Comparing the Effect of Ranibizumab vs. Aflibercept on Visual Acuity Outcomes for Neovascular AMD

Code for: CPH200B Project — Advaith Veturi, Sami Barrit, Marcus Moen

Overview

Causal inference analysis of anti-VEGF treatments for neovascular age-related macular degeneration, using the Moorfields Eye Hospital longitudinal cohort (Fu et al., JAMA Ophthalmology, 2021).

For now, this repository only implements a (very weak) fuzzy regression discontinuity design exploiting the October 2013 introduction of aflibercept at MEH as a natural experiment, with the treatment era as an instrumental variable for drug assignment.

Requirements

• Python 3.9
• Dependencies: pip install -r requirements.txt

Data

The analysis uses the MEH AMD Survival Outcomes Database, publicly available on Dryad:

Fu DJ et al. (2020). Insights from survival analyses during 12 years of anti-VEGF therapy for neovascular age-related macular degeneration. Dryad. https://doi.org/10.5061/dryad.nvx0k6drg

Download MEH_AMD_survivaloutcomes_database.csv and place it at input/MEH_AMD_survivaloutcomes_database.csv.

Reproduce

# 1. Preprocess raw data
python3 code/scripts/run_preprocess.py

# 2. Run fuzzy RDD analysis
python3 code/scripts/run_rdd.py

Results are written to output/tables/ (summary CSVs) and output/results/ (full datasets, not tracked in git).

The technical report is at output/RDD.md.

Repository Structure

alaa_3/
├── code/
│   ├── scripts/
│   │   ├── run_preprocess.py   # Data cleaning and curation
│   │   └── run_rdd.py          # Fuzzy RDD / IV analysis
│   └── plot/                   # (plotting scripts)
├── output/
│   ├── RDD.md                  # Technical report
│   └── tables/                 # Summary CSVs
├── requirements.txt
└── README.md

License

MIT