Effect of duration of employment in piecework on severe disability among female garment workers.

S. Effect of duration of employment in pieceworkon severedisabilityamong female garment workers. Scand J Work Environ Health 1989;15:329-334. This study measuredthe associationbetweenduration of employmentin piecework and the prevalenceof severe disabilityamong 533Quebecfemalegarmentworkerswholeftemploymentbetween1976 and 1985. Yearly information regarding occupational characteristics was available from a public agency for the 30-year period 1956-1985. Information on disability status was collected in a personal interview. Duration of employmentin pieceworkwas associatedwith the prevalenceof severedisability. A multivariate analysis found that the associationwasindependentof age, smokinghabits, education,type of task and total length of employment. Workers who spent 5-9, 10-14, 15-19, and ~20 years in pieceworkhad an increased prevalenceof severedisability in comparison with the base-linecategory of 0-4 years. The adjusted risk ratios were 2.2 [95 070 confidence interval (95 0/0 CI) 1.0-4.6], 3.3 (95 % CI 1.5-6.9), 3.6 (95 % CI 1.5-8.4), and 2.3 (95 % CI 0.8-6.6), respectively. The association wasparticularly strong for disability due to musculoskeletalproblems and disability due to cardiovascular disorders although small numbers were involved.

This is the second of two papers on long-term disability among female garment workers. The first showed an increased prevalence of severe disability relative to the corresponding prevalence for women employed in other occupations (I). In the present study we investigated the determinants of severe disability within the cohort of garment workers.
About one-half of the women employed in the Quebec garment industry receive a piecework remuneration (2). In other Canadian industries, such as textile, rubber, furniture, shoe, and garment manufacturing, a relatively large proportion of the work force is estimated to be employed in various systems of piecework (3). Therefore, the exposure of interest in the present study was the time pressure experienced by workers doing piecework, a type of paced work in which the employee is paid according to the number of units produced. The term "paced work" refers more generally to any job in which human or physical constraints determine the time in which a given task, and generally a specific quantity of production has to be accomplished by a worker (4). In addition to a high time pres-sure these jobs may involve repetitive motion and monotony.
While piecework has often been associated with a higher frequency of occupational injuries (5), little is known about the adverse chronic health effects possibly associated with this form of time pressure. Three experimental studies have assessed the health effects of piecework among factory and clerical workers by comparing the level of urinary catecholamine excretion of workers paid by piecework to that of workers in the same job paid an hourly wage (6)(7)(8). All three studies found a higher level of catecholamine excretion among the workers paid by piecework. Catecholamine excretion is usually considered a good biological indicator of the level of arousal in the presence of a stressor (9). However, the meaning of such an indicator in terms of future health effects remains largely unknown. Levi (7) suggested that workers paid by piecework may be more likely to deny their symptoms. Pieceworkers may be less inclined to adapt their pace when they experience pain or fatigue because of the monetary consequence of slowing down. This situation could lead to the accumulation of tension and fatigue and progress toward long-term adverse health effects.
Some cross-sectional studies have evaluated the health problems associated with other forms of paced work (eg, machine-paced work, assembly-line work, computerized control of individual production). One often-cited study of 2010 workers by Caplan et al (10) found that assembly-line workers had higher levels of depressive symptoms and somatic complaints (sleeping difficulties, digestive problems, loss of appetite, elevated heart rate , and sweating palms ) than 23 other occupational categories of blue-collar and white-collar workers. They also had the second highest level of anxiety and the third highest level of irritation symptoms. Results in line with the se previous findings have been reported when paced work was studied in automobile manufacturing (11), inspection on an assemblyline (12,13), fish processing (14), and service industries leg, telephone operation (15) and dat a entry (16,17)J.
Our pre vious study showed that garment workers who had left employment had an increa sed prevalence of severe disability compared with workers formerly employed in other occupations (1) . We hypothesized that the time pressure induced by piecework, when accumulated over several year s, had contributed to the development of disabling conditions among the garment workers. The purpose of the current study was thus to determine whether duration of employment in piecework was associated with the prevalence of severe disability within the cohort of garment workers.

SUbjects and methods
The Quebec women's ga rment industry is composed of more than 500 small factories , each emplo ying about 20-30 workers. Ninety percent of the se factories are in the Montreal area. Sewing-machine operators in the various factories manufacture the same products (women 's garments) with similar equipment. Little automation is used in women 's garment manufacturing because of the rapidly changing models and fabrics. All factories tend to be concentrated in one indu strial area of the city and located in similar buildings. Emplo yee turnover is high in the industry; 55 (1 10 of the sewing-machine operators in our study were emplo yed in three or more different factories during their worklife . Some differences ma y exist between factories in term s of dust, noise, light, and posture. However, site visits indicated that environmental factors did not appear to be systematically different in the factories using a piecework system as compared with factories using an hourly wage system. Thus it was po ssible to study the effect of piecework among workers doing essentiall y the same job in a comparable environment. The inclus ion criteria have been defined in our first paper (I). Given the focus on severe disability, the pr esent study was restricted to workers who had left employment at the time of the interview (N = 533).
Various systems of piecework are used in different indu stries (18). They can be classified according to the proportion of the salary dependent on the number of un its produced. Piecework systems can also be classified according to the type of rate given by unit produced. The rate can be fixed, but it can also decrease or increase above a certa in number of units produced. In the present study group, 100 070 of each wo rker' s sa lary was dependent on the quantity produced, according to a fixed rate per unit.

330
The classification of expo sure to piecework was based on the number of years in piecework bet ween 1956 and 1985. All occupational data were a bst racted from records kept by the Co mite paritaire du vetement pour dame s (Women's Garment Joint Committee). The se records conta ined a complete work history for each work er. This public agency is admini stered jointly by th e govern ment , employers, and union repr esent atives and is in charge of enforcing laws regarding wages and workhours in the garment industry. Th e inform ation is pro vided monthly by the employers. Inspectors review the info rma tio n regularly (on -site spot-checking) and in response to workers ' compl aints. The same dat a ha ve been collected and stored in a similar manner by the Cornite paritaire du vetement pour dames since 1955.
One record per worker per year was available. Each record contained a code for employer, job title, payment status (piecework versus hourly wage), the number of workhours, and the type of task . Th e repetitive ta sks were defined as those involvin g sewing onl y a section of each garment and nonrepetitive ta sks invo lved sewing the whole garment. Becau se of the incompleteness of th e records, info rmation regarding type of task was obtained by interview for 1980-1985. Expo sure and other work char act eristics were computed for years in which the worker was employed more than 1000 h to standardize the int ensity of expo sure between categories .
Severe disability was defined as one or more disabling conditions which pre vented employment or the conduct of hou sework for the 12 months preceding the interview. Data regarding disabili ty sta tus, smoking, and sociodemographic characteristics were collected by interview. A full description of the disability ind icator and the data colle ction has been presented in our previous paper (1).
The first step of the analysis ent ailed an analysis of the prevalence of disability from all cause s by occupational characteristics, education, and smoking status. Risk ratios and 95 010 Taylor-series con fidence intervals (19) were computed. A multivariate binomial regression (20) was performed to estimate the association between disability and duration of employment in piecework (and the variance of th e estim ate s) while pot ential co nfounders such as age, smo king status, educa tio n, type of task, and total length o f employment were controlled for. Potential confounders were assessed with the use of a bac kward elimination strategy, and the y were retained in the model when the ir exclusion resulted in a change of 15 0J0 or more in the effect estim ates. The second level of analysis focused on disability due to three gro ups of cau ses, ie, mu sculo skeletal [International Classification of Diseases (lCD) 71O-739J, cardiovascular (lCD 400-459), and other disorders co mbined (21). Analyses were performed with the statistical anal ysis system (22) and with the generalized linear interactive mod eling system (23).

Results
The study pop ulation was compo sed of 533 fema le sewing-machine operator s; this number repr esents 79.5 010 of the eligible population. The overall prevalence of severe disab ility was 18.6 % . Details of the population base, eligible popul ation, response rate, and a general description of the prevalence of disability have been presented in our previous paper (1) . Table 1 shows the crude risk ratios for various occupational characteristics, education , and smoking status. The prevalence of disab ility increased with the duration of emplo yment in piecework . A small increase in the risk ratio was observed for smokers and lessedu cated workers.
The regression analysis showed that the association with duration of emplo yment in piecework was confirmed after adjustment for age, smoking, total length of employment, and proportion o f years in repetitive tasks. Workers who spent 5-9, 10-14, 15-19 and <' :: 20 years in piecework had adju sted risk ratios of 2.2, 3.3,3.6, and 2.3, respectively, compared with the baseline category of 0-4 years in piecework (table 2).
We conducted a separate ana lysis for disability due to specific ca uses. Duration of emplo yment in piecework was associated with musculo skeletal disabilit y and, but perhaps to a less extent, with cardiovascular disability. There was no association with all other causes combined, and no other specific cause had sufficient numbers to warrant a separate anal ysis. The results remained essentially the same after confounders were controlled for, but the small numbers produced un stable effect estimates (table 2).

Discussion
This study examined the long-term effect of a specific form of time-related wor k pressure. Several crosssectional studies have measured the short-term symptoms or changes associated with garment assembly and various forms of paced work . However, this is the first time that the long-term adverse health effects have been directly investigated. The study design involved the following ad vantages over a cross-sectional study: (i) the study population consisted of work ers who had left employment during the last 10years; (ii) if the deceased cases are excluded and given that the withdrawal from emplo yment is assumed to be permanent, the pre valence of disability repre sents the cumulative incidence of disability for the l O-year study period (1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983)(1984)(1985); (iii) the cumulative exposure to piecework and other occupat ional char acteristics were measured retrospectively with objective records over the 30-year period preceding the interview.
The study used a self-repor ted indicator of disability. The limitations related to this indicator have been discussed in our prev ious paper (1) . Neither the expo sure status of the subject nor the exact hypothesis of the study was kno wn by the inter viewers and interviewees, and public and observable beha viors were reported which reduced the possibility of reporting bias. In addition, a self-reported cause of disability is a valid source of information with which to identify broad classes of disea se like those used in this study. This assumption is supported by Warren's finding (24) that the ability of impaired persons to report the un-Tabl e 1. Risk ratios for severe dis abil it y by occu pati onal cha ract erist ic s, education , and smoking status.    derlying medical condition is as high as 94 070 when compared with general pr actitioner's assessments.
The main limitation of our extern al comparison involved th e possible bias due to differences in job requirements. Howe ver, in th e present study, this limitation was overcome since this factor was controlled by the nature of the stud y population. The compared workers had all been emp loyed in the ga rment indu stry, and the y lived in the same metropolitan ar ea. Therefore, their employment opportunity either in piecework or hourly wage jobs in garment manufactu ring or in oth er indu stries, along with their jo b requirements, had been relatively similar. In addition, educational level was controlled in the an alysis, and thi s procedure minimi zed bias due to differences in qualification.
Duration of employment in piecework was associated with the prevalenc e of disabi lity. The association was independent of age, smo king status, education , type of task, and total length of employment. The data thus showed that the prevalence of disabilit y was specifi cally associated with duration of employment in piecework am ong workers previously employed as sewing-machine operato rs for similar durations.
Our results are compatible with finding s of previous cross-sectional studies of garment workers (25)(26)(27)(28)(29) and other groups of paced workers (6-8, 10-17, 30), which found a high frequency of ph ysiolog ical, musculo skeletal, and psychological symptoms. More specifically, the association with musculoskeletal disability 332 is compatible with the previous finding of a higher incidence of disability pensions due to musculoskeletal disorders amo ng slaughterhouse workers (work which also involves fast pacin g) when compared with the general population (3 I).
Among the 3I musculo skeletal disability cases, 19 were arthritic disorders (lCD 7 14-719), seven were back disorders (ICD 720-724) , and the remaining five cases were disper sed among other catego ries o r were unspecified. In another section of our questionnaire, a proportion of 58 to 77 070 of the women reported regular pain in the follo wing sites: (i) lower back , (ii) upp er back and neck , (iii) shoulders, (iv) hands, wrists or elbows, and (v) knees or ankle s. All but one of these women had at least one regularly painful site.
A high prevalence of mu sculo skeletal disord ers is compatible with the biomechanical stressors involved in a sewing-machine operator' s job. The most frequent type of arthritic disorder, osteo arthritis, is possibly associated with wear-and-tea r on joints (32). Th is conclusion is support ed by pre valence studies which have shown specific affected sites in various occupat ional groups (elbows and knees in miners, fingers in cotton pickers, and finger s, elbows and knees in dock wor kers) (32). In add ition , two po stural characteristics of sewing-machine op erators' jobs (working in the same position for a long period of time, including prol on ged sitting and bent -over position , and repetitive work) are two of the six most important work-related factors associated with ab sence from work due to low-back pain (32). Finally, repetitive hand motions are associated with soft-tissue disorders in garment manufacturing (29).
The fact that musculoskeletal disability was specifically associated with duration of employment in piecework among the sewing-machine operators who had performed identical tasks suggests a combined effect of the time pressure induced by piecework and the biomechanical stressors involved in performing the task.
Ergonomic studies (26,30) have suggested a possibly related phenomenon of a more bent and more rigid posture when time pressure intensifies.
In conclusion, the data show that the duration of employment in piecework was associated with an increased prevalence of severe disability among garment workers. The association was particularly high for disability due to musculoskeletal disorders and for cardiovascular disability, although small numbers were involved. For musculoskeletal disability, one possible mechanism could be the combined effect of the biomechanical stressors involved in performing the task and the time pressure induced by piecework. Further research is needed to determine whether or not pieceworkers in other industries and other groups of high-paced workers also experience disabling conditions and to investigate the possible mechanisms.