Application of lot quality assurance sampling for leprosy elimination monitoring--examination of some critical factors.

BACKGROUND: The concept of elimination of an infectious disease is different from eradication and in a way from control as well. In disease elimination programmes the desired reduced level of prevalence is set up as the target to be achieved in a practical time frame. Elimination can be considered in the context of national or regional levels. Prevalence levels depend on occurrence of new cases and thus could remain fluctuating. There are no ready pragmatic methods to monitor the progress of leprosy elimination programmes. We therefore tried to explore newer methods to answer these demands. With the lowering of prevalence of leprosy to the desired level of 1 case per 10000 population at the global level, the programme administrators' concern will be shifted to smaller areas e.g. national and sub-national levels. For monitoring this situation, we earlier observed that lot quality assurance sampling (LQAS), a quality control tool in industry was useful in the initially high endemic areas. However, critical factors such as geographical distribution of cases and adoption of cluster sampling design instead of simple random sampling design deserve attention before LQAS could generally be recommended. The present exercise was aimed at validating applicability of LQAS, and adopting these modifications for monitoring leprosy elimination in Tamil Nadu state, which was highly endemic for leprosy.

METHODS: A representative sample of 64000 people drawn from eight districts of Tamil Nadu state, India, with maximum allowable number of 25 cases was considered, using LQAS methodology to test whether leprosy prevalence was at or below 7 per 10000 population. Expected number of cases for each district was obtained assuming Poisson distribution. Goodness of fit for the observed and expected cases (closeness of the expected number of cases to those observed) was tested through chi(2). Enhancing factor (design effect) for sample size was obtained by computing the intraclass correlation.

RESULTS: The survey actually covered a population of 62157 individuals, of whom 56469 (90.8%) were examined. Ninety-six cases were detected and this number far exceeded the critical value of 25. The number of cases for each district and the number of cases in the entire surveyed area both followed Poisson distribution. The intraclass correlation coefficients were close to zero and the design effect was observed to be close to one.

CONCLUSIONS: Based on the LQAS exercises leprosy prevalence in the state of Tamil Nadu in India was above 7 per 10000. LQAS method using clusters was validated for monitoring leprosy elimination in high endemic areas. Use of cluster sampling makes this method further useful as a rapid assessment procedure. This method needs to be tested for its applicability in moderate and low endemic areas, where the sample size may need increasing. It is further possible to consider LQAS as a monitoring tool for elimination programmes with respect to other disease conditions.