Does the Upper-Limb Work Instability Scale Predict Transitions Out of Work Among Injured Workers?

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 <6 (negative predictive value, .81; 95% CI, .62e.94) and >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% of the scale), otherwise the UL-WIS was not scored. Evidence of internal consistency and construct validity of the UL-WIS have been previously demonstrated in WRUEDs.11 Other variables (descriptors, model covariates, or both) In terms of variables assessed at baseline, sociodemographic variables included age, sex, marital status, and level of education. Health-related variables included number of painful sites (15 upper-body regions assessed from a pain diagram), upper-limb pain intensity (5-item pain subscale of the Shoulder Pain and Disability Index),17-19 mental health status (5-item mental health subscale from version 2 of the Medical Outcomes Study 36-Item Short-Form Health Survey),20 and number of comorbidities based on a modified Sangha scale (11 conditions assessed).21 Job-related variables included several subscales from the Job Content Questionnaire22: psychological job demands (5 items), skill discretion (3 items), and decision authority (6 items), as well as upper-limb physical job demands (a 4-item checklist),23 coworker support (4 items from the Psychosocial Aspect of Work Questionnaire),24 and workplace organizational policies and practices (OPPs) (11 items).25 To further describe the cohort, injured workers were

1660 classified into 1 of 9 occupational types according to the National Occupational Classification developed by Human Resources and Skills Development Canada26 based on their job titles.

Analysis Descriptive statistics To describe the study participants, univariate statistics were applied to examine the distribution of baseline sociodemographic, health-related, and work-related variables. The baseline UL-WIS was further described at the item level. Logistic regression Given the small number of follow-up time-points and lack of precise information on the timing of work transitions (ie, large intervals between follow-up), logistic regression was ultimately preferred over a time-to-event analysis. A binary outcome specification was chosen (over a 4-level ordinal outcome) because of the small sample size in several of the initial outcome categories. To this end, we operationalized a binary outcome to differentiate between participants who had remained working at all 3 follow-up time points (nonevent), and those who had transitioned out of work at any point (event) by collapsing 3 of 4 initial outcome categories. First, a univariable logistic regression was performed with baseline UL-WIS as the sole predictor variable. For this model, the concordance (c) statistic was specifically observed to inform the predictive ability of the UL-WIS as a standalone tool. Then, a multivariable logistic regression was performed with the main intent of assessing whether the relationship between the ULWIS and study outcome is independent (ie, free from significant confounding bias). To this end, our strategy was to consider a broad range of personal (worker-related) and environmental (workplace-related) factors for inclusion into the multivariable model, since both types of factors have postulated relevance in occupational disability.27-31 Adhering to current guidelines on events-per-variable ratio (ie, not exceeding 10:1),32,33 a total of 9 variables were selected for consideration, including 4 personal factors (age, sex, upper-limb pain intensity, mental health status) and 5 environmental factors (upper-limb physical job demands, coworker support, social support, OPPs, and job strain [psychological job demands divided by the sum of skill discretion and decision authority]). To identify confounders of the relationship between UL-WIS and exiting work, a “change-in-estimate” screening step was initially performed. By this approach, each of the 9 factors was individually added to a base model with only UL-WIS as the predictor. If the inclusion of a factor led to a >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 in multivariable regression), then cut-scores were sought to differentiate between low, moderate, and high risk of exiting work, similar to previous efforts for the original Work Instability Scale for Rheumatoid Arthritis.12,36 To this end, we assessed and compared the positive predictive value (PPV) and negative predictive value (NPV)37 at each UL-WIS score increment across the scale range. In this context, the PPV would represent the probability of exiting work over the course of this study given a “positive” test (ie, UL-WIS above a given cut-score). The NPV, on the other hand, would represent the probability of not exiting work given a “negative” test (ie, UL-WIS below a given cut-score). Guided by these predictive values, our aim was to determine 2 UL-WIS cut-scores: a first threshold where scores greater than this value would be associated with a fairly high likelihood of subsequently exiting work (eg, PPV0.8, high risk), and a second threshold where scores below this value would be associated with a fairly high likelihood of not exiting work (eg, NPV0.8, low risk). Two main considerations were made in determining such cut-scores: (1) the magnitude of the PPV/NPV (ie, higher values are more desirable), and (2) the overall trend of these values across the scale rangedfor example, whether there are dramatic “step ups” in PPV/NPV between adjacent cut-scores that would warrant additional considerations. All study analyses were conducted with R version 3.1.a

Results Baseline characteristics Study participants had a mean age  SD of 44.99.2 years (range, 19e68) and 55.7% were women (table 1). Injured workers were most commonly employed in 2 sectors: trade, transport, and equipment operators (25.5%) or processing, manufacturing, and utilities (33.2%). On average, workers reported 5.03.8 painful upper-body sites, and the most commonly affected regions were the shoulder (75.0%), elbow (52.0%), and upper arm (49.7%). Bilateral upper extremity pain was also common (41.0%), suggestive of complex etiology. The Shoulder Pain and Disability Index pain subscale mean  SD was 30.711.4, and the Medical Outcomes Study 36-Item Short-Form Health Survey mental health subscale mean  SD was 64.921.4, indicating moderate pain intensity and mental health. The mean  SD UL-WIS, scored in 309 (86.8%) of 356 injured workers, was 10.13.9 (fig 1). The following 4 UL-WIS items were most often affirmed:  Item 8: I’ve got to watch how much I do certain things at work (95.7%);  Item 7: I have to say no to certain things at work (85.1%);  Item 12: I get on with work but afterwards I have a lot of pain (84.1%);  Item 3: I have pain or stiffness all the time at work (82.4%) (table 2).

Extent of transitioning out of work Of the 356 eligible participants, 280 (78.7%) could be classified into our initial outcome categories, as some had completely dropped out of the study after baseline (nZ55, 15.4%) or provided

www.archives-pmr.org

Predicting work transitions Table 1

1661

Baseline characteristics of injured workers

Variable

Instrumentation

Scale Range

n Available

Mean  SD

%

Age (y) Sex Marital status Education

Single Single Single Single

353 352 351 351

44.99.2 NA NA NA

NA 55.7 70.9 33.3

No. of painful sites

347

5.03.8

NA

30.711.4 64.921.4

NA NA

356 309

0.80.9 10.13.9

NA NA

Skill discretion Decision authority

JCQ subscale (6 items) JCQ subscale (3 items)

336 336

2.50.6 2.50.7

NA NA

Psychological job demands Supervisor support OPPs Upper-limb physical job demands Coworker support

JCQ subscale (5 items) JCQ subscale (5 items) OPP-11 (11-items) Checklist (4 items)

0e50 (50, high pain intensity) 0e100 (100, best mental health status) Out of 11 health conditions 0e17 (17, high work instability) 1e4 (4, high skill discretion) 1e4 (4, high decision authority) 1e4 (4, high demands) 1e4 (4, high support) 1e5 (5, favorable OPPs) 0e4 (4, high demands)

349 346

No. of comorbidities Work instability

Body diagram (15 upper-body sites assessed) SPADI pain subscale (5 items) SF-36 mental health subscale (5 items) Modified Sangha scale UL-WIS (17 items)

Numeric % Women % Married or common law % Completed postsecondary level 0e15 painful sites

317 303 346 349

2.70.5 2.80.7 3.10.7 2.41.2

NA NA NA NA

PAWQ subscale (4 items)

1e5 (5, high support)

340

3.70.9

NA

Upper-limb pain intensity Mental health status

item item item item

Abbreviations: JCQ, Job Content Questionnaire; NA, not applicable; OPP-11, 11-item Organizational Policies and Practices Scale; PAWQ, Psychosocial Aspect of Work Questionnaire; SF-36, Medical Outcomes Study 36-Item Short-Form Health Survey; SPADI, Shoulder Pain and Disability Index.

incomplete data at follow-up (nZ21, 5.9%). Among the 280 individuals who were classified, 78 (21.9%) had initially exited work at the 3-month follow-up, 40 (11.2%) had initially exited work at 6 months, and 25 (7.0%) had initially exited work at 12 months, whereas 137 (38.5%) had stayed working at all time points. To summarize, 51.5% (143/280) had exited work within the study period. The baseline UL-WIS means stratified by the outcome categories are presented in figure 2.

Fig 1

Predictive ability of UL-WIS Applying our binary outcome (“event” refers to transitioning out of work at any point; nZ143/280 [51.5%]), univariable logistic regression (nZ243 after listwise deletion) revealed baseline ULWIS to be a statistically significant predictor (crude OR for 1-point increase, 1.21; 95% CI, 1.12e1.31; P<.001). The model c statistic was .69. Screening of covariates by the “change-in-estimate”

Frequency distribution of baseline UL-WIS total score (range, 0e17) among injured workers with chronic WRUEDs (available in nZ309).

www.archives-pmr.org

1662 Table 2

K. Tang et al Item-level distribution of UL-WIS

Item No.

Label

Yes*

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Reduce hours Worried Pain or stiffness Stamina Holiday Push myself Say no Watch how much Opening doors Extra time Give up work Get on Feeling tired Restricted Getting up earlier Very stiff Stress

75 230 285 234 61 240 291 333 140 225 115 285 188 164 169 242 191

(23.2) (66.9) (82.4) (68.2) (17.7) (70.2) (85.1) (95.7) (41.1) (74.6) (34.0) (84.1) (57.5) (48.8) (49.4) (71.2) (56.7)

No

Missing

247 114 61 109 283 102 51 15 201 87 223 54 139 172 173 98 146

34 12 10 13 12 14 14 8 15 14 18 17 29 20 14 16 19

NOTE. Values are n (%) or n. * Percent calculations (in parentheses) exclude missing values.

approach identified 4 confounders: age, upper-limb pain intensity, mental health status, and OPPs. With all confounders included in our multivariable logistic regression (nZ233 after listwise deletion), the UL-WIS remained a significant predictor (adjusted OR for 1-point increase, 1.18; 95% CI, 1.07e1.31; PZ.001) (table 3). In addition, age (adjusted OR for 1-year increase, .95; 95% CI, .92e.98; PZ.003) and OPPs (adjusted OR for 1-point increase, .64; 95% CI, .43e.94; PZ.02) were also revealed to be significant predictors. Adequate model fit was verified for all regressions based on the Hosmer-Lemeshow goodness-of-fit statistic (P>.05). No differences in any baseline characteristics were found between observations included (nZ233) and excluded (nZ123) from the multivariable model (all comparisons 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<.0001. Hosmer-Lemeshow goodness of fit: 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-WIS<6 (NPVZ.81; 95% CI, .62e.94). These were ultimately considered sensible cut-score choices after inspecting the trends in predictive values over the UL-WIS range. By this, 3 bands of scores emerged as follows: low (<6), moderate (6e15), and high risk (>15) for a subsequent transition out of work.

Discussion In this study, transitioning out of work was revealed to be quite common among working individuals with chronic/complex WRUEDs. Also, the level of WI, measured by the UL-WIS, was shown to be an independent predictor of this outcome. As such, where work sustainability is a concern, current results suggest that the experience of WI among working individuals is worthy of some attention among clinician and workplace parties.

Fig 2 Baseline UL-WIS total score mean (range, 0e17) stratified by outcome categories (error bars represent SD). Abbreviations: f/u, follow-up; n, sample size available for calculating UL-WIS means.

www.archives-pmr.org

Predicting work transitions

Fig 3

1663

PPVs (A) and NPVs (B) across the UL-WIS scale range. Points of highest PPV (UL-WIS>15) and NPV (UL-WIS<6) were identified.

Our efforts to determine UL-WIS cut-scores were intended to help benchmark the tool to enhance its future use. We believe cutscores can be useful for a number of potential applications, such as triaging injured workers (eg, identifying those at high risk for exiting work), goal-setting during patient care (eg, achieving a score threshold), or research purposes (eg, stratification of a continuous measure). That said, we caution that limits to the predictive accuracy of the UL-WIS need to be recognized when applied as a standalone tool. Despite being a statistically significant predictor in our regressions, a c statistic of .69 is less than impressive (equivalent to area under the receiver operating characteristic curve, general guidelines: .60e.69, poor; .70e.79, fair), and the maximum PPV and NPV attained across the scale range were .80 and .81, respectively (ie, w20% probability of incorrect prediction). Also, the 95% CIs associated with PPV/NPV estimates for the 2 proposed cut-scores were quite wide since these cut-scores were approaching the scale ceiling or floor (see appendix 1 for raw data used in calculation). To further ascertain these cut-scores, replication of the current analysis would be worthwhile in the future, particularly in samples where extreme UL-WIS scores (both floor and ceiling) are not rare. Since there is room to augment predictive accuracy, future research may consider exploring the development of a prognostic “index” consisting of other measures alongside the UL-WIS, as additional predictive factors of work sustainability are uncovered over time. A key strength of our study is the focus on a cohort of injured workers with a common area of injury, as well as stage of recovery. In recent years, scholars have argued for the importance of www.archives-pmr.org

considering disease and phase specificity in occupational disability research, since unique factors may be relevant at different stages of recovery and also for different disorders.28,38-43 We believe research attention to this population is particularly worthwhile because chronic/complex WRUED cases are very burdensome to the workers as well as the workers’ compensation and health care systems. A key methodological strength is that our multivariable regression considered a broad range of personal and environmental variables as potential confounders, which strengthened the case for the independent predictive effect of the UL-WIS (ie, not spurious).

Study limitations Four key study limitations should be considered: (1) our results may not be generalizable to milder or fast-resolving WRUED cases, or both, where exiting work is likely much less common and high WI immediately after injury may not have long-term consequences; (2) because predictive values are influenced by the prevalence of the event of interest, caution is needed when interpreting our PPV/NPV estimates outside the current context. For example, a much lower prevalence of exiting work might be expected in the general labor force; (3) any work transitions taking place between time points would not have been captured in our survey; and (4) a nontrivial proportion (34.6%) of the eligible sample had to be excluded from our multivariable regression because of missing data on the outcome or any of the covariates; however, no evidence of significant selection bias was found, as

1664

K. Tang et al

baseline characteristics did not differ between those included and excluded from the final multivariable model.

Occupational injuries; Predictive value of tests; Rehabilitation; Upper extremity; Work; Workers’ compensation

Conclusions We conclude that the level of WI, quantified by the UL-WIS, is an independent predictor of a subsequent transition out of work among injured workers recovering from chronic/complex WRUEDs. However, as a standalone tool, the predictive accuracy of the UL-WIS is rather modest, which points to the need to (1) apply the scale and proposed cut-scores with care, and (2) continue to uncover other predictive factors of work sustainability beyond WI. Recognizing a connection between high WI and poor work sustainability, we further recommend an increased awareness for the extent of WI experienced by injured workers, as well as continued efforts to identify interventions that can lessen WI.

Supplier a. R version 3.1; R Foundation for Statistical Computing. Available at: http://www.r-project.org.

Appendix 1

Keywords

Corresponding author Kenneth Tang, PhD, School of Rehabilitation Science, Faculty of Health Sciences, McMaster University, Room 403, 1400 Main St. W., Hamilton, Ontario, Canada L8S 1C7. E-mail address: ktang@ mcmaster.ca.

Acknowledgments We thank the clinical and research staff at the 2 participating WSIB specialty clinics for their support: the Shoulder and Elbow Specialty Clinic at the Sunnybrook Holland Orthopaedic and Arthritis Centre in Toronto, and the Hand and Upper-Limb Clinic at St. Joseph’s Health Care London. We are especially grateful for their efforts to help coordinate the collection of patient data for this study.

Raw study data used for determining UL-WIS cut-scores (usable nZ243)

UL-WIS Categories (Determined From This Study) Band 1 (low risk) UL-WIS total score <6

Band 2 (moderate risk) UL-WIS total score 6e15

Band 3 (high risk) UL-WIS total score >15

UL-WIS Total Score

n

Exited Work (Event)

Did Not Exit Work (Nonevent)

0 1 2 3 4 5 Subtotal 6 7 8 9 10 11 12 13 14 15 Subtotal 16 17 Subtotal Total

3 2 3 5 7 7 27 16 12 19 23 22 21 29 20 19 20 201 11 4 15 243

1 0 1 1 1 1 5 4 5 7 9 10 14 20 13 11 12 105 9 3 12 122

2 2 2 4 6 6 22 12 7 12 14 12 7 9 7 8 8 96 2 1 3 121

www.archives-pmr.org

Predicting work transitions

1665

References 1. Workplace Safety and Insurance Board of Ontario. By The Numbers: 2014 WSIB Statistical Report (Schedule 1); 2014. Available at: http://www. wsibstatistics.ca/wp-content/uploads/2015/05/WSIB_BTN_SCHED1.pdf. Accessed June 23, 2015. 2. Bureau of Labor Statistics. Nonfatal occupational injuries and illnesses requiring days away from work, 2012. Washington (DC): U.S. Department of Labor; 2013. 3. Huisstede BM, Bierma-Zeinstra SM, Koes BW, Verhaar JA. Incidence and prevalence of upper-extremity musculoskeletal disorders. A systematic appraisal of the literature. BMC Musculoskelet Disord 2006;7:7. 4. Keogh JP, Nuwayhid I, Gordon JL, Gucer PW. The impact of occupational injury on injured worker and family: outcomes of upper extremity cumulative trauma disorders in Maryland workers. Am J Ind Med 2000;38:498-506. 5. Mayer TG, Gatchel RJ, Polatin PB, Evans TH. Outcomes comparison of treatment for chronic disabling work-related upper-extremity disorders and spinal disorders. J Occup Environ Med 1999;41:761-70. 6. Baldwin ML, Butler RJ. Upper extremity disorders in the workplace: costs and outcomes beyond the first return to work. J Occup Rehabil 2006;16:303-23. 7. Pransky G, Benjamin K, Hill-Fotouhi C, et al. Outcomes in workrelated upper extremity and low back injuries: results of a retrospective study. Am J Ind Med 2000;37:400-9. 8. Burton WN, Pransky G, Conti DJ, Chen CY, Edington DW. The association of medical conditions and presenteeism. J Occup Environ Med 2004;46(6 Suppl):S38-45. 9. Burton WN, Chen CY, Conti DJ, Schultz AB, Pransky G, Edington DW. The association of health risks with on-the-job productivity. J Occup Environ Med 2005;47:769-77. 10. Hemp P. Presenteeism: at workdbut out of it. Harv Bus Rev 2004;82: 49-58. 11. Tang K, Beaton DE, Gignac MA, Bombardier C. Rasch analysis informed modifications to the Work Instability Scale for Rheumatoid Arthritis for use in work-related upper limb disorders. J Clin Epidemiol 2011;64:1242-51. 12. Gilworth G, Chamberlain MA, Harvey A, et al. Development of a work instability scale for rheumatoid arthritis. Arthritis Rheum 2003; 49:349-54. 13. Gilworth G, Emery P, Gossec L, et al. Adaptation and cross-cultural validation of the Rheumatoid Arthritis Work Instability Scale (RAWIS). Ann Rheum Dis 2009;68:1686-90. 14. Gilworth G, Emery P, Barkham N, Smyth MG, Helliwell P, Tennant A. Reducing work disability in ankylosing spondylitis: development of a work instability scale for AS. BMC Musculoskelet Disord 2009;10:68. 15. Gilworth G, Carey A, Eyres S, et al. Screening for job loss: development of a work instability scale for traumatic brain injury. Brain Inj 2006;20:835-43. 16. Gilworth G, Bhakta B, Eyres S, Carey A, Anne Chamberlain M, Tennant A. Keeping nurses working: development and psychometric testing of the Nurse-Work Instability Scale (Nurse-WIS). J Adv Nurs 2007;57:543-51. 17. Roach KE, Budiman-Mak E, Songsiridej N, Lertratanakul Y. Development of a shoulder pain and disability index. Arthritis Care Res 1991;4:143-9. 18. Roy JS, MacDermid JC, Woodhouse LJ. Measuring shoulder function: a systematic review of four questionnaires. Arthritis Rheum 2009;61:623-32. 19. MacDermid JC, Solomon P, Prkachin K. The Shoulder Pain and Disability Index demonstrates factor, construct and longitudinal validity. BMC Musculoskelet Disord 2006;7:12. 20. Ware JE Jr. SF-36 health survey update. Spine (Phila Pa 1976) 2000; 25:3130-9. 21. Sangha O, Stucki G, Liang MH, Fossel AH, Katz JN. The SelfAdministered Comorbidity Questionnaire: a new method to assess comorbidity for clinical and health services research. Arthritis Rheum 2003;49:156-63. 22. Karasek R, Brisson C, Kawakami N, Houtman I, Bongers P, Amick B. The Job Content Questionnaire (JCQ): an instrument for internationally

www.archives-pmr.org

23.

24.

25.

26.

27.

28.

29.

30. 31.

32.

33.

34. 35.

36.

37. 38.

39.

40.

41. 42.

43.

comparative assessments of psychosocial job characteristics. J Occup Health Psychol 1998;3:322-55. Eriksen W, Natvig B, Knardahl S, Bruusgaard D. Job characteristics as predictors of neck pain. A 4-year prospective study. J Occup Environ Med 1999;41:893-902. Symonds TL, Burton AK, Tillotson KM, Main CJ. Do attitudes and beliefs influence work loss due to low back trouble? Occup Med (Lond) 1996;46:25-32. Tang K, MacDermid JC, Amick BC III, Beaton DE. The 11-Item Workplace Organizational Policies and Practices Questionnaire (OPP11): examination of its construct validity, factor structure, and predictive validity in injured workers with upper-limb disorders. Am J Ind Med 2011;54:834-46. Human Resources and Skills Development Canada. National occupational classification 2011. Available at: http://www5.hrsdc.gc.ca/ NOC/English/NOC/2011/Welcome.aspx. Accessed October 26, 2014. Schultz IZ, Stowell AW, Feuerstein M, Gatchel RJ. Models of return to work for musculoskeletal disorders. J Occup Rehabil 2007;17:327-52. Krause N, Frank JW, Dasinger LK, Sullivan TJ, Sinclair SJ. Determinants of duration of disability and return-to-work after workrelated injury and illness: challenges for future research. Am J Ind Med 2001;40:464-84. Feuerstein M, Shaw WS, Lincoln AE, Miller VI, Wood PM. Clinical and workplace factors associated with a return to modified duty in work-related upper extremity disorders. Pain 2003;102:51-61. Sandqvist JL, Henriksson CM. Work functioning: a conceptual framework. Work 2004;23:147-57. Tang K, Escorpizo R, Beaton DE, et al. Measuring the impact of arthritis on worker productivity: perspectives, methodologic issues, and contextual factors. J Rheumatol 2011;38:1776-90. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 1996;49:1373-9. Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361-87. Greenland S, Neutra R. Control of confounding in the assessment of medical technology. Int J Epidemiol 1980;9:361-7. Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic research: principles and quantitative methods. Belmont: Lifetime Learning Publications; 1982. Tang K, Beaton DE, Gignac MA, Lacaille D, Zhang W, Bombardier C. The Work Instability Scale for Rheumatoid Arthritis predicts arthritisrelated work transitions within 12 months. Arthritis Care Res (Hoboken) 2010;62:1578-87. Altman DG, Bland JM. Diagnostic tests 2: predictive values. BMJ 1994;309:102. Krause N, Ragland DR. Occupational disability due to low back pain: a new interdisciplinary classification based on a phase model of disability. Spine (Phila Pa 1976) 1994;19:1011-20. Krause N, Dasinger LK, Deegan LJ, Rudolph L, Brand RJ. Psychosocial job factors and return-to-work after compensated low back injury: a disability phase-specific analysis. Am J Ind Med 2001;40:374-92. Dasinger LK, Krause N, Deegan LJ, Brand RJ, Rudolph L. Physical workplace factors and return to work after compensated low back injury: a disability phase-specific analysis. J Occup Environ Med 2000;42:323-33. Lotters F, Burdorf A. Prognostic factors for duration of sickness absence due to musculoskeletal disorders. Clin J Pain 2006;22:212-21. Sinclair SJ, Sullivan TJ, Clarke JA, Frank JW. A framework for examining return to work in workers’ compensation: a view from one North American jurisdiction. International examinations of medical-legal aspects of work injuries. Lanham and London: Scarecrow Pr; 1998. Lotters F, Franche RL, Hogg-Johnson S, Burdorf A, Pole JD. The prognostic value of depressive symptoms, fear-avoidance, and selfefficacy for duration of lost-time benefits in workers with musculoskeletal disorders. Occup Environ Med 2006;63:794-801.