Associations between soil-transmitted helminthiasis and viral, bacterial, and protozoal enteroinfections: a cross-sectional study in rural Laos

Study setting and design

This cross-sectional study was nested within the Water, Sanitation, and Hygiene for Health and Education in Laotian Primary Schools (WASH HELPS) study, a longitudinal cluster-randomized trial evaluating a comprehensive school-based water, sanitation, and hygiene (WASH) intervention in 100 schools in Saravane Province, Lao People’s Democratic Republic (Lao PDR; Laos). Detailed methods of the parent study are described elsewhere [21]. The WASH HELPS study is registered at clinicaltrials.gov (NCT02342860).

Of the 100 schools participating in the WASH HELPS study, 50 (25 intervention and 25 comparison) were selected using stratified random sampling based on district size and WASH HELPS study intervention status. In each school-hosting village (there is only one school per village), we randomly selected 25 households. Households were eligible for inclusion if they had a child attending the primary school participating in the WASH HELPS study, and that pupil had a sibling < 5 years-old living in the household. At each household, the female head of household was surveyed on household demographics, asset and animal ownership, recent illness among household members, and WASH access and behaviors. Structured observations of WASH facilities were made when available.

In conjunction with the household survey, we collected stool samples from the pupil, the pupil’s parent/caregiver (preference was given to female parent/caregiver), and the pupil’s sibling < 5 years-old (if there were multiple siblings, preference was given to youngest sibling). To collect the stool samples, the female parent/caregiver was given three pre-labeled, resealable plastic bags each containing a plastic spoon. Caregivers were given diapers to collect stool from infants, when applicable. Written and pictorial instructions for stool collection were printed on the plastic bag, and participants were also provided verbal instructions. Participants were instructed to collect their first morning stool, and were informed that the field team would return to the household the following morning to collect all samples. If households did not return all three stool samples on the first day, participants were reminded of the stool collection procedures, provided new bags and spoons if needed, and a second return visit was made the following day. Stool samples were transported with a cold chain to the field laboratory within two hours of collection.

Upon collection, all samples were tested for STH using the Kato-Katz method [22]. For this sub-study, stool samples from a subset of 297 households were randomly selected for additional enteropathogen analysis via quantitative reverse transcription polymerase chain reaction (qRT-PCR). Households were eligible for inclusion in this sub-study only if all three subjects in the household (adult, school-aged child, and child < 5 years-old) returned their stool sample on the same day. Households were randomly selected, proportional to district size, village size, and WASH HELPS intervention status, from households participating in the household survey and STH testing by Kato-Katz.

All data were collected between February-April 2017 (dry season), prior to annual school-based preventative chemotherapy (PC) for STH. The time frame corresponded with the final round of data collection and conclusion of the WASH HELPS study [21].

Laboratory analysis

Following analysis for STH via Kato-Katz, 200 mg of stool was aliquoted into a DNA/RNA Shield Collection and Lysis Tube (Zymo Research, Irvine, CA, USA) containing a lysis buffer and bead beating system, and beaten for 20 min using a Disrupter Genie vortexer (Scientific Industries, Bohemia, NY, USA) [23]. One field control was processed each day using DNA/RNA-free water to evaluate the possibility of false positives from contamination in the field laboratory during sampling. Samples were kept frozen at − 20 °C until transported to a laboratory at Emory University, where they were subsequently stored at − 80 °C until extraction.

Total nucleic acid was extracted from samples using the ZymoBIOMICS DNA/RNA Mini Kit (Zymo Research, Irvine, CA, USA), according to manufacturer instructions. Samples were spiked with bacterophage MS2 (ZeptoMetrix, Buffalo, NY, USA), an external control, to monitor extraction and amplification efficiency [23]. One extraction blank was included per batch to exclude the possibility of false positives from contamination during extraction. Extractions were stored at − 80 °C until transported on dry ice to the University of Iowa for qRT-PCR analysis.

We created a custom TaqMan Array Card (TAC) (Thermo Fisher, Carlsblad, CA, USA) with compartmentalized, probe-based qPCR assays for 25 enteropathogens, including: five STH (Ancylostoma duodenale, Ascaris lumbricoides, Necator americanus, Strongyloides stercoralis and Trichuris trichiura); six viruses (astrovirus, adenovirus, norovirus GI, norovirus GII, rotavirus, sapovirus); nine bacteria (Aeromonas spp., Campylobacter jejuni, Clostridium difficile, enteroaggregative Escherichia coli (EAEC), enterohemorrhagic E. coli (EHEC), atypical or typical enteropathogenic E. coli (EPEC), heat-labile- (LT) or heat-stable (ST) enterotoxigenic E. coli (ETEC), Salmonella enterica and Shigella spp./Enteroinvasive E. coli (EIEC); and five protozoa (Cryptosporidium spp., Cryptosporidium hominus, Cryptosporidium parvum, Entamoeba histolytica and Giardia intestinalis) [24, 25]. The TAC included probes for the MS2 external control, as well as an 18S rRNA internal control. The TAC primer and probe sequences are listed in Additional file 1: Table S1.

TAC preparation was prepared based on the protocol described by Liu et al. [24]. Ag-Path-ID One-Step RT-PCR kit (Thermo Fisher, Waltham, MA) was used as the master mix reagent for the TAC analysis. Bovine serum albumin (BSA) was also applied into the TAC master-mix to prevent the possibility of PCR inhibition that may arise in nucleic acids extracted from stools [26, 27]. For each sample, 40 µl of DNA/RNA extract of equal volumes of DNA and RNA was mixed with 50 µl of 2× RT-buffer, 4 µl of 25× AgPath enzyme, 5.4 µl of nucleic acid-free water, and 0.6 µl of 50 mg/ml BSA to a total volume of 100 µl. All TAC runs were completed in a ViiA7 instrument with QuantStudio 7 software (Thermo Fisher, Waltham, MA), and the cycling conditions were as follows: holding stages of 45 °C for 20 min and 95 °C for 10 min, followed by 45 cycles of 95 °C for 15 s and 60 °C for 1 min.

TAC data were manually read by two independent researchers. True amplification was validated by inspecting the multicomponent plot for increases in fluoresce for the FAM-based gene-specific probe. Conflicting results were resolved by a third independent researcher. Samples were considered positive only when the corresponding field and extraction blanks were negative, otherwise the data were considered invalid [25].

Measures

Adult participants reported the age and sex of themselves, their primary school-aged child, and their child under five years-old. The following variables were reported by the female head of household: household ethnicity, in which households of non-Lao-Tai ethnicity were considered ethnic minorities; the number of household members, which was derived by listing and counting all people currently living in the household full time; animal ownership, which was defined as owning any cows, goats, sheep, poultry (chickens or ducks), or pigs; and the main source of household drinking water, which was further classified as improved/unimproved according to the World Health Organization/United Nations International Children’s Fund (WHO/UNICEF) Joint Monitoring Programme (JMP) standard definitions [28].

The following variables were reported by enumerators using structured observation: the presence of a household toilet, which was further classified as improved/unimproved according to WHO/UNICEF JMP standard definitions [28]; and the presence of a household basic handwashing facility, classified according to WHO/JMP standard definitions as having soap and water [28]. Socioeconomic status was determined through a series of questions and observations about household construction materials (roof, floor, and walls), ownership of a mobile phone, and presence of electricity. These variables were chosen based on those used in the Demographic and Health Surveys for measures of wealth in Laos [29]. We used principal components analysis to derive one single wealth metric from all of the wealth assets combined [30].

Escherichia coli pathotypes were classified according to the following gene targets: EAEC (aatA and/or aaiC), EHEC (eae with stx1 and/or stx2, and without bfpA), typical EPEC (bfpA with or without eae), atypical EPEC (eae without bfpA, stx1, or stx2), ETEC (eltB for heat-labile toxin [LT] and estA with or without eltB for heat-stable toxin [ST]) [31]. The number of microparasite infections was derived by summing all positive pathogens (range: zero to 20). We chose to use the ipaH gene target to be consistent with approaches used in other recent enteric disease studies of under-five children. However, ipaH occurs in Shigella spp. and EIEC, and does not validate the presence of the large virulence plasmid of other virulence genes that are unique to Shigella spp.

Statistical analysis

All data were analyzed using Stata Statistical Software: Release 15 (StataCorpLP, College Station, TX, USA).

We estimated the odds of concurrent microparasite infection using three separate logistic regression models for viral, bacterial and protozoal infection outcomes. For the primary analysis, the main exposure of interest was any STH infection, as determined by qRT-PCR detection. Secondary analyses examined specific STH species (i.e. hookworm, A. lumbricoides, T. trichiura or S. stercoralis) as main predictors. We controlled for the presence of the non-outcome microparasite taxa (e.g. the model of the association between STH and viral infection also controlled for concurrent bacterial and protozoal infection), as well as the following covariates determined a priori based on biological plausibility of affecting odds of both outcomes and STH infection: age group (i.e. adult, school-aged child, child < 5 years-old), sex, socioeconomic status, ethnic minority status, household population size, improved household toilet, improved household drinking water source, basic household handwashing facility, household animal ownership, and whether the school in the village was a beneficiary of a UNICEF WASH in Schools intervention. Random intercepts were included at the village and household levels to account for clustering.

The associations between STH infection or STH species and the number of concurrent microparasite infections were determined using separate Poisson regression models and are reported as beta coefficients representing the change in number of microparasite infections among subjects with STH (or specific STH species) infection compared to those without. Models included random intercepts at the village and household levels, and included the same covariates as the logistic regression models.

All models were assessed for effect modification by age group. All analyses were evaluated for statistical significance at P < 0.05.

Strengths and limitations

Strengths of this study include the random selection of participating villages and households. Also, pathogens in stool samples were detected and quantified using qPCR, which provides a higher sensitivity (98%) and specificity (100%) than conventional methods [24]. Further, the multi-target detection capacity of this method allowed us to examine 25 infectious pathogens [24], whereas most existing studies on pathogen co-infection have focused on pairs of agents [16]. Additionally, there is a dearth of clinical data on helminth co-infection, and most studies have relied on mouse models [35]. The predominant species involved in human STH and enteric microparasite infection is influenced by a range of factors, including age and WASH access [63, 64]. We examined human subjects from three distinct age groups, adults, school-aged children and children under five years-old, and controlled for potential confounding WASH and demographic variables to provide a more externally valid picture of STH and microparasite co-infection.

Our study is subject to a number of limitations. First, our data are cross-sectional so we do not know whether the STH or microparasite infection occurred first. However, STH are endemic in this population, our data were collected prior to annual primary school-based PC, there is no routine community-based PC in this population, and re-infection often occurs rapidly after PC [65]. Thus, it is likely that STH infections commonly preceded microparasite infections, particularly viral and bacterial infections which do not tend to be chronic. However, it is possible that some persistent protozoal infections (e.g. Giardia) preceded STH infection, and our results reflect an inverse association. Secondly, the high sensitivity of the TAC and other molecular assays may lead to the detection of prolonged shedding by attenuated pathogens and we cannot distinguish between symptomatic and asymptomatic infections [66]. However, even asymptomatic infections may lead to interactions within the host as well as other sequalae such as environmental enteropathy, malnutrition, and growth stunting [67–69]. Thirdly, evidence suggests that the outcomes of helminth-microparasite co-infection are context dependent and may depend on helminth infection intensity [19, 44]. Based on the Kato-Katz results from these samples, helminth infections were predominately of low infection intensity, so we were unable to stratify by infection intensity to evaluate differences in co-infection by infection intensity. Fourthly, we discarded 144 samples due to suspected contamination, which may have limited statistical power. Household toilet ownership and use of an improved water source were lower among participants whose samples were discarded. While these factors may be associated with the pathogen profile of the participants, contamination was a random event unassociated with the participants and would not confound the relationship between STH infection and odds of microparasite infection. Last, we did not measure cytokines, interferon, or other measures of immune response so we are unable to elucidate exact mechanisms of helminth-microparasite interaction.