Ancylostomatoidea

Statistical analysis was performed using SAS 9.2 (Cary, NC, USA). Statistical significance was assessed at the 0.05 level unless otherwise noted. Infection groups were defined by the presence or absence of Ascaris infection and/or malaria infection at each time point, resulting in the following four groups: (1) no infection; (2) infection with malaria only; (3) infection with Ascaris only; and, (4) infection with both Ascaris and malaria (co-infection). Haemoglobin values were analysed as a continuous variable and then categorized into multinomial responses for mild, moderate or severe anaemia.
The categorization of haemoglobin allowed for calculations of odds ratios between infection groups. Univariate analysis was performed to calculate descriptive statistics of the predictors and assess normality of the outcome variables. Due to the highly skewed distribution of parasite intensity in children, the ranks of the data were used in place of the actual values of parasite intensity. The ranks were determined by ordering the data from least to greatest and replacing each observation by its relative position in the order. Assuming no ties, the smallest parasite intensity would receive a rank of 1 while the largest intensity would receive a value of n (the number of observations). The demographics from each infection group were compared at baseline to describe the overall sample of children. Comparisons between the four groups were made using chi-squared tests, analysis of variances models, or Kruskal-Wallis tests. The Tukey-Kramer multiple comparison procedure was used to determine the significance of pair-wise comparisons.
Parasite intensity and haemoglobin levels were analysed using repeated measures linear mixed models using the MIXED procedure in SAS to account for unbalanced data and multiple observations per child. An infection group-by-time variable was initially included in the haemoglobin model but was removed after it failed to retain significance. Similarly, an anaemia severity group-by-time variable was included in the parasite intensity model but was not retained. The original study treated children with albenzadole or placebo according to the study design; however this factor was removed from the model because it failed to show an association with anaemia severity and did not significantly change the estimates in the model. Means and standard deviations were used to describe haemoglobin levels within each group. Medians and ranges were used to describe the parasite intensities within each anaemia severity group.
The effect of infection group on anaemia severity was analysed with generalized linear mixed models using a cumulative logit function with the assumption that the data are from a multinomial distribution. The link function allows for the ordinality of the outcome variable anaemia severity. Thus the parameter estimates represent the log odds of being anaemic vs normal and the estimates have been transformed to odds ratios for ease of interpretation. A random effect for each individual was added to the model to account for the repeated measures study design. The effect of sex, age, and SES were also included in the model. Since hookworm infection is known to cause anaemia [22 (link)], the model adjusted for the effect of hookworm in our initial model; however, this was dropped because this factor conferred little to no change in the odds ratio estimates.

In the years 2011, 2012 and 2013, we conducted three randomised controlled trials which evaluated new anthelminthic drugs or drug combinations against soil-transmitted helminths on Pemba Island, United Republic of Tanzania [21 (link)-23 ]. Children attending the schools in Wawi and Al-Sadik (both in 2011), Mchangamdogo and Shungi (both in 2012 and 2013) were invited to participate in the clinical trials. Within the frame of these randomised controlled trials, a total of 14,855 Kato-Katz thick smears were prepared and examined under a microscope at the Public Health Laboratory-Ivo de Carneri (PHL-IdC) by experienced laboratory technicians. Kato-Katz thick smears were prepared according to standard protocols. In brief, we used 41.7 mg templates, and the slides were read within 60 min to avoid over clearing of hookworm eggs [5 (link),24 (link),25 (link)]. Soil-transmitted helminth eggs were counted for each species separately (i.e. A. lumbricoides, hookworm and T. trichiura).
Figure 1 shows the number of Kato-Katz thick smears examined in each of the three trials before and after anthelminthic drug administration. Approximately 10% of the Kato-Katz thick smears were randomly selected and re-examined for quality control, on a day-to-day basis. As hookworm eggs disintegrate rapidly on Kato-Katz thick smears [24 (link)], quality control was restricted to A. lumbricoides and T. trichiura. Quality control was carried out within 24 hours after preparation of slides.Figure 1

Total number of Kato-Katz thick smears read in three randomised controlled trials conducted on Pemba Island, United Republic of Tanzania. Flow chart detailing the number of Kato-Katz thick smears which were re-read for quality control, and hence were used for our analysis.

The results from the initial reading and from the subsequent quality control were compared. In case of discordant results (i.e. positive versus negative for a specific soil-transmitted helminth species and if the investigator judged the difference in FECs subjectively as too large) between the first two readers, a third microscopist was asked to re-examine the respective slide. Quality control (i.e. the second reading) was performed by a senior laboratory technician or by an investigator of the clinical trial. If a third reading was necessary, a third technician who did not previously examine the Kato-Katz thick smear was randomly chosen to re-examine the slide. The only exception was that in cases where a false-positive result was suspected, the first reader was asked to re-read the slide and show the observed egg to the investigator. All microscopists were blinded to previous results.

The mathematical model used describes the evolution of the parasite distribution in different host age groups and the impact of periodic chemotherapy on host burdens, incorporating the key epidemiological and biological processes influencing transmission. Building on past research
[15 (link),16 ], it includes the observed features of sexual reproduction by the dioecious helminths, heterogeneity in exposure to infection by host age, variation in the intensity of transmission in different human communities, aggregated distributions of worm numbers per host and a decline in fecundity as a function of worm burden (density dependence)
[16 -18 (link)]. The dynamics of transmission under repeated rounds of treatment is examined for the three main intestinal nematodes, Ascaris lumbricoides, Trichuris trichuria and hookworms (Necator americanus and Ancylostoma duodenale). The model is described in detail in the Additional file
1 available online.
Although a full age distribution is embedded in the model, we employ the key age groupings described above to define intervention coverage levels and illustrate their effect. These are infants (0–1 years of age) who cannot be treated under current licensure of the main anthelmintic drugs in wide use (e.g. albendazole and mebendazole), pre-school aged children (pre-SAC, 2–4 years of age), school aged children (SAC, 5–14 years of age) and adults (15+ years of age). Varying combinations of the fraction treated in each age grouping, treatment frequency and duration of treatment are explored. The fraction in each grouping effectively treated is a product of the fraction given treatment and drug efficacy (defined as the proportion of worms expelled). Within the current model, these two aspects of treatment are inseparable, and coverage of the population is represented as a proportion of worms treated. Drug efficacy is typically in the region of 90% or more for Ascaris and hookworms, but somewhat less for Trichuris[19 (link)-22 (link)]. It should be noted that the fraction treated is effectively chosen at random from the subpopulation. This model does not address systematic non-compliance.
The life cycles of these parasites involve free living stages that are passed in the faeces of the human host and mature to infective stages in the external habitat (eggs for Ascaris and Trichuris and larvae for hookworms). The infective stages of the parasite in the environment are represented in the model by a common pool of infectious material. The life span of these stages is typically weeks to months under favourable environmental conditions, and they are excreted in very large numbers
[23 -26 (link)]. Although this duration is short by comparison with adult worm life expectancies in the human host, infectious material in the environment acts as a reservoir which is unaffected by chemotherapy and can play a significant role in the dynamics of treatment. Dynamics of a range of parasites within the host population can be represented by the same model, with distinct parameter ranges for different species (See Additional file
1: Table S1).
Different age groups are thought to both contribute to, and be exposed to, this infective pool to varying degrees. An indication of this is provided by the changes in the intensity of infection by age; the patterns are typically convex for Ascaris and Trichuris, but continue to rise for hookworms as individuals age
[27 (link)-29 (link)] (Figure 
2). The respective roles of age related exposure to infection versus acquired immunity remains uncertain, but rapid re-infection by all three parasites post treatment points to the former as the main driver of age-intensity of infection profiles. On this basis, MCMC methods
[30 ] are employed to fit the model to these age related patterns of infection, to estimate both transmission intensity (measured by the basic reproductive number R₀ - the average number of offspring produced by one female worm that survive to reproductive maturity) and age related exposure. We have endeavoured to choose typical or characteristic infection profiles for the parasite species investigated in the hope that our results will be broadly applicable.

Ancylostoma ceylanicum hookworm-positive samples were further characterized to a haplotype level by the mitochondrial gene (cox 1); a forward primer (Aceycox1F) and a reverse primer (Aceycox1R) to amplify a region of 377 bp as previously described [7 (link)] (Table 1). Briefly, 25 µL of the PCR reactions containing 16 µL of ddH₂O, 0.5 unit of Taq DNA Polymerase 0.5 µL (Taq DNA Polymerase, Applied Biological Materials (ABM^®) Inc., Canada), 10 pmol of each forward/reverse primer 1 µL, 10xPCR buffer 2.5 µL, 25 mM MgSO₄ 1.5 µL, 10 mM dNTPs 0.5 µL and 2 µL of DNA template. The amplification was initial denaturation at 95 °C for 5 min, followed by 50 cycles of denaturation at 94 °C for 30 s, annealing at 58 °C for 30 s, extension 72 °C for 30 s and a final extension at 72 °C for 7 min. The amplification conditions were controlled by a thermocycler (Mastercycler^® nexus gradient, Eppendorf AG, Germany). PCR products were identified by electrophoresis with 1.5% agarose gel at 100 V for 40 min.

All procedures used for this study were approved by the Research Ethics Committee of the Federal University of Sergipe (CAAE: 66074017.6.1001.5546) and the Federal University of Minas Gerais (CAAE no. 55239522.3.0000.5149). Parents/guardians and participating children were informed about the purpose and procedure of the study. Written informed consent was obtained from the parents/guardians of the children agreeing to participate in the study prior to enrolment. All human stool samples obtained in this study were coded and treated anonymously.
The participants were informed of their stool test results, and all of the infections were treated with praziquantel. The dosing utilized included 40 mg/kg for adults and 60 mg/kg for children, according to Brazilian technical regulations [20 ]. Individuals infected with Ascaris lumbricoides, Trichuris trichiura and hookworms were treated with a single dose of 400 mg albendazole.