Predictors of Time Loss After Back Injury in Nurses

Predictors of Time Loss After Back Injury in Nurses SPINE Volume 24, Number 18, pp 1930 1936 1999, Lippincott Williams & Wilkins, Inc. Robert B. Tate, MSc,* Annalee Yassi, MD,* and Juliette Cooper, PhD Study Design. A 2-year prospective inception cohort study of back injury in nurses. Objectives. To determine the extent to which characteristics of nurses, of the injury, and of the workplace predict occurrence and duration of time loss from work after back injury. Summary of Background Data. During 2 years, 320 nurses incurred 416 back injuries at a large teaching hospital in Winnipeg, Canada. Nurses injured on preselected wards were targeted for early intervention, including provision of modified work, whereas nurses injured on other wards received the usual care. Methods. Time loss attributable to the back injury during the 6 months after injury was analyzed. Three statistical models were used to examine occurrence of time loss (logistic regression), duration of time loss (Tobit regression), and duration of time loss once an injury incurring time loss had been documented (least-squares regression). Results. In 218 of the 416 injuries, the injured nurse consented to interview. Whereas perceived disability was related to whether a time loss injury would ensue, selfreported pain was strongly related to the duration of time loss once an injury had become a time loss injury. Duration of time loss was reduced by participation in the return-to-work program. Mechanism of injury, specifically injury occurring while lifting patients, resulted in greater time loss. Conclusions. Focusing on reducing the perception of disability at the time of injury is critical to preventing time loss, but once time loss has occurred, offer of modified work and attention to pain reduction are warranted. The findings add to the evidence that workplace-based intervention programs can be effective in reducing the morbidity resulting from back injury. [Key Words: low back injury, nurses, time loss] Spine 1999;24:1930 1936 From the *Department of Community Health Sciences, and the Division of Occupational Therapy, University of Manitoba; and the Departments of Occupational and Environmental Medicine and Occupational Therapy, Health Science Center, Winnipeg, Manitoba, Canada. Supported in part by the Workers Compensation Board of Manitoba and National Centres of Excellence-Health Evidence Application (NCE HEALNet) (funded by Medical Research Council and Social Science and Health Research Council [SSHERO]). Acknowledgment date: April 2, 1998. First revision date: August 14, 1998. Acceptance date: November 9, 1998. Device status category: 1. Work-related lower back injuries are frequent and costly. 2,9,11,16,20,24,28 Employees in health care settings, specifically ward nurses in hospitals, have been identified as workers at very high risk of work-related back injury, primarily because of lifting and transferring of patients. 1,12,13 The seriousness of this problem was recognized at the Health Sciences Centre in Winnipeg, Manitoba, Canada, where a 2-year pilot program was designed and implemented in 1990 to promote early intervention and modified work alternatives to injured nurses, to reduce TL caused by back injury. 30 A description of the epidemiology of back injury in this cohort and the effectiveness and cost benefit of the early intervention have been reported. 29,30 As noted by the Institute for Work and Health 7 and others, including Spitzer et al 25 in their task force report, approximately 7% of workers who sustain back injury are absent from work for an extended time, collectively incurring 70% of the total costs associated with this disorder. As noted by Deyo and Diehl, 5 many clinicians believe that aggressive early intervention might preclude the emergence of chronic pain syndromes if patients at high risk for disability could be identified. These investigators have been attempting to elucidate the factors associated with the prognosis of patients who have uncomplicated mechanical low back pain, to determine the predictors of chronic pain syndrome. Lehmann et al 17 designed a prospective study to develop a predictive model that would ascertain whether risk for long-term disability can be predicted in patients who have workrelated acute low back pain. Although limited by a small sample (45 people), the results of their study showed that patients who had a high rate of previous low back problems, whose injuries were associated with lifting, and whose pain and function scores indicated moderate to low pain magnitude but moderate to high levels of disability all had poor prognoses. Although personal characteristics, working conditions, history of prior back injury, and severity and early treatment of injury have all been identified as factors that relate to likelihood of incurring TL from work because of back injury and delaying return-to-work, as noted by Hazard et al, 15 cohorts have varied so much in selection techniques, pain duration, disability levels, and multiple socioeconomic factors, that initial comparisons and generalizations to clinical practice have been difficult. The outcome measurements of pain, subjective disability, and functional status are inconsistent among studies, complicating comparisons of reported predicted values of specific variables. Findings in numerous studies, summarized by Krause and Ragland, 16 Frank et al, 8 and others, have shown that provision of modified work is effective in facilitating return to work. Results in our studies 4,29,30 have also confirmed this. However, there is little known about how the provision of modified work interacts with other predictors of TL and how important this variable is compared with worker and injury characteristics in predicting 1930

Predictors of Time Loss After Back Injury Tate et al 1931 Table 1. Number of Reported Back Injuries and Back Injuries Incurring Time Loss by Ward and Extent of Program Participation Designation of Ward Where Injury Occurred Program Participation Total Injuries Injuries Incurring Time Loss Control ward Consented 158 33 Refused 12 2 Ineligible 115 29 Total 285 64 Study wards Consented 60 11 Refused 33 2 Ineligible 38 12 Total 131 25 Injuries on all wards 416 89 long-term disability. The Institute of Work and Health 8 has developed a broad-based framework acknowledging the multifaceted nature of the influences on return to work, in which factors have been clustered into four groups. 21 These include worker characteristics, including the worker s demographics and clinical manifestation of injury; rehabilitation interventions, including medical, physical, and vocational interventions; work and workplace factors, including the characteristics of the work itself and workplace organizational characteristics; and social policy and legislation, in which the structure of the compensation system and other incentives are taken into consideration. Despite the clarity in the framework presented, they concluded that further investigation is needed, particularly focusing on the interaction among the various factors, to increase understanding of how best to promote recovery of injured workers and optimize return-to-work programming. The objectives of the analysis to be presented were to determine the extent to which characteristics of nurses (known or obtained at the time of back injury), characteristics of the injury, and workplace factors are useful in predicting which nurses will incur TL and the duration of lost time from work during the 6 months after injury. Specifically, the purpose was to determine how participation in a workplace-based early intervention returnto-work program influences TL, when other predictors of TL are taken into consideration. Methods The Nurse Follow-up Study. A prospective inception cohort study was conducted during a 2-year period, beginning on October 1, 1990, at the Health Sciences Centre (HSC), a large acute and tertiary care teaching hospital in Winnipeg, Manitoba. A cohort comprising all nurses who sustained occupationally induced soft tissue back injuries during this period of observation was assembled. A 6-month follow-up period at the end of the observation period was necessary to document lost time for nurses injured near the end of the 2-year program. Hospital wards were designated as high risk (study) or lower risk (control) for back injury based on ergonomic assessments and review of previous injury data. 29 Study wards were targeted for an early intervention program that included provision of modified work. 4,30 All injured nurses were contacted as soon as possible after the reported injury. Nurses either consented to be interviewed (Consented), refused to be interviewed (Refused), or were ineligible to participate in the program because of planned departure from the HSC, pregnancy, or the presence of potentially confounding concurrent medical or chiropractic conditions (Ineligible). Further details concerning these three categories have been presented earlier. 29,30 Demographic data including age, gender, nursing experience, experience working on the ward where injured, history of prior back injury, and nature and cause of the injury were available for all injured nurses. However, only those nurses consenting to be interviewed provided information regarding self-rated pain and perceived disability at the time of injury. Pain was measured on a visual analog scale 10 cm long, ranging from 0 to 100 with 0 indicating no pain and 100 indicating extreme pain. Disability was measured using the Oswestry Low Back Pain Questionnaire. 6 This score ranges from 0, no perceived disability, to 100, complete disability. Every effort was made to interview nurses within 2 working days after injury. There were, however, some instances of delay in filing injury reports on the control wards. Although this would not affect the demographics of the injured nurse, it could have resulted in the reporting of lower pain and disability scores in these nurses, because the time delay might have rendered the injury less acute. Statistical Modeling of Time Loss. Demographic variables including age, nursing experience, ward experience, perceived pain, and perceived disability were compared between nurses on study wards and control wards in nurses who did and did not incur TL using two-way analysis of variance models. Categorical variables including mechanism of injury, full- or parttime employment, and sudden or gradual onset of injury were compared in those nurses with and without TL stratified by study and control ward using the Mantel Haenszel 2 statistic. A 5% level of significance was preset for each of these comparisons. Time loss attributable to the back injury during a 6-month interval after the injury was the dependent variable analyzed. Three statistical modeling techniques were used to examine occurrence of TL, duration of TL, and duration of TL once an injury incurring lost time had been documented. First, the logistic regression model 27 was used to relate characteristics at the time of injury to a binary indicator of any TL after the injury. The exponentiation of the coefficients of this model can be interpreted as the relative likelihood (relative odds) of TL associated with a unit change of each independent variable in the model. For binary variables indicating the presence or absence of a characteristic, the relative odds (with 95% confidence interval) of TL for those with or without the characteristic were calculated. Second, the duration of TL was modeled using the Tobit regression model. 26 This model accounts for censoring of TL at zero, appropriate in situations in which there is an imposed lower limit to the dependent variable. This model is appropriate for duration of TL when some injured nurses had no TL and others could have TL ranging from 1 day to 6 months. Third, ordinary least-squares linear regression 14 was used to examine the factors related to duration of TL in nurses who incurred at least one day of TL. Thus, this model included only the subset of nurses who incurred TL. All models were fitted using forward stepwise algorithms. A 10% level of significance for inclusion of variables in each model was used as

1932 Spine Volume 24 Number 18 1999 Table 2. Characteristics of Back Injured Nurses With and Without Time Loss Who Consented to Interview at Time of Injury by Ward Where Injury Occurred Characteristic Study Ward No Time Loss Time Loss Control Ward No Time Loss Time Loss No. of injuries 49 11 125 33 Age (yr) (mean SD) 31.4 9.2 34.4 9.6 33.8 7.5 33.4 10.1 Nursing experience 7.9 9.3 10.5 10.0 9.6 7.5 10.3 10.0 (mean SD) Ward experience 3.9 4.5 4.0 4.6 4.3 4.6 5.2 7.6 (mean SD) Employed full time (%) 89.8 54.5 65.6 57.6 Sudden onset of 87.8 81.8 81.6 75.8 injury (%) Mechanism: lifting (%) 34.7 9.0 32.0 21.2 Mechanism: patient 34.7 54.5 23.2 30.3 transfer (%) Perceived pain 21.8 20.8 57.8 26.5 16.0 19.1 31.6 27.0 (Mean SD)* Perceived disability (Mean SD)* 11.8 11.5 44.2 17.2 10.2 11.2 29.1 18.1 * The pain and disability scales range from 0 to 100, with zero being no pain or disability and 100 being extreme pain or complete disability. a guideline to allow the possibility of including marginally significant variables. Results As shown in Table 1, 416 back injuries were reported during the 2-year prospective period. Eighty-nine (89) of these injuries incurred TL from work because of back pain during the 6 months after injury, with TL ranging from 1 to 185 days. For 18 other back injuries, TL was documented in the 6 months after back injury, but this TL could not be attributed to the back injury in question, usually because of subsequent intervening injury. There were 44 TL injuries among the 218 nurses who were eligible and consented to be interviewed. Characteristics of the nurses, with or without TL, who were eligible and consented to be interviewed are presented in Table 2. Age and nursing and ward experience were not significantly different in study ward nurses and control ward nurses, nor were these characteristics different in those nurses incurring TL and those with no TL. Perceived pain and perceived disability were both greater in study nurses than in control ward nurses, perhaps because of the greater physical demands and therefore potentially greater severity of injury on these wards. Study ward nurses were more likely to be employed full time. An equal proportion of injuries occurring with sudden rather than gradual onset was found on both study and control wards. Because of patient care requirements, patient transfer was implicated as the mechanism of injury in more study ward injuries than control ward injuries. As shown in Figure 1, there was a very strong gradient between perceived disability at the time of injury and occurrence of TL. Sixteen (80%) of 20 nurses, who scored 40 or more on the disability scale incurred TL. Approximately 10% of nurses who indicated a pain score below 40 on the 0 100 range of the visual analogue scale in any of the 10 unit categories incurred TL, in contrast to 45% of back-injured nurses with pain scored above 40. Table 3 lists the variables that were statistically significant for the logistic regression model of occurrence or nonoccurrence of TL. A prior back injury during the pilot program carried with it an almost threefold increase in risk of another back injury during the 2-year follow-up period. Each point on the disability scale carried with it a 13% increase in risk of TL; when previous injury was used as a control, nurses who reported greater disability at the time of back injury for a given level of pain were more likely to incur TL. Figure 1. Percentage of nurses incurring time loss during the 6 months after back injury by perceived disability and pain at the time of injury.

Predictors of Time Loss After Back Injury Tate et al 1933 Table 3. Logistic Regression Modeling Time of Loss Versus No Time Loss due to Back Injury Variable Coefficient Standard Error P Value Odds Ratio 95% Confidence Interval Intercept 3.163 0.932 0.001 Previously injured during pilot program 1.086 0.565 0.054 2.962 (0.979,8.962) (yes 1, no 0) Study ward (yes 1, no 0) Disability score* (Oswestry) (per unit 0.131 0.024 0.001 1.140 (1.088,1.194) Pain score* (VAS) (per 10 unit 0.356 0.147 0.015 0.700 (0.525,0.934) * The pain and disability scales range from 0 to 100, with zero being no pain or disability and 100 being extreme pain or complete disability. not significant at P 0.10 in final model. Table 4 presents the results of modeling duration of TL using the Tobit regression model. Prior back injury increased the predicted duration of TL by 39 days in injured nurses, independent of a nurse s perceived disability at the time of injury, whereas each point on the Oswestry scale incurred an additional 3 days of TL. Participation in the early intervention program offered to study ward nurses decreased TL considerably. Although significant in univariate analysis, pain at time of injury did not predict duration of TL in this model beyond that explained by the other three significant variables. However, conditional on incurring at least 1 day of TL, the multiple linear regression model presented in Table 5 shows that although the early return-to-work program on study wards was still an important determinant of reduced TL, the pain score was independently more important for prediction of TL than was perceived disability, once TL had been established. Mechanism of injury was also significant in this model, with patient lifting being the activity at time of injury resulting in the greatest TL. The significant predictors of TL versus no TL, duration of TL, and duration of TL once TL occurs are summarized in Table 6. In all models, no statistically significant interaction between pain and disability was found; nor was any significant interaction found with either Table 4. Tobit Regression Modeling of Duration of Time Loss due to Back Injury Variable Coefficient Standard Error P Value Intercept 114.9 17.30 0.001 Previously injured 39.29 17.78 0.027 during pilot program (yes 1, no 0) Study ward (yes 44.75 17.39 0.010 1, no 0) Disability score* 3.19 0.45 0.001 (OSW) (per unit Pain score* (VAS) (per 10 unit * The pain and disability scales range from 0 to 100, with zero being no pain or disability and 100 being extreme pain or complete disability. not significant at P 0.10 in final model. pain or disability and prior back injury or program participation. Four factors (TL, pain, disability, and ward) are presented in Figure 2. Time loss for study and control ward injuries is plotted along the major axes of pain and disability. It is clear that most injured nurses without TL reported low levels of pain and disability. The greatest duration of TL was among control ward nurses with high levels of pain and high disability. However, there was a cluster of nurses whose pain and disability scores lay within the center of the field (with pain scores below 40 units and disability scores between 20 and 60 units) with substantial TL. Discussion Although previous studies have shown that prior back injury, lifting tasks, and high levels of disability are risk factors for subsequent back injury 10,18,19,22,23,31 very little is reported about factors predisposing to TL and duration of TL. The current study study is one of the few in which this correlation has been examined. Bigos et al 2 examined total incurred cost of back injury in 857 indus- Table 5. Linear Regression Modeling of Duration of Time Loss in Nurses Who Incurred Time Loss due to Back Injury Variable Coefficient Standard Error P Value Intercept 8.72 14.94 0.563 Previously injured during pilot program (yes 1/no 0) Study ward (yes 45.18 17.92 0.016 1/no 0) Disability score* (OSW) (per unit Pain score* (VAS) 13.42 2.77 0.001 (per 10 unit Injury occurred during 5.54 15.96 0.730 patient transfer (yes 1/no 0) Injury occurred during patient lifting (yes 1/no 0) 45.84 20.14 0.029 * The pain and disability scales range from 0 to 100, with zero being no pain or disability and 100 being extreme pain or complete disability. not significant at P 0.10 in final model.

1934 Spine Volume 24 Number 18 1999 Table 6. Significant Predictors in Back Injured Nurses of Occurrence of Time Loss, Duration of Time Loss, and Duration of Time Loss Among Nurses Who Incur Time Loss Predictors of a Back Injury Becoming a TL injury Predictors of Duration of TL Predictors of Duration of TL Once an Injury Incurs TL Perceived disability Perceived disability Perceived pain Perceived pain Perceived pain Mechanism of injury (lifting) Previous back injury Mechanism of injury (lifting) Participation in early RTW program Participation in early RTW program TL time loss; RTW retrun to work. trial workplace claimants; TL from work was a component of the calculation of total incurred cost. They found that improper lifting was associated with high-cost injuries; the current results support this finding. Lehmann et al 17 found that duration of pain severe enough to prevent a worker from returning to the workplace predicted high-cost injuries. This is consistent with the current finding that pain is predictive of duration of TL. In addition, there were injured nurses with moderate pain and disability who also incurred TL. It would be important to conduct long-term follow-up of these nurses to document their course of return to work and therefore to determine whether their prognosis is different from nurses with high pain and high disability scores. The current results showed that the offer of a modified work program, history of prior injury, and the extent of perceived pain and disability at the time of the injury were more important than the demographic characteristics of injured nurses for the prediction of TL. The study was unique in its ability to examine prospectively the effect of an early return-to-work program in context with these recognized risk factors for TL. Moreover, it was found that these significant factors had varying effects and therefore had to be considered differently in characterizing occurrence and duration of TL. At the time of injury, perception of disability in addition to pain level was a strong independent predictor of whether a TL injury would ensue. Further, TL was more likely to occur in an injured nurse who had a prior back injury. An injury occurring on a ward targeted for participation in the early return-to-work program (i.e., study ward) did not distinguish TL injuries from non-tl Figure 2. Duration of time loss in days during the 6 months after back injury by perceived disability and pain at the time of injury. Diamond, injury on study ward; square, injury on control ward.

Predictors of Time Loss After Back Injury Tate et al 1935 injuries. This indicates that the occurrence of TL may be a consequence of the characteristics of the injury itself, more so than proposed management of back injury. However, the offer of an early intervention program to study ward injured nurses was a powerful factor in reducing duration of TL. Also, mechanism of injury, specifically those injuries incurred during patient-handling tasks, caused a longer period of TL once TL had occurred. The interrelation between pain and disability, and the effect of early intervention on that relation in these nurses has been the subject of recent investigation. 3 In the current study, early intervention significantly tempered the extent to which pain and disability at the time of injury predicted future disability. The results of the analysis showed that although perceived disability was related to whether a TL injury would ensue, self-reported pain was strongly related to the duration of TL, once an injury had become a TL injury. That no statistically significant interaction was found between pain and disability or between either pain or disability and prior back injury or program participation is evidence that the effects of these variables are independent. The findings show that focusing on reducing the perception of disability at the time an injury occurs is critical to preventing TL, but that once TL has occurred, offer of modified work and attention to pain reduction are highly warranted. The findings add to the evidence that workplace-based intervention programs can be effective in reducing the morbidity resulting from back injury. References 1. Agnew J. Back pain in hospital workers. Occup Med 1987;23:609 16. 2. Bigos SJ, Spengler DM, Martin NA, et al. Back injuries in industry: A retrospective study. II: Injury factors. Spine 1986;11:246 51. 3. Cooper JE, Tate RB, Yassi A, Khokhar J. Effect of an early intervention program on the relationship between subjective pain and disability measures in nurses with low back injury. Spine 1996;21:2329 36. 4. Cooper JE, Tate R, Yassi A. Work hardening in an early return to work program for nurses with back injury. Work 1997;8:149 56. 5. Deyo RA, Diehl AK. Psychosocial predictors of disability in patients with low back pain. J Rheumatol 1988;15:1557 64. 6. Fairbank JCT, Couper J, Davies JB, O Brien JP. The Oswestry Low Back Pain Disability Questionnaire. Physiotherapy 1980;66:271 3. 7. Frank JW, Kerr MS, Brooker A S, et al. 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 1996;21:2908 17. 8. Frank JW, Brooker A S, DeMaio SE, et al. Disability resulting from occupational low back pain. Part II: What do we know about secondary prevention? A review of the scientific evidence on prevention after disability begins. Spine 1996; 21:2918 29. 9. Frymoyer JW, Cats Baril W. Predictors of low back pain disability. Clin Orthop 1987;221:89 98. 10. Fuortes LJ, Shi Y, Zhang M, Zwerling C, Schootman M. Epidemiology of back injury in university hospital nurses from review of workers compensation records and a case control study. J Occup Med 1994;36:1022 6. 11. Galka ML. Back injury prevention program on a spinal cord injury unit. Spinal Cord Injury Nursing 1991;8:48 51. 12. Greenough CG, Fraser RD. Assessment of outcome in patients with lowback pain. Spine 1992;17:36 41. 13. Harber P, Billet E, Vojtecky M, Rosenthal E, Shimozaki S, Horan M. Nurses beliefs about cause and prevention of occupational back pain. J Occup Med 1988;30:797 800. 14. Hassard TH. Understanding Biostatistics. St. Louis: Mosby Yearbook, 1991. 15. Hazard RG, Haugh LD, Reid S, Preble JB, MacDonald L. Early prediction of chronic disability after occupational low back injury. Spine 1996;21:945 51. 16. Krause N, Ragland DR. Occupational disability due to low back pain: A new interdisciplinary classification based on a phase model of disability. Spine 1994; 19:1011 20. 17. Lehmann TR, Spratt KF, Lehmann KK. Predicting long-term disability in low back injured workers presenting to a spine consultant. Spine 1993;18:1103 12. 18. Nuwayhid IA, Stewart W, Johnson JV. Work activities and the onset of first-time low back pain among New York City fire fighters. Am J Epidemiol 1993;137:539 48. 19. Papageorgiou AC, Croft PC, Thomas E, Ferry S, Jayson M, Silman AJ. Influence of previous pain experience on the episode incidence of low back pain: Results from the South Manchester Pain Study. Pain 1996;66:181 5. 20. Sinclair SJ, Hogg Johnson S, Mondloch MV, Shields SA. The effectiveness of an early active intervention program for workers with soft-tissue injuries: The early claimant cohort study. Spine 1997;22:2919 31. 21. Sinclair SJ, Sullivan TJ, Clarke JA, Frank JW. Framework for examining return-to-work in workers compensation: A view from one North American jurisdiction. Toronto: Institute for Work and Health, 1997. 22. Smedley J, Egger P, Cooper C, Coggon P. Manual lifting activities and risk of low back pain in nurses. Occup Environ Med 1995;52:160 5. 23. Smedley J, Egger P, Cooper C, Coggon P. Prospective cohort study of predictors of incident low back pain in nurses. BMJ 1997;314:1225 8. 24. Snook SH, Webster BS. The cost of disability. Clin Orthop 1987;221:77 84. 25. Spitzer WO, LeBlanc FE, Dupuis M, et al. Scientific approach to the assessment and management of activity-related spinal disorders. A monograph for clinicians. Report of the Quebec Task Force on Spinal Disorders. Spine 1987; 12(Suppl 7):1 59. 26. Tobin J. Estimation of relationships for limited dependent variables. Econometrica 1958;26:24 36. 27. Walker SH, Duncan DB. Estimation of the probability of an event as a function of several independent variables. Biometrika 1967;54:167 79. 28. Webster BS, Snook SH. The cost of 1989 workers compensation low back pain claims. Spine 1994;19:1111 16. 29. Yassi A, Khokhar J, Tate R, et al. The epidemiology of back injury in nurses at a large Canadian tertiary care hospital: Implications for prevention. Occup Med 1995;45:215 20. 30. Yassi A, Tate R, Cooper J, et al. Early intervention for back-injured nurses at a large Canadian tertiary care hospital: An evaluation of the effectiveness and cost benefit of a 2-year pilot project. Occup Med 1995;45:209 14. 31. Zwerling C, Ryan J, Schootman M. A case-control study of risk factors for industrial low back injury. Spine 1993;18:1242 47. Address reprint requests to Dr. Annalee Yassi Director, Occupation and Environmental Unit Department of Community Health Sciences S112 750 Bannatyne Avenue Winnipeg, MB Canada E-mail yassi@cc.umanitoba.ca

1936 Spine Volume 24 Number 18 1999 Point of View James Ryan, MD, MPH, FACOEM Harvard School of Public Health Boston, Massachusetts Dr. Yassi and her associates have identified a number of useful predictors of lost work time due to low back injury among nurses. Some of these may provide support for intervention programs directed at the risk factors themselves. Others may offer insights to guide further research into the control of morbidity from occupational back injuries. Consistent with previously reported studies, they found that one frequent nursing activity, the manual transfer of patients, was disproportionately associated with back injuries resulting in time lost from work. This finding supports the development of improved ergonomic methods to reduce the frequency and severity of lifting injuries. What combination of mechanical hoists, 2 dedicated lifting teams, 3 or use of improved lifting techniques 4 will prove most effective remains to be demonstrated. The findings, measured at the time of injury, that selfperceived level of disability is an independent predictor that the injury will result in lost time and that selfreported level of pain is an independent predictor of duration of disability, may be difficult to apply to lost work time reduction. While original and seemingly intuitive, the underlying nature of these associations deserves further study. If such self-assessments primarily reflect the severity of the actual injury, (as the higher levels of both in the high risk group might indicate), then addressing pain and perceived disability directly might contribute little to work retention or return. This cohort also experienced a markedly higher risk (Odds Ratio 2.96) for sustaining lost time in the event of a second back injury among nurses who had reported an earlier back injury during the study period. Other studies have found similar elevated risks for back injury in postal workers with any history of prior low back injury (Odds Ratio 2.9) 5 and among those who had filed any workers compensation claim within three years (Rate Ratio 3.4). 6 The elevated reinjury rate found in this study might justify the exercise of caution in the selection of assignments on return to work. Such an approach, however, might be limited by the requirements of the Americans with Disabilities Act and the Family and Medical Leave Act in the U.S. The four papers based on this early intervention program have emphasized its secondary prevention aspects, i.e., decreased lost work time for injured nurses on the study wards. However, the research protocol also included a primary prevention module for the study wards consisting of information about body mechanics, lifts and transfers, and healthy lifestyles. 7 The authors appear to downplay the significance of the primary prevention module describing it as poorly attended and involving less than one-half of the injured nurses. 8 However, the study wards experienced a 23% decrease in the rate of back injuries during a period when the rate on control wards increased by 42%. 9 It would be useful if the authors gave further consideration to the contributions of their primary prevention activity, especially as this low-cost module may have also contributed to decreased lost work time by reducing the severity of injuries. The demonstration of a cost-effective primary prevention program would constitute a major contribution to the body of medical knowledge as other well designed efforts at primary prevention have not always proven effective. 6 References 1. Tate RB, Yassi A, Cooper J. Predictors of Time Loss After Back Injury in Nurses. Spine 1999;24:1930 1936. 2. Garg A, Owen B, Beller D, Banaag J. A biomechanical and ergonomic evaluation of patient transferring tasks: Bed to wheelchair and wheelchair to bed. Ergonomics 1991;34:289 312. 3. Charney W, Zimmerman K, Walara E. The lifting team: A design method to reduce lost time back injury in nursing. AAOHN J 1991;39:231 4. 4. Winkelmolen GH, Landeweed JA, Drost MR. An evaluation of patient lifting techniques. Ergonomics 1994;37:921 32. 5. Zwerling C, Ryan J, Schootman M. A case-control study of risk factors for industrial low back injury: The utility of preplacement screening in defining high-risk groups. Spine 1993;18:1242 7. 6. Daltroy LH, Iverson MD, Larson MG, et. al. A controlled trial of an educational program to prevent low back injuries. New Engl J Med 1997;337:322 8. 7. Cooper JE, Tate RB, Yassi A, Khokhar J. Effect of an early intervention program on the relationship between subjective pain and disability measures in nurses with low back injury. Spine 1996;21:2329 43. 8. Yassi A, Khokhar J, Tate R, Cooper J, Snow C, Vallentyne S. The epidemiology of back injuries in nurses at a large Canadian tertiary care hospital: Implications for prevention. Occup Med 1995;45:215 20. 9. Yassi A, Khokhar J, Tate R, Cooper J, Snow C, Vallentyne S. Early intervention for back-injured nurses at a large Canadian tertiary care hospital: an evaluation of the effectiveness and cost benefits of a two-year pilot project. Occup Med 1995;45:209 14.