Individual supported work placements (ReISE) for improving sustained return to work in unemployed people with persistent pain: an internal pilot study of a cohort randomised controlled approach

Abstract

Background

Persistent pain is a frequent cause of sick leave and work disability in Norway. A return-to-work intervention featuring supported work placements, developed in the UK, demonstrated feasibility, and a return-to-work rate of 20% within 6 months was observed in the sample. We sought to adapt the intervention for delivery in Norway and to confirm feasibility prior to a full-scale trial.

Methods

In this internal pilot, we used a pragmatic cohort randomised controlled approach with national recruitment in Norway. We recruited people who were unemployed (for at least 1 month), having persistent pain (for at least 3 months), aged between 18 and 64, and wanting to return to work. We initially recruited people to an observational cohort study of the impact of being unemployed with persistent pain. After baseline measurement, we randomly sub-sampled participants to whom we offered the intervention, which featured individual case management and support, work-familiarisation sessions, and the offer of a 6-week part-time unpaid work placement. We assessed recruitment rates (aiming to recruit 66, and sub-sample 17 within 6 months); optimal recruitment pathways; intervention acceptance rates; the feasibility of data collection; using video links for work-familiarisation sessions and remote case manager support.

Results

The pilot ran from June to November 2022. Of 168 people expressing interest, 94 consented. Recruitment posts on Facebook yielded the most ‘expressions of interest’ (66%, n = 111). After screening for eligibility, we included 55 participants. Of these, 19 were randomised to be offered the intervention. Of these, less than half (n = 8) consented to intervention participation. Remote case manager and work-familiarisation sessions appeared feasible. Following a delay in identifying placements, three participants received offers of work placements, with one starting and completing during the pilot period. Data collection methods were feasible, and no adverse events were reported.

Conclusions

Recruitment and logistical processes, such as remote management by video link, are feasible. However, delivery of the intervention is challenging. In particular, sourcing placements and the time required for identifying appropriate placements was more challenging than anticipated. A full-scale trial is feasible but will require improvements to the placement identification processes.

Trial registration

ISRCTN85437524 (Referring to the ReISE trial, of which this internal pilot was a part), Registered 31 of May 2022 https://doi.org/10.1186/ISRCTN85437524

Trial funding

Norwegian Research Council.

Key messages regarding feasibility

• What uncertainties existed regarding the feasibility? 

  A case manager supported work placement intervention for helping people who are unemployed with persistent pain to return to work was acceptable and feasible to deliver in the UK, notwithstanding some recruitment and placement acquisition challenges. However, it was not clear whether recruitment, intervention acceptance, and delivery would be feasible in Norway within a cohort randomised approach as culture and welfare contexts differ from the UK. Additionally, the feasibility of video management sessions to enable national recruitment needed to be explored.

• What are the key feasibility findings?

  Recruitment, randomised subsampling within a cohort randomised approach, the use of video sessions to replace face-to-face sessions, and data collection methods were all feasible. However, the intervention acceptance rate was lower than anticipated, and finding placements in a timely manner was challenging, which, without attention, could threaten delivery of the intervention.

• What are the implications of the feasibility findings for the design of the main study?

  A higher subsampling rate would allow more people to receive the intervention in a full-scale trial. As intervention effects within an intention-to-treat analysis would be diluted by a low acceptance rate, a secondary complier average causal effect (CACE) analysis may aid interpretability of the causal effect of the intervention, and compliment testing of the effect of the offer. Streamlining the placement identification processes, and ensuring sufficient human resources for placement identification, may be key to the success of a full trial.

Background

Persistent (chronic) pain, most of which has a musculoskeletal origin, is a frequent cause of sick leave and work disability in Norway [1, 2]. Between 2006 and 2019, the total health burden related to musculoskeletal disorders in Norway, measured in disability-adjusted life years (DALYs), increased by 14% [3]. In 2022, around 10% of the Norwegian population of working age (18 to 67 years) received disability benefits, with around one-third being due to musculoskeletal disorders [4]. Moreover, the Norwegian Labour and Welfare Administration (NAV) has estimated an increase in people on disability benefits by 2.7% in 2023 and 2.5% in 2024, which will be an increase of around 20,000 people [5, 6]. Being unemployed increases the risk of poor physical and mental health [7, 8]. Conversely, return-to-work (RTW) may improve recovery and quality of life and reduce poverty [8,9,10]. The total socio-economic costs of musculoskeletal disorders in Norway were estimated to be EUR 22 billion in 2016; a substantial increase from a 2009 estimate of EUR 6 billion [11, 12].

Despite an increasing body of research related to RTW interventions, it is uncertain how to best help people with pain return to work [13, 14]. For a sick-listed pain population, there are indications that interventions of low intensity [15], multidisciplinary [13, 16], or work-focused care [10] may be beneficial. However, it is unclear if this will work in a population of people who are unemployed with persistent pain (including those on full disability or work-allowance benefits).

For a 2020 UK feasibility study, we developed a novel RTW intervention for unemployed people with pain [17]. Prior underpinning work for the developed intervention included (1) a systematic review of supported employment interventions [18], (2) a meta-ethnography on the perceptions of obstacles to RTW of people with persistent pain and employers [19], and (3) in-depth interviews with people off work due to chronic pain, and employers about their perceptions of obstacles to RTW [20]. The resultant intervention addresses the challenges individuals with chronic pain report facing when returning to work. It is grounded in supported employment principles, incorporating components such as pain management, workplace adjustments, and relationship-building activities [21, 22]. The core of the intervention involved training case managers to perform work ability assessments with participants to identify obstacles and then collaborative work on navigating obstacles, within a temporary work placement that is matched to skills and aspirations. We observed a 20% return-to-work rate in a 6-month follow-up from the piloting of the developed intervention, and the intervention was acceptable and valued by the people involved. Lessons learned included avoiding recruitment from pain clinics and the need for early engagement and identification of placements to secure matching of participants’ skills and aspirations to a placement job.

Our aim was to adapt the intervention for Norwegian consumption and to assess the feasibility of processes on a small scale ahead of delivering a full-scale cohort randomised approach testing a supported work placement intervention for unemployed people in Norway with persistent pain.

Objectives

Our objectives for this pilot study were to adapt the intervention from a UK to a Norwegian context and to assess whether we were able to (1) recruit participants with persistent pain who are unemployed, (2) use face-to-face sessions and sessions with video to enable national recruitment, (3) assess the feasibility of data collection, (4) determine the intervention acceptance rate, and (5) confirm we could deliver the intervention within a cohort randomised design.

Methods

Study design and setting

We adapted study material used in the UK feasibility study, engaging user representatives during the various stages of adapting the intervention to the Norwegian setting [23, 24]. User representatives were involved in the design of the proposal, advising on intervention adaptation (e.g. one view of the user group was that it would be beneficial that case managers were unaffiliated with NAV) and study design (e.g. contextual and institutional (NAV) obstacles of conducting the intervention nationally). A group of six lay members contributed to the development of participant facing material (e.g. information leaflets, consent forms, website content, and recruitment posters) and with additional recruitment strategies through biannual meetings and via email as needed.

The study protocol is described in detail elsewhere [23]. Briefly, we used a cohort randomised controlled approach to avoid contamination and ethical issues of non-consent that can be present in other designs performed for similar reasons (e.g. those associated with Zelen’s design) [25, 26]. Additionally, there is a lack of clarity surrounding what constitutes ‘usual care’ for this population and the impact of being out of work with chronic pain in Norway. Thus, all were initially recruited to an observational cohort study on the impact of being unemployed with persistent pain. After baseline measurement, we randomly selected one in three (altered to one in two during the pilot due to low intervention acceptance rate) to be offered the intervention.

The Southeast Regional Committee for Medical and Health Research Ethics assessed the pilot study was outside of the remit of its governance (record no. 384060), i.e. the Norwegian Act on Medical and Health Research, although they gave a favourable view of a protocol for the overall full-scale trial (record no. 402918) [27]. A Data Protection Impact Assessment was conducted and reported to Sikt (formerly, the Norwegian Centre for Research Data) for review, as required in Norway (record no. 693603). Data were stored and analysed using the University of Oslo’s Services for Sensitive Data (TSD, Oslo University, Oslo). TSD provides a secure platform for researchers to collect, store, and analyse sensitive data, integrating data collection tools such as the survey tool ‘Nettskjema’ (Oslo University, Oslo). The ReISE trial (including internal pilot) was prospectively registered (ISRCTN85437524). Written (electronically signed/paper) informed consent to participate was obtained from all cohort and intervention participants.

Participants

Eligibility

We recruited people aged between 18 and 64, who had been unemployed for at least 1 month, had pain for more than 3 months, and who wanted to return to work. Our protocol includes provisions to exclude those who had a job to go back to (e.g. those on sick leave), and those with insufficient English or Norwegian language skills to give consent, to reflect the fluency required to enter the open labour market in Norway.

Recruitment

We recruited participants using social media, charities (including pain and rheumatological charities), Finn.no (a job advertisement platform), and frisklivs-sentraler (community health centres). We offered NOK 200 (approximately EUR 17) in universal gift cards that can be redeemed in various retail outlets in Norway, which we have previously found to help recruitment [28, 29]. Potential participants were directed to either an online website containing the participant information leaflet and an expression of interest form, or to contact the study team for further information. Potential participants also had the option of receiving the information leaflet and consent form by post.

Training case managers and facilitators

We intended to recruit three case managers and two facilitators with clinical backgrounds and experience of working with people with persistent pain. Our training package was based on the Psychosocial Flags Framework—a practical framework for understanding and evaluating psychosocial influences in musculoskeletal problems [21]. The approach involves identifying obstacles to work and then collaboratively developing and agreeing solutions to those obstacles across the personal, workplace, and social domains. We adjusted the content of the UK’s 2-day work-familiarisation course to fit a Norwegian context and gave additional training on the work-health interface (i.e. the relationship between work and health) to the facilitators who led these sessions.

Intervention

A case manager contacted consenting participants for an initial discussion to explore work aspirations and skills and to collaboratively identify obstacles to work. Our partner, Manpower (a nationwide recruitment company), then aimed to match the participant with a work placement, based on skills, aspirations, and suitability. Once a placement was arranged, the case manager, participant, and placement manager met to collaboratively agree an individualised work plan, including any temporary work accommodations needed to help navigate individual work obstacles [30, 31]. The case manager also aided clinicians in providing work-focused healthcare by providing evidence-based leaflets [32]. As part of adapting the intervention into a Norwegian context, we culturally adapted three leaflets: (1) Advising Patients About Work (aimed at general practitioners), (2) Health and Work (aimed at the general population), and (3) Work and Health (aimed at employers/HR), originally commissioned to support the UK government work ability initiatives [33,34,35]. These cultural adaptations are described elsewhere [24]. These leaflets were given to all relevant participating parties.

Prior to the 6-week part-time (up to 16 h a week) unpaid work placement, participants attended a 2-day work-familiarisation session in groups of two to five people. The aim of the session was to increase self-efficacy and to help embed the person into their role. Within a biopsychosocial framework, the session contained practical advice about working with persistent pain including self-reflection exercises, myth-busting, work adaptations, and the management of pain in the workplace.

We used the online video platform Zoom (Zoom Video Communications, San Jose, CA) for case manager and work-familiarisation sessions, in cases where face-to-face meetings could not be arranged. Any placement-specific training required by the placement provider was given by the placement provider. The case manager provided regular and as required support to the participant and placement manager and referred to work-focused healthcare if necessary.

Cohort (comparator group)

In line with the pragmatic trial design, we did not restrict access to usual support nor exclude those participating in other return-to-work schemes [36]. Participants in the observational cohort study were free to follow any usual care as may have been prescribed by their healthcare provider and/or NAV.

Data collection

We recorded the number of people (1) filling out the expression of interest form, noting where they heard about the study, and their preference of postal or online information, (2) consenting to participate in the cohort study, (3) being screened for eligibility, and (4) completing baseline measurement. After baseline measurement, we recorded (1) the number of participants randomised to be offered the intervention and (2) the acceptance rate of intervention offers. Specifically for the intervention group, we asked for (1) reasons for declining the offer and recorded (2) case manager assessments completed, (3) work-familiarisation session attendance, (4) offers of work placements, (5) work placement attendance, and (6) details of support given (e.g. case manager report forms). Questionnaire data were collected at baseline. We explored the feasibility of delivering supportive intervention components (i.e. case manager assessments and work-familiarisation sessions) by video link that would enable national recruitment, by collecting feedback from participants and discussing with case managers, facilitators, the user group, and the trial management group.

We tested the feasibility of administering the outcome measures, which in addition to demographic data contained the following validated questionnaires: health-related quality of life (EQ-5D-5L) [37, 38], the short Warwick-Edinburgh Mental Well-being Scale [39, 40], PROMIS-29 [41, 42], and the Work Ability Score (as a single item taken from the Work Ability Index) [43, 44] (Table 1).

Table 1 Outcome domains and measurement instruments (Norwegian/English versions as required) at baseline

Sample size

Our aim was to recruit 66 cohort participants with 17 participants sub-sampled and randomised to be offered the intervention within 6 months. This was partially based on testing recruitment pathways (i.e. using different means, ways, and places to approach potential participants for the study) and that we would be able to recruit at a sufficient rate to enrol 228 over a recruitment period of 18 months, which would be required in a full trial. The sample size calculation for a full trial is described in details elsewhere [23]. Additionally, we estimated that 17 sub-sampled participants would permit sufficient throughput for us to regularly form small groups of people for the work preparation session [45].

Randomisation

Randomisation with minimisation was automated using a mix of Bash (Free Software Foundation, Boston, MA) and Stata scripting (StataCorp, College Station, TX). At the core of the process, the Stata program ‘rct_minim’ was used [46, 47]. The first participant was randomly allocated with equal probability, and then for each subsequent participant, there was a two-thirds probability that allocation was to the group that best minimised the imbalance on gender on three levels (male, female, prefer not to say) and three levels of age (18 to 29, 30 to 49, and 50 and over) and with an overall 1:3 ratio of allocation to intervention [48,49,50].

Data analysis

As this was a pilot trial, data analysis was descriptive only. We had no intention to undertake between-group comparisons. Participant characteristics from baseline measurement are summarised as mean and standard deviation for normally distributed data, or median and range, or counts and percentage for categorical data, as appropriate. To assess trial data completeness, the proportion of participants contributing data and completeness of case manager report forms (e.g., details of interaction, activities, and durations) were reviewed by the trial working group.

Results

Recruitment

Between June 2022 and November 2022, 168 people filled out the expression of interest form, with five reporting no interest and three not meeting the inclusion criteria. Most of those expressing interest were recruited from ‘work-allowance/disability’ groups on social media platforms (Fig. 1). We displayed posters and left information leaflets in seven community health centres and engaged with other private clinics (e.g. physiotherapy and osteopathy) indirectly by newsletters and online. Relevant charities and organisations were asked to share a link to our expression of interest on their member platforms and newsletters. We trained two osteopaths and one psychologist as case managers, each with a minimum of 8 years of clinical experience in managing persistent pain. Two physical therapists were trained to function as facilitators; one with over 10 years of experience in delivering RTW courses to people on sick leave, and the other with over 10 years of experience in managing patients with persistent pain.

Fig. 1

Recruitment pathways

Of those expressing interest, 94 (56%) reviewed information and signed the consent form online, self-reporting eligibility, while six (4%) wanted to receive the information leaflet and consent form by post. We contacted those 68 (40%) participants who expressed an interest, but who neither completed online consent nor asked for a postal package. This resulted in seven (4%) additional consents. Seventy people were sent the baseline questionnaire; one replied by post, and 69 responded using the online Nettskjema software (Oslo University, Oslo), giving a response rate of 93% (Fig. 2). When offering the intervention to those randomised, we discovered four participants were ineligible and subsequently screened all participants for eligibility by phone rather than depending upon self-report. We then attempted to contact all existing self-reported consenting participants to confirm eligibility. After excluding nine ineligible people, 19 (29%) had been randomised to be offered the intervention. Of these, eight were accepted, giving an acceptance rate of 42%. Reasons for declining included perceptions of being too incapacitated (n = 5), descriptions of having tried many unsuccessful NAV schemes (i.e. descriptions that were suggestive of intervention fatigue) (n = 2), having insecurities about potential financial consequences in relation to support schemes (n = 2), and planned surgery (n = 1). Mid-way through our recruitment period, we changed the allocation ratio from 1:3 to 1:2, due to lower than expected intervention acceptance rates, to aid the throughput of participants, and to facilitate the organisation of the work preparation sessions.

Fig. 2

Consort diagram adapted from the CONSORT 2010 statement: extension to randomised pilot and feasibility trials [51]. The diagram shows the flow of participants through the study

Participant characteristics

Table 2 shows the baseline characteristics of the two samples. The median number of body areas where participants reported at least ‘moderately bothersome pain’ was six in both groups. The distribution of pain by body area is reported in Fig. 3. Fourteen (74%) people in the intervention and 26 (72%) people in the cohort reported additional health concerns within their mental, neurological, hormonal, reproductive, cardiovascular, rheumatological, and/or musculoskeletal health.

Table 2 Baseline sample characteristics (n = 55)

Fig. 3

Percentage of sample reporting areas of at least moderately troublesome pain over the past month

Intervention delivery

One participant (12%) withdrew from the intervention prior to the allocation of a case manager. Seven initial case manager sessions were conducted using Zoom, whereupon one participant was unable to be contacted after the session. Seven participants (88%) attended the 2-day work-familiarisation session online using Zoom. Although acknowledging that meeting face-to-face has potential added value on a social/interaction level, the overall feedback from participants, facilitators, and the trial management group suggested that it was a feasible way to help those who would otherwise be outside of our geographical reach. Indeed, during the pilot, we were not able to conduct any face-to-face sessions due to all participants being recruited outside of our geographical reach.

Three participants (38%) received an offer of a placement during the pilot period. Of these, one was unable to be contacted, one accepted and started the placement, and one accepted but could not be contacted prior to the work plan meeting. Our placement partner experienced difficulty in finding placements for participants to the extent that without modification to this part of the protocol, a full-scale trial would unlikely be feasible.

Data quality and adverse events

Only one participant (2%) responded by post. The remaining 54 participants (98%) used Nettskjema. These digitally returned forms contained mandatory required fields, thus there were no missing data, although some free-text responses under the category of ‘health resources used’ were incomplete. In contrast, the one postal return had 3% of data missing. Case manager report forms were completed as per protocol. No adverse events were reported from intervention participation.

Discussion

The results demonstrate the feasibility of many aspects of our cohort randomised controlled approach ahead of a full-scale trial while also highlighting important challenges in acceptance rates for intervention offer, identifying work placements, and placing participants. Additionally, participants were primarily recruited through social media, which may pose challenges for generalisability without further efforts to diversify recruitment in a full-scale trial. Feasible aspects include cohort recruitment, automated minimisation randomisation, the workability of video sessions, and data collection methods, suggesting that conducting the intervention nationally would be feasible in the case that placement challenge can be overcome.

In contrast to the UK feasibility study, where a portfolio of placements was set up prior to the intervention, we collaborated with a recruitment company with the strategy being to identify placements following case manager assessment [17]. As a result of the unforeseen inefficiency of this strategy, our partner dedicated two (total FTE = 0.40) recruitment consultants to work closely with the trial working group to identify and help match placements in a time-efficient manner. However, this was implemented after the 6-month pilot phase. As an additional strategy, the case managers collected more precise work-related information from participants to aid identification of suitable work placements; a refinement was also implemented after the pilot phase. This yielded placement offers for all enrolled participants, including those in the pilot, although with major delay.

Several minor challenges were resolved during the piloting, such as minor data collection-related issues (e.g. double-entries causing computer script issues), time delays in intervention offer, and participants’ interpretations of ‘unemployed’ within the inclusion criteria. We focused on refining elements related to the intervention during the piloting phase as an adaptive response to context and system events [52]. We refined logistical processes, such as recruitment strategy, participant flow through the intervention and trial manager role, in collaboration with participants, user group, case managers, and our recruitment partner. No changes were made compared to the intervention, as described in the protocol, during the piloting for the full-scale trial. However, we note that if these data are analysed as part of a trial, the time periods before and after the change in the allocation ratio may need to be modelled as a random effect [23, 53].

While the raw recruitment rate was encouraging, we fell slightly short of target. However, there are indications from trial sampling from similar populations that recruitment is difficult. For example, Sveinsdottir et al., in 2022, included just 58 participants of a planned 80, recruiting from a pain clinic [54]. In our UK-based feasibility study, we found that recruiting from pain clinics was not likely feasible, and the majority of recruitment occurred through ‘job-shops’ [17]. Such job shops do not exist in Norway. While we included pain clinics, as well as Finn.no (a work advertising platform) and newsletters from charities and organisations, most of our participants were recruited from social media. We identified additional strategies to aim to diversify recruitment such as targeted newspaper advertisements, essays in newspapers or similar outlets, and more collaboration with general practitioners’ offices, although time did not permit the testing of these strategies within the pilot period. We acknowledge the potential for a risk of selection bias if sampling were to be disproportionally from a population of those active online within such a group; such people may differ from the general population and indeed those more physically pro-actively seeking work, as might be more comparable to those attending UK job shops. Most expressions of interest from social media came from work-allowance or disability-benefit groups, which may focus more commonly on matters such as benefits system operation and pain/diagnosis rather than returning to work.

The pilot sample was predominantly female (80%), leading us to consider whether we should aim to targeting more male participants in the next phase as the proportion of females with persistent pain in Norway is estimated to be only 57 to 59% [55,56,57]. However, we note that Sveinsdottir et al. (using self-reported gender) and another Norwegian trial by Gismervik et al. (using registry data) respectively reported gender as being 74% and 79% female in their samples from unemployed and sick-listed pain population [54, 58]. It is unclear whether the distribution of pain by sex/gender may need re-evaluation in Norway, or if females may be overrepresented in these samples; if for example, males are less likely to want to join.

We wanted to explore the feasibility of delivering the intervention nationally and opened recruitment nationally expecting to recruit a majority of participants from the Oslo area, where the trial is managed. Conversely, recruitment from Oslo was lower than from outside Oslo, and we were unable to test the feasibility of face-to-face case manager and work-familiarisation sessions during the pilot period, although these procedures were well-tested during our UK feasibility work, and we do not anticipate that this will differ by country.

We observed a low acceptance rate for the intervention during the pilot, leading us to change the subsampling ratio. Intervention effects are likely to be diluted in intention-to-treat analysis of a full-scale trial due to less than half accepting the offer of the intervention; therefore, our view is that incorporation of a secondary CACE analysis will be worthwhile [59]. The context and/or the system within which participants are ‘embedded’ appeared to be a common reason for declining the intervention. Examples of this include intervention fatigue, fear of consequences from NAV (e.g. insecurities about benefit withdrawal/consequences), timing (e.g. in a process of being considered eligible for permanent disability benefit), healthcare providers discouraging participation, and a perception of their work ability and/or pain being a definitive barrier. These observations will be explored further in a planned process evaluation within the full trial. An economic evaluation was unable to be piloted due to an inability (known a priori) to produce registry data within the timeframe of the pilot.

In contrast to some other internal pilot studies, no definitive progression criteria were set a priori; rather, we convened independent Data Monitoring Committee and Trial Steering Committees to review study progress and a view with regard to continuation or stopping [60]. Progression to a full-scale study was ratified by the Trial Steering Committee with a caveat that recommended finding additional sources for placement provision.

Conclusion

Testing a novel intervention, featuring supported case management to identify work obstacles and agree solutions for implementation in work placements, within a cohort randomised controlled approach in Norway is likely to be feasible in many respects. Some aspects will require specific attention in a full-scale trial. These include efforts to diversify recruitment pathways such that not all participants are recruited by social media and identifying and matching participants to placements in a timely fashion.