Lot sampling plans in the measure of quality of care indicators

Application of the lot sampling technique to measure indicators of quality of care on screening activities directed toward women: detection of risk factor for cardiovascular diseases, screening for breast cancer and for cervical cancer.

The sampling plan adopted involved, in each centre, a sample size of 15 and a maximum of four non-compliers that the random sample would tolerate. The binomial distribution was used. This 15-4 sampling plan sets the probability of concluding that an indicator is performing well for a lot or population that has an 80% compliance level at 0.84, keeping the alpha error or provider's risk at 0.16, while those for a population that has a 50% compliance is set at only 0.059, keeping the beta error or consumer's risk at 0.059.

Ten primary care centres of the Catalan Health Institute.

Random sample of all women aged 40-65 years enrolled in each centre.

For indicators of risk factors for cardiovascular disease detection, six out of 10 centres performed well for hypertension and obesity, and five out of 10 centres performed well for hypercholesterolaemia. However, no centre was acceptable either in cervical cancer or breast cancer screening.

Rather than seeking to obtain precise estimates, this technique aims to facilitate the decision-making process regarding the quality levels of the indicators examined. Despite some limitations, the technique is a good test for detecting gross departures from stated compliance thresholds. It has several advantages for application in health services: (i) the use of small sample sizes; (ii) the fact that it can detect deficiencies in small areas; and (iii) the simplicity of applying this technique by relying on ready-made tables.