Table 2.

Causal effect estimates of poverty status on likelihood of common mental disorder (with 95% confidence intervals), stratified by gender, highest education and age

	Sample size	Odds ratio	Absolute % diff.	Prev. in unexposed	PAF
Men	20 891 (90 369 obs)	1.16 (1.07, 1.26)	1.73% (0.72, 2.74)	17.15% (16.75, 17.54)	4.64% (2.23, 7.05)
Women	24 634 (111 928 obs)	1.17 (1.10, 1.25)	2.34% (1.41, 3.26)	23.25% (22.85, 23.65)	4.62% (2.87, 6.36)
High education	19 987 (82 080 obs)	1.18 (1.07, 1.30)	2.13% (0.85, 3.42)	18.53% (18.16, 18.91)	5.25% (2.22, 8.28)
Medium education	19 576 (75 559 obs)	1.18 (1.09, 1.28)	2.14% (1.06, 3.22)	20.39% (19.91, 20.86)	4.84% (2.70, 6.99)
Low education	12 636 (44 658 obs)	1.17 (1.07, 1.28)	2.20% (0.90, 3.49)	24.11% (23.38, 24.83)	4.23% (1.96, 6.51)
Younger working-age	20 384 (71 777 obs)	1.16 (1.07, 1.25)	2.05% (0.96, 3.14)	20.52% (19.97, 21.08)	4.59% (2.27, 6.91)
Older working-age	29 711 (130 520 obs)	1.18 (1.10, 1.26)	2.14% (1.25, 3.03)	20.68% (20.34, 21.01)	4.75% (2.94, 6.56)

Adjusted for both tim- invariant (gender, education, ethnicity) and time-varying confounders [age, age squared, employment status (current and 1 year lagged), benefit status (lagged), home ownership status (lagged), marital status (lagged), number of children (lagged), government office region (lagged), 12-item Short Form survey (SF-12) physical health component (lagged), SF-12 mental health component (lagged), and previous indication of common mental disorder (1 year lagged GHQ-12 caseness)]. Absolute risk difference indicates %-point change.

Obs, observations; Prev., prevalence; diff., difference; PAF, population attributable fraction; GHQ-12, General Health Questionnaire.