Introduction The three major Soil-Transmitted Helminths (STH), Ascaris lumbricoides (roundworm), Trichuris trichiura (whipworm) and Necator americanus/Ancylostoma duodenale (the hookworms) are amongst the most widespread parasites worldwide. An estimated 4.5 billion individuals are at risk of STH infection and more than one billion individuals are thought to be infected, of whom 450 million suffer morbidity from their infection, the majority of who are children. An additional 44 million infected pregnant women suffer significant morbidity and mortality due to hookworm-associated anemia. Approximately 135,000 deaths occur per year, mainly due to infections with hookworms or A. lumbricoides [1]. The most widely implemented method of controlling STH infections is through periodic administration of anthelmintics. Rather than aiming to achieve eradication, current control programs are focused on reducing infection intensity and transmission potential, primarily to reduce morbidity and avoid mortality associated with the disease [2]. The benzimidazole (BZ) drugs, i.e. albendazole (ALB) and mebendazole, are the most widely used drugs for the control of STH. While both show broad-spectrum anthelmintic activity, for hookworms a single dose of ALB is more effective than mebendazole [3]. The scale up of chemotherapy programs that is underway in various parts of Africa, Asia and South America, particularly targeting school children, is likely to exert increasing drug pressure on parasite populations, a circumstance that is likely to favor parasite genotypes that can resist anthelmintic drugs. Given the paucity of suitable alternative anthelmintics it is imperative that monitoring programs are introduced, both to assess progress and to detect any changes in therapeutic efficacy that may arise from the selection of worms carrying genes responsible for drug resistance. The well documented occurrence of resistance to anthelmintics in nematode populations of livestock [4], highlights the potential for frequent treatments used in chemotherapy programs to select drug resistant worms. Such an eventuality threatens the success of treatment programs in humans, both at individual and community levels [5]. Although some small scale studies [6], [7], have suggested emerging drug resistance in human STH, these studies should be interpreted with some caution, since suboptimal efficacy could have been due to factors other than drug resistance. Moreover, although for the BZ drugs there are many published studies reporting the Cure Rate (CR) and the Fecal Egg Count Reduction (FECR), the two most widely used indicators for assessing the efficacy of an anthelmintic in human medicine, comparison of such studies is difficult, largely because there is no widely accepted standard operating procedure for undertaking such trials [8]. Published studies are confounded by methodological variations including treatment regimens, poor quality of drugs, differing statistical analyses used to calculate therapeutic efficacy, as well as a range of other problems in study design, such as small sample size, diagnostic methods, variation in pre-intervention infection intensities and confounding factors related to geographical locations. Such variation among studies greatly complicates direct comparison [3]. A World Health Organization-World Bank (WHO-WB) meeting on “Monitoring of Drug Efficacy in Large Scale Treatment Programs for Human Helminthiasis”, held in Washington DC at the end of 2007, highlighted the need to closely monitor anthelmintic drug efficacy and to develop standard operating procedures for this purpose. In a systematic meta-analysis of published single-dose studies, Keiser and Utzinger [8], confirmed that there was a paucity of high quality trials, and that the majority of trials were carried out more than 20 years ago. They recommended that well-designed, adequately powered, and rigorously implemented trials should be undertaken to provide current data regarding the efficacy of anthelmintics against the main species of STH. These should be designed to establish benchmarks (including standard operating procedures) for subsequent monitoring of drug resistance. The objective of the present work was to validate a standard protocol that has been developed for monitoring efficacy of anthelmintics against STH. To give the study wide relevance, we conducted the trial in seven populations in different geographic locations in Brazil, Cameroon, Cambodia, Ethiopia, India, Tanzania and Vietnam. In each of the study sites, different epidemiologic patterns of infection prevail, including different combinations of STH. We assessed the efficacy of a single dose (400 mg) of ALB in terms of the CR and the FECR in school children between 14 and 30 days following treatment. The McMaster egg counting technique was used in a standardized fashion, with rigorous quality control. Levecke et al. [9] reported that the McMaster holds promise as a standardized method on account of its applicability for quantitative screening of large numbers of subjects. This method is the recommended method for measuring fecal egg counts (FEC) when performing FECR for the detection of anthelmintic resistance in veterinary medicine [10], [11]. Methods Study sites This study was carried out in seven different countries covering Africa (Cameroon, Ethiopia and Tanzania), Asia (Cambodia, India and Vietnam) and South-America (Brazil). However, it is important to note, that while we refer to individual countries to identify results from particular trials, we do not make any conclusions about any country as such. Here, names of countries are used only to distinguish between 7 separate trials that were conducted in 7 geographically disparate regions of the world. In total ten study sites with varying STH and treatment history were included. These seven STH endemic countries were selected because of the presence of investigator groups with previous extensive experience in the diagnosis and control of STH. Table 1 provides their specific locations (district/province/state) and treatment history. Both species of hookworms (N. americanus and A. duodenale) were present in all study sites in varying degree with the exception of Brazil where only N. americanus was present. 10.1371/journal.pntd.0000948.t001 Table 1 The location and treatment history of the ten study sites. Country District/Province/State Treatment history Brazil Minas Gerais LSAT since 2007 (ALB) Cambodia Kratie LSAT since 1997, last in 2007 (MBD) Cameroon Loum LSAT (MBD/ALB) since 1999, last in 2008 (MBD) Yoyo No LSAT Ethiopia Jimma No LSAT India Vellore LSAT, since 2001, last in 2008 (ALB) Thiruvanamalai No LSAT Tanzania (Zanzibar) Pemba Island LSAT since 1994, last in 2006 (PZQ, IVM, ALB) Vietnam Thái Nguyên LSAT since 2005 Tuyên Quang No LSAT LSAT: large scale anthelmintic treatment, MBD: mebendazole, PZQ: praziquantel, IVM: ivermectine, ALB: albendazole. Trial design During the pre-intervention survey, school children aged 4 to 18 years at the different study sites were asked to provide a stool sample. For the initial sampling the aim was to enroll at least 250 infected children with a minimum of 150 eggs per gram of feces (EPG) for at least one of the STH. This sample size was selected based on statistical analysis of study power, using random simulations of correlated over-dispersed FEC data reflecting the variance-covariance structure in a selection of real FEC data sets. This analysis suggested that a sample size of up to 200 individuals (α = 0.05, power = 80%) was required to detect a 10 percentage point drop from a null efficacy of ∼ 80% (mean percentage FEC Δ per individual) over a wide range of infection scenarios. Standard power analyses for proportions also indicated that the detection of a ∼10 percentage point drop from a null cure rate required sample sizes up to 200 (the largest samples being required to detect departures from null efficacies of around 50%). Given an anticipated non-compliance rate of 25%, a sample of 250 individuals with >150 EPG pre-treatment was therefore considered necessary at each study site. Fecal samples were processed using the McMaster technique (analytic sensitivity of 50 EPG) for the detection and the enumeration of infections with A. lumbricoides, T. trichiura and hookworms [9]. None of the samples were preserved. Samples which could not be processed within 24 hours were kept at 4°C. A single dose of 400 mg ALB (Zentel) from the same manufacturer (GlaxoSmithKline Pharmaceuticals Limited, India) and same lot (batch number: B.N°: L298) was used at all trial sites. No placebo control subjects were included in the trial for ethical and operational reasons. Between 14 to 30 days after the pre-intervention survey, stool samples were collected from the treated subjects and processed by the McMaster technique. All of the trials were carried out in a single calendar year (2009). Subjects who were unable to provide a stool sample at follow-up, or who were experiencing a severe concurrent medical condition or had diarrhea at time of the first sampling, were excluded from the study. The participation, the occurrence of STH and sample submission compliance for pre- and post-intervention surveys are summarized in Figure 1. 10.1371/journal.pntd.0000948.g001 Figure 1 The participation, occurrence of STH and sample submission compliance for pre- and post-intervention surveys. Subjects who were not able to provide a sample for the follow-up, or who were experiencing a severe current medical condition or had diarrhea at the time of the first sampling were excluded from the trial. The McMaster counting technique The McMaster counting technique (McMaster) was based on the modified McMaster described by the Ministry of Agriculture, Fisheries and Food (UK; 1986) [12]. Two grams of fresh stool samples were suspended in 30 ml of saturated salt solution (density = 1.2). The suspension was poured three times through a wire mesh to remove large debris. Then 0.15 ml aliquots were added to each of the 2 chambers of a McMaster slide. Both chambers were examined under a light microscope using a 100x magnification and the FEC for each helminth species was obtained by multiplying the total number of eggs by 50. Statistical analysis The efficacy of the treatment for each of the three STH was evaluated qualitatively based on the reduction in infected children (CR) and quantitatively based on the reduction in fecal egg counts (FECR). The outcome of the FECR was calculated using three formulae. The first two formulae were based on the mean (arithmetic/geometric) of the pre- and post-intervention fecal egg count (FEC) ignoring the individual variability, whereas the third formula represented the mean of the reduction in the FEC per subject. The latter is the only quantitative indicator of efficacy for which the importance of confounding factors can be assessed by statistical analysis. The CR and the FECR (1-3) outputs were calculated for the different trials, both sexes, age classes (A: 4–8 years; B: 9–13 years and C: 14–18 years) and for the level of egg excretion intensity at the pre-intervention survey. These levels corresponded to the low, moderate and high intensities of infection as described Montresor et al. [13] For A. lumbricoides these were 1–4,999 EPG, 5,000–49,999 EPG and >49,999 EPG; for T. trichiura these levels were 1–999 EPG, 1000–9,999 EPG and >9,999 EPG; and for hookworms these were 1–1,999 EPG, 2,000–3,999 EPG and >3,999 EPG, respectively. In addition, the robustness of the three FECR formulae was explored by comparing the FEC reduction rate obtained from all samples containing STH and those obtained from samples containing more than 150 EPG as recommended in the anthelmintic resistance guidelines of the World Association for the Advancement of Veterinary Parasitology [9]. Finally, putative factors affecting the CR and the FECR (3) were evaluated. For the CR, generalized linear models (binomial error) were built with the test result (infected /uninfected) as the outcome, ‘trial’ (7 levels: trials in Brazil, Cambodia, Cameroon, Ethiopia, India, Tanzania and Vietnam) and ‘sex’ (2 levels: female and male) as factors, and ‘age’ and the log transformed pre-intervention FEC as covariates. Full factorial models were evaluated by the backward selection procedure using the likelihood ratio test of χ2. Finally, the CR for each of the observed values of the covariate and factor was calculated based on these models (The R Foundation for Statistical Computing, version 2.10.0 [14]). For analysis of the data from FECR (3), non-parametric methods were used, because models based on parametric statistics, even with negative binomial error structures, or based on transformed data would not converge satisfactorily as a consequence of the high proportion of FEC with zero EPG. Hence, the impact of the factors ‘trial’ and ‘sex’ were assessed by the Kruskal-Wallis test (for more than 2 group comparisons) and the Mann-Whitney U test, respectively. The correlation between the outputs of FECR (3) and the covariates (age and pre-intervention FEC) was estimated by the Spearman rank order correlation coefficient (SAS 9.1.3, SAS Institute Inc.; Cary, NC, USA). Ethics statement The overall protocol of the study was approved by the Ethics committee of the Faculty of Medicine, Ghent University (Nr B67020084254) and was followed by a separate local ethical approval for each study site. For Brazil, approval was obtained from the Institutional Review Board from Centro de Pesquisas René Rachou (Nr 21/2008), for Cambodia from the National Ethic Commitee for Health Research, for Cameroon from the National Ethics Committee (Nr 072/CNE/DNM08), for Ethiopia from the Ethical Review Board of Jimma University, for India from the Institutional Review Board of the Christian Medical College (Nr 6541), for Tanzania (Nr 20) from the Zanzibar Health Research Council and the Ministry of Health and Social Welfare, for Vietnam by the Ministry of Health of Vietnam. An informed consent form was signed by the parents of all subjects included in the study. This clinical trial was registered under the ClinicalTrials.gov Identifier NCT01087099. Results The cure rate (CR) Overall, the highest CRs were observed for A. lumbricoides (98.2%), followed by hookworm (87.8%) and T. trichiura (46.6%). However, as shown in Table 2, the CRs varied across the different trials, age classes and pre-intervention FEC levels. The differences in CRs between trials were most pronounced for T. trichiura, ranging from 21.0 (Tanzania) to 88.9% (India). The T. trichiura CRs of 100% for the trials in Brazil and Cambodia are not considered here as they were based on only 1 and 2 individuals, respectively. For hookworms and A. lumbricoides, the CRs varied from 74.7 (India) to 100% (Vietnam) and from 96.4 (Tanzania) to 99.3% (Ethiopia and Cameroon), respectively. The CRs for A. lumbricoides in Cambodia (100%) and India (95.2%) are not considered here as they were based on fewer than 50 individuals. The CRs increased over the three age classes (A. lumbricoides: 95.8 to 100%; T. trichiura: 44.7 to 54.1%), except for hookworms where the CRs ranged from 86.1 to 88.3, and then to 87.5%. For each of the three STH, there was a decline in the CR with increasing levels of infection intensities at the pre-intervention survey. The largest drop was observed for T. trichiura, which decreased from 53.9 to 12.5%. For the two other STH, the drop in the CR was less pronounced, ranging from 88.6 to 76.9% for hookworms and only from 98.3 to 95% for A. lumbricoides. The observed differences between sexes were negligible for all three STH. 10.1371/journal.pntd.0000948.t002 Table 2 The cure rate (CR) for treatment with a single dose of albendazole against soil-transmitted helminths. A. lumbricoides T. trichiura Hookworms n CR (%) N CR (%) n CR (%) Country Brazil 50 98.0 1* 100 52 88.5 Cambodia 5* 100 2* 100 127 87.4 Cameroon 298 99.3 386 47.4 140 87.1 Ethiopia 151 99.3 105 85.7 91 98.9 India 21* 95.2 18* 88.9 95 74.7 Tanzania 279 96.4 396 21.0 349 86.8 Vietnam 148 98.6 138 81.2 58 100 Age class A (4–8) 215 95.8 219 44.7 173 86.1 B (9–13) 669 98.8 753 46.3 643 88.3 C (14–18) 68 100 74 54.1 96 87.5 Sex Female 462 98.1 503 48.5 393 89.1 Male 490 98.4 543 44.8 519 86.9 Pre-intervention infection intensity Low 662 98.3 823 53.9 859 88.6 Moderate 270 98.1 215 19.5 40 75.0 High 20 95.0 8 12.5 13 76.9 Total 952 98.2 1046 46.6 912 87.8 *Due to the low number of infected subjects ( 75%). The pre-intervention FEC was probably the most important as it had a considerable effect on the CR of A. lumbricoides (χ2 1 = 4.14, p 99.3%). The results of FECR (3) mostly yielded comparable or lower values than those from FECR (1). The low values (sometimes negative) can be explained by subjects for whom the post-intervention FEC exceeded the pre-intervention FEC. These subjects contributed to a negative FEC reduction rate which had a significant impact on the final FEC reduction rate calculated with FECR (3). This became apparent in the FEC reduction rate for A. lumbricoides, where a Cameroonian male subject of 7 years with a pre-intervention FEC of 100 and a post-intervention FEC of 22,050 EPG, contributed markedly to lowering the overall values for the data-set from the trial in Cameroon (FECR (1): 99.2%; FECR (3): 26.0%). This lowering of FECR (3) compared to FECR (1) for A. lumbricoides also occurred with age class A (FECR (1): 98.9%; FECR (3): −2.7%) and the low pre-intervention infection intensity level (FECR (1): 97.8%; FECR (3): 66.6%), but not for the remaining variables. The number of negative individual FEC reduction rates, and the magnitude of the difference between pre- and post-intervention FEC, both contributed to the discrepancies found for T. trichiura (176 subjects) and hookworms (10 subjects). Robustness of FECR formulae Table 5 summarizes the FEC reduction rates restricted to samples of more than 150 EPG indicating that the results of FECR (1) and FECR (2) remained roughly unchanged. The values from FECR (3) increased and were mostly comparable with those obtained by FECR (1). This change in the results of FECR (3) is due to the exclusion of negative individual FEC reduction rates which mostly occurred among the subjects with low pre-intervention FEC (see also Table 4). Differences of more than 5% between the results of FECR (3) and FECR (1) were limited to T. trichiura (country: Cameroon, India, Tanzania and Vietnam; age class: A and C). 10.1371/journal.pntd.0000948.t005 Table 5 Fecal egg count reduction for samples with a pre-intervention FEC of more than 150 EPG. A. lumbricoides T. trichiura Hookworms n FECR(1)(%) FECR(2)(%) FECR(3)(%) n FECR(1)(%) FECR(2)(%) FECR(3)(%) n FECR(1)(%) FECR(2)(%) FECR(3)(%) Country Brazil 47* 100.0 100.0 100.0 0* _ _ _ 46* 97.5 99.6 97.5 Cambodia 1* 100.0 100.0 100.0 1* 100.0 99.7 100.0 100 97.7 99.6 96.7 Cameroon 266 99.8 100.0 100.0 233 39.9 93.4 50.4 71 93.6 99.5 95.1 Ethiopia 145 100.0 99.9 100.0 72 92.3 99.2 92.6 66 99.6 99.7 99.8 India 17* 98.9 99.9 99.6 11* 72.0 99.1 87.0 83 87.8 99.3 84.2 Tanzania 266 100.0 100.0 99.9 325 58.3 86.6 36.4 281 95.4 99.7 93.1 Vietnam 130 100.0 99.9 99.9 71 93.1 99.2 88.0 19* 100.0 99.7 100.0 Age class A (4–8) 196 99.9 100.0 99.8 153 65.1 94.8 57.2 130 94.7 99.6 94.4 B (9–12) 613 99.8 100.0 99.9 515 48.4 94.1 51.8 460 94.9 99.6 93.2 C (13–18) 63 100.0 100.0 100.0 45 60.2 94.4 46.4 76 96.4 99.6 97.1 Sex Female 428 100.0 100.0 99.9 343 54.0 94.7 57.7 286 95.2 99.6 93.8 Male 444 99.7 100.0 99.9 370 53.0 93.8 48.0 380 94.8 99.6 94.0 Pre-intervention infection intensity Low 582 99.9 99.9 99.9 490 49.0 95.1 50.2 613 94.1 99.6 93.6 Moderate 270 100.0 100.0 100.0 215 58.7 92.2 58.7 40 97.6 99.9 97.1 High 20 99.5 100.0 99.6 8 40.0 88.6 40.1 13 95.9 99.9 96.4 Total 872 99.9 100.0 99.9 713 53.5 94.3 52.7 666 95.0 99.6 93.9 FECR(1): group based and arithmetic mean; FECR(2): group based and geometric mean; FECR(3): individual based and arithmetic. *Due to the low number of infected subjects ( 95% for A. lumbricoides and >90% for hookworms are appropriate thresholds, and that efficacy levels below this should raise concern. The great variability of the FECR for T. trichiura and the relatively low efficacy of ALB, confirmed in this present study, indicate that it is not possible to propose an efficacy threshold for this parasite based on our data. In conclusion, the present study is the first to evaluate drug efficacy of a single-oral dose of ALB on such a scale and across three continents. The results confirm the therapeutic efficacy of this treatment against A. lumbricoides and hookworms, and the low efficacy against T. trichiura. Efficacy varied widely across the seven different trials, particularly in the case of T. trichiura and it remains unclear which factors were principally responsible for this variation, although pre-intervention FEC and age played clear roles in this respect. The FEC reduction rate based on arithmetic means is the best available indicator of drug efficacy, and should be adopted in future monitoring and evaluation studies of large scale anthelmintic treatment programs. Finally, our findings emphasize the need to revise the WHO recommended efficacy threshold for single dose ALB treatments. Supporting Information Checklist S1 CONSORT Checklist (0.22 MB DOC) Click here for additional data file. Protocol S1 Trial Protocol (1.17 MB PDF) Click here for additional data file.
