Workplace factors and care seeking for low-back pain among female nursing personnel

Workplace factors and care seeking for low-back pain among female nursing Objectives Low-back pain is common among nursing personnel, and its origin is multifactorial. The present study focused on physical and psychosocial work load. The objectives of the study were to estimate the relative risk for ~lursing personnel, cornpared with other occupational groups, to seek health care for low-back pain, and to identify risk factors. Methods This study is a part of a population-based case-referent study in thc municipality of Norrtalje, situated north of Stockholm. Altogether 333 women served as cases and 733 served as referents. Eighty-one cases and 188 referents were employed ill nursing work. The cases had sought health care for low-back pain during the study period, November 1993 to November 1996. The referents were randomly selected from the same population. The subjects filled out 2 questionnaires and participated in interviews about physical exposures and psychosocial factors. Results When the female nursing personnel were cornpared with other employed women, no increased risk of consultation for low-back pain was found. According to a multivariate logistic regression, nursing perso~lnel exposed to forward-bending working positioas, high energetic work load, perceived physical exertion, or insufficient social support had the highest risk estimates. In univariate analyses, the combination of physical and psychosocial risk factors was associated with a particularly high risk. C O ~ C ~ U S ~ O ~ S In nursing work, physical load seems to be more significant than psychosocial factors when a worker seeks health care for low-back pain. The results of did not support the hypothesis that nursing work is a risk occupation for seeking care for low-back pain when compared with other occupations. Low-back pain is common among nursing personnel. In a previous study, 53% of the nurses in a regional hospital reported that they had experienced low-back pain during the previous 12 months, and 16% reported current intense symptoms in the low-back region (1). a high level of physical exertion at work (27). In another study, including 4 annual cross-sectional surveys in a hospital setting, a higher prevalence of symptoms from the neck, shoulders, and back was observed among the persons who reported high physical exertion. In the I-, 2-, and 3-year follow-up analyses the crude point estimate of the risk ratio for having current symptoms was between 1.3 and 1.6 for the high-exertion group (1). high physical exertion in nursing work lnay partly direct


Results
When the female nursing personnel were cornpared with other employed women, no increased risk of consultation for low-back pain was found. According to a multivariate logistic regression, nursing perso~lnel exposed to forward-bending working positioas, high energetic work load, perceived physical exertion, or insufficient social support had the highest risk estimates. In univariate analyses, the combination of physical and psychosocial risk factors was associated with a particularly high risk.
C O~C~U S~O~S In nursing work, physical load seems to be more significant than psychosocial factors when a worker seeks health care for low-back pain. The results of did not support the hypothesis that nursing work is a risk occupation for seeking care for low-back pain when compared with other occupations.
Key terms case-referent study, epidemiology, health care, physical load, psychosocial factors, women.
Low-back pain is common among nursing personnel. In a previous study, 53% of the nurses in a regional hospital reported that they had experienced low-back pain during the previous 12 months, and 16% reported current intense symptoms in the low-back region (1).
The origin of low-back pain is multifactorial, and the present study focused on workplace factors related to physical and psychosocial load (2,3). In studies on psychosocial dimensions, physical load is often assessed at the group level (4). In studies on physical factors, the psychosocial conditions have often been ignored or crudely assessed (5,6). When exposures are assessed with different precision, it is difficult to estimate the relative effect of the physical and psychosocial load.
In our study, we assessed physical and psychosocial factors at the individual level. Both interview data with a descriptive focus on the work environment and questionnaire data focused on perceived work load were considered. The objectives of the study were to estimate the relative risk of nursing personnel seeking health care because of low-back pain when compared with other occupatifinal groups and to identify risk factors for health care consultations for low-back pain among nursing personnel.

Subjects and methods
This research was a part of the Norrtalje study, a population-based, case-referent iilvestigation of low-back disorders in the municipality of Norrtalje, situated north of Stockholin. The study base comprised about 17 000 persons between 20 and 59 years of age, who lived and wol-lted within the municipality. The data collection took place from Novernber 1993 to November 1996. The cases had sought health care from any of the about 70 care givers in the region during the study period. All care givers in the region, such as physicians, physiotherapists, chiropractors, osteopaths, homeopaths and others, took part in the study, and none refused to participate. The referents were selected from the population register as a stratified random sample, taking age (5-year intervals) and gender into account. All the care givers reported patients with low-back problen~s to the project group. The case patients were contacted by telephone as soon as possible, and a time for the examination was reserved, usually within 1 or 2 weeks and no longer than 2 tnoilths after the first contact. The referents were contacted by letter. An inclusion criterion was that the cases and referents should not have sought care for low-back or neck and shoulder pain during the 6 months prior to the study occasion. Altogether 694 cases and 1423 referents, who were gainfully employed at least 18 hours per week for at least 2 months during the previous 12 months, were investigated.
The participation rate for all the formal cases in the study is unknown, but, according to interviews with the care givers, only a few refused to participate. Among the female referents, 31% could not or did not want to participate; the corresponding frequency was 32% for the male referents.
The classification as to occupational sectors was based on the Nordic Occupation Classification (7) at the 3-digit level. Nursing occupations included registered nurses, assistant nurses, attendants in psychiatric care, home-care workers, and assistants for the mentally retarded. The registered nurses, the assistant nurses, and the attendants in psychiatric care worked rnaiilly in a hospital setting, in medical, surgical, psychiatric or geriatric wards. The home-care workers were employed by the local municipal authority and worked with elderly people in their private homes. Assistants for the mentally retarded were also employed by the local cnuilicipal authority and took care of the mentally retarded in nursing homes. Since there were only 7 male referents and 12 male cases working in nursing occupations, they were excluded from further analysis. In the present study, when self-employed women with occupations requiring at least 6 years of education after compulsory school were excluded, there were 333 womeil who formed cases and 733 women who acted as referents. Eighty-one of the cases and 188 referellts were gainfully employed in nursing worlt. Of the cases, 1 womail became a case twice according to the i~lclusion criteria, and 4 referents became cases during the study period.
The subjects took part in a clinical examination, an interview by a physiotherapist about physical exposures, and an interview by a behavioral scientist about psychosocial factors. Furthermore, the participailts filled out q~~estiollilaires about living conditions, occupational exposure, and general health. The physiotherapist and the behavioral scientist did not know the case-referent status of the subject during the interviews. The assessed exposures and the cut-off points between the exposed and unexposed subjects were determined on the basis of experience, the results of previous studies, knowledge of risk factors and work conditions, and the exposure distribution in the reference group.

Physical factors
As an overall proxy for the energetic work load, an estimate of the energy expenditure was used. For each task the energy expenditure, expressed as multiples of the resting metabolic rate (MET), was esti~ilated by the physiotherapist. A coding scheme, modified from 2 review articles (8,9), was used for classifying physical activities by rate of energy expenditure. A time-weighted MET (TWA MET) was calculated for a typical worliday. A TWA MET of 2.9 or more was the cut-off point for a relatively high energetic load. An energy expenditure of 2.9 MET for an average middle-aged woman represented around 30% of her maximal aerobic power (10). An upper general limit of around 30-35% of the maximal aerobic power has been recommended for an 8-hour workday for mixed physical work (1 1).
A rating of perceived exertion (RPE) was used as an item assessing perceived physical exertion during a normal workday in the questionnaire (12). The RPE scale was set as 6 (resting) to 20 (maximal exertion). Nursing personnel who reported physical exertion higher than "somewhat hard" (score 214), approximately the highest textile among the female referents in the Norrtalje study, were categorized as exposed.
Manual lifting of 10 kg during at least 5 minutes of a typical workday and lifting of at least 30 kg once a week were considered to be exposure criteria. In addition working in forward-bent positions with the hands below knee level for 1 hour during a typical workday was considered to be high-level exposure; work in the same positions for 1--59 mi~lutes was regarded as medium exposure.

Psychosocial factors
The occupational tasks, the duration of each task, and the mental requirements of the work were assessed from the description in the interview. 'The interviewer calculated the time spent ~I I routine work, the active application of knowledge, and problem solvii~g for a typical workday (13). The underlying hypolhesis was that worli requiring little thinliing and planning bas negative consequences on health in the long-term (14). "Mainly routine worli" was defined as 50% of the workhours spent in routine work and no tasks requiring problem solving (13).
According to the job-strain model, the combination of high psycl~ological demands, low intellectual discretion, and low authority over decision malting was considered to be a job-strain situation and a potential risk factor (15). The index for psycl~ological demands included the following 5 items: excessive work, conflicting demands, time to do work, fast work, and hard work. The score variatioil was 5 to 20: the higher the score, the higher the demands. Iritellectual discretion included the following 4 items: learning new things, high levels of skill, high levels of creativity, and repetitious job. The score variation was from 4 to 16: the lower the score, the less the skill discretion. The index of authority over decisions included the 2 items of questions about influence over what to do and how to perform the work. The score variation was 2-8: the lower the score, the less the authority over decisions. The indices were divided into tertiles according to the distribution among the male and female referents in the Norrtalje study. A score of 14---20 for psychological demands, a score of 4-10 for intellectual discretion, and a score of 2-5 for authority over decisions were considered potential risk factors. A sum score of 6-16 for low intellectual discretion and authority over decisions (approximately the lowest tertile) combined with high psychological demands (approximately the highest tertile) was categorized as a job-strain situation. The possibilities for social support and satisfying social relations at work were assessed both in the interview (2 questions) and by questionnaire (6 items). The possible score variation was fro111 8 to 32: the higher the score, the more satisfied with the social support. A score of 8 to 22 was considered a potential risli factor. The i~lterilal consistency of the psychosocial indices was considered acceptable, the Cronbach alpha values were between 0.72 to 0.87 when calculated for the reference nursing personnel (16,17).
Terms of employment, living conditions, family situation, and the amount of domestic work were also considered both in the iilterview and in the questionnaire.

Data analysis
When nursing person~lel wit11 occupatioris requiring the same level of education were compared, the categorization was based on the Swedish socioeconomic index classification (7). Risk analyses, stratified into 2 age groups by the Mantel-Haenszel method, and a logistic regression analysis were conducted. In the logistic regression factors with an odds ratio (OR) of 1.5 or greater in the univariate analysis were included, along with age and smoliing. The stability of the logistic regression analysis was tested by the Hosmer & Lemeshow test (18). The highest correlated exposure variables included in the multivariate analysis were workhourslweeli and nightshift work (Spearman rank correlation coefficient of -0.37). In the multivariate analysis all the variables were dichotomized excepl age, which was treated as a continuous variable. In the Norrtalje study the data were frequency-matched for age to increase the efficiency of the study. In the analyses, age was considered to be a potential confounder or effect modifier (19).

Results
The majority of the nursing persoinlel was always or almost always satisfied with the support from their workmates and supervisors, and only 6 reported mainly soutine work and no problen~ solving. The median number of workhours per week was 34 for the refereilts in nursing work and 30 for the cases in nursing work. Part-time (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29) hourslweek) worli was associated with night shifts. Seventy-seven percent of the referents and 93% of the cases on night shift were employed part-time.
A comparison of nursing personnel with the other occupational groups did not show any increased risk of consultatioil for low-back pain (table 1). On the contrary, the risk was lower for the registered nurses than for the other employed women in occupations requiring the same level of education (table 1 ).
High energetic work load, work in forward-bent positions, low intellectual discretion, low authority over decisions, part-time work, temporary employment, and night shifts were less freque~lt among the registered nurses than among the other nursing groups (table 2).  Table 2. Distribution of exposure factors among the referents. risk estimate of 6.2 (95% CI 1.7-22). However, no mul-The differences between the registered nurses and the other nursing personnel (assistant nurses, attendants in psychiatric care, home-care workers and assistants for the mentally retarded) in proportion exposed (proportion of exposed registered nurses minus proportion of other exposed nursing personnel) and 95% confidence intervals (95% CI) of the differences.

Discussion
In this study, nursing work, especially work as a registered nurse, seemed to decrease the relative risk of consultation for low-back pain. An increased prevalence of low-back pain has been reported for nursing personnel in a comparison with the general population (20). Offi-(weekly) 27 18 9 cial Swedish statistics for 1995-1996 show an increased However, high psychological demands were Inore frequent among the registered nurses (table 2). The assistant nurses, attendants in psychiatric care, home-care workers, and assistants for the mentally retarded were compared with the registered nurses, and the risk estimate for seeking care for low-back disorders was 7.2 (95% CI 2.5 -21). The highest risk estimate for the nursing personnel was found for those highly exposed to work in forwardbent positions (table 3). Except for low intellectual discretion, the increased relative risk estimates for different exposure factors in the univariate analyses were confirmed in the multivariate ailalysis (table 3).
Living conditions, current smoking, and age were not associated with an increased estimated relative risk (table 3). Of the psychosocial factors, insufficient social support was associated with the highest relative risks (table 3). In the univariate analyses, combined exposure to forward-bent work positions and low intellectual discretion gave an increased risk estimate of 9.0 (95% CI 2.5-33). Nursing personnel with insufficient social support in combination with forward-bent work positions had a ulation (21). Presumably, the relatively low-risk level for registered nurses can partly be explained by a low physical work load. According to previous studies, assistant nurses have a higher physical work load and also a greater risk of low-back pain when compared with registered nurses (22,23).
Work in forward-bent positions and with a high energetic load, but without lifting, was associated with an increased relative risk in our study. Several studies on nursing personnel have shown an association between the amount of patient-handling and the risk of low-back pain (24). Patient-handling involves work in forward-bent positions, lifting, and also manual handling such as pushing and pulling. In the present analyses pushing and pulling were not considered, and it could be that energetic load and the time spent in bent positions were better estimates than lifting for the time in patient-handling.
If the observed association between work in bent positions and consultation for low-back pain is causal, an attributable fraction can be calculated [(RR-IIRR) x proportion of cases exposed] (19). The attributable fraction can be explained as the proportion of cases whose disorder was caused by the examined exposure. If work in forward-bent positions is not considered part of the work load, 73% of the cases of consultation for low-back pain could have been prevented in our study.
A high RPE was associated with an increased relative risk in our study. In previous studies, ratings on the RPE scale have been useful for listing the patient-handling tasks perceived as most stressful for the low back (25). In our study the RPE scale was used as an item in a questionnaire, and the instructions to the subjects were to report the perceived physical exertion in their daily work. In accordance with previous studies, the variation in the RPE not only seemed to mirror the physical activity, but maybe also the worker's physical capacity and the psychological load (26,27). If the RPE scale is used Table 3. Number of exposed subjects among the referents and cases. The estimated odds ratios (OR) with 95% confidence intervals (95% CI), adjusted for age by the Mantel-Haenszel method, and the estimated OR in a multiple invariate logistic regression analysis. In the logistic regression, besides age and smoking, factors with a crude risk estimate of 21 as an item in a questionnaire to identify those with a high physical load, the extent of independent lnisclassification may be substantial. In a previous study among nursing aides in geriatric care, the cross-sectional analysis and the 6-month followup demonstrated a weak tendency towards a higher prevalence of low-back pain among the persons who reported a high level of physical exertion at work (27). In another study, including 4 annual cross-sectional surveys in a hospital setting, a higher prevalence of symptoms from the neck, shoulders, and back was observed among the persons who reported high physical exertion. In the I-, 2-, and 3-year follow-up analyses the crude point estimate of the risk ratio for having current symptoms was between 1.3 and 1.6 for the high-exertion group (1). Perceived high physical exertion in nursing work lnay partly be an effect of current low-back pain, but our study and previous studies indicate that high perceived physical exertion is a risk indicator of a condition preceding low-back pain.
Insufficient social support is strenuous in itself and may probably also increase the negative effect of other strenuous work conditions (28). Studies concerning the influence of social support on low-back pain are not consistent. One showed that nursing personnel who reported low-back pain were more satisfied in their relationship with workinates (29). In a Swedish study among nursing personnel low support from superiors was related to low-back pain (22), while, in another study (30), low support was related to neck and shoulder symptoms but not with symptoms from the low back. In a study concerning home-care workers, social support had no main effect on low-back symptoms, but it modified the effect of physical and psychological work load (31). Our study indicated evidence for a direct association between insufficient social support and low-back pain. However, the frequency of insufficient social support was low and the association was not significant.
Job strain was less frequent among the registered nurses than among the women in the nursing occupations with a relatively increased risk of low-back pain. On the other hand, in the analyses on the individual level, when nursing personnel who reported job strain were compared with nursing personnel not reporting job strain, no increased relative risk was shown. In 1 study, including 4 annual cross-sectional surveys in a hospital setting, the Scand J Work Environ Health 1998, vol24, no 6 risk ratio for current syrnptorns from the neck, shoulders, or back was between 1.1 and 1.5 for the job strain group when compared with that of the other rnernbers of the nursing staff (1). In 2 cross-sectional studies among nursing personnel, a significant relationship between job strain and low-back pain was observed ( 5 , 30). In our study the outcoine definition was incident cases of care sought for low-back pain, whereas in the studies with a positive relationship, the outcome was the prevaleilce of self-reported symptoms. The prevalence of self-reported low-back pain is dependen1 both on the incidence and the duratioil of the symptoms, and job strain inay be nlore relevant for the duration of symptoms than for the onset of low-back pain.
Worlcing part-time seemed to imply an increased rislc for low--back pain. In a previous cross-sectional study on female hospital staff, f~111-time and part-time workers did not differ in the prevalence of low-back pain. However, a multiple analysis indicated that psychosocial job factors were associated with the intensity of low--back pain for f~111-time workers but not for part-time personnel (32). In order to test the same hypothesis, we stratified the u~livariate analyses for world~ours, but no differences in rislc were observed.
Our study indicated a positive relationship between night shifts and low-.back pain and supported the hypothesis of an indepeilderlt association between night shifts and low-back pain. Night shifts had been shown to be connected with reduced sleeping time, and, in addition, subjective sleep quality is reduced (33). One hypothesis is that sleep disturbance leads to increased muscle ten-sion that contributes to pain in the low back (34.).
Forward-bent positions in coinbination with strenuous psychosocial conclitio~ls or night shifts increased the estimate for low-back pain. Different explanations have been suggested for a combined effect of psychosocial and physical risk factors. Increased muscle teilsion due to strenuous psychosocial conditions inay lead to illcreased vulnerability to physical demands (34). Furtherlnore, low control in the plan~ling of patient-handling tasks may lead to unsuitable and strenuous work positions (4, 3 6 3 6 ) . However, in the risk calculations of the effects of the combination of poteiltial risk factors cornpared with no exposure, the statistical power became low since few cases were exposed to a potential physical or psychosocial risk factor. It was not possible to estimate reliably wl-~ether the coinbination of risk factors exceeded their additive effect.
Our study focused on risk factors for health care consultations for low-back probleins and did not include the actual presence of disorders, symptoms, and pain froin the low baclc. In previous studies concerning low-back pain and work factors the definitions of disorders have not been consistent. Pain, pain leading to incapacity, sickness absence, incapacity for daily tasks, the duration of pain, the intensity of the pain, and baclc problems at the workplace are examples of outcome definitions (3). In our study health-care seeking was used as a proxy for the incidence of low-back disorders. All the subjects were examiiled by a physiotherapist according to a standardized procedure. The cases were classified into a group with clinical signs of low-back pain or sciatica and a group without any objective clinical signs. Of the health care personnel, 47 of 81 cases had objective signs and a diagnosis. The risk estimates did not vary a great deal between the cases with and without diagnosis.
One conceivable selection bias for this study could be that health care personnel, especially registered nurses, do not seek care for their pain. However, among the female referents, 34% reported current low-back pain. Of the registered nurses 22% of the referents reported cur-rent pain, the corresponding figures being 32% for the assistant nurses and 38% for the honie-care workers. The data did not support the hypothesis that health-care personnel suffer from low-back disorders without seeking care to a greater extent than do other e~nployed women.
On the contrary, one possible selection bias could be that ~l u r s i~l g perso~lnel seek inedical treatlnent for disorders inore often than do other occupational groups (37). Nursing personnel possibly rely more on professional care than women enlployed outside the health-care sector. An inclusion criterion of tliis study was that t l~e cases and the referents shoulcl not have sought care for lowback and neck or shoulder pain during the 6 inonths prior to the study occasion. Chronic cases currently under treatment for low-back pain were excluded. Thus a possible referral bias was that nursing personnel were currently under treatment more often than the other occupational groups.
A possible cause of the low risk for registered nurses when coinpared with that of the other groups could be that the health care sector in Norrtiilje was better supplied with transfer equiplnent or that the patients' requirements for care involved less load than in other regions in Sweden. None of the official statistical data, interviews with employees, or visits to workplaces supported this explanation.
The physical and psychosocial work factors were not assessed independently of the persoil in that the assessments were based on interviews and questionnaires. Differential recall bias is a validity problein in case-referent studies. It is possible that low-back pain biases the selfassessment of the work load. If the cases recalled the exposure with better accuracy or exaggerated the exposure when co~upared with the referents, it led to an overesti-rnatior~ of the observed risk factors. In the analyses, the individual reporting and the exposure distribution at the occupational level were considered. Differences between the registered nurses and the other nursing occupations ia the proportio~ls exposed in the reference group, the risk of seeking care for low-back pain in different occupational groups, and the risk factors at the individual level were presented. When the exposirre pattern aInong the referents is considered, potential 'ecall bias due to lowback disorders is reduced. Furthermore, when the reporting of exposure at the occupational level is examined, the importance of individual biasing factors decreases. On the other hand, exposure assessments based on occupation ignore differences in work conditions for different persons in the same occupation (38).
Perceived physical load and psychosocial factors, reported in the question~iaires, reflected the work conditions but were probably also associated with the person's psycl~ological and physical status and factors outside work. The interview lllethods were developed on the basis of the statement that an interview gives Inore descriptive information than self-reported questionnaires (13,39).