LQAS

What Is LQAS?

Lot Quality Assurance Sampling (LQAS) is a classification method used to prioritize areas, facilities, or programs for action. Its design specifies lower and upper performance levels, acceptable errors at both levels, a sample size, and a decision rule. Small samples may support a useful classification rule but do not provide a precise area coverage estimate.

LQAS originated in industrial quality control and was subsequently adapted for health program monitoring. It has become particularly popular in maternal and child health programs, immunization monitoring, and nutrition surveillance where programs need to quickly identify which areas need intervention.

Why Does LQAS Matter?

LQAS addresses a critical gap in program monitoring: the need for rapid, actionable decisions without the resource burden of precise estimation. Traditional survey methods require large samples to estimate coverage rates with narrow confidence intervals, but programs often don't need precision - they need to know whether an area is above or below a threshold that triggers action.

LQAS can reduce fieldwork when a classification decision is sufficient. The time and sample size depend on the separation of the two performance levels, acceptable errors, and sampling design; there is no universal efficiency ratio.

A pass/fail label must be communicated with its design thresholds and decision errors. Between the lower and upper performance levels lies an indifference region where neither error guarantee applies. Passing does not certify that the true prevalence exceeds a single threshold.

How Is LQAS Used in Practice?

Consider an illustrative coverage design with 19 independently sampled eligible individuals and a pass rule of at least 15 vaccinated individuals. Take 50% as the illustrative lower level and 90% as the upper level. The risk of passing a lot at the lower level is about 0.0096; the risk of failing a lot at the upper level is about 0.0352. Those meet illustrative maximum error tolerances of 1% and 5%, respectively, under this model. At 80%, the pass probability is about 0.6733, so the same rule cannot certify coverage above 80%. These are binomial operating characteristics for this illustration, not a validated immunization protocol. Define eligible individuals and household/within-household selection explicitly; clustering or unequal probabilities can change the calculations.

Before fieldwork, specify the lower and upper performance levels and acceptable errors, then select a sample size and decision rule that meet them. In the illustration, a fail label triggers additional investigation or support; a pass label supports routine follow-up with the stated residual risk. Check error probabilities at both design levels and inspect behavior throughout the indifference region. Use an appropriate sampling model rather than assuming binomial independence.

Programs using LQAS typically collect data through rapid household surveys or facility assessments, analyze results using decision tables, and produce maps or dashboards showing which areas pass or fail. Results inform resource allocation - failing areas receive additional support, training, or resources. The method can be repeated quarterly or semi-annually to track whether interventions are moving areas from fail to pass status.

Method Source

WHO, World Health Statistics Quarterly 44 (1991), pp. 171-181, selection of sample size and performance levels. This is methodological guidance, not a funding requirement.