Prognostic factors in short-term disability due to musculoskeletal disorders

Authors


  • ISRCTN: 38768940.

Abstract

Objective

To identify factors associated with poor outcome in temporary work disability (TWD) due to musculoskeletal disorders (MSDs).

Methods

We conducted a secondary data analysis of a 2-year randomized controlled trial in which all patients with TWD due to MSDs in 3 health districts of Madrid (Spain) were included. Analyses refer to the patients in the intervention group. Primary outcome variables were duration of TWD and recurrence. Diagnoses, sociodemographic, work-related administrative, and occupational factors were analyzed by Cox proportional hazards models.

Results

We studied 3,311 patients with 4,424 TWD episodes. The following were independently associated with slower return to work: age (hazard ratio [HR] 0.99, 95% confidence interval [95% CI] 0.98–0.99), female sex (HR 0.84, 95% CI 0.78–0.90), married (HR 0.90, 95% CI 0.83–0.97), peripheral osteoarthritis (HR 0.77, 95% CI 0.6–0.9), sciatica (HR 0.59, 95% CI 0.54–0.65), self-employment (HR 0.56, 95% CI 0.48–0.65), unemployment (HR 0.41, 95% CI 0.28–0.58), manual worker (HR 0.86, 95% CI 0.79–0.94), and work position covered during sick leave (HR 0.84, 95% CI 0.77–0.92). The factors that better predicted recurrence were peripheral osteoarthritis (HR 1.75, 95% CI 1.14–2.6), inflammatory diseases (HR 1.66, 95% CI 1.009–2.72), sciatica (HR 1.30, 95% CI 1.08–1.56), indefinite work contract (HR 1.43, 95% CI 1.14–1.75), frequent kneeling (HR 1.39, 95% CI 1.15–1.69), manual worker (HR 1.19, 95% CI 1.003–1.42), and duration of previous episodes (HR 1.003, 95% CI 1.001–1.005).

Conclusion

Sociodemographic, work-related administrative factors, diagnosis, and, to a lesser extent, occupational factors may explain the duration and recurrence of TWD related to MSD.

INTRODUCTION

Musculoskeletal disorders (MSDs) have a high prevalence and morbidity (1–3), and are one of the main causes of activity restriction, functional loss, and short- or long-term disability (3–6). MSDs generate a high demand for health care resources and social services (4, 7), and produce a great socioeconomic impact in Western countries (8).

In the active working population, MSDs represent the second largest cause of short-term or temporary work disability (TWD) after the common cold. In fact, 20% of all sick-leave days in a year may be related to MSDs (9), which has a tremendous yet growing impact in the workplace (10, 11).

Work disability is a complex health problem. It is the result of clinical, demographic, psychological, social, and work factors, all of which interact and make the return to an optimal functional status difficult (12). Although a variety of actions in the field of occupational health have been promoted to reduce the burden of specific MSDs (13), global approaches that are easy to incorporate at the first step of care are lacking.

Nine years ago, our group started a community-based program for recent-onset TWD due to MSDs, based upon clear-cut protocols, which were carried out by rheumatologists who had been trained in the basics of sick-leave administrative tasks as well as in rehabilitation and occupational aspects of MSDs (14). The program resulted in a significant increase in the rate of return to work (relative efficacy of 39%), an almost 50% decrease in permanent work disability proposals, and reached a net benefit of $7.8 million, increasing patients' satisfaction (15). However, the program was unable to attain the same effectiveness in all cases. Therefore, the purpose of the present study was to analyze the factors associated with MSD-related TWD in order to identify the subgroup of patients with poor outcomes in terms of TWD duration (defined as days of sick leave per episode) and number of recurrences (defined as any successive TWD episode occurring during the study period, regardless of the musculoskeletal cause). This will provide new insights into the course of MSD-related TWD and help to better design future specific pathways for these patients.

PATIENTS AND METHODS

Setting.

The population in Madrid, Spain receives universal coverage, and care is organized into 11 health areas, with access through primary care physicians (PCPs). Workers on sick leave are given a TWD initiation form, which entitles them to receive compensation payments; these TWD forms are renewed weekly by the PCP until the worker either recovers or reaches a maximum of 18 months in TWD. The PCP may refer the patient to specialized care at any point in time.

Study design.

We conducted a secondary data analysis of an intervention study in which a specific program was offered to workers with MSD-related TWD from 1998 to 2001. The intervention study has been described in detail previously (14, 15).

Patients.

All patients from health areas 4, 7, and 9 of Madrid, Spain who received a TWD initiation form from a PCP due to a diagnosis of MSD during the study period were included, except for those in whom the MSD was related to traumatism, surgery, or work accident. Every day, patients fulfilling the inclusion criteria were randomized into the control or intervention group. Patients in the control group simply followed the standard care by PCPs as described above. Patients assigned to the intervention group were contacted and invited to attend a specific care program based upon detailed proceedings and carried out by trained rheumatologists. Inclusion and followup periods lasted 12 months each. Any time a patient previously assigned to the intervention group had an MSD-related TWD, the patient was offered the followup of the new episode.

Dates and cause of the TWD episodes as well as patients' age, sex, and work type were retrieved from the National Institute of Social Security database in both groups. Other variables pertinent to the work disability process, such as detailed sociodemographic characteristics and work and occupational data, could only be gathered for the patients attended by the program rheumatologists through a structured interview at first visit. Therefore, the patients included in this analysis are those who were seen at least once by a program's rheumatologist.

Variables.

Two primary outcomes were established: 1) duration of TWD, defined as days of sick leave per episode, and 2) recurrence of TWD, defined as any successive TWD episode that occurred during the study period, regardless of the musculoskeletal cause.

A list of all explanatory variables studied is shown in Table 1. These include 1) sociodemographic variables, 2) work-related administrative variables and occupational variables (only referred to physical activity at work), and 3) comorbidity. All variables were retrieved from the patient reports at the structured interviews at baseline (first TWD episode).

Table 1. Baseline description of the patients on work disability due to musculoskeletal disorders*
VariablesValue
  • *

    Values are the percentage unless otherwise indicated. The number of patients with available data for each variable is indicated in parentheses.

Age, mean ± SD years (n = 3,217)41±12
Female sex (n = 3,307)49.8
Marital status (n = 3,115) 
 Married64.7
 Divorced4.6
 Single28.4
 Widowed2.3
Education (n = 3,073) 
 None1.6
 Primary school41.7
 Secondary school42.8
 University13.8
Housekeeping chores (n = 3,022) 
 None22.4
 Some34.3
 Most18.0
 All25.3
Any comorbidity (n = 1,059)24.7
Work-related administrative variables 
 Employment (n = 3,025) 
  General workers93.3
  Self-employment6.7
 Job stability (n = 2,902) 
  Temporary contract19.6
  Indefinite contract75.7
  Functionary4.7
 Work type (n = 3,129) 
  Management/intellectual7.6
  Services22.3
  Secretarial25.7
  Specialized or nonspecialized manual work44.4
 Work position covered by temporary worker (n = 2,927)18.5
 Unemployed (n = 3,084)1.3
Occupational variables 
 Must kneel frequently at work (n = 3,095)15.1
 Must raise arms frequently at work (n = 3,097)30.8
 Must perform frequent postural changes (n = 3,081)70.6
 Must stand up for long periods (n = 3,098)76.0
 The job is physically demanding (n = 3,108)63.0
  Amount of physical demand (n = 1,917) 
   Little26.5
   Intermediate56.8
   Large16.6
 Must pull objects frequently (n = 3,017)63.5
  The objects are heavy or very heavy (n = 1,947)73.5
 Must stay squatting at times (n = 3,095)20.3
 Must stay seated for long periods (n = 3,098)44.5
 Must perform lateral flexion of the trunk (n = 3,097)36.9
 Must perform anterior flexion of the neck (n = 3,097)49.6
 Must perform anterior flexion of the trunk (n = 3,097)58.3
 Must perform repetitive movements (n = 3,063)70.0
 Exposed to frequent vibrations (n = 3,025)17.0

Each TWD episode was linked to a main diagnostic code (International Classification of Diseases, Ninth Revision coding system), which was the cause of the TWD as reported by the PCP. The cause of the episode was revised and modified if deemed necessary by the program rheumatologist. All diagnoses were further grouped into 11 categories based on the syndrome or localization of the MSD, and based on common clinical management. After the rheumatologist visit, 48% of the PCP diagnoses and 30% of the diagnosis categories changed. These categories were neck pain, back pain (which included low back pain and thoracic back pain), sciatica (discopathies and sciatica), muscular pain (muscle spasms, joint pain without arthritis, fibromyalgia, and other noninflammatory muscle disorders), peripheral osteoarthritis, tendinitis (tendinitis, bursitis, peripheral sprains, tenosynovitis, and ganglions), inflammatory diseases (rheumatoid arthritis, spondylarthritis, systemic lupus erythematosus, and vasculitis), microcrystalline and undifferentiated arthritis, knee pain (not osteoarthritis), carpal tunnel syndrome, and other (which included isolated cases of avascular necrosis, osteoporosis, or septic arthritis).

Statistical analysis.

To describe the sample, we used either the distribution of frequencies, the mean ± SD, or the median and interquartile range, depending on the distribution. To examine the association of explanatory variables with the duration of TWD or with recurrences, we calculated Kaplan-Meier curves (allowing for multiple failure in the case of recurrence), log rank tests, and Cox bivariate and multivariate regression models. All multivariate models included the variables that reached a P value <0.10 in the bivariate analysis, and were additionally adjusted by sex, age, and health district. In the Cox models exploring the duration of TWD, the dependent variable was days off work, and the results are expressed as a hazard ratio that can be interpreted as the relative rate of return to work per day with respect to the referent category. The Cox models exploring the recurrences of TWD allowed for the occurrence of multiple episodes, as in a Poisson model. In these models the dependent variable was recurrence, and the results are expressed as a hazard ratio that can be interpreted as the relative risk of recurrence per year with respect to the referent category.

To test the effect of diagnostic categories on TWD duration as well as on recurrences, we ran models for each of the diagnostic categories plus the variables remaining in the best explanatory models from previous analyses. Missing data were not replaced by any method. All analyses were performed using Stata 9.0 (StataCorp, College Station, TX). For all comparisons, a 2-tailed P value <0.05 was considered to indicate statistical significance.

RESULTS

We analyzed the data on 3,311 patients who had 4,424 MSD-related TWD episodes during the 2 years of the study. The mean time between the first day of TWD episode and the referral to the program rheumatologist was 5 days (range 1–477 days). The studied patients were, on average, middle-aged men and women, most of whom were general workers and were actively employed, many of them as manual workers. A full description of the patients is depicted in Table 1. Besides working, one-quarter of the patients, the large majority of whom were women, were also responsible for carrying out all housekeeping chores on a regular basis.

The causes of the TWD episodes were, in order of frequency, back pain (n = 1,262 [34.0%]), tendinitis (n = 801 [21.5%]), sciatica (n = 665 [17.8%]), neck pain (n = 446 [12.0%]), microcrystalline and undifferentiated arthritis (n = 216 [5.8%]), muscle pain (n = 84 [2.3%]), peripheral osteoarthritis (n = 78 [2.1%]), knee pain (not OA) (n = 72 [1.9%]), inflammatory diseases (n = 56 [1.5%]), carpal tunnel syndrome (n = 24 [0.7%]), and other (6 avascular necrosis, 10 osteoporosis, and 1 septic arthritis [0.1%]).

Duration of TWD.

The median duration of the TWD episodes was 13 days (P25-75: 8–26). Most patients (96%) returned to work within the first 3 months. The results of the bivariate and multivariate Cox regression analyses are shown in Table 2. Only variables that reached a P value <0.10 are presented. Being unemployed or self-employed and, as expected, age were the factors that prolonged the TWD episodes the most. No independent association was found between factors related to the job characteristics and TWD duration.

Table 2. Results of the Cox adjusted regression analyses for duration of temporary work disability episodes*
VariablesBivariateMultivariate
  • *

    Values are the hazard ratio (95% confidence interval). Hazard ratios must be interpreted as the relative rate of return to work per day. Only variables that reached a P value <0.10 at the bivariate analysis are exposed.

  • P < 0.001.

  • P < 0.01.

Age (per year)0.99 (0.98–0.99)0.99 (0.98–0.99)
Female sex0.88 (0.83–0.93)0.84 (0.78–0.90)
Self-employed0.51 (0.45–0.59)0.56 (0.48–0.65)
Married0.86 (0.81–0.92)0.90 (0.83–0.97)
Low educational level (none or primary)0.88 (0.83–0.94) 
Responsible for most or all house chores0.91 (0.85–0.97) 
Work position covered0.87 (0.80–0.94)0.84 (0.77–0.92)
Unemployed0.50 (0.34–0.66)0.41 (0.28–0.58)
Manual worker0.90 (0.83–0.97)0.86 (0.79–0.94)
Seated for long periods1.05 (0.99–1.13) 
Must adopt squatting position1.08 (1.01–1.16) 
Must stand up for long periods0.89 (0.84–0.97) 
Physically demanding job0.94 (0.88–1.01) 
Must perform anterior flexion of the neck1.06 (0.99–1.13) 

The effect of each diagnostic category on the duration of TWD is shown in Table 3. The diagnoses that significantly prolonged the TWD episodes were other, sciatica, and peripheral osteoarthritis (mean TWD duration 76.8 days, 42.6 days, and 42.6 days, respectively). The shortest TWD episodes were those related to microcrystalline arthritis (mean 16.4 days) (Figure 1).

Table 3. Effect of diagnosis on temporary work disability duration*
Diagnostic categoriesHR (95% CI)P
  • *

    Results of multivariate models in which each diagnostic category plus the variables remaining in the best explanatory models from previous analyses (Table 2). Hazard ratios (HRs) must be interpreted as the relative rate of return to work per day. 95% CI = 95% confidence interval.

Neck pain1.24 (1.12–1.38)< 0.001
Back pain1.26 (1.18–1.35)< 0.001
Sciatica0.59 (0.54–0.65)< 0.001
Knee pain (not osteoarthritis)0.91 (0.70–1.17)0.500
Peripheral osteoarthritis0.77 (0.60–0.90)0.040
Muscle pain0.91 (0.73–1.14)0.400
Tendinitis1.07 (0.97–1.15)0.160
Inflammatory diseases0.83 (0.62–1.10)0.210
Microcrystalline and undifferentiated arthritis1.78 (1.53–2.07)< 0.001
Carpal tunnel syndrome1.15 (0.75–1.77)0.500
Other0.32 (0.19–0.54)< 0.001
Figure 1.

Days per temporary work disability (TWD) depending on the diagnosis. Boxes represent the interquartile range. Lines outside the boxes represent the 10th and 90th percentiles. Lines inside the boxes represent the median. Circles indicate outliers. NP = neck pain; BP = back pain; Sci = sciatica; KP = knee pain; POA = peripheral osteoarthritis; SIA = inflammatory diseases; MA = microcrystalline arthritis; MM = muscle pain; Ten = tendinitis; CTS = carpal tunnel syndrome; oth = other (avascular necrosis, osteoporosis, and septic arthritis).

Risk of recurrence.

During the study period, 795 patients (24%) had at least 1 TWD recurrence. The diagnosis sometimes was the same in all recurrences of TWD in the same patient. For example, inflammatory diseases and knee pain (not osteoarthritis) diagnoses were present at least twice per patient in 45% and 44%, respectively, of the patients who had a first episode due to these conditions. We also found 209 patients (6.3%) with ≥3 episodes of TWD. In these cases the diagnoses that were repeated the most were inflammatory diseases (in 71% of the cases), followed by knee pain (not osteoarthritis) (50%), microcrystalline arthritis (50%), and sciatica (49%).

The factors that best predicted the recurrences were those related to work, such as having an indefinite contract or being functionary, being a manual worker, or kneeling frequently at work (Table 4). The diagnoses associated with the largest risk for recurrence were inflammatory diseases and peripheral osteoarthritis. The risk was moderate for sciatica and was smallest for tendinitis and back pain (Table 5).

Table 4. Results of the Cox adjusted regression analyses for temporary work disability relapses*
VariablesBivariateMultivariate
  • *

    Values are the hazard ratio (95% confidence interval). Hazard ratios must be interpreted as the relative risk of relapse per year. Only variables that reached a P value <0.10 at the bivariate analysis are shown. TWD = temporary work disability.

  • P < 0.05.

  • P < 0.001.

  • §

    P < 0.01.

Age (per year)1.01 (1.01–1.01)1.00 (0.90–1.01)
Female sex1.08 (0.96–1.21)1.13 (0.97–1.32)
General worker regimen1.49 (1.10–2.04)1.67 (0.96–2.94)
Married1.15 (1.01–1.30) 
Low educational level (none or primary)1.02 (0.90–1.16) 
Responsible for most or all house chores1.11 (0.85–0.97) 
Indefinite contract or functionary1.43 (1.19–1.72)1.43 (1.14–1.75)§
Unemployed0.50 (0.24–1.09) 
Manual worker1.20 (1.04–1.38)1.19 (1.00–1.42)
Must stay seated for long periods0.88 (0.78–0.99) 
Must adopt squatting position1.29 (1.13–1.49) 
Must kneel frequently1.37 (1.18–1.60)1.39 (1.15–1.69)§
Physically demanding job1.16 (1.02–1.32) 
Must perform anterior flexion of the neck1.14 (1.01–1.28) 
Must perform anterior flexion of the trunk1.18 (1.04–1.34)§ 
Must perform lateral flexion of the trunk1.13 (1.003–1.28) 
Duration of previous TWD episode (per day)1.00 (0.99–1.00)1.00 (1.00–1.00)
Table 5. Effect of diagnosis on the risk to relapse*
Diagnostic categoriesHR (95% CI)P
  • *

    Results from multivariate models for each diagnostic category plus the variables remaining in the best explanatory models from previous analyses (Table 4). Hazard ratios (HRs) must be interpreted as the relative risk of relapse per year. 95% CI = 95% confidence interval.

Neck pain1.30 (1.06–1.58)0.013
Back pain0.82 (0.70–0.96)0.016
Sciatica1.30 (1.08–1.56)0.005
Knee pain (not osteoarthritis)1.46 (0.90–2.37)0.124
Peripheral osteoarthritis1.75 (1.14–2.61)0.010
Muscle pain0.55 (0.28–1.06)0.070
Tendinitis0.79 (0.65–0.95)0.016
Inflammatory diseases1.66 (1.01–2.72)0.040
Microcrystalline and undifferentiated arthritis0.99 (0.89–1.54)0.250
Carpal tunnel syndrome1.02 (0.42–2.47)0.950
Other0.51 (0.13–2.05)0.340

DISCUSSION

Work disability caused by MSD is a relevant health problem in which many factors play a part (16). Our results show that sociodemographic factors, work-related administrative factors, and diagnosis are the major components explaining the duration and recurrence of work disability, with occupational variables being less important to these outcomes.

Patients included in our analyses are representative of workers with TWD related to MSD (middle-aged persons with back pain or soft tissue rheumatism who are employed by others [17]). Our patients attended a specific care program, which demonstrated a high efficacy, increasing the rate of return to work, reducing permanent work disability proposals, and reaching a net benefit of $7.8 million, increasing patients' satisfaction (15). This intervention program integrated different clinical, diagnostic actions and treatments for each diagnosis that have already been proven beneficial in the management of MSDs.

The relevance of sociodemographic factors, specifically age and sex, to TWD duration has been shown persistently (18–23). We found that older married women have a higher probability of staying on sick leave. This knowledge may help physicians to identify patients at high risk for poor-outcome work disability.

We agree with other authors that work-related factors also influence TWD outcomes. The probability of staying on sick leave was greater for self-employed workers compared with general workers, as well as for patients whose position was covered while on sick leave and unemployed patients. Similar results have been found in other studies (18, 22–26). Bartley et al (24) concluded that secure employment increases the likelihood of recovery, and Stover et al (23) and Cheadle et al (18) found that predictors of prolonged disability include a high unemployment rate and specific types of work. In our setting, self-employed workers do not receive any compensation payments until the fourteenth day on sick leave, which is, very conceivably, the underlying reason for few but long TWD episodes in these workers. Besides, it is understandable that unemployed workers stay on sick leave longer because there is no motivation to return to no job. In contrast, we could not demonstrate a clear association between occupational factors and the duration of TWD.

Neither age nor sex was a significant determinant for repeated TWD episodes, although this association has been described. However, most of these studies have focused on low back pain (26, 27), and it may be that we did not find an association of recurrences with sociodemographic factors because we included all types of MSDs.

One of the most important determinants of recurrences of TWD found in back pain is the duration of the previous episode (27). Our results confirm the role of this factor in all MSDs. The risk of recurrence increases by 7% per 1-week duration of the previous TWD episode, independently of the diagnosis. This could imply either that shorter episodes of care and earlier return to work contribute to better outcomes or that the previous episode was not really cured and worsened when the patient returned to work. We must say that, in the program we run, no patient is ever forced to return to work if the patient feels that returning would worsen the problem.

The probability of recurrences over time varied mainly depending on work characteristics. In agreement with Evans et al (28), we have demonstrated that work stability is a risk factor for recurrence. Moreover, the type of work, or level of responsibility, is also a risk factor for recurrence. Nonspecialized manual workers have almost 20% more probability of recurrences than the rest of workers. Wasiak et al (27) found that the more physical the nature of one's work is, the larger the likelihood of work disability recurrences. However, we are not sure whether this is the explanation for manual workers because having a physically demanding job was not found to be independently associated with recurrence in our study.

Finally, regarding occupational variables, we found an association between kneeling frequently at work and the recurrence of TWD. Manual workers are at a greater risk of recurrence than other workers, although being a manual worker may be a factor that fits better under work responsibility than under occupational factors. Contrary to other studies, we cannot conclude that occupational factors are particularly important to the outcome of TWD (21, 25, 29–32).

Another key to understanding the outcome of work disability, and related to the biopsychosocial model, is diagnosis (33). The clinical entities independently associated with poor outcome of TWD in terms of duration and recurrences are peripheral osteoarthritis, sciatica, and inflammatory diseases. In contrast, back pain, tendinitis, and microcrystalline arthritis are the diagnoses with the best results. The results for gouty arthritis may be related to the emphasis of our program on a good control of diet, uric acid level, hypertension, and diuretics (9, 34). Regarding back pain, we confirm the observation by Pengel et al (35) that most people with acute back pain have rapid improvements in disability within a month, but we disagree with them and with many other authors on the importance of back pain as a predictor of future episodes of back pain (30, 35). Our back pain results may be related to our specific intervention, which is a balanced program of reassurance, intensive clinical treatment, education, and exercise, as shown previously (15). Of course, our results may also be due to considering sciatica as a distinct clinical entity, which actually showed a much poorer outcome.

Inflammatory diseases have a great disability potential with a significant impact on workplace productivity (36, 37); actually, many individuals with rheumatoid arthritis would like to work but are unable to, not only because of their disease, but because of work conditions (38). Our results reflect the intrinsic, chronic, fluctuant course of most of the inflammatory diseases, with short but recurrent episodes of TWD. A combination of demographic characteristics, consequences of the illness, and workplace conditions is known to contribute to disability in patients with rheumatoid arthritis (39–41). There are many studies on work disability in rheumatoid arthritis that corroborate the idea that to reduce disability, patients need education, improved workplace conditions, and a more aggressive and earlier control of the disease (42–44).

Our study may have some limitations. The main limitation is that we only analyzed patients who attended at least once. This fact should be taken into account because it might imply that risk factors for persons not participating in an intervention course might be different. Occupational factors were self-reported, extracted from a clinical nonexplicit-occupational questionnaire, not grouped specifically, and presented many missing values. However, occupational factors may not have a large influence on the duration or recurrence of TWD, but they are perhaps more modifiable than any other factors with a larger association, namely, sociodemographic and employment-related factors. Despite these limitations, we are confident that our study offers useful and relevant information for clinicians involved in the care of people with MSD-related TWD, although it should not be considered a substitute for more focused occupational or sociologic studies.

Despite the number of studies that have examined predictors of disability in MSDs, most have focused on specific diseases, mainly rheumatoid arthritis or back pain, or on occupational factors, but few have integrated multiple domains, as we have done. We hope that our findings help to improve the treatment of patients with recent-onset work disability in different ways. The more obvious way would be to reinforce the clinical treatment of patients with inflammatory conditions, osteoarthritis, and sciatica, who had a poorer short-term work disability prognosis despite our intervention program. Such reinforcement might include the need for early formal rehabilitation for such patients and the prompt initiation of either combination therapy or biologic agents in inflammatory diseases.

Another way our study might help clinicians would be to detect and evaluate the relative weight of the different sociodemographic, administrative, and occupational variables that negatively affect the recovery of patients with TWD. The presence of such variables, sometimes identified as “yellow flags,” should warrant a less aggressive clinical approach, in order to avoid unnecessary and/or invasive diagnostic or therapeutic procedures. Moreover, the eventual correction of specific occupational risk factors might help to avoid the recurrence of TWD episodes once the worker has returned to work.

In conclusion, sociodemographic factors, underlying disease, employment-related administrative factors, and occupational factors influence the persistence and recurrence of work disability related to MSDs. Our data stress the need to improve the knowledge of other psychosocial factors not studied here, and highlight the importance of secondary disability prevention to public health policy.

AUTHOR CONTRIBUTIONS

Dr. Jover had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Abásolo, Hernández-García, Jover.

Acquisition of data. Abásolo, Lajas, Candelas, Blanco, Jover.

Analysis and interpretation of data. Abásolo, Carmona, Lajas, Candelas, Blanco, Loza, Hernández-García, Jover.

Manuscript preparation. Abásolo, Carmona, Lajas, Candelas, Blanco, Loza, Hernández-García, Jover.

Statistical analysis. Abásolo.

Ancillary