ALERT — Action Leveraging Evidence to Reduce perinatal mortality and morbidity: Stata analysis syntax for a stepped-wedge cluster-randomised trial in 16 hospitals across Benin, Malawi, Tanzania and Uganda.
Project website:
alert.ki.se
This release archives the complete do-file syntax used for the main results paper, covering the full pipeline from raw-data cleaning through outcome derivation, per-hospital modelling, and meta-analysis.
Contents
0_ALERT_pipeline_master.do — master runner documenting and chaining the stages.
Data cleaning/ — preparation pipeline (0a → 0b → 1a → 1a.i → 1b): prepare Benin data, append the four countries, clean per woman, clean baby morbidity, and reshape wide → long.
Main analyses/ — outcome derivation, descriptive tables, per-hospital mixed-effects models (melogit/mixed/xtgee), and random-effects meta-analysis with metan (one forest plot per outcome).
README.md, LICENSE (MIT), Authors.docx.
Outcomes analysed
Fresh stillbirths, early perinatal mortality, low Apgar (<7), neonatal seizures, caesarean section, severe maternal morbidity, fetal-heart and fetal-monitoring process indicators, resuscitation, companionship, early ambulation, breastfeeding, and the responsiveness and respectful-care scores from exit interviews — with per-protocol and hospital-exclusion sensitivity analyses.
Notes
Syntax only — no individual-level participant data are included, for ethical and data-protection reasons.
Requires Stata (≥17 recommended) with the user-written metan and table1 commands and the classify_sga program.
The main analysis is run section by section (single-letter x/xx/XXXX lines are deliberate stop markers). See README.md for run instructions and reproducibility notes.
Citation
Akuze, J., Alsina, M. del R., Straneo, M., Sun, S., Annerstedt, K., Waiswa, P., ALERT study team, & Hanson, C. (2026). ALERT stepped-wedge trial in Benin, Malawi, Tanzania and Uganda — full Stata pipeline (data cleaning, analysis and meta-analysis) (v1.0.0). Zenodo.
doi.org