Antimicrobial Stewardship in Pluralistic Health Systems: Why Local Evidence Must Guide Implementation
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Antimicrobial resistance (AMR) strategies frequently emphasize provider knowledge and public awareness, but prescribing behavior can also reflect practice routines, information sources, and facility context. We examined these relationships in a pluralistic primary-care system in three districts of Chhattisgarh, India.
Methods
We analysed three independently sampled cross-sectional components: a provider knowledge, attitudes, and practices (KAP) survey (n=171), a WHO/International Network for Rational Use of Drugs (WHO/INRUD) prescribing assessment (600 encounters from 20 government facilities), and 371 patient exit interviews. For providers, proportional-odds logistic regression modelled ordered cold/sore-throat antibiotic provision (none, some, most, all), adjusting for provider cadre, work setting, and years in practice; binary any-versus-none models were sensitivity analyses. We summarized WHO/INRUD indicators overall and by facility, and examined exploratory encounter-level antibiotic correlates with unadjusted and facility-adjusted models. We analysed patient satisfaction using multivariable binary logistic regression and perceived clarity using proportional-odds logistic regression.
Results
Provider AMR awareness was 91.4% (149/163), 95.8% (160/167) agreed that antibiotics were overused nationally, and 58.3% (95/163) reported guideline familiarity. Guideline familiarity was not associated with lower cold/sore-throat antibiotic frequency (adjusted common OR 1.13, 95% CI 0.57-2.24; p=0.725). Each additional 5 years in practice was associated with greater odds of being in a higher prescribing-frequency category (adjusted common OR 1.36, 95% CI 1.09-1.70; p=0.006 before multiplicity correction). Medical-representative literature was a routine information source for 39.6% (65/164); its association with prescribing frequency was directionally positive but imprecise (adjusted common OR 1.89, 95% CI 0.88-4.07; p=0.104). In government facilities, 54.8% (329/600) encounters included an antibiotic, ranging from 16.7% to 80.0% by facility. Antibiotic-containing encounters had more drugs (2.88 vs 2.47; p<0.001), lower generic proportions (73.6% vs 80.7%; p=0.003), lower Essential Drugs List concordance (48.9% vs 63.8%; p<0.001), and more co-prescribed injections (15.2% vs 7.7%; p=0.007); several associations attenuated after accounting for facility. In exit interviews, 92.3% (131/142) of informal-provider encounters involved a reported injection, and informal providers accounted for 72.4% (131/181) of all reported injections. Clear/very clear instructions were strongly associated with satisfaction (adjusted OR 13.66, 95% CI 6.73-27.71; p<0.001; n=369).
Conclusions
In these three districts, high provider AMR awareness coexisted with prescribing patterns not explained by awareness or guideline familiarity alone. The findings support prospective evaluation of point-of-care decision support, independent drug information, facility audit and feedback, targeted informal-provider engagement, and communication-focused stewardship. Generalization beyond the study districts requires caution.
