A randomised survey of the quality of antibiotics and other essential medicines in Indonesia, with sales volume-adjusted estimates of the prevalence of substandard medicines

Abstract

AbstractIntroductionMost substandard and falsified medicines are at best not optimally effective, and at worst fatal. While the World Health Organisation and others warn they are a major threat to public health in low and middle income countries, little is known about their true prevalence. Authors of meta-analyses universally warn that survey data are not generalisable, because of unrepresentative study designs and variations in medicines included; tests performed; reference standards and pharmacopeia used; and definitions used when translating multiple quality parameters into a single pass/fail measure.We hypothesised that weighting for sales volume of different products and brands would increase accuracy of estimates of medicine quality.Methods and FindingsWe collected samples of allopurinol, amlodipine, cefixime and dexamethasone, as well as amoxicillin in 2 formulations in seven districts across Indonesia, the world’s fourth most populous country. Outlets, including retail pharmacies, over the counter medicine shops, public and private hospitals, primary health centres, doctors and nurses were randomised. We also sampled from the internet. Retail samples were collected by mystery shoppers, other samples overtly.We tested 1274 samples for identity and assay, and all relevant samples for dissolution and uniformity of content, using USP reference standards and monographs. Samples that failed any laboratory test were considered out of specification. We calculated prevalence per product and brand, and weighted the results by the sales volume of each product, using sales data from IQVIA and the national public procurement system.The weighted prevalence of out-of-specification products was 4.9%, 40.3% lower than the raw estimate (8.2%). Antibiotics were more likely to be substandard (weighted prevalence 6.8 vs 3.1; raw prevalence 13.6 vs 4.9, p<0.000). There was no relationship between quality and any of the following: price; branded status; public procurement status; outlet type. Our estimate compared with the Indonesian medicine regulator’s estimate of 4.0%, calculated based on unweighted analysis of 13,539 samples of a wider variety of medicines, collected overtly nation-wide.ConclusionsWhere data are available, weighting survey results by sales volume is a cost-effective way of improving the accuracy of estimates of out-of-specification medicines measured in field surveys.