METHODS

Findings are based on data reported in 1990/91 by 858 respondents born in the early 1930s who are participating in the West of Scotland Twenty-07 study, a longitudinal study of social inequalities in health. Further details of the design and sampling have been described elsewhere.³ Ethical approval was obtained for the study. The main outcome measure was locomotor disability; those reporting positively to any of the 17 items of a questionnaire adapted from the Office of National Statistics surveys of disability⁴ were classified as disabled. These items incorporated, ability to walk, climb stairs, bending and straightening, and number of falls. The measures of socioeconomic position that were used in the analysis included: (1) occupational class—occupation of head of household ranked into five classes from social class (SOC) I (highest) to SOC V (lowest); (2) car ownership—yes/no; (3) housing tenure—owner occupier/other; (4) household income—in pounds per week adjusted for household size and composition; (5) area deprivation assessed using the Carstairs Index, which was derived from 1991 census variables (from 1 (least deprived) to 7 (most deprived)). Respondents were asked to estimate the number of years they had been exposed to 13 conditions relating to the physical and psychological workplace environment during their working lives and included: work in very noisy conditions, work in very dusty conditions, work in very hot conditions, work in very cold conditions, work in very wet conditions, work with fumes or chemicals, work with a lot of vibration, work in a bent or uncomfortable position, work which was monotonous and repetitive, work which was too hectic or fast, work causing a lot of stress and worry, hard tiring physical work, and outdoor work. For this analysis, these data were categorised into none compared with one or more years of exposure.

The relation between locomotor disability and the measures of occupational exposure, socioeconomic position, and gender were assessed using χ² or t tests and univariable logistic regression. All variables on occupational exposure were tested for interaction with gender. A series of logistic regressions were then conducted to consider the relations between social class and gender with locomotor disability controlling for other measures of socioeconomic position (including household income) and those occupational exposures significantly associated with locomotor disability (for either men or women) in the univariable analysis. All multivariable analysis was conducted using complete data.

RESULTS

Prevalence of locomotor disability among this cohort (aged about 58 years) was 27.2% (95% CI 22.7 to 31.7). The odds of locomotor disability were increased by all measures of lower socioeconomic position and all measures of more adverse occupational exposures relating to the “physical” environment in univariable analyses (see table 1).

When adjusting for gender, those in manual social classes were more likely to have locomotor disability than those in the non-manual social classes (OR 2.74, 95% CI 1.90 to 3.93 p<0.001). This relation was attenuated after adjustment for additional measures of socioeconomic position (OR 1.67, 95% CI 1.12 to 2.50, p<0.012). When those occupational exposures found to be associated with locomotor disability in the univariable analysis were then included in the model, the association between social class and locomotor disability was further attenuated and was no longer statistically significant at the 5% level (OR 1.39, 95% CI 0.91 to 2.11, p=0.128). This would suggest that most of the observed association between social class and locomotor disability, at early old age, is explained by a combination of degree of affluence and occupational exposures.

Women were about 60% more likely to report locomotor disability compared with men, after adjusting for household social class (OR 1.63, 95% CI 1.17 to 2.27, p=0.004). Adjustment for additional measures of socioeconomic position slightly attenuated this association (OR 1.49, 95% CI 1.06 to 2.10, p=0.023). However, with further adjustment for cumulative occupational exposures, the association became much stronger, and women were over two times more likely than men to report locomotor disability (OR 2.27, 95% CI 1.47 to 3.49, p<0.001).

There was no evidence of any interactions between gender and the occupational exposures with the exception of exposure to dust. This may have led to some residual confounding for this exposure in the logistic regression analysis and may have resulted in a slight over-estimation of the increased risk of locomotor disability among women. However, it is unlikely that any residual confounding relating to exposure to dust would explain all of the observed gender difference. Given the lack of previous research postulating this interaction, this result should be interpreted with caution and may have occurred by chance given multiple significance testing.

COMMENT

The magnitude of gender differences in locomotor disability may be masked by variables relating to socioeconomic position and occupational exposures. Our findings are consistent with other studies in showing that women in early old age are at greater risk of disability than men,^2,5 but controlling for socioeconomic position and occupational exposures shows that the extent of the association between gender and locomotor disability is underestimated. Our findings suggest that social class inequality in locomotor disability is largely explained by adverse socioeconomic factors and occupational exposures. By contrast, the gender difference in locomotor disability is augmented after adjustment for these variables. Making allowance for the effects of adverse occupational exposure increases the gender difference as these exposures are themselves strongly associated with disability but are much more common among men than women.

While locomotor disability towards the end of their working lives is better explained by occupational exposures among men than women, it is probable that adverse non-occupational factors (for example, obesity) play a part in explaining gender differences in locomotor disability. These gender differences in the explanatory power of occupational exposures may reflect the kinds of jobs available to men and women in the second half of the 20th century.