kaxixi/Keheala_RCTII

Replication data and code for a cluster-RCT evaluating digital adherence technologies for TB treatment in Kenya # Keheala Study 2 — Data Analysis in Python Data and analysis code for a cluster-randomized controlled trial (RCT) evaluating digital adherence technologies for tuberculosis (TB) treatment in Kenya, conducted in partnership with USAID. The study randomized patients across four arms — Control, SMS Reminders, a digital platform (SBCC), and Keheala — and measured treatment outcomes including unsuccessful outcomes, loss to follow-up (LTFU), and death. ## Quick Start **Requirements:** Python 3 with `pandas`, `numpy`, and `statsmodels`. ```bash pip install pandas numpy statsmodels ``` **Run the analysis** (scripts can be run in any order): ```bash python3 Python_Analysis/analysis_manuscript.py # Tables 1–4 python3 Python_Analysis/analysis_si.py # Supplementary Tables SI5–SI13 python3 Python_Analysis/analysis_dqa.py # Data Quality Assessment (SI14) python3 Python_Analysis/generate_consort.py # CONSORT flow diagram numbers ``` Output: LaTeX `.tex` files are written to `Python_Analysis/output/`. ## Repository Structure ``` Keheala_RCTII/ ├── README.md ├── deidentified_data/level2/ # De-identified datasets for analysis │ ├── study2_cleaned.csv # Study data (N=17,160 rows; mITT N=14,962) │ ├── TIBU_firstnm_deidentified.csv # Kenya national TB registry │ ├── Urine_Test_Results.csv # Urine verification data │ └── DQA_combined.csv # Data quality assessment records ├── Python_Analysis/ │ ├── analysis_manuscript.py # Tables 1–4 (main manuscript) │ ├── analysis_si.py # Supplementary Tables SI5–SI13 │ ├── analysis_dqa.py # Data Quality Assessment (SI14) │ ├── generate_consort.py # CONSORT flow diagram │ ├── prepare_study_data.py # Data cleaning [reference only] │ ├── prepare_TIBU_data.py # TIBU aggregation [reference only] │ ├── deidentify_data.py …