Simplifying a Complex Diagnosis-Related Group Classification: The French Case

Context: The French Diagnosis-Related Group (DRG) classification has become a complex system of 2629 groups. The challenge lies in determining the appropriate number of DRGs to balance classification precision with administrative feasibility. Objectives: This paper investigates a complementary and largely unexplored question: whether the observed distribution of hospital stays justifies the current level of DRG classification granularity in France. To address this question, we propose a novel approach to optimizing the DRG classification by identifying the minimum number of groups needed to capture the majority of hospital stays. To our knowledge, this is the first study to quantify hospital activity concentration across DRGs in France and examine whether low-use DRGs are systematically associated with patient and institutional characteristics. Research Design: We analyzed PMSI-MCO administrative databases (18.6 million stays, 2009–2022) to derive the cut-off defining the minimum number of DRGs. We estimate logistic regression models with fixed effects to identify determinants of low-use DRG assignment. Subjects: All hospital stays assigned to DRGs in French administrative databases (2629 DRGs, 3698 DRG-fees). Results: Fewer than 500 DRGs (19.4%) code 84.1% of hospital activity—a substantial reduction from 2629 DRGs. Rare DRGs disproportionately include very severe cases (p < 0.001) and are concentrated in university hospitals (p = 0.002). Conclusions: The DRG classification should be simplified. For rare, high-cost cases, T2A should be supplemented with fee-for-service reimbursement. This study provides the first evidence-based framework for DRG simplification in France.