Novel insights in the fecal egg count reduction test for monitoring drug efficacy against soil-transmitted helminths in large-scale treatment programs

Abstract

Background: The fecal egg count reduction test (FECRT) is recommended to monitor drug efficacy against soil-transmitted helminths (STHs) in public health. However, the impact of factors inherent to study design (sample size and detection limit of the fecal egg count (FEC) method) and host-parasite interactions (mean baseline FEC and aggregation of FEC across host population) on the reliability of FECRT is poorly understood.

Methodology/principal findings: A simulation study was performed in which FECRT was assessed under varying conditions of the aforementioned factors. Classification trees were built to explore critical values for these factors required to obtain conclusive FECRT results. The outcome of this analysis was subsequently validated on five efficacy trials across Africa, Asia, and Latin America. Unsatisfactory (<85.0%) sensitivity and specificity results to detect reduced efficacy were found if sample sizes were small (<10) or if sample sizes were moderate (10-49) combined with highly aggregated FEC (k<0.25). FECRT remained inconclusive under any evaluated condition for drug efficacies ranging from 87.5% to 92.5% for a reduced-efficacy-threshold of 90% and from 92.5% to 97.5% for a threshold of 95%. The most discriminatory study design required 200 subjects independent of STH status (including subjects who are not excreting eggs). For this sample size, the detection limit of the FEC method and the level of aggregation of the FEC did not affect the interpretation of the FECRT. Only for a threshold of 90%, mean baseline FEC <150 eggs per gram of stool led to a reduced discriminatory power.

Conclusions/significance: This study confirms that the interpretation of FECRT is affected by a complex interplay of factors inherent to both study design and host-parasite interactions. The results also highlight that revision of the current World Health Organization guidelines to monitor drug efficacy is indicated. We, therefore, propose novel guidelines to support future monitoring programs.

Figure 1. The classification tree of the factors affecting FECRT specificity (TDE ≥90%).

The classification tree of the factors affecting FECRT specificity (%) (correct detection of a ‘true’ drug efficacy (TDE) ≥90%); factors included mean fecal egg count (FEC) before administration of drugs (pre-DA FEC), aggregation of FEC (k), sample size, detection limit, and TDE. N = number of combinations.

Figure 2. The classification tree of the factors affecting FECRT sensitivity (TDE <90%).

The classification tree of the factors affecting FECRT sensitivity (%) (correct detection of a reduced efficacy when ‘true’ drug efficacy (TDE) <90%); factors included mean fecal egg count (FEC) before administration of drugs (pre-DA FEC), aggregation of FEC (k), sample size, detection limit, and TDE. N = number of combinations.

Figure 3. The detection of reduced efficacy when defined as ‘true’ drug efficacy <90% and <95%.

Only combinations of sample size (n), mean fecal egg count before drug administration (mean pre-DA FEC), detection limit and aggregation (k), which resulted in a reliable FECRT (sensitivity (Se) and specificity (Sp) ≥85%) are shown. Also shown are TDE limits for Se and Sp, which caused Se and Sp <85%.