“…Multivariate hazard ratios (HR) and confidence intervals (CI) for the incidence of ovarian cancer were calculated using an inverse probability weighted (IPW) [19,20] competing risk model [21] in which we treated incidence of cancer other than ovarian cancer and death as a competing events [22]. To develop the generalized propensity score, we conducted a multinomial logistic regression analysis using variables for all demographic information [23], including age (a continuous variable), body mass index (BMI; <18.5, 18.5-24.9, or ≥25.0 kg/m 2 , or missing), educational level (school up to age 15, 15-18, or ≥19 years, or missing), smoking status (never, former, or current smoker, or missing), alcohol drinking status (never, former, or current alcohol drinker, or missing), average daily sleeping time (<6.5, 6.5-8.4, or ≥8.5 hours/day, or missing), average daily walking time (<1, ≥1 hours/day, or missing), age at menarche (<13, 13-15, or ≥16 years, or missing), age at menopause (<44, 44-48, 49-52, or ≥52 years, or having periods/missing), parity (none, 1-2, ≥3 times, or missing), use of hormone therapy ever (never, user, or missing), family history of breast, ovarian, or prostate cancer (yes or no/missing) [24]. To assess the covariate balance, we showed the propensity score overlap using kernel density plots.…”