The Relationship Between Work-Disability Duration and Claimant’s Expected Time to Return to Work as Recorded by Workers’ Compensation Claims Managers

With an increasing number of studies finding that a worker’s medical condition incompletely explains return to work (RTW) following occupational injury, the role of psychosocial influences has come into question [1, 2]. Of the psychosocial variables that have been examined, “RTW expectation” has frequently been found to relate to outcomes [3, 4]. Questions are now being raised regarding whether a worker’s expectations for RTW can be used in a clinical setting to gain an understanding of likely future outcome.

The relationship between RTW expectations and RTW outcomes has been observed not only in numerous geographical and social settings, but also across a variety of health conditions, disability durations, and methods of scientific inquiry [5]. However, while study results have indicated a relationship between expectations and RTW, our understanding of the relationship is still limited. The reasons for this are numerous, with some of the most important being: inconsistency in findings, study sample sizes that have limited the inclusion of covariates, and data being collected by researchers rather than those involved in the claims process.

Regarding inconsistency of findings, research has indicated that a claimant’s self-assessment may be more or less accurate depending on their condition. For example, in their analysis of 1040 workers’ compensation claimants, Gross and Battié found that recovery expectations predicted recovery in workers filing injury claims for back pain, but not other musculoskeletal conditions including sprains, strains, or pain of other body parts besides the back, other injuries such as fracture, dislocation, or amputation, nor other compensable conditions such as carpal tunnel [6].

Regarding sample size, studies have tended to have sample sizes that have restricted their analytical options and the number of covariates that could be included. Of the studies that have investigated the relationship between expectations and RTW outcomes, the maximum sample size was 1566 [7]. This study was conducted in Canada and found that four measures of recovery explained one-sixth of the variation in time receiving benefit. While expectations regarding RTW were not found to be individually predictive of time receiving benefits, this is likely due to the question about whether or not the respondent thought they would recover enough to return to their usual job. The next largest sample comprised 1068 people with a workers’ compensation claim for back pain in Washington State [8]. In this study, they found that very low recovery expectations (operationalized as being very uncertain about whether or not they would be working in 6 months’ time) were significant independent predictors of chronic work disability. Four other studies have had samples sizes of approaching 1000 [6, 9‐11]. In the majority of the remainder of studies, sample size has tended to be around 500 [e.g., 12‐17].

Regarding the impact of those collecting the data, to date, data has mainly been collected by research staff. It is possible that claimant responses are influenced by who is asking the question. For example, workers may be more willing to give an honest response to an un-invested party. Or, they may give what they believe to be a socially desirable (biased) response to someone with an interest in their specific case. While there are some studies in which those collecting the data have been directly involved in the RTW process [e.g., 18‐20] in these cases data has been collected by a treating health care professional: in the study by Waylett-Rendall and Niemeyer [20], data were collected at the point of care by therapists working in the hand therapy program; in the case of Nieuwenhuijsen et al. [18], data were collected by an occupational physician; and in the study by Gross and Battié [19], data were abstracted from the treating facility’s database. As yet, there has been no study of the relationship between RTW expectations data as collected by an insurer representative as part of the case management process and work-disability duration. Given that this stakeholder group is one of the most likely to implement interventions based on the information received, understanding the impact of who is asking is important for both research and clinical reasons.

In the current study, we utilized the WC data from a large, United States-based insurance company. The data covers claims from a variety of organizations with different workforce sizes and from various industries. We focused on data pertaining to claimants aged 18–80 who had at least 7 days of compensated temporary total disability (TTD) and who reported expecting to return to work. The reason for focusing on people with at least 7 days of TTD was that this timeframe represents a substantial period of time away from the workplace. As such, findings are likely to be more applicable to the subset of injured workers that has the potential to benefit most from work-disability prevention initiatives. So as to allow for at least 18 months of claims maturation, data were extracted for persons with an accepted claim that occurred from January 1, 2010 until December 31, 2013.

As part of the insurer’s claims management process, claimants who are off work, but not necessarily receiving indemnity payments, were asked about their RTW expectations. For the current study, we restricted our sample to claimants who provided an estimated RTW date within 30 days of the date that their injury was first reported to the workers’ compensation insurer. The reason for this is that our focus is on early risk prediction. We also restricted the sample to claims that had just one episode of TTD in the 365 days following RTW expectation data collection. Our justification for this is that while it is likely that the majority of respondents would reference their most immediate RTW when asked about their expectations, we could not be sure of this. Excluding people with multiple episodes of work-disability (TTD) means that we could be confident of the RTW the claimant was referencing. Similarly, in instances where a claimant had multiple claims within our data collection period (2010–2013), all claims for this individual were also excluded so as to avoid confusion about what incident a person was referencing when making RTW expectations.

Consistent with prior studies of disability recurrence, we considered a new episode of TTD to have occurred if there was more than a 7-day gap between TTD payments [21]. If the duration between TTD payments was 7 days or less, it was considered a single episode. We also excluded claimants who received a lump sum payment within 365 days of the date that the prediction was made of when a claimant would return to work. The reason for this is that in such cases, the end of TTD payments is a less reliable proxy for RTW. Claimants with missing data were also excluded. After the sample restrictions, 15,277 claims were included in our analyses. Figure 1 illustrates where sample was lost due to the various restrictions.

Three key dates were utilized in this study. These were the date of injury report, the date the RTW estimate was made, and the date that consecutive temporary total disability (TTD) payments ended (note that gaps of 7 days or less were ignored). The date that the injury was reported was used to assess whether an estimate for expected time to RTW was made within 30 days of the injury report day. The date the RTW estimate was made was then used to calculate (1) the expected length of time (days) to RTW and (2) the actual length of time (days) to RTW, estimated using the date that TTD payments ended. The key dates and associated measures are illustrated in Fig. 2.

In total, 15,277 claims were included in the analyses. The mean work-disability duration was 78.7 days (SD 102.8), and the mean expected time to return to work was 33.4 days (SD 42.8). Slightly more than two-thirds (67.8 %) of the claims were for men. Claimants’ ages ranged from 18 to 80 years with an average age of 42.6 years and the average length of tenure was 5.8 years. Approximately a fifth (21.7 %) of participants reported having a comorbid health condition (having hypertension and/or diabetes). The average perceived pain level was 5.4. The majority of claimants did not have a prior injury (70.7 %), while 11.1 % of claimants had a prior injury that was related to their current claim and 18.2 % had a prior injury that was not related to their current claim. The average number of days between the report date for the claim and when a claimant made a prediction about the date at which he/she would RTW was 9.7 days. Additional descriptive statistics can be found in Table 1.

Although the average expected time to RTW was less than the time it took for TTD payments to cease (33 vs. 79 days), analysis revealed a significant relationship between expected time to RTW and work-disability duration. The unadjusted correlation between work-disability duration and the expected time to RTW was .25 (p < .001). To examine how closely work-disability duration aligned with the expected time to RTW, we categorized both variables into eight categories and conducted cross-tabulations among the categories (see Table 2). In line with our expectations, within each of the length-of-disability categories, the largest percentage of claims were in the same or similar number-of-days category for the expected time to RTW. The most accurate group were those who expected to be back at work in 1-7 days, with 43 % of this group making a correct estimation. Approximately 10 % of the sample had an expected time to RTW of 7 days or less and a work-disability duration of 7 days or less. In contrast, less than 1 % of the sample had an expected time to RTW of 181 days or longer and a work-disability duration of 7 days or less. Those who expected to return to work in more than 7 days tended to be less accurate, but in most cases the rate of accuracy was still greater than 20 %. The rate of accuracy for those who expected to be off work for more than 7 days was 22 %. Overall, 28 % of the sample made estimates that were within 7 days (±) of their TTD payments ending; and 41 % were within 14 days (±) of when their TTD payments ended.

The adjusted relationship between expected time to RTW and work-disability duration while controlling for selected variables contained within the WC insurer’s administrative database was tested using a two-step analytical procedure. In the first step of the analyses, the covariates which represent the variables typically collected in administrative databases explained 20.2 % of the variance in work-disability duration (see Table 1). In the second step of the analyses, when the expected time to RTW was added to the model, the model then explained 29.8 % of the variance, with the expected time to RTW explaining an additional 9.5 % of the variance in work-disability duration beyond what was explained by the covariates. As hypothesized, the claimants’ expected time to RTW was related to work-disability duration, with a greater expected time to RTW being associated with an increase in work-disability duration (β = .346, p < .001).

We also conducted sensitivity analyses limiting our sample to claimants who had finished receiving indemnity payments for TTD within 365 days of RTW expectation data being collected. Findings remained consistent. The correlation between expected time to RTW and TTD duration was .26 (p < .001), and the expected time to RTW explained an additional 10.6 % of the variance in work-disability duration after adjusting for the covariates (full model explained variance = 27.4 %). Finally, we compared the main sample, with claimants who were excluded from the sample due to missing covariate data. Claimants with missing data had significantly (p < .001) shorter lengths of disability (mean = 60 days) and expected time to RTW was sooner (mean = 30 days) than those in the main sample (work-disability duration—mean = 79 days; expected time to RTW—mean = 33 days). For the claimants who were excluded as a result of missing information, the correlation between work-disability duration and expected time to RTW was slightly higher than for the study sample (r = .28 as compared to r = .25).¹

While past research has consistently found that workers’ expectations for return to work are predictive of their eventual outcome, understanding of the relationship has been limited by inconsistency in findings, small and restricted study sample sizes and the potential influence of the person collecting the data. Although estimates tended to be more optimistic than was actually the case, the current study’s findings support the hypothesis that claimant RTW estimates as recorded by claims managers are significantly related to compensated-disability duration, and that relationship is maintained after controlling for variance that can be explained by other variables contained within workers’ compensation data. With being true regardless of whether or not the claimant was in receipt of wage replacement payments at the time of being asked about their RTW expectations. Accuracy was highest for people who expected to RTW within a week. Although accuracy was found to decrease in cases where the estimated time to RTW was greater than 7 days, in more than 40 % of the sample the predicted time to RTW was within 14 days of the end of their TTD payments.

Study findings are consistent with a relationship observed in a recent investigation into RTW following carpal tunnel release [29] where it was found that expected time to RTW explains 18 % of the variance when compared to time to a full RTW. This is despite the current sample’s estimated time to RTW being longer than was the case in the earlier study (33.4 days vs. 18.9 days) and our larger sample size (N = 15,277 vs. N = 65). Such findings add further support to the idea that injured worker’s expectations for RTW can be used in a clinical or insurance setting to gain an understanding of likely future outcome. Further, current study findings also indicated that expected time to RTW, based on our earlier review of the literature [5], allows for the collection of RTW expectation data that is related to disability duration. As such, findings support the suggested measure as being effective for work-disability risk prediction purposes.

When conceptualizing this study we saw the potential for the relationship between RTW expectation and outcome to be influenced by who was questioning the workers about their expectations. More specifically, we saw the potential for workers to give a more “socially desirable” (and thus less accurate) response when they were asked about their RTW timeframes by an insurance representative involved in their case management. While we cannot be sure that this was not the case, we did observe that workers’ RTW estimates as recorded by their claims managers were highly predictive of compensated work-disability duration.

Our finding that some of the relationship between RTW expectations and work-disability duration was accounted for by demographic and injury variables is consistent with the idea that RTW expectations represent a self-assessed summary of claimant’s individual and contextual biopsychosocial influences [4, 30]. This finding also adds support to the idea that unpacking the reasoning behind one’s expectations, has the potential to assist in the identification of RTW facilitation opportunities, as well as obstacles that may be amenable to intervention with the aim of improving the sickness-absent worker’s RTW outcome [31].

In terms of implications for work-disability prevention, our finding that expected RTW was highly predictive of work-disability duration indicates that claims managers can use this information to gain an understanding of likely outcome. It provides a starting point for discussion of what might be done to assist workers to achieve a timely, safe and sustained RTW. If someone expects to RTW in the near future, this suggests that the injured worker has the necessary resources to return without the need for assistance or intervention (making them a “low touch” claim). However, it should be noted that this is not necessarily the case. There may be instances where expectations are unrealistic or forced, in which case there is the potential that the RTW could result in adverse effects [32]. If this is suspected, a low touch approach may not be the appropriate course of action. For persons who provide a far off estimate, this could indicate that help is needed, especially if that person’s health condition is relatively minor.

Study findings add to the body of knowledge indicating a relationship between RTW expectations and RTW outcomes. They demonstrate a relationship that is present when type of condition is varied, and when the sickness-absent worker is questioned by an insurance representative involved in the worker’s claim management. While reported RTW expectations share explanatory power with other variables found in administrative databases, findings indicate that additional insight into likely work-disability duration can be gained through asking the workers when they expect to be back at work.