Verified National Cholera Surveillance Dataset for Nigeria, 2021-2025

This dataset contains verified, state-level cholera surveillance records for Nigeria from 2021 through to 2025, extracted from the public Cholera Situation Report archive published by the Nigeria Centre for Disease Control and Prevention (NCDC). Of the 93 published reports retrieved from the national archive, 84 (90.3%) were parsed using a pipeline combining direct PDF text extraction with an optical character recognition (OCR) fallback to resolve documents where internal text layers decoupled state names from numeric values. Every extracted record underwent an arithmetic integrity check requiring reported deaths and suspected cases to reconcile with the published case-fatality ratio to within 0.15 percentage points prior to inclusion. Records failing this threshold were withheld rather than corrected by assumption. The dataset contains 1,233 verified state-level cumulative records spanning all 36 states and the Federal Capital Territory. Within this series, 1,153 records (93.5%) passed a within-year monotonicity check, confirming internal temporal consistency. National annual totals were independently verified against NCDC primary benchmark reports (such as Epi Week 52, 2021). Files included: final_state_cholera_dataset_v2.csv : The verified dataset comprising 1,233 state-level cumulative records. README_final_dataset_v2.md : Full data dictionary, column definitions, data provenance, and documented limitations. data_pipeline_log.md : Complete methodology log detailing all identified PDF failure modes, diagnostic steps, and algorithmic corrections. needs_manual_review_v2.csv : Supplementary dataset containing 210 records withheld due to unreconciled arithmetic discrepancies or structural ambiguities, preserving raw source strings for independent review. This repository accompanies a methodology manuscript describing the extraction and validation pipeline in full (in preparation). This Zenodo record will be updated with the formal journal citation upon publication.