff with a CMD may not have consented to being referred to OH by line managers (possibly owing to a fear of disclosing a CMD to their employer or because they did not want contact with the workplace during their sickness period).

• Employees may not have felt ready to engage with the workplace or OH at a particular point in their sickness absence episode.

• Line managers were adhering to existing organisational policy guidelines about the timing of when OH referrals should be made (i.e. preferring to wait until 4 weeks of sickness absence had passed).

• Line managers felt confident and competent in managing the sickness absence locally (including the eventual RTW process) without the need for OH guidance.

• Possible preferences to delay engagement with OH, the purpose being to seek OH validation once proposed RTW plans had been formulated and agreed between the sick-listed staff and their line manager.

• Misconceptions and a lack of knowledge and awareness among line managers and staff of the role of OH and the support they are able to offer.

• Negative past experiences involving OH discouraging engagement.

• The widespread need, recognised by the Study Steering Committee, to raise the profile of OH across the NHS and improve managers’ OH referral practices.

In addition, this study found that there was a large proportion of cases where referrals to OH coincided with an employee’s imminent or actual RTW, rather than occurring earlier in the period of sickness absence. Unfortunately, the study did not capture the duration of sickness absence for staff who were already on a RTW plan at the time that OH referrals were made. Therefore, it was not possible to determine how many additional participants may have been previously eligible for entry into the study had they been referred to OH at an earlier time point, or whether or not intervening earlier in their episode of sickness absence may have facilitated an earlier RTW.

Moreover, many participating sites were known to offer a dedicated managers’ telephone advice line where managers could obtain initial OH advice on current employee-related matters, including discussions on whether or not a formal OH referral is necessary for an individual case. It is likely that these initial consultative interactions between line managers and OH may have mitigated the need for formal OH referrals to be made. Contrary to this, participants (line managers) in the focus group sessions conducted during the qualitative process evaluation often reported poor or no knowledge of the availability of the managers’ advice line service.

The study also found wide variation across participating sites in the method used for screening participants for entry into the study. For example, at some sites, the lead field worker (OH nurse) reviewed all OH management referral forms as they were received into the department and attempted to identify staff who were on sick leave with a CMD from the details provided by line managers. Those identified as meeting the inclusion criteria were sent a copy of the participant information sheet along with their OH appointment details. However, this was not found to be a wholly reliable approach for screening prospective participants because either ambiguous information was provided on the management referrals forms (e.g. CMD masked as a physical complaint such as back pain or fatigue) or the information described non-specific health reasons (e.g. broad psychosocial factors) as the reason for the sickness absence. In short, this method relied solely on the completeness of background case information provided by managers. This meant that there were occasions where eligible participants were assessed as eligible only during their first OH appointment. There were also several occasions where it was deemed inappropriate to proceed with taking participant consent just prior to the first OH appointment because of the emotional distress individuals were displaying when being approached by the field workers.

Table 12 presents OH referral data from the University Hospital Southampton NHS Foundation Trust (a non-participant recruiting site). We collated these data because we wanted to gain referral information from a trust with excellent data recording systems and that already had an early referral to OH system in place. Moreover, this NHS trust was the site of the pilot case management approach on which our intervention was based. 24 According to feedback from this site, in a 7-month period, 72 staff members on sick leave with a CMD at the time of referral to OH were likely to have been identified as eligible for entry into the study prior to their first OH appointment based on information recorded on management referral forms. Furthermore, there were five cases where it was not clear that the staff members were on sick leave with a CMD until they were first seen by an OH practitioner. In addition, 143 cases were referred with a CMD (confirmed at the first OH appointment) but were not on sick leave at the time of their first OH appointment [i.e. they had returned to work (of any form)].

Data collection tools

We found that the data collection tools developed or adapted and administered during this study were fit for purpose, and completion rates for participants and field workers were high. We concluded that in a future feasibility trial we would use the self-report and ESR data collection methods to capture sickness absence data.

For the health economic data collection tools, a reduced version of the CSRI may be more appropriate.

Response rate to study questionnaires: sample and lost to follow-up

In summary, 18 out of 24 participants returned the baseline questionnaire. Among these, 10 (56%) participants returned the intermediate questionnaire (3 months after entry into the study) and 9 (50%) returned the final questionnaire (6 months after entry into the study). One participant returned both follow-up questionnaires without having returned a baseline questionnaire. The counts of participants by time of questionnaire and intervention arm are shown in Table 13. Several reminders were used to encourage participants to return completed questionnaires to optimise the response rate.

As shown in Table 14, 94% of participants were female, with a mean age of 43 years. In terms of main job role, the largest proportion of participants were categorised as nursing, midwifery and health visiting staff. On average, participants worked > 30 hours per week, and the majority reported working day shifts only. Feedback on consumption of alcohol and history and impact of CMDs is also detailed in Table 14.

The impact of CMDs on occupational functioning was assessed at baseline only. For those participants who reported previous episodes of sickness absence (for any health reason) in the previous 12 months and prior to this current episode, the median number of days on sick leave was 3.8. Seven (39%) participants reported that they expected that the duration of their sick leave would be < 4 weeks and the remaining 11 (61%) participants expected that the duration would be ≥ 4 weeks. Moreover, more than half of participants (59%) reported overall positive job satisfaction, and 12% of participants reported a level of dissatisfaction (Table 15).

Figure 6 presents changes in participants’ perceived capacity to function in the workplace (workability). Higher scores indicate greater difficulty in each domain. Data are presented for the CAU and intervention arms. The results showed that participants were able to answer the relevant questions. The small numbers mean that it is difficult to draw reliable inferences from the results, but there appears to be a trend in the CAU arm to perceive reduced workability at 3 months compared with baseline. This trend is not apparent in the intervention arm.

FIGURE 6.

Self-report measures of participants’ capacity to function in the workplace across different domains. (a) CAU arm; and (b) intervention arm.

In some categories the numbers are very small, resulting in wide CIs.

Return-to-work self-efficacy was assessed at the three time points during the study (Figure 7), with higher scores indicating more positive perceptions of personal capabilities in relation to RTW self-efficacy. The results showed that the participants were able to answer the questions. While the numbers are small, there was a trend towards greater improvement in RTW self-efficacy at 3 months in the intervention arm compared with the CAU arm.

FIGURE 7.

Self-report measures of participants’ self-efficacy associated with RTW across different domains. (a) CAU arm; and (b) intervention arm.

Several measures of self-efficacy were assessed, with higher scores indicating greater levels of self-efficacy. There was an improvement in most measures of self-efficacy for both the CAU and intervention arms. In those measures, improvement seemed more gradual in the CAU arm, from baseline to 3 months and 3 months to 6 months, while improvement in the intervention arm was stronger at 3 months following baseline and scores remained stable at the subsequent follow-up. A few measures of self-efficacy (coping with setbacks and emotional state) showed the opposite effect (i.e. lower scores over follow-up time; see Figure 7).

Perceptions of the availability of collegial support and access to supportive working environments remained relatively static across the six domains in both groups. Higher scores indicate more positive perceptions (Figure 8).

FIGURE 8.

Self-report measure of perception of workplace support across different domains. (a) CAU arm; and (b) intervention arm.

Distribution of outcome measures

Anxiety was assessed using the GAD-7. One participant did not answer one of the seven assessment tool questions at baseline and another did not answer one of the seven assessment tool questions at 6 months. Missing data were imputed as the mean value from the other six questions for those two participants. A score was generated from the seven questions for each of the three time points of the study (baseline and 3 and 6 months). A categorical variable of anxiety was generated using cut-points suggested by the NHS Data Model and Dictionary. 106

Depression was assessed using the PHQ-9. One participant did not answer one of the nine assessment tool questions at baseline and another did not answer one of the nine assessment tool questions at 6 months. Missing data were imputed as the mean value from the other eight questions for those two participants. A score was generated from the nine questions for each of the three time points of the study (baseline and 3 and 6 months). A categorical variable of depression was generated using cut-off points suggested by the NHS Data Model and Dictionary. 107

Differences in the continuous (anxiety and depression) scores from baseline to 3 and 6 months of follow-up were assessed using Wilcoxon matched-pairs signed-ranks tests, and differences in the scores between the two study arms were explored using Wilcoxon rank-sum tests.

Table 16 describes the distribution of outcome measures across the three time points of the study. Over half (77%) of participants reported moderately severe to severe anxiety at baseline. Moreover, there was a trend that this measure had improved marginally by 3 and 6 months. Similarly, over half (61%) of participants reported moderately severe to severe depression at the time of their entry into the study, with a similar trend for improvement at 3 and 6 months, with milder symptoms reported in the CAU arm at 6 months.

Baseline anxiety score was similar between the two arms of the feasibility study [CAU arm, median 15 (IQR 10–17) vs. intervention arm, median 14 (IQR 10–19); p-value for difference = 0.62]. Anxiety score was lower at 3 months (p-value for difference from baseline score = 0.005) and 6 months (p-value for difference from baseline score = 0.02), but the drop in the score was greater at 3 months. Reduction in anxiety score was more pronounced in the intervention arm at 3 months. However, at 6 months, the drop in anxiety score was similar between the two study groups.

As with the anxiety score, the depression score assessed at baseline was similar between the two arms [CAU arm, median 17.5 (IQR 11.5–22.0), vs. intervention arm, median 20.0 (IQR 13.0–21.0); p-value for difference = 0.9]. Depression score decreased at 3 months (p-value for difference from baseline score = 0.005) and decreased further at 6 months (p-value for difference from baseline score = 0.008). The depression score at 6 months was lower among those in the CAU arm than among those in the intervention arm. However, this difference was not statistically significant (p = 0.60).

Use of antidepressants was found to have increased slightly during the study period (Table 17).

Among the 18 participants who returned baseline data, 11 completed the intermediate questionnaire. Among those who completed the intermediate questionnaire, nine (82%) reported returning to work. Among those who returned to work, seven (78%) continued at work for ≥ 4 weeks without further sickness absence. All participants who returned to work returned to modified duties/hours, with most participants subsequently returning to their normal duties (78%) and working hours (89%). Among those who returned to their normal duties/hours, most (63%) stayed at work in their normal duties/hours for > 4 weeks without further sickness absence.

Among the 18 participants who returned baseline data, five completed the 6-month follow-up questionnaire. Among those who completed the 6-month follow-up questionnaire, four (80%) reported that they had returned to work and all reported remaining at work for ≥ 4 consecutive weeks without further sickness absence. Most (3/4) returned to modified duties/hours, with two returning to normal duties and one returning to normal hours. Among the four participants who returned to work, two returned from modified duties/hours to normal duties/hours for > 4 weeks without further sickness absence.

Very few missing data were noted. However, associated dates for events were more poorly filled in. The outcomes relating to RTW (any RTW, RTW and continue for ≥ 4 weeks without any further sickness absence, return to modified duties, return to unmodified duties, return to normal hours, return to unmodified duties/normal hours and continue for ≥ 4 weeks without further sickness absence) were also reported by case managers. Table 18 shows the agreement between self-report and case manager forms for different RTW outcomes, with an indication of missing data. Overall, there was relatively poor agreement between the two data sources.

Comparison between self-reported time off and sick absence from records

Figure 9 shows the time periods of sickness absence for 10 study participants. Light blue lines represent periods of sickness absence reported by participants and dark blue lines represent periods of sickness absence calculated from dates recorded in their ESRs. For each of the 10 participants (pairs of lines) in the figure, the duration of sickness absence in weeks according to self-report and ESRs are shown. In most cases, there is very good agreement between the self-reported dates of sickness absence and dates of sickness absence extracted from ESRs. In a few cases it was observed that, using ESRs, it was possible to extract information on multiple spells of sickness absence.

FIGURE 9.

Line plot showing periods of sick leave with a CMD as reported by participants and as extracted from ESRs for 10 participants (corresponding to the 10 pairs of lines).

Table 19 presents data on consultation times and number of follow-ups. Among those in the case management group (n = 11), consultation time for at least one of the appointments was reported for 10 participants. Among these 10 participants, two had missing information on consultation time for the first appointment. Among those in the CAU arm (n = 12), consultation time for at least one of the appointments was reported for 11 participants. Among these, one participant had missing information on consultation time for the first appointment. Longer consultation times at first and subsequent appointments occurred for participants who received the case management intervention than for participants receiving CAU. Table 20 presents the number of follow-up appointment sessions per case. Those in the case management intervention had more appointments than those in the CAU arm.

Table 21 presents data from the 6-month questionnaire relating to self-reported uptake and acceptability of components of the case management intervention. Overall, the majority of participants who accessed and used the resources found them useful. Although the numbers are small, the RTW leaflet and the MAP were generally well received; there was less agreement about the usefulness of the other resources.

Health economics results

Client Service Receipt Inventory data were available for 18 participants at baseline and 10 participants at follow-up, with five participants from each arm (Table 22). The baseline data are considered as one group in the reporting of these results. At baseline, data on GP contacts were available for all participants. Data on the use of other services were available for 13 or 14 participants, depending on the service. At follow-up, data on GP contacts were available for all participants who completed the CSRI. Data on other services were available for more participants in the intervention arm than in the control arm. Data on other talking therapies in particular were missing for the control arm.

TABLE 22 - Total number of respondents providing service use data from the CSRI

Service	Baseline		Follow-up
	Intervention arm	CAU arm	Intervention arm	CAU arm
GP	10	8	5	5
Practice nurse	8	4	5	4
Psychiatrist	8	5	5	3
Other doctor	8	8	4	3
Counsellor	9	5	5	3
Psychologist	8	5	5	3
Social worker	8	5	5	3
Community mental health worker	8	6	5	3
Occupational therapist	8	5	5	3
IAPT	8	6	5	3
Other talking therapy	8	5	5	2
Online therapy	8	5	5	3
Fit for Work service	8	5	5	3

Table 23 shows the number of participants using specific services. At baseline, all participants had GP contacts. Most other services were used by relatively few participants. None of the participants reported seeing psychologists or social workers. At follow-up, all participants reported having seen a GP. There were relatively high levels of use of counsellors and occupational therapists. However, a large number of services were not used.

TABLE 23 - Total number of respondents providing service use data from the CSRI at baseline and follow-up by study arm

Service	Intervention arm				CAU arm
	Baseline		Follow-up		Baseline		Follow-up
	Yes	No	Yes	No	Yes	No	Yes	No
GP	10	0	5	0	8	0	5	0
Practice nurse	2	6	0	5	1	3	2	2
Psychiatrist	1	7	0	5	0	5	0	3
Other doctor	1	7	1	3	1	5	0	3
Counsellor	3	6	4	1	1	4	3	0
Psychologist	0	8	0	5	0	5	0	3
Social worker	0	8	0	5	0	5	0	3
Community mental health worker	0	8	0	5	1	5	1	2
Occupational therapist	2	6	3	2	1	4	2	1
IAPT	0	8	0	5	1	5	0	3
Other talking therapy	0	8	0	5	1	4	0	2
Online therapy	1	7	1	4	0	5	0	3
Fit for Work service	0	8	0	5	1	4	1	2

Completion rates for the EQ-5D-5L were very good. All 18 participants provided scores at baseline and at follow-up; only the usual activities domain had a missing value (for the control arm).

Figures 10 and 11 present the distribution of EQ-5D-5L responses at baseline. For mobility and self-care, most participants reported ‘no problem’ and the only other response was ‘slight problem’. There were greater distributions of responses for the other three dimensions, although no participants reported having extreme problems.

FIGURE 10.

Distribution of EQ-5D-5L responses at baseline: intervention arm.

FIGURE 11.

Distribution of EQ-5D-5L responses at baseline: CAU arm.

In Figure 12 we see that, for the intervention arm at follow-up, there is limited variation for mobility and self-care but a wider distribution for the other three dimensions. At follow-up, participants in the control arm reported ‘no problem’ or ‘slight problem’ for three dimensions (Figure 13). Usual activities showed most variation.

FIGURE 12.

Distribution of EQ-5D-5L data at follow-up: intervention arm.

FIGURE 13.

Distribution of EQ-5D-5L data at follow-up: CAU arm.

The mean EQ-5D-5L weights for the intervention arm were 0.786 (SD 0.133) at baseline, increasing to 0.836 (SD 0.109) at follow-up. For the CAU arm, the mean EQ-5D-5L weights were 0.649 (SD 0.155) at baseline and 0.870 (SD 0.166) at follow-up.

The intervention costs comprised £10,788 for training, £5025 for materials and £954 for therapy delivery. For the 11 recipients in the intervention arm, the therapy costs per person were £87.

In summary, this feasibility study has shown that the CSRI was reasonably well reported. However, it was apparent that most services were not used and, therefore, in a full trial a reduced version of the CSRI may be appropriate. The EQ-5D-5L was very well completed, but for some dimensions there was very little variation in responses. The intervention costs consisted of fixed elements for training and materials. It is unclear how these should best be apportioned, and they may be best considered as ‘sunk costs’ that do not recur. The costs of the therapy sessions per person were relatively low.

Discussion

Method

We worked with our stakeholder group, including our PPI representatives, to review the information from the feasibility study and to determine whether or not a cluster RCT could be designed that was free from unacceptable contamination and avoidable logistical constraints. To inform finalisation of the manualised intervention and formulation of recommendations for design of a RCT, we reviewed the findings from the various qualitative, quantitative and economic analyses that were undertaken in this feasibility study. We paid particular attention to the response rate and likelihood of contamination to determine whether an individual or a cluster trial would be the more appropriate method for a definitive trial.

Proposed revised recruitment method

In Kenning et al. ’s study,23 potentially eligible staff who were on sick leave were identified by scanning organisations’ sickness absence reporting databases and sending out participant information sheets directly. A similar approach could be readily adopted for a future evaluation of our case management intervention while complying with current common law duty for confidentiality. Identification of prospective eligible study participants from the screening of routine records followed by sending study invitations is a method that has been used successfully in previous studies. 109,110 During data analysis, and at the time that this report was produced, the Guy’s and St Thomas’ NHS Foundation Trust Information Governance Committee (represented by the chief medical officer, the newly-appointed Caldicott guardian and others) was consulted and provided confirmation that the following method to identify and screen sick-listed staff with a CMD would be acceptable to implement:

• HR colleagues will identify sick-listed staff with a CMD by searching the code S10 in ESRs.

• HR will be responsible for sending out the participant information sheet to sick-listed staff with a CMD to inform them of the study and to opt in, should they wish to participate.

It is notable that we approached the former Caldicott Guardian at Guy’s and St Thomas’ NHS Foundation Trust in 2017 to explore the proposal of identifying sick-listed staff with a CMD for our study from ESR data and, at this time, we were advised that this approach would not be possible owing to data protection restrictions; more specifically, the OH department would not have authority to access these data for the purpose of identifying and approaching potentially eligible sick-listed staff for inclusion in a research study. In 2019, however, this method was presented to the Caldcott Guardian and Information Governance Committee with the suggestion that HR be responsible for sending out the participant information sheet directly to sick-listed staff. This revised method was approved for use in the future evaluation study on the basis that it did not contravene current data protection regulations. Unfortunately, we are unable to provide any reliable recruitment figure estimates using this alternative method without first undertaking further development work.

Outputs

• Complete specification of the intervention package and potential staff and cost implications.

• Recommendations for future development work and design for a main study.

Results

Finalisation of the manualised intervention

Figure 14 provides an overview of the final manualised case management intervention, taking into account the results from the feasibility study.

FIGURE 14.

Overview of the final manualised case management intervention.