Risk factors & evidence

How individual patient factors change risk for a given operation

Outcomes Studies
How to read this page▸
  • Each finding is one study's result in its own cohort. Findings are not a patient-specific prediction and do not change the STAT 2020 values.
  • Forest plot: each finding with a hazard, odds or risk ratio and its 95% CI, grouped by outcome, on a log scale. A finding without a printed 95% CI is drawn as a point only. Findings given per unit (per 1 kg, per 1 mm Hg) or without a stated comparison group are listed below the plot only. Rows under one heading can cover different time windows and populations. A pooled estimate appears only where 2 or more independent cohorts report an adjusted hazard ratio, or an adjusted odds ratio (never mixed), for a comparable exposure and outcome; the differences between them are listed under each pooled estimate. HeartCalc calculates it (DerSimonian-Laird random effects on the log scale). It is not a published meta-analysis, and with few studies its interval is likely too narrow. Strongest associations ranks this operation's adjusted findings within each outcome by the end of the 95% CI nearest 1: higher means a clearer association in that study, not a bigger effect for an individual child. A pooled estimate stands in for its studies; findings whose CI includes 1 are folded below each list; race, ethnicity, insurance and neighbourhood findings are ranked separately under Social context. Left out: per-unit findings, models adjusted for year only or with covariates not stated, cohorts that include other operations. Each finding's adjustment is graded full or limited (2 or fewer other variables, or only demographic and administrative ones). Study cards are folded: select a study name in the plot or in the "Not in the plot" list to open its card, or Show all cards.
  • Direction: ↑Higher risk the exposed group did worse; ≈No clear difference the study found none; ?Unclear numbers too small or not tested; ↓Lower risk the exposed group did better.
  • neonates · any operation and any operation mark findings from cohorts of neonatal, or all, congenital heart surgery rather than this operation. Neonatal findings are shown with operations usually done in the newborn period; for other operations, and for findings from all congenital heart surgery (for example the STS and PC4 risk-model covariates), select + Any operation.
  • Rates are shown as % (events/n). small numbers marks groups under 20 patients.
  • unadjusted results are crude comparisons; adjusted results account for the covariates the study chose.
  • Within a factor, studies of this operation come first, then neonatal-surgery cohorts; each set is ordered by design (registries, then single-centre series, then case series).
  • Every number from a study is checked against the article's full text before it is published here; pooled estimates are calculated from those numbers.
  • AI-generated summary Behind the speech-bubble icon next to a heading. These short summaries are written by an AI model (Claude) from the published findings shown, one for each operation and each factor, and are rewritten automatically whenever those findings change. Every number in a summary is checked by software against the findings before it appears. They describe the studies; they are not pooled estimates or advice.