The manuscript contains a minor error regarding the coding of one of the covariates. One of the levels of the smoking variable (never smoked) was coded as missing data, resulting in a loss in degrees of freedom for the regression models of the primary and secondary outcomes (~35%). When the smoking variable is replaced with a proper variable that correctly assigns the participants, the primary outcome (frailty) results are nearly identical, only the 95% confidence intervals for many of the contrasts are narrower and one odds ratio estimate (hysterectomy classification for odds of being frail using Fried’s criteria) is more significant. For the secondary outcomes (HRT-related variables), similar minor adjustments in the estimates and confidence intervals are apparent and the age of HRT onset was found to be significantly associated to the frailty index.
None of the conclusions drawn in the study are affected by this error.