Prevalence, incidence, risk and odds ratios with confidence intervals from a research dataset (API key)
POST /epidemiology/analyse · API key required
Send a dataset, the study design, an outcome and optionally an exposure and strata. The design decides what is reported. cross_sectional: prevalence with Wilson or exact intervals, prevalence ratio, difference and odds ratio. cohort: cumulative incidence (risk), risk ratio, difference and odds ratio; with a person_time column, incidence rates (exact Poisson interval) and rate ratios; baseline_outcome removes people who already have the outcome. case_control: the odds ratio only; prevalence, risk, incidence and risk ratios are refused. Outcome and exposure are binary rules on a column (positive values, or below / at_or_below / above / at_or_above / between for a numeric cut-off such as haz below -2). Unrecognised or missing values are excluded, counted and shown, never guessed. strata (up to 3, categories or numeric bands) give stratified prevalence, stratum-specific effect measures, Mantel-Haenszel pooled estimates, a Cochran-Mantel-Haenszel test, a heterogeneity test and a crude-versus-adjusted comparison; adjust_for cross-classifies several strata. indicators and means add other binary indicators and group means. design_effect widens intervals for cluster surveys; weights and clustering are not modelled. Up to 5000 records. Nothing is stored. Research aid, not a diagnosis.
Request body
Content type: application/json
Responses
200400429