Low-back pain in commercial travelers

MORCET BLONDET M. Low-back pain in commercial travelers. Scand J Work Environ Health 1992;18:52-8. The role of occupational environ ment in the occurrence of low-back pain was analyzed for 1719 commercial travelers (1376 men and 343 women). At the beginning of the study (To)the group was interviewed to determine current lifestyle and occupational factors associated with low-back pain during the previous 12 months. Twelve months later (at T 1), a subsample of 1118persons (893 men, 225 women) was reinterviewed to study the asso ciation between the incidence of low-back pain during the year of follow-up and the risk factors at To. At Tolow-back pain was significantly related with time spent driving a car at work, comfort of the car seat, carrying loads during work, standing for long periods at work, smoking, and psychosomatic fac tors. At T 1 driving 10 h/week or more, seat comfort, and psychosomatic factors were associated with first occurrence of low-back pain.

Low-back pain is considered an important public health problem because of its high frequency and its heavy socioeconomic burden (1-3). According to various studies, 60 to 800/0 of the general population has or has ever had low-back pain (4)(5)(6)(7)(8), and back problems are responsible for between 9 and 19.5% of all sick leave days (9). Several studies have been carried out in the general population and in the work environment in order to determine the role of various individual, social , and occupational factors in the occurrence of this disorder. In the literature, the most consistent result is a nonlinear increase in prevalence with age (5, [10][11][12]. Although some authors found a higher prevalence among women (5,10,13), on e study showed that men had a higher prevalence of the lumbar disc syndrome (12), and others did not find an y relat ionship with the gender of the indiv idual (8,14). With regard to marital status, the prevalence of low-back pain seems to be higher for people living alone (10,II,13,14). However people who have never been married were found to be at the lowest risk in three of the studies (10,11,13) and at the highest risk in one (14). Other factors such as edu cational level or tobacco consumption have been studied and will be discussed later. Most of the etiologic factors have been studied in the work environment, and the most consistent results are related to the lifting of heavy loads (7,8,(15)(16)(17)(18)(19), and the driving of motor vehicles (8,(19)(20)(21). Other factor s, such as sedentary work or con- 52 siderable physical effort at work, have also been studied, but the results are less consistent (I I, 13,14). A national survey of a group of 1719 commercial tra velers and 861 sedentary workers was carried out in France between I October 1985 and 30 September 1986 in order to compare the occupational risk factors and morbidity of commercial travelers with those of a reference group of white-collar workers. The results of this study have been reported elsewhere (22). The homogeneity of the group of commercial travelers with regard to the kind of work constraints they experienced led us to carry out· a complementary study limited to this occupational group. This study focused on low-back pain in relation to lifestyle and the work environment. In this article we report the prevalence and incidence of low-back pain and their association with other factors for the group of commercial travelers.

Study design
This national survey was conducted in 11 medium-size towns in France by a group of 74 physicians in cooperation with research workers . The ph ysicians worked with the same organization [Centre d'Information des Services Medicaux d'Entreprises et Interentreprises (Information Center for Medical Services in Small or Medium Size Firms)] and were in charge of the medical surveillance of employees who worked in a wide range of occupations. In France, most employees undergo a compulsory medical examination each year.
Each person in the survey was interviewed during the annual medical examination. The interview, lasting 10-20 min, was performed by the physician. It included questions about life-style, work conditions, and a list of health problems. For nonmedical variables, the physician asked questions exactly as they were in the questionnaire, and the measures of exposure were based on self-declarations of the subjects. For the health variables, possible sources of bias due to physicians were discussed with them before the survey and at the beginning of the data collection, in order to standardize their approach.
The sampling procedure was as follows: each physician participating in the survey hada quota of sedentary workers and commercial travelers to interview in each quarter of the year. The first person to meet the inclusion criteria in each half-day of routine examination was included until the quota for the quarter was reached. This sampling procedure was simple and avoided recruitment bias because the arrival order of employees relies on an administrative procedure and is not under the control of the physicians.
The inclusion criteria for the commercial travelers were: (i) employed in a job involving soliciting orders and selling goods, products, and services to retailers, industrial consumers, institutional and professional establishments, or private persons; (ii) employed in a job involving visiting clients for more than half of the time.
The International Standard Classification of Occupations (23) would classify persons in this group in categories 4.32 (commercial travelers and manufacturers' agents), 4.31 (technical salesmen and service advisors) , and 4.41 (insurance, real estate and securities salesmen).
All of the individuals in this group had similar work conditions. They were responsible for an allotted geographic area, which involved many driving hours and working far from home, sometimes for more than a day. They had to eat at restaurants alone or with clients. Some of them had heavy samples of goods to carry, although the physical constraints were not as severe as for many groups of blue-collar workers.
We intended to recruit physicians for the survey in every large and medium-size town in France, but it was not possible in Paris and Marseille. As a consequence, the sample is representative of commercial travelers in medium-size towns . Another limitation is that employees in the largest firms (which have their own medical services), and self-employed workers (who do not have annual medical visits as a benefit) were excluded. This was a cross-sectional and longitudinal study based on two interviews conducted as has already been described. The first one was carried out at the beginning of the survey (To) and concerned health problems during the past 12 months and risk factors at To (cross-sectional study). The second one was conducted among the same workers 12 months later (T I ) and mainly concerned health problems during the year of follow-up (longitudinal study). Most of the risk factors were not recorded at T l ' In all, 1719 commercial travelers (1376 men and 343 women) were interviewed in the cross-sectional phase (To). Some of the physicians did not volunteer for the longitudinal phase of the study (T.), which reduced the sample to 1381 subjects, who had the same characteristics as the group as a whole. Of these, only 1118 persons (893 men and 225 women) could be interviewed. The other 263 subjects could not be traced by the physicians, mainly because they were no longer employed in the same firm or they had moved. The workers lost to follow-up were younger and slightly healthier than the rest of the sample.
Ten subjects were excluded from the analysis because their age was unknown. The numbers of subjects were thus reduced to 1709 at To and 1115 at T[.

Analysis
The effects of various factors, such as sociodemographic characteristics, life-style [especially tobacco and alcohol consumption, sports), work environment (eg, hours spent driving, lifting and standing), and psychosomatic factors] were studied.
The following two measures of low-back pain were used in the study: (i) one-year prevalence of low-back pain in the cross-sectional study, defined as the proportion of persons who had had low-back pain during the 12 months prior to the interview at To, irrespective of whether the symptoms were new, and (ii) one-year cumulative incidence of low-back pain, which was determined from a question of whether the persons had had new symptoms of low-back pain during the year of follow-up. The annual incidence was calculated for the subjects who had never had low-back pain until To. Among them, incident cases were defined as persons having had low-back pain for the first time in their life between To and T r-All types of low-back pain were recorded, irrespective of whether the symptoms were severe or not.
In the analyses, risk factors were determined at To only. The following two analyses were performed: (i) prevalence of low-back pain in relation to risk factors at To and (ii) incidence of low-back pain in relation to risk factors at To.
The study on prevalence was cross-sectional and performed on the sample as a whole. The study on incidence was a cohort study with a prospective followup period of one year, restricted to the subs ample of people interviewed in the longitudinal study (T I ) who had never had low-back pain before To.
For the study on incidence, it was decided to limit the list of risk factors to those recorded at To in order to be sure that the risk factor was present before the onset of the disease and therefore avoid record bias.
The variables used in the analysis are described in tables 1-3. They concern sociodemographic data (age, marital status, number of children, educational level), life-style (playing a sport, consumption of coffee, tobacco and alcohol), weight, height, body mass index, general work conditions (duration of employment as commercial traveler, time per week spent at work and driving, number of kilometers per year), physical con-   (24). The number of answers expressing psychosomatic symptoms, from 0 to 10, was used in the analysis to evaluate the magnitude of psychosomatic problems as a whole.
The logistic model was used for the two analyses utilizing BMDP (biomedical data package) computer software (25). Odds ratios were calculated in both analyses, and relative risks were calculated in the incidence analysis.
First, all of the variables were entered, one by one, with age in a model. Only those significantly correlated with low-back pain were entered together in the multivariate logistic regression presented in this report .
Since the multivariate analysis by gender showed the same trends in risks and were consistent with the overall analysis, and since the number of women was small, especially in the longitudinal study, both genders were grouped in the analysis. Possible interactions between gender and other risk factors were not studied since gender was not significantly associated with low-back pain in the multivariate analysis.

Description of the sample
A description of the sample, by gender at To, is presented in tables 1-3. Some differences were observed between the men and women. The men had worked as commercial travelers longer than the women, but they did not differ in age. Most of the men did not live alone. Alcohol consumption was higher among the men, and there were more smokers and exsmokers among the men than among the women . Coffee consumption and regular participation in a sport did not differ between the men and women. The men spent more time at work, especially driving, but the women complained more often about the physical constraints of their job.

Results
The prevalence and incidence of low-back pain are shown for the sample in table 4. At To and T 1 , the women had a higher prevalence of low-back pain than the men. The incidence was also higher for the women than the men.

Prevalence of low-back pain in relation to other factors at To
Using a multivariate logistic regression model, we compared the subjects with low-back pain in the 12 months preceding To with those without low-back pain during this period (table 5). The strongest relationships with low-back pain were observed for driving (only the risks associated with driving more than 20 h/week being significant), comfort of the seat, and psychosomatic factors , the odds ratio increasing with the number of symptoms. The odds ratio was also significantly different from unity for the smokers. Standing and carrying loads was found to be associated with low-back pain with an odds ratio close to unity. Age and gender were not significantly associated with lowback pain when the other risk factors were taken into account.

Incidence of low-back pain in relation to other factors at To
The incidence of low-back pain in relation to risk factors at To is shown in table 6. Because the analysis was restricted to people never having experienced lowback pain before To, only 627 subjects were included.
An association was found between the comfort of the car seat and the occurrence of low-back pain as a new disorder during the year of follow-up. For time spent driving there was an association, but it was significant only for driving between 10 and 20 h/week . A significant excess risk also remained for subjects having a sum of three or four psychosomatic symptoms. Carrying loads, standing, and smoking did not prove to be predictive factors of low-back pain.

Discussion
-In our stud y, the one-year prevalence of low-back pain varied from 25.1 to 28.2070 for the men and from 34.9 to 37.5% for the wo men . In the literature, the oneyear prevalence varied between 14 and 45% for the general population (5,13,26) and between 20 and 62% for workers (7,11,14). However, it is difficult to compare these results since the definition of the cases varied considerably according to the author.
The one-year incidence was estimated as 13% in our study. Only a few other studies have estimated the incidence of low-back pain. The incidence has been shown to range between 6% for the general population (5) and 11.7% for company employees (11).
With regard to the sociodemographic factors associated with low-back pain in the cross-sectional study, we found a borderline significant increase in the one-year prevalence with age (table 5). However, this association was observed for the men only (results not presented here), and it slightly differs from results of other authors (6, II). Table 4. Prevalence and cumulative incidence of low-back pain among the men and women separately.
Cross-sectional study (To) Long itudinal study b Prevalence during the 12 months preceding the beginning of the study. e Incidence from the beginning of the study (To) to the end of the tollow-up (T l ). Table 5. Unconditional odds ratios (OR) for the one-year prevalence of low-back pain during the 12-month period preceding the cross-sectional phase of the study (To) and the sociodemographic data, life-style, work environment, and psychosomatic factors at that time . Gender was also associated with low-back pain, with a higher risk among the women, but this result was not significant when other factors were taken into account. This finding does not disagree with the results of other authors (5, 8, to, 13, 14).
Most of the other nonoccupational factors were not associated with low-back pain in our study . We did not find an association with marital status or educational level. Educational level has been found to be correlated with low-back pain in other studies (to, 11, 13, 14). The lack of this association in our study could be due to the fact that the educational level of our group was rather homogeneous . With regard to marital status, the results of various studies are not consistent (to, 11, 13, 14). Life-style factors associated with marital status could playa more important role than marital status per se. Although height (11, 14) and a 20070 excess in relative weight (11) have been found to increase the risk of low-back pain, we did not find any association ' with height, weight, or the body mass index. 56 Of the nonoccupational factors significantly associated with low-back pain in the cross-sectional study, tobacco consumption showed a significantly elevated risk among current and ex-smokers compared with nonsmokers. This finding is coherent with those of other studies (6,14,21,27). No obvious explanation has been found for this association, and it is not certain that tobacco consumption plays a causal role. The occupational factors associated with low-back pain in the cross-sectional study were the time per week spent driving, the comfort of the car seat, carrying loads, and standing. Kelsey and her co-workers found that the driving of motor vehicles was associated with an increased risk for developing acute herniated lumbar intervertebral disc (20,21,28). An association between driving vehicles and low-back pain was also shown by Walsh et al (8) and Damkot et al (19). For Walsh and his co-workers driving more than 4 h daily in the year prior to the onset of symptoms was found to be associated with the occurrence of low-back pain. The study of Damkot et al was cross-sectional, and no de- Table 6. Unconditional odds ratios and relative risks for the one year cumulative incidence of low-back pain in the longitudinal phase of the studyand the sociodemographic data , life-style, work environment, and psychosomatic factors in the cross- tails were given about the time spent driving. Mechanical stresses on the spine from the sitting posture and from vibration have been incriminated in this association. The prolonged sitting posture has been found to be associated with low-back pain (8,28,29), but no reference was made to the comfort of the seat. Because of the cross-sectional design of this part of the study, it was difficult to determine whether an uncomfortable seat was responsible for low-back pain or whether the subjects found their seat uncomfortable because of their back pain. In the literature, heavy lifting is one of the most constant factors associated with low-back pain (6-8, 18, 19). With regard to carrying loads, we found a significant increase in risk of the same magnitude among the men and the women. However, no details were available about the weight lifted. Prolonged standing also represented a slightly increased risk, and this finding is supported by those of other studies (7,29). As has already been discussed, the direction of the association between these factors and low-back pain cannot be established from cross-sectional data. Moreover, commercial travelers do not undergo very high levels of physical constraints. Finally, a strong association with psychosomatic factors was observed, and corroborated by the trend in the odds ratios according to the magnitude of the disorders . Although there is some evidence of an association between low-back pain and psychological factors such as anxiety, depression, or hysteria, no consistent results emerge from the literature (30)(31)(32). Most studies on psychological factors and low-back pain have been carried out retrospecti vely, in which case it is not possible to conclude whether the psychological troubles are the cause or the consequence of low-back pain . A recent study using longitudinal data concluded that stress symptoms and musculoskeletal disorders are reciprocally related (33).
In the longitudinal part of the study, we attempted to determine which factors could be considered predictive of low-back pain. Except for age, carrying loads, and stand ing, the associations found for prevalence were also observed for incidence. The consistency of 57 our results at To with those at T t is an argument in favor of a causal relation between low-back pain and some of the risk factors . For example, the number of hours spent driving and the comfort of the car seat remained risk factors and thus supported the hypothesis of driving as a causal risk factor of low-back pain. The results suggest that there could be a threshold limit for the incidence of low-back pain (10 h driving/week). The fact that a dose-response relationship with prevalence was observed could be coherent with the hypothesis that the duration of symptoms increases with the number of hours driven.
For the study on incidence , the risk factors were recorded at To. If the risk factors have an effect without a latency period, it might be better to assess them immediately before the onset of the symptoms. The differences between these two ways of assessing exposure are probably not important, but they could have biased the odds ratios of our study toward unity.
Most of our results confirm those of other authors, especially concerning driving and psychological factors. Even if it is not possible to draw formal conclusions with regard to causal associations, the consistency of the results with time (at To and T t ) and with findings reported in the literature indicate that these associations are probably not spurious . Further longitudinal studies are needed to confirm the causality so that preventive measures can be adopted in occupations at risk of low-back pain.