Results of survival models
The covariates included in the final model were age at first HRT
prescription, birth cohort, HRT type, hypertension and its treatment,
CHD, type 2 diabetes, oophorectomy or hysterectomy, BMI, smoking, and
deprivation. All significant variables in the full model were also
significant in all age subgroup models. There was no significant
interaction of HRT with other variables, which means that the survival
effect of HRT on the hazards of all-cause mortality were the same across
different subgroups.
The adjusted hazard ratios for all-cause mortality associated with HRT
were time-invariant. Overall, the hazard of death was lower in combined
HRT users compared to non-users and there were no significant increased
or decreased hazards associated with oestrogen-only HRT (Figure 2). The
HR for combined HRT was 0.91 (95% CI, 0.88 - 0.94), and for oestrogen-
only users was 0.99 (0.93 - 1.07). In age subgroups, the HRs in combined
HRT for women who received first treatment at age 46-50, 51-55, 56-60,
and 61-65 were 0.98 (0.92-1.04), 0.87 (0.82 - 0.92), 0.88, (0.82 -
0.93), and 0.92 (0.85 - 0.98), respectively. See Table S4 for all
results.
Oophorectomy and hysterectomy were associated with improved survival
prospects, in which the highest reduction of hazards was in the 61-65
age cohort and lowest in the 46-50 cohort (Figure S2). Both treated and
untreated hypertension increased the hazards of all-cause mortality and
the findings did not differ much in the age subgroup models. Overall,
living in more deprived areas was associated with 42% higher hazard of
death than living in less deprived areas. The interaction of BMI and
smoking also had a considerable impact on women’s survival. The HRs of
all-cause mortality in current smokers compared to non-smokers were
higher in healthy weight and overweight women than in obese women in all
age cohorts.
As birth cohort was time-variant, we calculated the cumulative hazards
for each birth cohort, and found that longevity increased in women born
in the later birth cohorts across all age and HRT type subgroups (Figure
3). Survival prospects also significantly varied by GP practice with the
variance of the frailty term 0.16 (0.14 - 0.19) in the full model.