Pilot flight experience and accident severity in Brazilian general aviation: a register-based analysis of CENIPA final reports (2007– 2024)

General aviation is used in a broad operational sense; narrower operational definitions are examined in sensitivity analyses. The package includes regression coefficients, confidence intervals, sensitivity analyses, multiple-imputation summaries, missingness statistics, annual report-linkage and inclusion counts, calibration summaries, and model-derived probability curves. It also provides statistical code, variable and output dictionaries, descriptive metadata, file checksums, and instructions for regenerating figures.

The primary complete-case model includes 764 records and 192 fatal outcomes. The adjusted odds ratio per approximate doubling of total flight hours is 0.93 (95% CI 0.85–1.01; p=0.100). These observational results do not establish a causal or statistically conclusive inverse association.

To minimise indirect-identification and redistribution concerns, the package excludes record-level analytical data, aircraft registrations, occurrence identifiers, individual predictions, report narratives, original CENIPA PDFs, and third-party reference PDFs. The deposited summaries support inspection of the estimates and regeneration of six figures. Independent model refitting and reconstruction of the analytical cohort require additional record-level inputs that are not included.

Packaging and automated checks do not replace human validation of source extraction and linkage.