Papers

Prospective cohort study of predictors of incident low back pain in nurses

BMJ 1997; 314 doi: https://doi.org/10.1136/bmj.314.7089.1225 (Published 26 April 1997) Cite this as: BMJ 1997;314:1225
  1. Julia Smedley, consultant occupational physiciana,
  2. Peter Egger, medical statisticiana,
  3. Cyrus Cooper, reader in rheumatologya,
  4. David Coggon, reader in occupational and environmental medicinea
  1. a MRC Environmental Epidemiology Unit, University of Southampton, Southampton General Hospital, Southampton S016 6YD
  1. Correspondence to: Dr Coggon
  • Accepted 24 January 1997

Abstract

Objective: To assess the impact of handling patients and indicators of individual susceptibility on risk of low back pain in nurses.

Design: Prospective cohort study with follow up by repeated self administered questionnaires every three months over two years.

Setting: NHS university hospitals trust.

Subjects: 961 female nurses who had been free from low back pain for at least one month at the time of completing a baseline questionnaire.

Main outcome measures: Incidence of new low back pain during follow up and of pain leading to absence from work.

Results: Of 838 women who provided data suitable for analysis, 322 (38%) developed low back pain during follow up (mean 18.6 months), including 93 (11%) whose pain led to absence from work. The strongest predictor of new low back pain was earlier history of the symptom, and risk was particularly high if previous pain had lasted for over a month in total and had occurred within the 12 months before entry to the study (incidence during follow up 66%). Frequent low mood at baseline was strongly associated with subsequent absence from work for back pain (odds ratio 3.4; 95% confidence interval 1.4 to 8.2). After adjustment for earlier history of back pain and other potential confounders, risk was higher in nurses who reported frequent manual transfer of patients between bed and chair, manual repositioning of patients on the bed, and lifting patients in or out of the bath with a hoist.

Conclusions: Of the indicators of individual susceptibility that were examined, only history of back trouble was sufficiently predictive to justify selective exclusion of some applicants for nursing posts. The main route to prevention of back disorders among nurses is likely to lie in improved ergonomics.

Key messages

  • A history of back trouble, particularly if recent and prolonged, is highly predictive of new episodes of back pain

  • There are grounds for excluding nurses with recent and prolonged back pain from the most physically demanding jobs

  • Age, height, and weight are not sufficiently discriminatory for risk of back pain to influence selection and appointment of nurses

  • Back pain is more common in nurses who lift and move patients frequently without the use of mechanical aids

  • Controlled trials are needed to assess the benefits of ergonomic intervention aimed at prevention of back pain in nurses

Introduction

Low back pain is common in the general population, affecting more than 60% of people at some time in their lives and often causing appreciable disability.1 2 It is particularly common in nurses. In a recent survey 10% of 1616 female nurses employed by a large NHS trust reported having lost more than a month in total from work because of back problems.3 This high incidence is not only a burden on the many nurses who develop back pain but also a substantial cost to employing hospitals in lost efficiency, lost time, wasted training, and claims for industrial injuries.

The high rate of back disorders in nurses is associated with heavy physical workload,4 particularly in lifting and moving patients,5 6 7 and with adverse postures.8 9 One approach to prevention, therefore, is through improvements in ergonomics and training, with avoidance or modification of the tasks that carry the highest risks. In addition, exclusion of people who are specially vulnerable to back injury from the most hazardous jobs may be justified. To optimise preventive strategies, however, more information is needed about the levels of risk associated with specific nursing activities and about the influence of individual susceptibility on risk. To examine these questions we carried out a two year longitudinal study of back pain in a cohort of nurses working in hospitals.

Subjects and methods

In 1993 we sent a baseline questionnaire to the 2405 hospital based nurses employed by Southampton University Hospitals Trust. The trust provides inpatient facilities in most clinical specialties other than psychiatry. The questionnaire asked about various non-occupational risk factors for back pain, including age, height, and weight; about activities in the nurse's current job; and about past and recent low back pain and other symptoms. Throughout the study low back pain was defined as pain in an area (illustrated in a diagram) between the 12th ribs and the gluteal folds that lasted for longer than a day and occurred other than in association with pregnancy, menstruation, or febrile illness. The results of a cross sectional analysis of this initial survey have already been published.3

Of 1616 women who completed the baseline questionnaire, 1336 (83%) agreed to take part in the longitudinal phase of the study. This paper presents results for the subset of 961 women who had not had any low back pain in the month before they completed the baseline questionnaire. Their ages ranged from 19 to 64 years with a mean of 38 years. Three hundred and forty seven were auxiliary or enrolled nurses, 573 were staff nurses or sisters, and 41 worked in administrative or specialist posts.

Each woman was sent a short follow up questionnaire every three months for two years. This asked about any change in manual handling activities and about the occurrence of any low back pain and resultant loss of time from work since the last contact. Women who did not respond to a questionnaire were sent a single reminder and were also mailed again at the next three monthly follow up. Those who failed to respond to two successive three monthly follow ups were regarded as having dropped out. At the end of the two year study period we checked the personnel records of those who dropped out and to those whose addresses were known we sent a further questionnaire asking about low back pain since the last contact and whether this had led to time off work.

In our analysis we used a discrete-time logistic-normal survival model to explore risk factors for the incidence of new low back pain during follow up.10 11 This was chosen in preference to a proportional hazards model because symptoms were timed by calendar month, leading to multiple ties in the recorded onset of back pain. We examined the risk of all low back pain and of back pain leading to loss of time from work. The risk factors studied included constitutional and non-occupational attributes measured at baseline and occupational activities at the time back pain began.

Results

Table 1) summarises the response to follow up. Altogether, 843 women (88%) returned at least one follow up questionnaire, and 446 (46%) were still under follow up after 24 months. Addresses were available for 291 of the 397 women who dropped out, and 176 answered the final questionnaire at the end of the study period. Of these, 42 reported low back pain since dropping out, and 10 had had to take time off work because of back pain. Their two year cumulative incidence of symptoms (47%) was similar to that of those who remained under follow up throughout the study (46%). Table 1) also shows the numbers of questionnaires that women returned while under follow up. Further analysis is based on the 838 women who completed at least one follow up questionnaire and who provided usable information about back pain. Of these, 322 (38%) developed low back pain while under follow up (33.1 new episodes per 100 woman years), including 93 (11%) whose pain was bad enough to require time off work. Table 2) shows the risk of incident low back pain according to age, height, and weight. Symptoms were significantly more common in the tallest women, but this excess did not lead to much more absence from work. No clear trends were apparent in relation to age or weight. Neither was there an association with body mass index (data not shown).

Table 1

Completeness of follow up

View this table:
Table 2

Risk of low back pain during follow up according to age, height, and weight. For each outcome all risk estimates are mutually adjusted

View this table:

All of the women included in the analysis had been free from back pain for at least one month at the time they completed the baseline questionnaire, but almost half had suffered from symptoms earlier. Table 3) summarises the risk of low back pain during follow up according to the duration of previous pain and the time since last symptoms. Risk tended to increase with the total duration of previous pain and was highest in those who had experienced symptoms during the year before entry to the study. Of 92 women who reported pain in the 12 months before answering baseline questionnaire and a history of pain for at least one month in total, 61 (66%) developed further symptoms during follow up (91.0 new episodes per 100 woman years) and 21 (23%) required time off work as a consequence.

Table 3

Risk of low back pain during follow up according to earlier history of low back pain. All risk estimates are adjusted for age and height (classified as in table 2)

View this table:

Table 4) shows the association of incident low back pain with other complaints reported at baseline. After adjustment for age, height, and earlier history of back pain women with frequent low mood at baseline were significantly more likely to require time off work for back pain during follow up (odds ratio 3.4; 95% confidence interval 1.4 to 8.2).

Table 4

Risk of low back pain during follow up according to symptoms other than back pain at baseline. For each outcome all risk estimates are mutually adjusted and adjusted also for age, height, and earlier history of low back pain (classified as in tables 2 and 3)

View this table:

When we looked at associations of back pain with occupational activities we adjusted for age, height, earlier history of back pain, and reports of other symptoms at baseline. Table 5) presents risk estimates for eight tasks commonly carried out by nurses that entail handling patients. With all low back pain as the outcome, exposure-response trends were observed for manual transfer of patients between bed and chair; transfer of patients between bed and chair with a hoist; manually moving patients around—that is, repositioning them—on the bed; and lifting patients in or out of the bath with a hoist. Associations with back pain leading to absence from work were less clear. Exposures to the various tasks tended to correlate with each other, and when we analysed all of the activities in a single statistical model risk estimates were generally reduced and less significant. The association with frequent transfers between bed and chair with a hoist completely disappeared. Otherwise, however, the pattern was similar.

Table 5

Risk of low back pain during follow up according to occupational activities at time of onset. Risks are estimated separately for each outcome with adjustment for age, height, earlier history of low back pain, and symptoms other than back pain at baseline (classified as in tables 2, 3, and 4)

View this table:

Discussion

Most previous studies of low back pain have been retrospective or cross sectional. By using a longitudinal design we avoided having to rely on women's distant memory for the ascertainment of symptoms. Moreover, because risk factors were assessed before the onset of symptoms there was less opportunity for bias. In particular, any errors in the reporting of nursing activities would be expected to obscure rather than exaggerate associations with back pain. The incompleteness of follow up was a potential weakness, but we were able to contact almost half of those who dropped out at the end of the study, and their reported incidence of back pain was similar to that of women who remained under follow up.

Handling patients

Our results confirm the high incidence of back disorders among nurses and support a relation with tasks entailing the handling of patients. The need for nurses to carry out lifting and other manual handling tasks is determined by their patients' mobility, and, not surprisingly, many of the handling activities were intercorrelated. This made it more difficult to distinguish their individual contributions to risk, but the most hazardous activities seemed to be manual transfer of patients between bed and chair, manual repositioning of patients on the bed, and lifting patients in and out of the bath with a hoist. These findings accord with those of our earlier cross sectional survey in the same population.3 Manual transfers and repositioning would be expected to stress the spine, but the association with lifting by hoist is harder to explain. It may reflect confounding by other tasks that are associated with bathing patients.

Other risk factors

Of the other risk factors examined, three—height, earlier history of back symptoms, and low mood—were significant predictors of incident back pain. The association with height was relatively weak and apparent only in the tallest women. This pattern has been observed previously in one study,23 although not in others.16 24 25

Earlier history of back trouble was by far the strongest predictor of new symptoms. This is not surprising given the chronicity and recurrent nature of back disorders, but our data illustrate clearly how risk increases with both the duration and recency of previous symptoms. Other studies have also found that risk of back pain is increased in people with previous back trouble,12 27 28 29 although the finding has not been universal.21 Few have examined the influence of earlier history in more detail, but in Denmark Biering-Sørensen found that the probability of developing further pain within the next year fell from 76% in people with back pain in the past week to 28% in those who last had back pain more than five years earlier.12

Several previous studies have linked back complaints with low mood, stress, and job dissatisfaction.9 13 30 31 32 33 34 Most, however, have been cross sectional, and it is unclear to what extent the psychological complaints were secondary to the back problem rather than antecedent. Our analysis, which was restricted to women who were free from pain at baseline and which adjusted for earlier history of back complaints, indicates that low mood does predict future back problems. It is notable that the association was particularly with back pain leading to loss of time from work. This might reflect an influence particularly on more severe disease or an effect on women's ability to cope when symptoms occurred.

Implications for prevention

Our findings have important implications for the prevention of occupational back pain, especially in nurses. One approach to prevention is through screening before employment and selective recruitment of staff who are at lower risk. Some hospitals have rejected applicants for nursing posts because they were obese, but we found no increase in risk of back disorders with weight or body mass index. Neither were the risks in taller women sufficient to warrant selective exclusion from employment. There may be justification for excluding women with a history of prolonged and recent back pain from the most physically demanding nursing jobs, but this would eliminate only a small proportion of cases.

Thus, the main route to preventing back disorders among nurses is likely to lie in improved ergonomics. In the past three years many NHS trusts have invested substantially in aids for handling patients, such as sliding sheets and hoists, but the outcome has yet to be properly assessed. Our findings point to nursing tasks that might most usefully be eliminated or modified. There is now an urgent need to evaluate such ergonomic interventions in a controlled trial.

Acknowledgments

We thank Graham Wield, who carried out the computing; the management of Southampton University Hospitals trust; and the nurses who participated in the investigation.

Funding: This study was funded in part by a grant from Wessex Regional Health Authority.

Conflict of interest: None.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
View Abstract