Figure 3. Associations between early-life risk factors and early menopause.

Results are for the logistic regression model including one early-life risk factor at a time, adjusted for the potential confounding variables Townsend deprivation index, BMI, smoking status, smoking pack-years, frequency of alcohol intake, number of live births, educational level and whether the participant ate meat. Analyses are in all ages, except for year of birth which was analysed in women aged 60 and over at recruitment. Smoking status is included for reference. Confidence intervals are 99.995%. To achieve the same effect size as being a current smoker (the binary variable with nominally the largest effect size), for the continuous variables the required change in risk factor would be: age at menarche, 11.1 years (7.0 s.d.); birth weight 2.8 kg (4.7 s.d.); year of birth, 16.9 years (2.1 s.d.).