Interpreting mutual adjustment for multiple indicators of socioeconomic position without committing mutual adjustment fallacies

Abstract

Research into the effects of Socioeconomic Position (SEP) on health will sometimes compare effects from multiple, different measures of SEP in “mutually adjusted” regression models. Interpreting each effect estimate from such models equivalently as the “independent” effect of each measure may be misleading, a mutual adjustment (or Table 2) fallacy. We use directed acyclic graphs (DAGs) to explain how interpretation of such models rests on assumptions about the causal relationships between those various SEP measures. We use an example DAG whereby education leads to occupation and both determine income, and explain implications for the interpretation of mutually adjusted coefficients for these three SEP indicators. Under this DAG, the mutually adjusted coefficient for education will represent the direct effect of education, not mediated via occupation or income. The coefficient for occupation represents the direct effect of occupation, not mediated via income, or confounded by education. The coefficient for income represents the effect of income, after adjusting for confounding by education and occupation. Direct comparisons of mutually adjusted coefficients are not comparing like with like. A theoretical understanding of how SEP measures relate to each other can influence conclusions as to which measures of SEP are most important. Additionally, in some situations adjustment for confounding from more distal SEP measures (like education and occupation) may be sufficient to block unmeasured socioeconomic confounding, allowing for greater causal confidence in adjusted effect estimates for more proximal measures of SEP (like income).

Potential for non-combustible nicotine products to reduce socioeconomic inequalities in smoking: a systematic review and synthesis of best available evidence

“…We also note that, in studies using more than one indicator of SES (e.g. education and income), multivariate regression analyses were sometimes adjusted for the other indicator, which may mask the true magnitude of the socioeconomic gradient in product use [112]. This limitation applies only to the lowest tier of our ‘best evidence’ gradient, i.e.…”

Socioeconomic and ethnic differences in children’s vigorous intensity physical activity: a cross-sectional analysis of the UK Millennium Cohort Study

“…Multivariable linear regression models were fitted to analyse differences in absolute mean daily minutes of VPA achieved across socioeconomic and ethnic groups, adjusting for mean daily minutes of MPA, mean accelerometer wear time, season of measurement, age and sex. Separate models were run for each exposure variable (maternal education, equivalised household income, ethnicity) to assess the effects of each seperately 44. Models were also run separately for week and weekend days as there is evidence that children accumulate physical activity differently on weekdays and weekend days 45.…”

Trends in socioeconomic inequalities in premature and avoidable mortality in Canada, 1991–2016

“…We included both indicators as they capture different dimensions of socioeconomic inequality. 2,[28][29][30][31][32][33] We constructed household income by calculating the sum of all self-reported sources of income during the preceding calendar year. Following recent Statistics Canada and Organisation for Economic Co-operation and Development reports, [34][35][36][37][38] we adjusted for household size by dividing total household income by the square root of the number of household members.…”