FGCIFPLOT: Stata module to draw Fine-Gray cumulative-incidence plots with conventional and subdistribution risk-set summaries

fgcifplot draws model-based cumulative-incidence curves after Fine-Gray competing-risks regression and can place one or more risk-set summaries below the curve. The command distinguishes three different quantities that are often conflated in competing-risks figures: Conventional N at risk: subjects whose observed follow-up interval still contains time t. Subjects leave after the event of interest, a competing event, or censoring. FG retained N: the conventional risk set plus subjects who experienced a competing event before t. Each retained competing-event subject counts as one. This is the unweighted head count of the extended Fine-Gray risk set. FG weighted total: the conventional risk set plus retained competing-event subjects weighted by the inverse-probability-of-censoring weight used by stcrreg. For a subject j with a prior competing event, the contribution at time t is the censoring-survival function evaluated at t divided by that function evaluated at the subject's competing-event time. This quantity may be noninteger. It is a weighted risk-set total, not a survey-style effective sample size. The cumulative-incidence curves are adjusted model predictions produced by stcurve. Covariates not explicitly set in at() follow stcurve's usual postestimation defaults. In contrast, the risk-set summaries are unadjusted observed quantities calculated within the requested groups; they are not model-adjusted predictions.