Logo Lanfrica

bwelson/NHIS-ENROLLMENT-GAP-ANALYSIS

Domaine:

healthcare

Type de record:

dataset
Créateur:
bwe
Hôte:
Survey-weighted analysis of the mismatch between self-reported and card-verified NHIS enrollment in Ghana. Identifying and discussing its implications on health insurance policies and previous published works # Measuring the Unmeasured: Self-Reported vs Card-Verified NHIS Enrollment in Ghana Survey-weighted analysis of the gap between self-reported and card-verified National Health Insurance Scheme (NHIS) enrollment in Ghana, using the 2022 Ghana Demographic and Health Survey (GDHS). ## Research purpose Ghana's National Health Insurance Scheme (NHIS) is central to the country's health financing strategy, but coverage statistics are almost always based on self-report. The 2022 GDHS is unusual in that enumerators also recorded whether they physically verified a respondent's NHIS card. This project uses that feature to ask: 1. How large is the gap between self-reported and card-verified NHIS enrollment, and how consistent is it across sociodemographic subgroups and regions? 2. Do the same predictors (wealth, education, residence, religion, pregnancy) show different associations depending on which outcome definition is used? 3. What characterizes people who self-report coverage but cannot produce a valid card ("gap members")? Two outcomes are modeled in parallel: - **Card-verified enrollment** (enumerator-confirmed) - **Self-reported enrollment** (respondent-declared) plus a third model of **gap membership** (self-reported but not card-verified). Analysis uses standard survey-weighted logistic regression (`survey`/`srvyr` packages) — no Bayesian/hierarchical modeling (that is a separate companion paper). ## Data - Ghana DHS 2022, Women's (IR) and Men's (MR) recode files (`.DTA` format) - Not included in this repository due to DHS data-use restrictions. Access requires registration at dhsprogram.com. ## Repository contents - `analysis code.R` — full analysis script: data cleaning, survey design specification, descriptive statistics, subgroup enrollment gaps, regional variation, logistic regression models, and all figures/tables. - `fig.png` — forest plot comparing adjusted odds ratios for NHIS enrollment under the card-verified (M1) vs self-reported (M2) outcome def …