Background

Musculoskeletal pain is one of the leading contributors to disability worldwide [1]. Low back pain has been the single leading cause of disability globally since 1990, and musculoskeletal pain global prevalence and incidence are increasing [1]. In Europe, musculoskeletal pain is the most common work-related health condition [2], causing 50% of sick leave episodes and 60% of permanent disabilities, generating an estimated cost between 2.6 and 3.8% of the gross domestic product [3, 4].

The prevalence of musculoskeletal pain in nursing staff (i.e. nurses and aides) is around 70–80% in several European countries [5, 6]. In particular, nursing staff who work in hospitals with daily patient handling have almost twice the risk of developing work-related back pain compared with nurses without daily patient handling [7, 8]. Moreover, musculoskeletal pain has a multi-causal origin and is influenced by other non-occupational, individual (e.g. sex, age, genetic factors), psychological, and cultural factors (e.g. health beliefs, somatization) that play an important role in its incidence and management [9].

Musculoskeletal pain clearly requires the adoption of preventive measures in the workplace and the management of nursing staff who already have some complaint, to allow them to continue working and, if necessary, to reintegrate them into the workplace after related sick leave. Several systematic reviews suggest that multicomponent interventions that combine several specific approaches and those that incorporate participatory elements in the intervention process can obtain better results [10,11,12,13]. Therefore, a combination of interventions including primary prevention (aiming to protect healthy people from the possibility of becoming ill or suffering an injury), secondary prevention (early diagnosis and action once the first symptoms or signs of the disease have appeared), and tertiary prevention (to manage long-term health problems, as chronic pain, functional capacity or sick leave) is needed.

In healthcare settings, adapting to the continuous development of new treatments, interventions, and approaches with limited resources is a crucial challenge [14]. Decision-makers benefit from economic evaluations on the efficiency of interventions [15]. In the last decades, the use of patient-reported outcome measures (PROMs) has been increasing. PROMS are usually questionnaires that capture a person’s perception of their own health, as, for example, musculoskeletal pain or quality of life [16]. Their use is highly recommended not only for effectiveness evaluation, but also for economic evaluations as these measures help to understand the patient-relevant outcomes and can be used to influence healthcare decisions on effective therapies, along with clinical and economic outcomes [17, 18].

This economic evaluation is framed in the “INTEVAL_Spain” intervention [19]. It was an intention-to-treat, cluster-randomized controlled trial that aimed to evaluate a multifaceted, evidence-based, workplace intervention comprising three levels of prevention. The intervention included participatory ergonomics: a healthy lifestyle promotion programme with Nordic Walking, mindfulness, and a healthy diet (primary prevention), and a tailored case management programme (secondary and tertiary prevention). Effectiveness evaluation showed that the intervention was effective to reduce neck, shoulder, and upper back pain [20] and suggested that a multifactorial approach is needed to reduce musculoskeletal pain.

The objective of this study was to evaluate the cost-effectiveness and cost utility of this complex workplace intervention designed to reduce musculoskeletal pain in nursing staff.

Methods

The INTEVAL_Spain Intervention

The study was designed as a two-armed, cluster-randomized, controlled trial with 1-year follow-up. Full details of the INTEVAL_Spain protocol, effectiveness analysis, and process results are available elsewhere [19,20,21].

Study Context and Population

The intervention was conducted from September 2016 to December 2017 in two main tertiary hospitals in Barcelona and Sabadell (Spain), each of which had a workforce of 4000 workers, of whom around 60% were nursing staff, and had an experienced in-house occupational health service (OHS).

Clusters were independent hospital units and participants were the nurses and aides working in these hospital units. A total of eight clusters were selected and randomized, with the respective nursing staffs numbering between 20 and 60 in each cluster. All nursing staff from these units were eligible for the study. Exclusion criteria included temporary nursing staff who had worked less than 3 months, those who worked in several units, and those who were on sabbatical leave.

Information briefings were held in person in each cluster before randomization, to collect informed consent and administer the ERGOPAR baseline questionnaire [22, 23]. The questionnaires were anonymous. Follow-up questionnaires were administered at 6 and 12 months. Moreover, the nursing staff could sign the informed consent and join the study at any time during that year (e.g. a nurse who did not take the baseline questionnaire could fill in the follow-up questionnaires and/or participate in the activities).

Interventions

Intervention Group

This study aimed to reduce musculoskeletal pain through an intention-to-treat intervention comprising three levels of prevention. It consisted of (1) participatory ergonomics (primary prevention), (2) a healthy lifestyle promotion programme composed of Nordic Walking, mindfulness, and a healthy diet (primary prevention), and (3) a tailored case management programme (secondary and tertiary prevention).

Participatory ergonomics training was based on the ERGOPAR method and was led by an expert ergonomist of the OHS of each corresponding hospital [22]. Briefly, an Ergo group was created and included the ergonomist, a volunteer nurse, and/or nursing aide from each shift (morning, afternoon, and two night shifts), the unit supervisor/s, one prevention delegate (i.e. representative of workers in the field of occupational risk prevention), and the project coordinator. They held 1-h weekly meetings for 3 weeks to identify the ergonomic problems in their hospital unit, prioritize them, and propose preventive measures. In between these meetings, the volunteers of the Ergo group held discussions with their co-workers in the work units about the subject discussed in the meetings, to provide feedback for the next meeting. Finally, the OHS led an Operative group composed by the key managers of each hospital (e.g. managers from the economics department, human resources, cleaning/maintenance, etc.). This group held monthly meetings to implement the preventive measures (e.g. purchase of sliding sheets and footrests, repair, and maintenance of the wheels of the medication carts, redistribution of the rooms in order to facilitate mobility, etc.).

The healthy lifestyle promotion programme included three group activities: Nordic Walking training [24], mindfulness course based on Mindfulness-Based Stress Reduction [25], and a 3-h formative session about the Mediterranean diet carried out by an expert chef [26].

The tailored case management programme consisted of a multidisciplinary care system for the early detection of disabling musculoskeletal pain and, if necessary, support for return to work. Nursing staff with musculoskeletal pain self-referred to the case manager or were referred by their supervisor or an occupational health physician. A case manager generated a risk profile (i.e. low, medium, or high risk for persistent musculoskeletal symptoms) through a telephone interview using a questionnaire composed of several scales [9, 27,28,29,30,31,32]. Low-risk cases attended a health-beliefs educational session related to musculoskeletal pain. Medium- and high-risk cases were discussed in the weekly clinical session of the OHS to decide the specific treatment, including medical consultation with an occupational physician, rehabilitation service (physician consultation and physiotherapy), and cognitive behavioural therapy by a psychologist.

The implementation of components was progressive to facilitate participation and logistics [21]. All the interventions were conducted during time off work.

Control Group

The OHS carried out the usual occupational health practices for the intervention and control units. These practices included investigation of occupational injuries, risk assessments, smoking cessation programmes, health surveillance, training, expert assessment in occupational health, recommendations of workplace adaptations, and clinical support and management of permanent disability.

Economic Analysis

An economic evaluation was conducted from both a societal and a health system perspective. The societal perspective included the costs of all types of health services (direct costs) and productivity loss costs (indirect costs). The health system perspective included only direct costs from the Spanish public healthcare services. The time frame of this study was 1 year, and consequently, no discount rate was applied. All costs are expressed in euros (€) for the reference year 2022.

Costs

Cost data were collected for the entire study period and for 1 year before the study from several sources specified below. Costs of 1 year before the study were collected in order to adjust the analyses. Table 1 shows the unit costs.

Table 1 Unit cost and adjusted means during the study period by group
Full size table

Direct Health Costs

Direct health costs included hospital costs and primary care use of resources.

Hospital costs were obtained from a detailed analytical accounting system implemented in Hospital del Mar (Barcelona) [33] for the reference year 2017 and updated to 2022 using the Spanish Healthcare Consumer Price Index. Hospital services were classified using homogeneous and functional groups and included traumatology surgery, traumatology hospitalization, hospitalized rehabilitation (i.e. physiotherapy sessions), ambulatory rehabilitation, ambulatory traumatology, traumatology emergencies, and diagnostic tests related to musculoskeletal pain.

Primary care costs were calculated by multiplying the number of health service provided by their unit cost. Primary care services included musculoskeletal pain-related physician visits, diagnostic tests, and medication. Data were obtained from the Shared Clinical History in Catalonia [34]. Unit costs of physician visits and diagnostic tests were obtained from the tariffs corresponding to the services provided by the public provider (Catalan Institute of Health) for the year 2022 [35]. Costs of medication were calculated as the mean price of each medical prescription; that is, when a medication had several formats, we summed up the prices and divided to obtain the mean cost of the medication (Supplementary Table 1). Also, when medication information specified only “Nonsteroidal Anti-Inflammatory Drugs”, the cost was calculated as the mean cost of a full medical prescription of ibuprofen, as it is the most prescribed Nonsteroidal Anti-Inflammatory Drug in Spain [36]. Costs were obtained from the Spanish Pharmaceutical Vademecum (hppt: //www.vademecum.es) for the year 2022.

Indirect Cost: Work Loss

Costs of work loss were calculated, according to the human capital approach of 2022, by multiplying the days on sick by the gross salary of nurses and aides in Spain, obtained from the Catalan Institute of Health tariffs for the reference year 2022 [37]. Sick leave data were available for all workers who signed the informed consent. Data were obtained from the human resources department of each hospital. The episodes were registered by the OHS using the group code of the International Classification of Diseases, 10th revision. Episodes related to musculoskeletal pain were included in the analyses. Red flags, including bone fractures and surgery in the previous 3 months, rheumatic pathologies, pregnancy (e.g. low back pain related to pregnancy), and recurrent episodes of sick leave were excluded from the analyses.

Intervention Costs

Costs of ergonomist, Nordic Walking trainer, mindfulness instructor, and chef were estimated as price per hour for the year 2017 and updated to 2022 using the Consumer Price Index (CPI). For each of these group activities, the costs were summed up and divided by the number of participants, thus assigning the same value to each participant.

Case management was an individual service, so the costs were calculated individually for each user according to the services they attended. Case manager cost was estimated as the cost per case and updated to 2022 using the CPI. Health education beliefs professional cost was calculated as price per hour using the minimum wage in Spain for the reference year 2022. Occupational physician visit, rehabilitation physician, physiotherapy, and cognitive behavioural therapy psychologist visits were obtained from the Catalan tariffs for 2022 [35].

Health Effects

Self-perceived musculoskeletal pain was measured using a validated Spanish version of the Standardized Nordic Questionnaire at baseline and 6- and 12-month follow-up [22, 23, 38]. Data were collected with the question “Do you have pain or discomfort in this anatomical site?”. There were three possible answers (1) yes, pain, (2) yes, discomfort, and (3) no pain. Answers were dichotomized (yes pain or discomfort, no pain). The question was asked at present. The seven anatomical sites were included in the questionnaire (i.e. neck, shoulders and upper back, elbows, hands, legs, knees, and feet). In addition, an overall dichotomous variable was created on whether the nurse/nursing aide reported pain or discomfort in at least one anatomical site.

Quality-adjusted life years were measured using a validated Spanish version the EuroQol 5-dimensional questionnaire [29]. This tool is composed of a self-reported description of health problems according to a 5-dimensional classification (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression). For self-reported health with the 5-dimensional classification, workers marked one of three levels of severity in each dimension (1 = no problems, 2 = some/moderate problems, and 3 = severe/extreme problems). Perfect health status is coded as “11,111”. Each health status has a utility based on a country’s median score, where 1 represents perfect health, 0 represents death, and negative numbers symbolize statuses which are considered worse than death. Spanish scores were used to estimate the utility of the health status described by the participants [39]. Quality-adjusted life years were calculated by multiplying the utility by the amount of time a patient spent in a particular health state. EuroQol 5-dimensional questionnaire was not included at baseline, and consequently, quality-adjusted life years’ data were available only at 6- and 12-month follow-ups.

Pairing Data

In our study, effects data could not be paired, as the questionnaires were anonymous and lacked identifiers. This decision was taken to encourage participation and ensure reliability of responses, as the questionnaires were self-administered at the workplace and included personal questions and questions about the respondent’s relationship with their co-workers and supervisors.

Statistical Analysis

Analyses were performed based on the intention-to-treat principle, analysing all participants according to the randomized group to which they belonged, whether or not they were actively participating in the intervention. All the analyses were performed from both the societal and health system perspectives. To evaluate the goodness of fit of the models, we used the Bayesian Information Criterion (BIC) in which a lower value indicates a better fit.

Costs

Primary care and total costs from both perspectives had missing values, which were missing at random and ranged from 23 (11.2%) to 26 (12.7%). Mean costs during the study period by group were calculated using a linear regression adjusted by age and occupation. The incremental cost (difference in cost between the intervention and the control groups) was calculated as the average marginal effects with a generalized linear model (family gamma and link log), adjusted for age, occupation, and the costs for 1 year before the study.

Effects

The effect of the intervention was obtained for the intervention and the control group, as the difference on prevalence data at 12-month follow-up compared to their baseline prevalence. This incremental effect on musculoskeletal pain was calculated as the difference in the effect obtained in the intervention group compared to the effect obtained in the control group. First, we estimated a generalized logistic model (family binomial and link logit) for pain, with an interaction between time and group as independent variables, and adjusted for age. The prevalence of pain for each group and time point was estimated from the GLM as the average marginal effects, i.e. averaging the observations (probability) predictions from the model.

The incremental effect on quality-adjusted life years was calculated by comparing 12-month follow-up data with 6-month follow-up data, since baseline data were not available, and the average marginal effects were estimated through a generalized linear model with family gamma and link log, adjusted for age.

Incremental Cost-Effectiveness Ratios (ICERs)

Multiple imputation by chained equations using predictive mean matching and 40 imputed databases was used to deal with the missing data. ICERs were calculated as the difference in the cost between the intervention and the control group, divided by the difference in the effect of each outcome (musculoskeletal pain and quality-adjusted life years). A probabilistic cost-effectiveness analysis was performed based on 5000 bootstrap replications and cost-effectiveness planes and acceptability curves were constructed for the ICERs from societal and health system perspectives to assess the uncertainty around the estimates [40, 41]. The proportions of (bootstrapped) ICERs in each quadrant were calculated.

The analyses were performed using STATA 17 (Stata Statistical Software Release 17; StataCorp, College Station, TX).

Results

Participants

The eight clusters included 473 nurses and aides. Eighteen workers were excluded because they were on a sabbatical leave. At baseline, 293 nurses and aides signed the informed consent to participate and 257 answered the baseline questionnaire, 138 in the intervention group, and 119 in the control group. Costs data were collected for 134 participants in the intervention group and 71 in the control group. During the fieldwork, one of the control clusters (n = 105; informed consent n = 53) started a period of reorganization and left the study (Fig. 1). No adverse effects were reported to the OHS.

Fig. 1
figure 1

Flow chart of recruitment, allocation, and outcome assessment

Full size image

Table 2 shows that, at baseline, the participants in both groups were similar with regard to sex (84% women), age (53% between 31 and 49 years), occupation (63% nurses and 37% nursing aides), and self-perceived musculoskeletal pain (ranged from 95% overall to 10% in elbows). Quality-adjusted life years were 0.757 and 0.751 in the intervention and control group, respectively, at 6 months’ follow-up (baseline data not available). Baseline characteristics of workers by cluster are described in Supplementary Table 2.

Table 2 Baseline characteristics of workers by group
Full size table

Costs

During the study period, the total mean cost per person in the intervention group was €614 and €216 in the control group with a societal perspective. Total mean direct health costs during the study period were also slightly higher in the intervention group (€95/person) than in the control group (€90/person). Work loss mean cost had a cost of €408/person for the intervention group and €115/person for the control group. The mean cost of the workplace intervention was €38/person.

Cost-Effectiveness Analysis

Table 3 shows the incremental costs, the incremental effects (i.e. the difference between the effect in the intervention group and the control group) and the ICERs. The ICERs showed that an additional €68 (from a societal perspective) and €8 (from a health system perfective) were required to achieve 1-extra-percentage-point reduction in overall musculoskeletal pain in the intervention group compared to the control group. The ICERs from the societal and health system perspectives were, respectively, €34 and €4 for musculoskeletal pain in neck, shoulders, and upper back; €53 and €7 in the low back; €179 and €23 in the hands; €39 and €5 in the legs; €115 and €14 in the knees; and €36 and €5 in the feet. Musculoskeletal pain in elbows and quality-adjusted life years worsened in the intervention group compared with the control group, and the costs were higher. The ICER for elbows and QALYs showed that the intervention group was dominated by the control group. The ICER for elbows and QALYs showed that the intervention group was dominated by the control group.

Table 3 Incremental cost-effectiveness ratio (ICER)-adjusted analysis of musculoskeletal pain and quality-adjusted life years from societal and health system perspectives
Full size table

Figure 2a shows the cost-effectiveness plane for the analysis of overall musculoskeletal pain with both societal and health system perspectives. In both perspectives, around 90% of the ICERs were distributed in the North-East quadrant (most effective and costly intervention) and 0 to 4% in the South-East quadrant (most effective and less expensive intervention). Figure 2b shows the cost-effectiveness plane for quality-adjusted life years. In both societal and health system perspectives, 60 to 70% of ICERS were distributed in the North-East quadrant and 0 to 3% for societal and for health system in the South-East quadrant. Acceptability curves (Supplementary Fig. 1) shows that for a willingness to pay of €100, the probability of the intervention being cost-effective for musculoskeletal pain was around 90% in both perspectives. For QALY, acceptability curves showed that with a willingness to pay of €25,000, the probability of the intervention being cost-effective was around 67% in both perspectives. Cost-effectiveness planes of musculoskeletal pain for each anatomical site (i.e. neck, shoulders, and upper back, low back, elbows, hands, legs, knees, and feet) from both societal and health system perspectives are shown in Supplementary Fig. 2.

Fig. 2
figure 2

Cost-effectiveness plane of a musculoskeletal pain in overall and b quality-adjusted life years of the intervention group compared to the control group at 12 months’ follow-up with societal and health system perspective. MSP Musculoskeletal pain. QALYs Quality-adjusted life years. % shows the percentages of dots in each quadrant. Blue dot shows the cost-effectiveness data of the original database

Full size image

Discussion

The intervention showed that to achieve a 1-extra-percentage-point reduction in musculoskeletal pain overall in the intervention group compared with the control group, an additional €12 was needed from a societal perspective and €6 from a health system perspective. For quality-adjusted life years, the usual care dominated.

Comparison with the Existing Literature

Economic evaluations of multifaceted interventions in the workplace to reduce and manage musculoskeletal pain in nursing staff are scarce, and we did not find any study that used musculoskeletal pain as outcome for a cost-effectiveness analysis.

Musculoskeletal pain, as well as the EQ-5D, is a patient-reported outcome measure (PROM). Hospitals and healthcare services can better meet the needs and preferences of their patients by responding to PROM’s questionnaires. These measurements are intended to close a critical gap in our understanding of patient-relevant outcomes by providing useful source of data to influence healthcare decisions on effective therapies, along with clinical and economic outcomes [16,17,18, 42]. Despite the recommendations of using PROMs in effectiveness and cost-effectiveness studies, musculoskeletal pain has not been used previously in cost-effectiveness analysis. Thus, there is no established threshold in the literature for the maximum amount of money that may be required for an intervention to reduce musculoskeletal pain and still be considered cost-effective, which can be a barrier to stablish recommendations.

We found two previous cost-effectiveness evaluations of a multicomponent intervention for musculoskeletal pain in workers. Suni et al. (2018) study consisted of neuromuscular exercise and back-care counselling in female healthcare workers with non-specific low back pain and lasted 12 months. The authors concluded that the combined arm (i.e. the study arm combining neuromuscular exercise and back-care counselling) was effective in reducing pain, but was not cost-effective on sick leave and quality-adjusted life years [43]. Otherwise, Tingulstad et al. (2023) concluded that their multicomponent intervention for workers on sick leave due to musculoskeletal disorders was cost-effective in terms of quality-adjusted life years [44]. In our study, the intervention was also effective to reduce pain, especially in those anatomical sites with a higher prevalence of pain (e.g. neck, shoulders, and upper back), but it entailed a higher cost than the usual care. However, acceptability curves showed that by increasing the willingness to pay, the probability of being efficient could reach 90% with a 100€/1-percentual-point reduction of musculoskeletal pain. In terms of quality-adjusted life years, our intervention was neither efficient as the mean quality-adjusted life years was lower in the intervention group compared with the control group. Despite this, this result should be interpreted carefully, as we did not have the baseline measure for quality-adjusted life years.

In relation to sick leave, the intervention evaluated by Suni et al. was not efficient [43]. In contrast, the intervention evaluated by Steenstra et al., [45] consisting in a multi-stage return-to-work programme for workers on sick leave due to low back pain, found that an additional expenditure of €19 was required in the workplace intervention to reduce sick leave due to musculoskeletal pain by 1 day compared with usual care [45]. In our study, sick leave was included in the societal perspective analysis, showing higher mean costs in the intervention group (€408) compared to the control group (€155). A longer follow-up is recommended to detect differences in sick leave [46]. Also, in our intervention, the case management programme was the component focused on return to work, but only 13 workers participated and only 5 finished it (data not shown) [21]. Perhaps, a higher implementation of this component could have decreased sick leave duration and improved the efficiency of our intervention.

Limitations

This study has some limitations. The main limitation is that the questionnaires were anonymous, and an identifier was not created to guarantee confidentiality, so effects data could not be paired. Second, the baseline questionnaire did not include the EuroQol 5-dimensional questionnaire, and therefore quality-adjusted life years’ data were only available for the 6- and 12-months follow-ups. Perhaps we would have found an effect on quality-adjusted life years by comparing baseline data with the 12-month follow-up. Third, effectiveness results could have been limited by the process implementation, as the intervention had a slightly lower implementation than expected [47]. Recommendations to improve the implementation have been published elsewhere [47]. Finally, other information regarding costs and effectiveness might be missing, as no information was available on whether nursing staff consumed over-the-counter medication, or whether they went to private hospitals or services (e.g. physiotherapy consultations, etc.).

Study Implications and Future Research

This intervention could be a first approach to develop efficient interventions to reduce and manage musculoskeletal pain in nursing staff at work. Given that this study shows that the willingness to increase the spending needed to improve musculoskeletal pain is currently low, health policy makers should consider these types of evidence-based, multifaceted workplace interventions to improve musculoskeletal pain in nursing staff.

Studies in the economic evaluation of multifaceted interventions with paired data are needed to reinforce the quality of the analyses and confirm the findings of this study. Quality-adjusted life years should be measured from the baseline, and a longer follow-up is recommended [46]. Also, as suggested by other authors, we recommend adding an economic evaluation to any future randomized controlled trial, and collaboration with health economists or researchers with expertise in this area. Finally, to establish a threshold for musculoskeletal pain PROM could be a useful basis to promote recommendations.

Conclusion

This intervention reduced musculoskeletal pain, but was not cost-effective in terms of quality-adjusted life years. ICERs, in terms of musculoskeletal pain, showed that an additional €68 from a societal perspective and €9 from a health system perspective were needed to achieve an additional 1 percentage point reduction overall in musculoskeletal pain in the intervention group compared to the control group. With a willingness to pay of €100, the probability of the intervention being cost-effective for musculoskeletal pain was around 90%. For quality-adjusted life years, the usual care dominated the intervention. Further studies incorporating our recommendations are needed to confirm these findings.