among female garment workers. A comparison with a national sample.

BRISSON A, M. Disability among female garment workers: a comparison with a national sample. Scand J Work Environ Health 1989;15:323- 328. An indicator of permanent disability was used to measure the chronic health problems associated with garment work. The study population comprised 800female sewing-machine operators employed in the Quebec garment industry between 1976 and 1985. The workers were identified from public agency records. Their disability status was obtained in a personal interview. The disability prevalence of these workers was compared to national disability data of women employed in clerical work, services, and manufacturing industries. The garment workers who had left employment had an increased prevalence of severe disability (in comparison with that of workers who had left other types of employment) and an increased prevalence of moderate and slight disability. Currently employed garment workers had an increased prevalence of moderate and slight dis ability when compared with workers currently employed in other occupations. The risk ratios remained similarly elevated when the data were adjusted for age and smoking status.

Little is known about the he alth consequences of wo rking as a sewing-m achine operator , but mu sculoskeletal co m pla ints and an xiety sym pto ms have been reported (3)(4)(5)(6)(7). Even though the ta sks o f th ese worker s invol ve rep etitive motions, ergonomic studi es (5) have sho wn that they ca nnot be performed automatically, as has often been sug gested for assembly lines o r other repetitive jobs. They require a high level of mental activity coupled with a high level of visual attention and accuracy of motion, with consta nt coordination between the eyes, hands, and feet. In addition, th ese jobs often involve a high level of time pressure, which has generally been considered an important cause of anxiety symptoms and mu sculoskeletal complaints (4-7) among workers.
Becau se pr eviou s stud ies included onl y ac tive worker s, they did not assess the more sever e effect s wh ich would lead to termination of employment. Th e pur-pose of the present study wa s to det ermine if garment wo rke rs, and par ticularly those who leave employment, have an increased risk of chronic health problems when compared with women employed in other industries.

Subjects and methods
The garment wo rke r population included all the women wh o were employed in unionized fa ctories in Quebec between 1976 and 1985 and who sa tisfied the followin g criteria : (i) age 45-70 years in 1985 (time of int erview) , (ii) sewing-machine operator fo r at least five cons ecutive years (at least 1000 workhours each year), (iii) Canadian-born and French-speaking, and (iv) resident of the Montreal metropolitan a rea at the time of the interview.
We used the first and second criteria to select a group of workers whose health problems were more likely to reflect the long-term effect of the work characteristics rather th an sho rt-term alterations in health sta tus . Immi grant workers were excluded to improve the comparabilit y of the gro u ps. The study was restr icted to the Montreal metropolitan area because 90 0J0 of the garment factories in Quebec are located there.
The coho rt was identified fro m list s of ga rment workers supplied by the Cornite Paritaire du Vetement pour Dames (Women' s Garment Joint Committee). All employers op erating in the women ' s garment industr y are required to provide thi s agency with monthly information regarding the wages and workhours of their employees. The ga rme nt workers were co m pa red with a subgroup of the Canad ian H ealth Survey, co nd ucted in 1980 (8). The subgroup wa s selected to have cha rac-teristics similar to those of the garment workers under study. Thus it was composed of women who were 45-70 years of age, had 12 or less years of schooling, were born in Canada, were French-or Englishspeaking, and were currently employed or had been employed in the last five years preceding the interview in clerical work, services, or manufacturing. The education and occupation criteria were used to minimize confounding by socioeconomic status. This selected subgroup could not be restricted to French-speaking women because the numbers available were too small.
For the determination of the differences in chronic health among the two groups of subjects an indicator of permanent disability was chosen. Disability was defined as the presence of one or more disabling conditions for at least 12 months. This indicator was divided into four categories according to the individual's ability to function. These categories were (i) no restriction of activity (no disability), (ii) slight restriction of activities such as hobbies, civic projects, church work, or club work, (iii) moderate restriction of the amount or kind of main activity, ie, working at a job or doing housework, and (iv) severe restriction (unable to carry on main activity).
We attempted to collect data on the garment workers in 1985 in a manner comparable to the Canadian Health Survey. In both groups data regarding disability status, smoking, and sociodemographic characteristics were collected in an interview with the same questionnaire and techniques. The interviews were conducted at home by an interviewer trained according to a standard protocol (9). In both groups reported health problems or diseases which caused disability were coded according to the International Classification of Diseases, ninth revision (ICD) (10). These health problems were  grouped into the following three categories: (i) musculoskeletal diseases (ICD 710-739), (ii) cardiovascular diseases (ICD 400-459), and (iii) other diseases combined. The analysis controlled for age, smoking status, and employment status (currently employed versus having left employment). In both groups the "having left employment" category was comprised of those who had not worked for the majority of the 12-month period prior to the interview. For comparisons which involved workers who had left employment, the group of garment workers was restricted to those who had left work during the five years prior to the interview. This restriction ensured similarity in respect to time since last employment among the two groups compared (the Canadian Health Survey subgroup was those who left employment during the last five years preceding the interview). Risk ratios were used as the measure of effect (11). Adjusted risk ratios were computed according to the Mantel-Haenzsel method (12), and confidence intervals were computed with Miettinen's approximation method (13). All the analyses were performed with the statistical analysis system (14). Table 1 shows the population base, eligible population, and response rate of the garment workers. A total of 79.5 010 of the eligible population was interviewed. Approximately 50 % of the obtained addresses proved to be incorrect, and applications were made to several public agencies in an attempt to locate the workers. The final untraced percentage was 10.0 11 10 , and the refusal rate was 10.5 %. The refusals may be partially explained by the fact that the study population was composed of women, many of whom were retired and living alone and who worried about admitting strangers to their homes. In addition, the tracing was complicated by the fact that women changed their names when they married.

Results
The interviewed garment workers thus comprised 800 sewing-machine operators between the ages of 45 and 70 years. Approximately 45 % of the workers were between the ages of 55 and 64 years. Women with an education of 12 years or less comprised 95 % of the study population, and nearly 66 1110 of the women had left employment at the time of the interview. Of the workers with disabilities, 9004 % of the conditions had been present for more than 12 months. This proportion is similar to that found for the general population in previous studies (8,15). As is standard practice with this indicator (8,15), all disabilities were considered permanent.
The prevalence of disability among the garment workers is presented in table 2 by age and employment status. By definition, severe disability (a condition which prevented employment or the conduct of house-work) was exclusively applicable to workers who had left employment. Among these workers the prevalence of severe disability was highest in the 55-to 64-year age group (24.6 070) as compared to 18.5 % for the 45to 54-year age gro up. Th e prevalence was lowest among the workers between 65 an d 70 years of age (13.9 % ). Moderate disabilities were very com mon . Among the inactive workers modera te disabilities increased with age. Amo ng the acti ve workers, all the age groups were equally affected .
Former garment workers had an increased prevalence of severe disability com pared with workers former ly employed in other occupations (table 3). The age-adjusted risk ratio was 5.0. This increased prevalence of severe disa bility was larger in the younger age gro ups. The former garment workers also had an increased pre valence of mod erat e and slight disab ility (age-adju sted risk ratio 1.3). Currently emp loyed garment workers had an increased prevalence of mo derate and slight disability compared with workers cur-  rently employed in other occupations; the age-adjusted risk ratio was 1.7. The risk ratios for all these associations remained similarly elevated when data were stratified according to smoking status. Analysis by cause of disability indicated that the association found for severe disability was present for all major categories of disabilities. The risk ratio for disability due to musculoskeletal disorders was 6.9, that for disability due to cardiovascular disorders was 4.8, and that for disability due to other causes combined was 4.9 (table 4).

Discussion
This study found an increased prevalence of disability among female garment workers as compared with women employed in other occupations. The prevalence was particularly elevated for severe disability, and all major categories of disability were involved. These findings indicate that short-term and nondisabling conditions identified by previous authors may have more significant long-term sequelae than had been previously documented.
For the determination of the differences in chronic health problems among the two groups of subjects, we chose an indicator of permanent disability. Disability indicators measure morbidity states, and more specifically, the consequences of morbidity states in terms of the individual's ability to function. They generally include measures of physical mobility, physical independence in basic actions, and the ability to carryon one's usual activities (16,17). We chose a disability indicator based on one's ability to perform usual activities because it encompasses both the highly disabling physical states and the less disabling physical and mental health problems. This indicator also had the advantage of having been used in the Canadian Health Survey (8); thus comparison with national data was possible.
The chosen disability indicator has had useful applications both as a single measure (18,19) and as a 326 component of an index of life expectancy free of disability (15,20,21). The latter-mentioned index combines mortality and disability data and is expressed as the average number of years a cohort may hope to live without disability. Its use has highlighted the importance of musculoskeletal problems which are not reflected in mortality indices (20, 21).
The indicator of disability was a self-reported one. Conceptual and operational limitations related to this indicator have been discussed by Sullivan (22). One possible source of bias was that our interviews were conducted directly with the workers, while in the Canadian Health Survey interviews were conducted with one member of the household, who provided information regarding other household members. Thus, in some instances, the disability status of the subjects selected from the national sample was assessed from an interview with a household member. This method of data collection could have led to an underestimation of the real prevalence of moderate and slight disability in the Canadian Health Survey. However, it is reasonable to assume that the reporting of severe disability was not markedly affected. It also seems unlikely that sociocultural differences would affect the reporting of severe disability.
Another limitation of the study is that the prevalence of severe disability, as defined in this study, may have been influenced by the availability of pension and other supports for illness. Indeed, subjects may be more likely to stop work or the conduct of housework because of a health problem if they have access to these services. However, in both of the groups compared, the availability of both types of public pensions (regular retirement pensions and disability pensions) was similar (23). Access to medical care should also have been similar in both groups since hospitalization, treatment, and visits to physicians are universally insured for Canadian citizens.
Another potential problem is that this study compared data collected at two different time periods. The garment workers were interviewed in 1985, while the Canadian Survey interviews were conducted in 1980. Ho wever , the interview procedure a nd op erational definitions of disability were ident ical. In addition factors possibl y related to the reporting of severe disability, such as the availability of a regular pen sion, a di sability pens ion, and medical ca re, were similar in both populat ions over that time period .
Admission crit eria were slightly different for the two population s. The garment workers had been employed five years or mo re in garment work. Th e Canadian Survey subgro up may ha ve been employed for shor ter per iods since the criterio n was em ployment an ytime between 1976 and 1980. The healthy worker effect is known to be stronger as length of employment increases (24). Thus, if anything, this difference would underestimate the true effect.
Fina lly, the prevalence of severe disability may be influenced by the requirements of jobs available. An illness which prevents work in certain types of jobs may not prev ent work in jobs that are less physically or psychologically demanding. In that sen se, th e severe disability indicator measures a morbidity level which is relativ e to the requirements of jobs available. In Qu eb ec the level of unemployment for wom en in the relevant age gro ups was similar to that in other areas of Ca na da (25). Thus the number of jobs available to both groups , and pre sumably their variety, tended to be similar. Ho wever, female garment work ers in these age groups may not be able to switc h to new types of work so rea dily as other women. Therefo re , we cannot conclude that the requirem ent s o f jobs available were similar in both gro ups compared. For thi s reason the ef fect mea sures pre sented in thi s report ma y no t represent differen ces in 'absolute " health stat us. They ma y, to some extent, be conditioned by the requirements of jobs.
Th e ob served prevalence of severe disability was lowest in the oldest age group. Thi s phenomenon ma y reflect early mortality among per sons with severe disability. Alternatively, it may refle ct the limitation of the indicato r. Regardless of their health status, individuals 65 years of age or older do not consider themselves as part of the work force be cause the Quebec standard public retirement pension is provided at this age (23). Thus th e main activity at th is age is generally hou sehold wor k. A chronic condition which prevent s a wom an fro m gaining employment may not pr event her from carrying out household activities. Thus it is not appropria te to use thi s indi cator to compare the pre valence of severe disability of women age d 45-64 years to th at o f wo men aged 65-70 years . However , this limitation o f the disabilit y indicator should not affect our compar ison with other female work ers since age-specific and age-adjusted anal yses were presented .
This study was the first to examine the chronic health problems of ga rment workers. Therefore , the results cannot be dire ctly compared to pr evious find ings. Given the limita tion s due to possible dif feren ces in the requirements o f job s, these results should be int er-preted with caution. However , our results are compatible with findings of previous cross-sectional studies which sho wed a high frequency of musculoskeletal and psychological symptoms among garment wo rkers (3)(4)(5)(6)(7). The present results are also consi stent with the internal an alysis performed within the group o f garment wo rkers. Thi s a nalysis found specific work conditions, par ticul arl y duration of employment in piecework, to be associated with an increased pre valen ce of severe disability. This finding is pre sented in a second pap er (26).