Drug Resistance Index Correlates with Antimicrobial Resistance and Declines Following Stewardship Implementation: A Multicentre Analysis of 20 Indian Hospitals (2016–2023)

Abstract

Background: Antimicrobial resistance (AMR) is a growing threat in India, where fragmented pharmacy, microbiology, and clinical data systems limit real-time stewardship. The Drug Resistance Index (DRI), a composite measure of antibiotic use weighted by resistance, is validated internationally but rarely applied across hospital networks in India.
Methods: We conducted a retrospective, multicentre, quasi-experimental before-after study across 20 Fortis Healthcare hospitals in India, analysing 133 site-year observations of institutional DRI and antimicrobial resistance data spanning 2016–2023. DRI and resistance rates were compared between a pre-intervention period (2016–2018) and a post-intervention period (2019–2023), during which a stewardship programme was implemented, comprising DRI-informed prescribing feedback, audit-and-feedback review, formulary controls, and clinician education. Associations were assessed using Pearson correlation and linear regression, and pre-post change was assessed using paired comparisons among the hospitals contributing data to both periods.
Results: Mean DRI declined from 0.524 (pre-intervention) to 0.498 (post-intervention), and mean resistance declined from 46.98% to 41.42%. DRI and resistance were significantly and positively correlated at the site-year level (r = 0.732, p < 0.001; R² = 0.536), consistent with the relationship reported in prior DRI validation studies. This association held across two additional DRI constructions computed from the same data: the organism-matched (Gram-negative bacilli) sub-index showed the strongest correlation (r = 0.774, R² = 0.599), while the restricted-antibiotic sub-index, only moderately related to the primary measure, provided independent confirmation (r = 0.559). Among the 15 hospitals contributing data to both periods, the decline in resistance following stewardship implementation was statistically significant (paired t-test, p = 0.009), while the corresponding decline in DRI did not reach significance (p = 0.444).
Conclusion: This multicentre analysis supports the Drug Resistance Index as a valid, positively correlated surveillance metric for antimicrobial resistance in Indian hospitals, and demonstrates a significant, network-wide decline in resistance following stewardship implementation. These findings support embedding DRI into routine antimicrobial stewardship dashboards, though clustering-adjusted, prospective validation is recommended before broader adoption.