Print Share

Library

APS Bulletin • Volume 8, Number 2, 1998

The Pain Facts

Ernest Volinn, Department Editor

Disability in Low-Back Pain:
What Do the Numbers Mean?

James P. Robinson, MD PhD

Low-back pain (LBP) is the most common condition leading to workers' compensation claims associated with time loss (i.e., injuries sufficiently severe to lead a worker to miss days from work [Cheadle et al., 1994]). Research has shown that among workers with compensable claims for low-back strain, rapid recovery is typical (Frank et al., 1996; Hashemi, Webster, Clancy, & Volinn, 1997; Spitzer 1987; Volinn, Van Koevering, & Loeser, 1991; Waddell, 1994). In a comprehensive study of work-related spinal disorders (70% of which involved the low back) in Quebec, Spitzer showed the recovery curve in Figure 1 and described the recovery process from a back injury as follows: "The duration of absence from work was short in most cases: 74.2% of workers were absent less than 1 month. This figure sharply reduced to 9.4% for the second month. The curve flattens out for absence of more than 3 months, indicating that workers still absent from work at that time tend to remain absent. After 1 year, 4.3% of workers remained absent from work" (p. S13).

FIGURE 1 Compensated Back Injury by Duration of Absence from Work, Quebec, 1981 (19% of compensated workers did not lose any time from work and are not included in this graph [compensated for medical care only]).
NOTE. From "Scientific Approach to the Assessment and Management of Activity Related Spinal Disorder," by W.O. Spitzer, 1987, Spine, 12, p. 514. Copyright 1987 by Lippincott-Raven Publisher. Reprinted with permission.

Data such as Spitzer's allow clinicians to reassure workers with acute back injuries that their problems are very likely to resolve quickly. On the other hand, the flattening of the recovery curve after about 3 months elicits concern by experts on work-related back pain. The shape of the recovery curve implies that if a worker has not recovered from a back injury within several months, the probability is high that he or she will go on to very protracted disability. McGill (1968) stated the issue this way: "Records indicate that workers with back complaints who are off work over 6 months have only a 50% possibility of ever returning to productive employment. If they are off work over 1 year this possibility drops to 25%, and if more than 2 years, it is almost nil" (p. 174). In essence, McGill contended that the tail of the recovery curve for LBP is flat, so that patients who have not gone off disability within the first 2 years remain disabled indefinitely.

Although others have replicated the general shape of Spitzer's recovery curve (Andersson, Svensson, & Olden, 1983; Cheadle et al., 1994; Oleinick, Gluck, & Guire, 1996), the interpretation of the curve is somewhat problematic. This article addresses four issues that contribute to ambiguities in interpretation: the difference between disability status and work status, the shape of the tail of the recovery curve, the difference between the prognosis for an episode of LBP, and the prognosis for a worker who has sustained such an episode.

Disability status versus work status

Industrial compensation law in the state of Washington* generally mandates that injured workers receive payments for time loss (sometimes called wage replacement payments or compensation) when they are judged to be totally disabled from work temporarily because of a work injury. Time-loss status is a dichotomous variable; at any given time, a worker either is or is not receiving compensation. Detailed information about the time-loss status of injured workers is typically available in the data banks of industrial insurers. Thus, it is easy for an investigator to get information about the time-loss status of 100% of workers in a cohort. However, a worker's time-loss payments can be discontinued even though he or she has not returned to work. The reason for this is that decisions about time-loss payments are made by the claims manager for the worker's industrial claim. Typically, the claims manager terminates time-loss payments when he or she judges a worker to be employable. There is no requirement that the worker actually be employed when the payments are terminated.

The concept of return to work after a disabling injury is intuitively clear. It means that the individual has returned to competitive employment. However, the apparent simplicity of return to work is deceptive for at least two reasons:

  1. Return to work is a complex variable with multiple values. At one extreme, an injured worker might return to unrestricted duty at his or her preinjury job. At the opposite extreme, the worker might remain completely out of the work force, but there are many gradations between these extremes. For example, a worker might return to part-time work, light-duty work, or a job with lower pay. Thus, a study that simply divides patients into those who have returned to work and those who have not risks oversimplification of a complex outcome. I am not aware of any well-designed questionnaire for assessing the many gradations in return to work.
  2. The collection of data regarding work status is difficult. In most studies, subjects are simply asked about their work status at some time after an injury (Jamison, Matt, & Parris 1988; Johnson & Baldwin, 1993). Problems in interpreting the results of such studies include the difficulty of finding all subjects in a cohort and the possibility that some subjects will have an incentive to give inaccurate information. One strategy is to use collateral information sources such as records of employers, treating physicians, attorneys, or (theoretically, at least) income tax records (Burke, Harms-Constas, & Aden, 1994; Frederickson, Trier, Van Beveren, Yuan, & Baum, 1988), but these methods also have practical and ethical limitations.

Given these ambiguities, one would expect research reports on recovery from a back injury to include a clear statement about whether discontinuation of time-loss benefits or return to work was the outcome variable and to describe measures of return to work in detail. Unfortunately, many studies do not provide sufficient detail. Spitzer's statement, presented earlier in this article, suggests that return to work was the outcome variable, but the methods section of the article did not include any details about how Spitzer actually computed the variable. From an inspection of the entire report, it seems much more likely that Spitzer measured discontinuation of time loss and used this as a proxy for return to work.

The failure of researchers to distinguish clearly between discontinuation of time loss and return to work is conceptually confusing, but it is unimportant empirically if the two variables correlate highly. Unfortunately, the issue of the correlation between discontinuation of time loss and return to work has received little attention in published studies. Mitchell and Carmen (1994) said, "The cessation of wage loss payments to the injured worker most commonly signaled a return to work, and our experience with acute soft tissue and back injuries showed that this was accurate in more than 95% of cases" (p. 634), but they gave no details about how the relationship between return to work and cessation of wage-loss payments was assessed. Oleinick et al. (1996) reported an 80% correspondence between the two variables, but their method for assessing return to work was of questionable validity. Thus, although some data suggest a close relationship between return to work and discontinuation of time loss, the issue remains unresolved.

Moreover, the relation between the two variables might vary with the chronicity of an industrial claim. Informal experience suggests that in injuries associated with short durations of time loss, a worker's return to work becomes the stimulus for discontinuation of time loss, so that the two variables correspond closely. In contrast, chronically disabled workers are frequently declared employable and have their time loss payments stopped even when they are still out of the work force.

Disability status versus pain

Von Korff (1994) has studied the reported symptoms and activity restrictions reported by LBP patients in a primary care setting. He noted that LBP is often a recurrent condition so that after an index injury, a patient is at increased risk for a return of symptoms. In fact, he points out that it is often somewhat arbitrary to determine whether symptoms in a patient after an index low-back injury should be construed as recurrent back pain or as chronic back pain. In any case, he found that almost half of the patients in one study were symptomatic 1 year after an index back injury (Von Korff & Saunders, 1996). He discussed the difference between recovery in the sense of return to work and recovery in the sense of symptom resolution as follows: "The large majority of workers' compensation back pain cases return to work within 3 months of injury.... But, high rates of return to work immediately after a back injury should not be interpreted as indicating that the back pain has necessarily resolved as many go back to work while still experiencing intermittent or chronic pain. There is an essential distinction between the functional outcome of back pain and the pain outcome; function may be restored even though pain continues" (Von Korff, p. 2043S).

In essence, Von Korff's analysis points to two complications in the interpretation of the recovery curve after back injury. First, recovery in the sense of return to work (or termination of time-loss benefits) is different from recovery in the sense of symptom resolution. Second, back pain tends to be a recurrent condition, and episodes might blend into each other. This can make recovery from an episode difficult to determine with certainty.

Is the tail of the recovery curve always flat?

The results that McGill reported (1968) indicate a bleak prognosis for patients with back pain who have been disabled for extended periods of time. Waddell (1987) reached similarly bleak conclusions in a literature review. Both investigators emphasized the flatness of the tail of the recovery curve after a back injury. However, several considerations mitigate their conclusions:

  1. McGill's study fails to describe the methods used to determine return-to-work rates for patients with back injuries. Thus, it must be viewed with skepticism.
  2. Waddell's 1987 article relies mainly on data from the Workmen's Compensation Board of British Columbia. The outcome variable actually studied was discontinuation of time loss rather than return to work. Also, the figure displayed by Waddell (Figure 2) does not show actual data about the long-term outcomes of low-back claims. Instead, the time-loss status of a large cohort of workers in British Columbia was followed for up to 455 days after injury, and a curve-fitting procedure was used to extrapolate to longer-term outcomes (W.P. Hrudey, personal communication, August 6, 1997).
  3. There is probably no single recovery curve for industrial back injuries. The recovery curve that a clinician observes may well vary from one time period to another and from one administrative system to another, so that broad generalizations from data are hazardous. For example, actuarial data from the Washington State Department of Labor and Industries (Shufelt et al., 1997) indicate that as of March 31, 1997, the "closing rate" for claims 21-24 months old was about 12% (Figure 3). This means that if a worker had a date of injury between January 1, 1995, and March 31, 1995, and if he or she was receiving time-loss payments during the last quarter of 1996, the probability was .12 that he or she would have had the time-loss benefits discontinued during the first quarter of 1997. Turning this around, the closing rate curve shows that if a person with a 2-year-old claim was on time-loss as of the last quarter of 1996, the probability was .88 that he or she would still be on time loss as of March 31, 1997. If the risk for discontinuation of time loss was iterated over four quarters, we see that by December 31, 1997, the hypothetical patient would have had approximately a 60% chance of still being on time loss and only a 40% chance of being taken off. Thus, Washington's Labor and Industries data indicate that even for 2-year-old claims, the "recovery" rate (i.e., discontinuation of time loss) is 40% during the third year. Although the closing curve is based on all compensable claims rather than only on back pain claims, it probably reflects the recovery curve for LBP pain fairly well because a large proportion of the patients with 2-year-old compensable claims have low-back injuries.

FIGURE 2 Return to Work as a Function of Time Away from Work Because of Low-Back Pain, Showing the Proportion of Patients Returning to Work with Time and the Diminished Probability of Returning to Work in the Short Term or Ever (based mainly on date from Hrudey and the Workmen's Compensation Board of British Columbia
NOTE. From "A New Clinical Model for the Treatment of Low-Back Pain," by G. Waddell, 1987, Spin, 12, p. 633. Copyright 1987 by Lippincott-Raven Publisher. Reprinted with permission.

The rate curve for closing claims also suggests the fascinating effects of changes in the way the Department of Labor and Industries claims are managed. In 1985, the Washington state legislature repealed a law that had required vocational rehabilitation services for every chronically disabled injured worker ("Digest of an Important Publication," 1994). In early 1989, the Department of Labor and Industries started the Yes We Can project, which assembled a task force of experienced claims adjudicators to review chronic claims and resolve them. As Figure 3 shows, there were sharp increases in closing rates almost immediately after both changes were made. These increases suggest that discontinuation of time-loss payments for chronically disabled workers is influenced significantly by systems issues such as legislative initiatives and claims management strategies. If this is true, recovery curves for conditions such as industrial back injuries would be expected to be dynamic. Although patients with chronic injuries clearly tend to remain disabled, the very pessimistic view expressed by McGill and Waddell with respect to discontinuation of time-loss payments probably does not apply to all injured workers.

The prognosis for return to work among people with long-standing back injuries is also not as bleak as previous investigators have suggested. Franklin, Haug, Heyer, McKeefrey, & Picciano (1994) telephoned injured workers who had undergone lumbar spinal fusions between August 1, 1986, and July 31, 1987. The survey was done an average of 7.4 years after the workers had sustained their back injuries. The great majority of the workers had been on time loss for extended periods in the interval between their injuries and the survey. However, 41% of the respondents indicated that they had worked during the previous 4 weeks. Thus, Franklin et al.'s survey contradicts the observation that the return-to-work rate is "almost nil" (McGill, 1968) after workers have been disabled for as long as 2 years.

In essence, the Washington Department of Labor and Industries actuarial data and the data of Franklin et al. (1994) suggest that the tail of the recovery curve is not as flat as the curve McGill, Waddell, and others have proposed. Whether one measures outcomes by discontinuation of time-loss or by return-to-work statistics, chronically disabled workers in Washington state demonstrate higher rates of recovery than the earlier investigators reported. Also, the actuarial data suggest that recovery rates in the tail of the recovery curve are sensitive to systems issues such as legislative changes or changes in claims management strategies.

What is the long-term prognosis for an episode of low-back pain?

Most studies of industrial injuries focus on the first several months after an injury has occurred (Cheadle et al., 1994; Spitzer, 1987). In this context, a study by Johnson and Baldwin (1993) on approximately 11,000 workers in Ontario is a notable exception. Participants had sustained injuries leading to time off work between 1974 and 1987. They underwent interviews in 1989 and 1990; that is, between 2 and 15 years after their injuries. Two findings are particularly significant. First, even among workers who eventually returned to work after their injuries, the return to work was often unstable. Only 39% of the participants remained in the work force continuously after their first return to work. In contrast, 29% dropped out of the work force after their first reentry attempt and remained unemployed through the time of the interview; 21% achieved fairly stable reentries into the work force but only after numerous attempts; and 11% left the work force after numerous unsuccessful attempts to return to work (Butler, Johnson, & Baldwin, 1996). Second, among patients who had returned to work, those with low-back injuries were less likely to be employed at the time of the interview than those who had originally sustained other kinds of injuries (Johnson, Baldwin, & Butler, in press). These findings indicate that many industrial injuries have sequelae that continue beyond the time of the worker's first return to work, and suggest that low-back problems are more likely than other conditions to lead to recurring difficulties in the workplace.

The results of Johnson and colleagues adds yet another layer of complexity to the recovery curve for industrial low-back injuries. Basically, they caution that the prognosis for a worker who has sustained a back injury may be very different from the prognosis for a specific episode of back pain. Johnson et al.'s data have important implications for the methodology of research on industrial low-back pain. They suggest that the unit of analysis for the study of industrial low-back pain should be an individual who has sustained a back injury rather than an episode of back pain and that follow-up should continue for several years rather than for several months.

Conclusions

The general shape of the recovery curve after an episode of disabling LBP has been established, but several issues remain unresolved by previous investigations and are often not clearly articulated in research reports on recovery from back pain episodes. In particular, these are the unresolved issues:

  1. Discontinuation of time-loss compensation is conceptually different from return to work as an index of recovery from an episode of back pain. The two measures of recovery are not interchangeable.
  2. Reported symptoms follow a different recovery curve from either return to work or discontinuation of time loss compensation. It is often difficult to determine whether a patient with ongoing symptoms has failed to recover from an index episode of back pain or has sustained a recurrence.
  3. Although the recovery curve clearly flattens out for workers who have not recovered within several months, the tail of the curve is not always as flat as McGill, Waddell, and others have stated. That is, under at least some circumstances, patients with long-term claims for back pain show substantially higher recovery rates than those suggested by McGill and Waddell.
  4. The flatness of the tail of the recovery curve may well be influenced by systems factors such as legislative initiatives and claims management policies. More generally, there is no single recovery curve after an episode of LBP. Specific statements about the probability of recovery among workers with chronic back claims may not be generalizable over time and across insurance systems.
  5. Multiyear follow-up data on patients after an index episode of disabling LBP suggest a significant risk of unsuccessful long-term reentry into the workplace. These data highlight the importance of distinguishing between the prognosis for an episode of back pain and the prognosis for an individual who has sustained such an episode.

I hope this discussion will help clinicians as they discuss the significance of an episode of work-related low-back pain with their patients.

Acknowledgments


I wish to express my appreciation to Ernest Volinn, PhD, Richard Chapman, PhD, and William Hrudey, MD, for their suggestions during the preparation of this manuscript.

References
Andersson, G.B.J., Svensson, H., & Olden, A. (1983). The intensity of work recovery in low back pain. Spine, 8, 880-884. Burke, S.A., Harms-Constas, C.K., & Aden, P.S. (1994). Return to work/work retention outcomes of a functional restoration program. Spine, 19, 1880-1886.

Butler, R.J., Johnson, W.G., Baldwin, M.L. (1995). Managing work disability: Why first return to work is not a measure of success. Industrial and Labor Relations Review, 48, 452-469.

Cheadle, A., Franklin, G., Wolfhagen, C., Savarino, J., Liu, P.Y., Salley, C., & Weaver, C. (1994). Factors influencing the duration of work-related disability: A population-based study of Washington state workers' compensation. American Journal of Public Health, 84(2), 190-196.

Digest of an important publication: Workers' compensation success stories. (1994). In J.F. Burton (Ed.), 1995 workers' compensation year book. (pp. I-211-I-212). Horsham, PA: LRP Publications.

Frank, J.W., Kerr, M.S., Brooker, A., DeMaio, S.E., Maetzel, A., Shannon, J.S., Sullivan, T.J., Norman, R.W., & Wells, R.P. (1996). Disability resulting from occupational low back pain, Part I: What do we know about primary prevention? A review of the scientific evidence on prevention before disability begins. Spine, 21, 2908-2917.

Franklin, G.M., Haug, J., Heyer, N.J., McKeefrey, S.P., & Picciano, J.F. (1994). Outcome of lumbar fusion in Washington state workers' compensation. Spine, 19, 1897-1904.

Fredrickson, B.E., Trier, P.M., Van Beveren, P., Yuan, H.A., & Baum, G. (1988). Rehabilitation of the patient with chronic back pain: A search for outcome predictors. Spine, 13, 351-353.

Hashimi, L., Webster, B.S., Clancy, E.A., & Volinn, E. (1997). Length of disability and cost of workers' compensation low back claims. Journal of Occupational and Environmental Medicine, 39, 937-945.

Jamison, R.N., Matt, D.A., & Parris, W.C.V. (1998). Treatment outcome in low back pain patients: Do compensation benefits make a difference? Orthopaedic Review, 17, 1210-1215.

Johnson, W.G., & Baldwin, M. (1993). Returns to work by Ontario workers with permanent partial disabilities (Report to the Workers' Compensation Board of Ontario). Ottawa: Workers' Compensation Board of Ontario.

Johnson, W.G., Baldwin, M, & Butler, R.J. (in press). Back pain and work disability: The need for a new paradigm. Industrial Relations.

McGill, C.M. Industrial back problems. (1968). Journal of Occupational Medicine, 10, 1740-1748.

Mitchell, R.I., & Carmen, G.M. (1994). The functional restoration approach to the treatment of chronic pain in patients with soft tissue and back injuries. Spine, 19, 633-642.

Oleinick, A., Gluck, J.V., & Guire, K.E. (1996). Factors affecting first return to work following a compensable occupational back injury. American Journal of Industrial Medicine, 30, 540-555.

Shulelt, G., Redding, R., Wolfhagen, C., Vasek, W., Mercier, M., Hauser, J., Britton, T., Wilkinson, L., Crooker, K., & Nylander, J. (1997). Quarterly report on analysis and trends of the operations of the Insurance Services Division, first quarter, calendar year 1997. Olympia, WA: Department of Labor and Industries.

Spitzer, W.O. (1987). Scientific approach to the assessment and management of activity-related spinal disorders. Spine, 12, S1-S57.

Volinn, E., Van Koevering, D., & Loeser, J. (1991). Back sprain in industry. Spine, 16, 542-548.

Von Korff, M. (1994). Studying the natural history of back pain. Spine, 19, 2041S-4046S.

Von Korff, M., & Saunders, K. (1996).The course of back pain in primary care. Spine, 21, 2833-2837.

Waddell, G. (1987). A new clinical model for the treatment of low-back pain. Spine, 12, 632-644.

Waddell, G. (1994). Epidemiology review: The epidemiology and cost of back pain. London: HMSO.

* This discussion is based on the legal and administrative structure of the Department of Labor and Industries, the major industrial insurance carrier in Washington state. It is possible that some of the points are not relevant to other industrial insurance carriers. Back to article

James P. Robinson is a clinical assistant professor in the department of rehabilitation medicine at the University of Washington in Seattle.

Issue Index