Real-World Evidence — Biologic Therapies in Rheumatoid Arthritis, GCC Registry
Simulated Case Study
This case study uses simulated data to demonstrate the statistical methodology, analysis workflow, and reporting standards we apply to real client projects. No actual patient or institutional data is represented.
PSM · IPTW · Cox Regression · Kaplan-Meier
Confidentiality Notice
All data presented in this case study is simulated and replicates the statistical structure of the original registry dataset. The CRO and hospital names are anonymized. No identifiable patient data is disclosed.
Study design
Multicenter retrospective cohort
Sample
892 RA patients
Sites
6 GCC hospitals
Median follow-up
36 months
Primary endpoint
Time to EULAR response
Bias control
PSM + IPTW (SMD < 0.10)
Project Overview
A Gulf-based CRO engaged Naggar Analytics to conduct a retrospective cohort study using registry claims data from 6 hospitals across the UAE, Saudi Arabia, and Kuwait. The study compared three biologic therapy classes — TNF inhibitors, IL-6 inhibitors, and JAK inhibitors — in 892 adult RA patients who initiated biologic therapy between 2020 and 2022.
Causal inference methods (propensity score matching and IPTW) were used to control for indication bias across treatment groups. The primary endpoint was time to EULAR moderate/good response at 6 months, analyzed via Kaplan-Meier curves and Cox proportional hazards regression. All analyses were pre-registered and conducted following STROBE reporting standards.
Analytical Methods
- ▹Cohort identification — ICD-10 coded RA diagnosis + biologic prescription from GCC registry claims
- ▹Propensity Score Matching (PSM) — 1:1 nearest-neighbor (MatchIt); 12 baseline covariates balanced
- ▹IPTW (stabilized weights) — covariate balance verified via SMD < 0.10 for all variables
- ▹Primary endpoint: time to EULAR moderate/good response — Kaplan-Meier + log-rank test
- ▹Cox proportional hazards — adjusted HR + 95% CI; proportionality assessed via Schoenfeld residuals
- ▹Subgroup analyses — stratified by biologic class (TNFi, IL-6i, JAKi), baseline DAS28, CCI score
- ▹Missing data — multiple imputation (m = 20 datasets, Rubin's rules)
- ▹Sensitivity analysis — on-treatment, ITT, and per-protocol populations
- ▹STROBE reporting checklist; all analyses pre-registered on ResearchRegistry.com