Monitoring a primary health care program with lot quality assurance sampling: Costa Rica, 1987

Abstract

The Lot Quality Assurance Sampling (LOAS) method was developed in industry to determine at reasonable cost whether each lot of a shipment is acceptable, based on samples of a few elements from the lot. Acting as inspectors of quality control, the field supervisors of the Costa Rican primary health care programme used the LOAS technique to assess performance in all of the programme's 758 delivery points (lots). They selected probability samples of 20 households and classified the lot as unacceptable when the sample contained more than four unserved households. This 20-4 LOAS rule was aimed to identify lots with less than 70% coverage. Forty-three percent of lots were found unacceptable in their home visit schedule, and 25% unacceptable in vaccination coverage. The probability of accepting defective lots, or consumer risk, was estimated at 4%, and the probability of rejecting acceptable lots, or provider rik, was at 17%. As side results, it was found that 84% of children aged 1–2 years were fully vaccinated, and an estimated 58% of households had been visited in the last six months. A household survey showed that clinic records, which serve as both a sampling frame and source of information for the LOAS assessment, produce accurate estimates of vaccination coverage but contain biased information about home visit dates. Given the chronic lack of timely and accurate information from service statistics, and the high costs of conventional sampling surveys, the LOAS technique appears to be a cost-efficient alternative for monitoring delivery points of primary health care in some circumstances.