Quality assurance in anti-tuberculosis drug procurement by the Stop TB Partnership-Global Drug Facility: Procedures, costs, time requirements, and comparison of assay and dissolution results by manufacturers and by external analysis.

ABSTRACT

BACKGROUND: Quality-assured medicines are a principal means of achieving health-related Sustainable Development Goals. An example of quality assurance/quality control (QA/QC) procedures in drug procurement is provided by the operation of the Global Drug Facility (GDF) of the Stop TB Partnership, the largest provider of tuberculosis (TB) medicines to the public sector worldwide. METHODS: Procedures and results of GDF's quality assurance/quality control (QA/QC) over the five-year period 2013-2017 were analysed retrospectively. 13,999 batches of 51 different medicines had been procured and reviewed within this period. 1,388 of these batches had been analysed in the laboratories of GDF's external quality control agent (QCA). Assay and dissolution results determined by the manufacturers and by the external QCA were compared using Bland-Altman analysis. RESULTS: All investigated batches of medicines were in specifications at the time of shipment. The costs for QA/QC were 0.8% of purchase costs. The median time required for chemical analysis was 10 working days. Comparison of the medicine quality analysis results showed for the poorly water-soluble compound rifampicin a bias of 4.4%, with the manufacturers reporting higher values than the external QCA, most likely due to different methods employed for the analysis. Overall 95% limits of agreement (LOAs) were -6.7 to +8.0% for assay, and -10.1 to +11.8% for dissolution. In case of kanamycin injections, 95% LOAs for assay reached -14.5 to +13.2%, largely attributable to samples from one manufacturer who had used a microbiological assay while the external QCA had used an HPLC assay. CONCLUSIONS: GDF's procedures represent a useful benchmark when evaluating QA/QC procedures of other medicine procurement operations. Inter-laboratory comparison using Bland-Altman plots allows to investigate bias and variability in medicine quality control and should be considered as a routine procedure by drug procurement agencies, to identify priorities for further improvements.