Archives of Physical Medicine and Rehabilitation journal homepage: www.archives-pmr.org Archives of Physical Medicine and Rehabilitation 2015;96:1658-65
ORIGINAL RESEARCH
Does the Upper-Limb Work Instability Scale Predict Transitions Out of Work Among Injured Workers? Kenneth Tang, PhD,a,b,c,d Dorcas E. Beaton, PhD,a,b,c Sheilah Hogg-Johnson, PhD,a,c,e Pierre Coˆte´, PhD,e,f Patrick Loisel, PhD,e Benjamin C. Amick III, PhDc,g From the aInstitute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada; bMusculoskeletal Health and Outcomes Research, Li Ka Shing Knowledge Institute of St. Michael’s Hospital, Toronto, Ontario, Canada; cInstitute for Work & Health, Toronto, Ontario, Canada; dSchool of Rehabilitation Science, Faculty of Health Sciences, McMaster University, Hamilton, Ontario, Canada; e Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario, Canada; fFaculty of Health Sciences and University of Ontario Institute of Technology - Canadian Memorial Chiropractic College Center for the Study of Disability Prevention and Rehabilitation, University of Ontario Institute of Technology, Oshawa, Ontario, Canada; and gDepartment of Health Policy and Management, Robert Stempel College of Public Health & Social Work, Florida International University, Miami, FL.
Abstract Objective: To investigate the predictive ability of the Upper-Limb Work Instability Scale (UL-WIS) for transitioning out of work among injured workers with chronic, work-related upper extremity disorders (WRUEDs). Design: Secondary analysis of a 12-month cohort study with data collection at baseline and 3-, 6-, and 12-month follow-up. Survey questionnaires were used to collect data on an array of sociodemographic, health-related, and work-related variables. Setting: Upper extremity specialty clinics. Participants: Injured workers (NZ356) with WRUEDs who were working at the time of initial clinic attendance. Interventions: Not applicable. Main Outcome Measure: Transitioning out of work. Results: Multivariable logistic regression that considered 9 potential confounders revealed baseline UL-WIS (range, 0e17) to be a statistically significant predictor of a subsequent transition out of work (adjusted odds ratio, 1.18; 95% confidence interval [CI], 1.07e1.31; PZ.001). An assessment of predictive values across the UL-WIS score range identified cut-scores of 15 (positive predictive value, .80; 95% CI, .52e.96), differentiating the scale into 3 bands representing low, moderate, and high risk of exiting work. Conclusions: The UL-WIS was shown to be an independent predictor of poor work sustainability among injured workers with chronic WRUEDs; however, when applied as a standalone tool in clinical settings, some limits to its predictive accuracy should also be recognized. Archives of Physical Medicine and Rehabilitation 2015;96:1658-65 ª 2015 by the American Congress of Rehabilitation Medicine
Work-related upper extremity disorders (WRUEDs) remain prevalent in many developed regions.1-3 Given the potential for persisting symptoms and the risk of chronicity associated with musculoskeletal disorders,4-7 injured workers recovering from WRUEDs can have difficulties sustaining their work role. One question faced by clinicians and workplace parties is, “Which
Supported by the Workplace Safety and Insurance Board Research Advisory Council (grant no. 05028), a Canadian Institutes of Health Research Fellowship, a Canadian Institutes of Health Research New Investigator Award, and the Canada Research Chairs program. Disclosures: none.
individuals are most likely to transition out of work?” The ability to predict this outcome could facilitate early identification of those at high risk for exiting work, and timelier applications of health or workplace interventions to mitigate such risk. In turn, this would also ensure scarce health care resources and disability management efforts are directed to those with the greatest need. In recent years, on-the-job problems among workers with existing musculoskeletal disorders are increasingly recognized,8-10 These problems often suggest some precariousness in a person’s ability to maintain his/her work roleda plausible precursor for a subsequent work exit. One instrument that taps into this concept
0003-9993/15/$36 - see front matter ª 2015 by the American Congress of Rehabilitation Medicine http://dx.doi.org/10.1016/j.apmr.2015.04.022
Predicting work transitions and has specific applicability for WRUEDs is the Upper-Limb Work Instability Scale (UL-WIS).11 Initially derived from the rheumatoid arthritis version of the scale (Work Instability Scale for Rheumatoid Arthritis),12 the UL-WIS is designed to quantify work instability (WI), defined by the developers as “a state of mismatch between a worker’s functional capabilities in relation to job demands due to a health disorder.”12(p350) High WI is thought to threaten continuing employment,12 and accordingly, we hypothesize that the UL-WIS will be able to predict poor work sustainability among WRUEDs. Our objectives are to (1) examine the propensity for exiting work among injured workers recovering from chronic/complex WRUEDs, (2) investigate whether the level of WI is predictive of this outcome, and if so (3) determine meaningful UL-WIS cut-scores.
Methods Study setting Study participants were injured workers attending 1 of 2 upper extremity specialty clinics operated by the Workplace Safety and Insurance Board (WSIB) of Ontario. The WSIB is a regional single-source workers’ compensation insurer, funded by employers and legislated by the Ontario provincial government. These specialty clinics are designed to provide specialized clinical consultations for claimants experiencing an atypical recovery course (ie, insufficient progress after w6mo) following WRUEDs (eg, repetitive strain injuries, acute or cumulative trauma disorders of muscle/tendon/ligaments, or uncomplicated fractures). Clinical consultations are provided by a multidisciplinary team of rehabilitation therapists, orthopedic surgeons, social workers, and case managers to evaluate recovery progress and prognosis, suitability for work, or candidacy for surgery or rehabilitative interventions. Research ethics board approval for this study was obtained at the Sunnybrook Health Sciences Centre, St. Michael’s Hospital, University of Toronto, and University of Western Ontario.
Study design and data collection Secondary data analysis was performed on a 12-month cohort study that followed up 614 attendees of specialty clinics in Toronto (nZ303) or London (nZ311), Ontario. In this study, survey questionnaires were fielded at initial specialty clinic attendance (baseline), and 3-, 6-, and 12-month follow-up. These surveys comprised questions on work status, sociodemographics, health-, and job-related variables based on a literature review of prognostic factors of work disability in musculoskeletal populations. Participants who had missed a follow-up survey were allowed to continue in the study (ie, complete the next survey).
List of abbreviations: CI NPV OPP OR PPV UL-WIS WI WRUED WSIB
confidence interval negative predictive value organizational policy and practice odds ratio positive predictive value Upper-Limb Work Instability Scale work instability work-related upper extremity disorder Workplace Safety and Insurance Board
www.archives-pmr.org
1659
Inclusion criteria Since our aim was to assess the predictive ability of a tool that is specifically applicable for working individuals, to be eligible for this analysis the injured worker must (1) have been working at study baseline, (2) have understood written English, and (3) have provided written consent for use of their deidentified survey data for research. All participants were explicitly assured that neither the WSIB nor their employer would have access to their surveys. In our dataset, 356 (58%) of 614 injured workers were working at baseline and therefore eligible for the current analysis.
Variables Outcome: transitioning out of work A transition out of work within the study period was our study outcome. At each of the 3 follow-up time points, a participant’s work status was determined by the survey question, “Are you currently working?” (yes or no). Four outcome categories were initially differentiated, depending on whether the participant had exited work and also the time of initial exit. These categories were (1) initial exit of work reported at 3-month follow-up; (2) initial exit of work reported at 6-month follow-up; (3) initial exit of work reported at 12-month follow-up; and (4) did not exit work (ie, working at all time points). For specialty clinic attendees, common reasons for not working include being on short or long sick leave, an inability to arrange suitable modified work with employer, or retirement. Main predictor of interest: UL-WIS WI was assessed at study baseline by the UL-WIS,11 which is part of a growing family of psychometrically sound and feasible measures applicable in various populations or settings.12-16 The UL-WIS consists of 17 items that assess perceptions of symptom control at work, work task performance, stamina at work, time management issues, cognitive distresses associated with work, and perceptions of sustainability of current work role.12 A dichotomous response option (yes [1]/no [0]) is provided for each item, and the scale is scored by summing all 17 items (range, 0e17). We allowed a maximum of 1 missing item (ie, 10% change in the UL-WIS estimate, then this factor would be considered a confounder and be included in the final multivariable model.34,35 No elimination steps were subsequently performed. Effects were expressed as unstandardized parameter estimates (b) and their associated SEs, Wald Z values, and adjusted odds ratios (ORs) with 95% confidence intervals (CIs). Proper model fit for each model was verified based on Hosmer-Lemeshow goodnessof-fit statistics (P>.05 was acceptable fit). For all regressions, complete case analysis (listwise deletion) was applied where observations with missing data on any covariates or study outcome would be excluded. To check for potential selection bias, the baseline characteristics were compared between subsets of observations included and excluded from the final multivariable model by way of t tests (for continuous variables) or chi-square tests (for categorical variables).
K. Tang et al Determination of UL-WIS cut-scores If the UL-WIS demonstrated independent predictive ability (ie, P.05).
Table 3 Multivariable logistic regression to examine the independent predictive ability of UL-WIS for a subsequent transition out of work Predictor
b
SE
Wald Z
P
Adjusted OR
Work instability Age Upper-limb pain intensity Mental health status OPPs
.17 e.05 .03
.05 .02 .01
3.22 e2.95 1.79
.001 .003 .07
1.18 (1.07e1.31) 0.95 (0.92e0.98) 1.03 (1.00e1.06)
.00
.01
e0.48
.63
1.00 (0.98e1.01)
e.45
.20
e2.25
.02
0.64 (0.43e0.94)
NOTE. nZ233, after listwise deletion of observations with missing data. Model likelihood ratio test Z46.36 (degrees of freedom, 5), P.05; c statistic Z.75.
Determination of UL-WIS cut-scores A plot of the PPVs and NPVs for each incremental UL-WIS cutscore over the scale range is illustrated in figure 3 (for raw data, see appendix 1). The largest PPV was observed at UL-WIS>15 (PPVZ.80; 95% CI, .52e.96), while the largest NPV was observed at UL-WIS15) and NPV (UL-WIS
ORIGINAL RESEARCH
Does the Upper-Limb Work Instability Scale Predict Transitions Out of Work Among Injured Workers? Kenneth Tang, PhD,a,b,c,d Dorcas E. Beaton, PhD,a,b,c Sheilah Hogg-Johnson, PhD,a,c,e Pierre Coˆte´, PhD,e,f Patrick Loisel, PhD,e Benjamin C. Amick III, PhDc,g From the aInstitute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada; bMusculoskeletal Health and Outcomes Research, Li Ka Shing Knowledge Institute of St. Michael’s Hospital, Toronto, Ontario, Canada; cInstitute for Work & Health, Toronto, Ontario, Canada; dSchool of Rehabilitation Science, Faculty of Health Sciences, McMaster University, Hamilton, Ontario, Canada; e Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario, Canada; fFaculty of Health Sciences and University of Ontario Institute of Technology - Canadian Memorial Chiropractic College Center for the Study of Disability Prevention and Rehabilitation, University of Ontario Institute of Technology, Oshawa, Ontario, Canada; and gDepartment of Health Policy and Management, Robert Stempel College of Public Health & Social Work, Florida International University, Miami, FL.
Abstract Objective: To investigate the predictive ability of the Upper-Limb Work Instability Scale (UL-WIS) for transitioning out of work among injured workers with chronic, work-related upper extremity disorders (WRUEDs). Design: Secondary analysis of a 12-month cohort study with data collection at baseline and 3-, 6-, and 12-month follow-up. Survey questionnaires were used to collect data on an array of sociodemographic, health-related, and work-related variables. Setting: Upper extremity specialty clinics. Participants: Injured workers (NZ356) with WRUEDs who were working at the time of initial clinic attendance. Interventions: Not applicable. Main Outcome Measure: Transitioning out of work. Results: Multivariable logistic regression that considered 9 potential confounders revealed baseline UL-WIS (range, 0e17) to be a statistically significant predictor of a subsequent transition out of work (adjusted odds ratio, 1.18; 95% confidence interval [CI], 1.07e1.31; PZ.001). An assessment of predictive values across the UL-WIS score range identified cut-scores of 15 (positive predictive value, .80; 95% CI, .52e.96), differentiating the scale into 3 bands representing low, moderate, and high risk of exiting work. Conclusions: The UL-WIS was shown to be an independent predictor of poor work sustainability among injured workers with chronic WRUEDs; however, when applied as a standalone tool in clinical settings, some limits to its predictive accuracy should also be recognized. Archives of Physical Medicine and Rehabilitation 2015;96:1658-65 ª 2015 by the American Congress of Rehabilitation Medicine
Work-related upper extremity disorders (WRUEDs) remain prevalent in many developed regions.1-3 Given the potential for persisting symptoms and the risk of chronicity associated with musculoskeletal disorders,4-7 injured workers recovering from WRUEDs can have difficulties sustaining their work role. One question faced by clinicians and workplace parties is, “Which
Supported by the Workplace Safety and Insurance Board Research Advisory Council (grant no. 05028), a Canadian Institutes of Health Research Fellowship, a Canadian Institutes of Health Research New Investigator Award, and the Canada Research Chairs program. Disclosures: none.
individuals are most likely to transition out of work?” The ability to predict this outcome could facilitate early identification of those at high risk for exiting work, and timelier applications of health or workplace interventions to mitigate such risk. In turn, this would also ensure scarce health care resources and disability management efforts are directed to those with the greatest need. In recent years, on-the-job problems among workers with existing musculoskeletal disorders are increasingly recognized,8-10 These problems often suggest some precariousness in a person’s ability to maintain his/her work roleda plausible precursor for a subsequent work exit. One instrument that taps into this concept
0003-9993/15/$36 - see front matter ª 2015 by the American Congress of Rehabilitation Medicine http://dx.doi.org/10.1016/j.apmr.2015.04.022
Predicting work transitions and has specific applicability for WRUEDs is the Upper-Limb Work Instability Scale (UL-WIS).11 Initially derived from the rheumatoid arthritis version of the scale (Work Instability Scale for Rheumatoid Arthritis),12 the UL-WIS is designed to quantify work instability (WI), defined by the developers as “a state of mismatch between a worker’s functional capabilities in relation to job demands due to a health disorder.”12(p350) High WI is thought to threaten continuing employment,12 and accordingly, we hypothesize that the UL-WIS will be able to predict poor work sustainability among WRUEDs. Our objectives are to (1) examine the propensity for exiting work among injured workers recovering from chronic/complex WRUEDs, (2) investigate whether the level of WI is predictive of this outcome, and if so (3) determine meaningful UL-WIS cut-scores.
Methods Study setting Study participants were injured workers attending 1 of 2 upper extremity specialty clinics operated by the Workplace Safety and Insurance Board (WSIB) of Ontario. The WSIB is a regional single-source workers’ compensation insurer, funded by employers and legislated by the Ontario provincial government. These specialty clinics are designed to provide specialized clinical consultations for claimants experiencing an atypical recovery course (ie, insufficient progress after w6mo) following WRUEDs (eg, repetitive strain injuries, acute or cumulative trauma disorders of muscle/tendon/ligaments, or uncomplicated fractures). Clinical consultations are provided by a multidisciplinary team of rehabilitation therapists, orthopedic surgeons, social workers, and case managers to evaluate recovery progress and prognosis, suitability for work, or candidacy for surgery or rehabilitative interventions. Research ethics board approval for this study was obtained at the Sunnybrook Health Sciences Centre, St. Michael’s Hospital, University of Toronto, and University of Western Ontario.
Study design and data collection Secondary data analysis was performed on a 12-month cohort study that followed up 614 attendees of specialty clinics in Toronto (nZ303) or London (nZ311), Ontario. In this study, survey questionnaires were fielded at initial specialty clinic attendance (baseline), and 3-, 6-, and 12-month follow-up. These surveys comprised questions on work status, sociodemographics, health-, and job-related variables based on a literature review of prognostic factors of work disability in musculoskeletal populations. Participants who had missed a follow-up survey were allowed to continue in the study (ie, complete the next survey).
List of abbreviations: CI NPV OPP OR PPV UL-WIS WI WRUED WSIB
confidence interval negative predictive value organizational policy and practice odds ratio positive predictive value Upper-Limb Work Instability Scale work instability work-related upper extremity disorder Workplace Safety and Insurance Board
www.archives-pmr.org
1659
Inclusion criteria Since our aim was to assess the predictive ability of a tool that is specifically applicable for working individuals, to be eligible for this analysis the injured worker must (1) have been working at study baseline, (2) have understood written English, and (3) have provided written consent for use of their deidentified survey data for research. All participants were explicitly assured that neither the WSIB nor their employer would have access to their surveys. In our dataset, 356 (58%) of 614 injured workers were working at baseline and therefore eligible for the current analysis.
Variables Outcome: transitioning out of work A transition out of work within the study period was our study outcome. At each of the 3 follow-up time points, a participant’s work status was determined by the survey question, “Are you currently working?” (yes or no). Four outcome categories were initially differentiated, depending on whether the participant had exited work and also the time of initial exit. These categories were (1) initial exit of work reported at 3-month follow-up; (2) initial exit of work reported at 6-month follow-up; (3) initial exit of work reported at 12-month follow-up; and (4) did not exit work (ie, working at all time points). For specialty clinic attendees, common reasons for not working include being on short or long sick leave, an inability to arrange suitable modified work with employer, or retirement. Main predictor of interest: UL-WIS WI was assessed at study baseline by the UL-WIS,11 which is part of a growing family of psychometrically sound and feasible measures applicable in various populations or settings.12-16 The UL-WIS consists of 17 items that assess perceptions of symptom control at work, work task performance, stamina at work, time management issues, cognitive distresses associated with work, and perceptions of sustainability of current work role.12 A dichotomous response option (yes [1]/no [0]) is provided for each item, and the scale is scored by summing all 17 items (range, 0e17). We allowed a maximum of 1 missing item (ie, 10% change in the UL-WIS estimate, then this factor would be considered a confounder and be included in the final multivariable model.34,35 No elimination steps were subsequently performed. Effects were expressed as unstandardized parameter estimates (b) and their associated SEs, Wald Z values, and adjusted odds ratios (ORs) with 95% confidence intervals (CIs). Proper model fit for each model was verified based on Hosmer-Lemeshow goodnessof-fit statistics (P>.05 was acceptable fit). For all regressions, complete case analysis (listwise deletion) was applied where observations with missing data on any covariates or study outcome would be excluded. To check for potential selection bias, the baseline characteristics were compared between subsets of observations included and excluded from the final multivariable model by way of t tests (for continuous variables) or chi-square tests (for categorical variables).
K. Tang et al Determination of UL-WIS cut-scores If the UL-WIS demonstrated independent predictive ability (ie, P.05).
Table 3 Multivariable logistic regression to examine the independent predictive ability of UL-WIS for a subsequent transition out of work Predictor
b
SE
Wald Z
P
Adjusted OR
Work instability Age Upper-limb pain intensity Mental health status OPPs
.17 e.05 .03
.05 .02 .01
3.22 e2.95 1.79
.001 .003 .07
1.18 (1.07e1.31) 0.95 (0.92e0.98) 1.03 (1.00e1.06)
.00
.01
e0.48
.63
1.00 (0.98e1.01)
e.45
.20
e2.25
.02
0.64 (0.43e0.94)
NOTE. nZ233, after listwise deletion of observations with missing data. Model likelihood ratio test Z46.36 (degrees of freedom, 5), P.05; c statistic Z.75.
Determination of UL-WIS cut-scores A plot of the PPVs and NPVs for each incremental UL-WIS cutscore over the scale range is illustrated in figure 3 (for raw data, see appendix 1). The largest PPV was observed at UL-WIS>15 (PPVZ.80; 95% CI, .52e.96), while the largest NPV was observed at UL-WIS15) and NPV (UL-WIS
