Determinants and Impact of Giardia Infection in the First 2 Years of Life in the MAL-ED Birth Cohort

1Giardia lamblia, also known as Giardia duodenalis and Giardia intestinalis, is the most common etiology of intestinal parasitic infection in the first 2 years of life in low-resource settings. Although Giardia is a recognized pathogen of waterborne diarrhea outbreaks and a common cause of diarrhea among travelers and after recreational water exposure, the impact of endemic pediatric giardiasis is less clear. Two large studies of global etiologies of endemic pediatric diarrhea, the Global Enterics Multicenter Study (GEMS) and the Etiology, Risk Factors, and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development Project (MAL-ED), found Giardia significantly more often in nondiarrheal than diarrheal stools. Similarly, Giardia was not associated with acute diarrhea in a meta-analysis of 12 acute pediatric diarrhea studies, and Giardia had a protective effect against acute diarrhea in 2 longitudinal studies.

2Evidence for an association between Giardia infection and child growth outcomes has been mixed. Giardia infection is associated with disrupted villus architecture, an elevated lactulose/mannitol ratio (a marker of intestinal permeability), and zinc and vitamin A deficiencies, which suggests gut dysfunction and inadequate nutrient uptake. These associations, however, have been inconsistent and limited in ascribing directionality, because vitamin A deficiency, for example, can increase susceptibility to Giardia infection.

3The multisite MAL-ED birth-cohort study provides high-resolution prospective data to clarify early-life Giardia epidemiology in high-prevalence settings. Longitudinal analysis specifically enables assessment of the temporality between detection of Giardia, diarrhea, micronutrient status, markers of intestinal permeability and inflammation, and the estimation of longer-term effects on growth. Here, we describe the determinants, burden, and impact of Giardia infection in the first 2 years of life in 8 low-resource sites.

4The MAL-ED study design and methods have been described. In brief, the study was conducted between November 2009 and February 2014 at sites in Dhaka, Bangladesh (BGD), Fortaleza, Brazil (BRF), Vellore, India (INV), Bhaktapur, Nepal (NEB), Naushahro Feroze, Pakistan (PKN), Loreto, Peru (PEL), Venda, South Africa (SAV), and Haydom, Tanzania (TZH). Children were followed from birth (<17 days of age) via twice-weekly home visits for illness surveillance, medicines, and breastfeeding practices and monthly for anthropometry until they reached 2 years of age. Nondiarrheal surveillance stool samples were collected and tested for 40 enteropathogens monthly in the first year (0–12 months) of life and quarterly in the second year (12–24 months) of life. Stool samples were collected and tested also during every diarrhea episode reported during the twice-weekly surveillance visits. Diarrhea was defined as maternal report of 3 or more loose stools in 24 hours or 1 stool with visible blood. Weight-for-age (WAZ) and length-for-age (LAZ) z scores were calculated using the 2006 World Health Organization child growth standards. Sociodemographic information was assessed biannually and summarized using the Water, Assets, Maternal Education, Income (WAMI) score, which is based on monthly household income, maternal education, wealth measured by 8 assets, and access to improved water and sanitation, as defined by World Health Organization guidelines. Plasma zinc and retinol concentrations were assessed at 7, 15, and 24 months of age. Urinary lactulose/mannitol excretion ratios, measured at 3, 6, 9, and 15 months of age, were converted into sample-based z scores (LMZs) using the BRF cohort as the internal reference population. All sites received ethical approval from their respective governmental, local institutional, and collaborating institutional ethical review boards. Informed written consent was obtained from the parent or guardian of each child.

5We included in the analysis all monthly surveillance and diarrheal stool samples that were tested for Giardia by enzyme immunoassay (EIA) (TechLab, Blacksburg, VA), the majority of which were also tested by wet-prep microscopy. The laboratory methods for detecting other enteropathogens and gut biomarkers, including α-1-antitrypsin (ALA), myeloperoxidase (MPO), neopterin (NEO), and α-1-acid glycoprotein (AGP), a marker of systemic inflammation, have been described.

6Definitions of incident Giardia-related diarrhea were defined with increasing specificity for diarrhea of true Giardia etiology as follows: (1) Giardia-positive diarrhea, Giardia was detected in a diarrheal stool sample; (2) new Giardia-positive diarrhea, Giardia was detected in a diarrheal stool sample, and the most recent previous stool sample tested negative for Giardia or was taken more than 2 months earlier; (3) Giardia-positive diarrhea-associated pathogens–negative diarrhea, Giardia was detected in a diarrheal stool sample, but no diarrhea-associated pathogens that were previously identified in MAL-ED were detected (13 of 40 pathogens tested, ie, norovirus GII, rotavirus, astrovirus, adenovirus, Campylobacter, Cryptosporidium, heat-stable enterotoxin-producing enterotoxigenic Escherichia coli, typical enteropathogenic E coli, heat-labile enterotoxin-producing enterotoxigenic E coli, Shigella, enteroinvasive E coli, Entamoeba histolytica, and Salmonella); and (4) Giardia-positive-only diarrhea, Giardia was detected in a diarrheal stool sample, and no other enteropathogens among all 40 tested were detected. Persistence of Giardia detection was defined as 2 consecutive stool samples that tested positive for Giardia (2 consecutive months in the first year of life or 2 consecutive quarters in the second year). Prolonged persistence was defined as 3 consecutive stool samples that tested positive for Giardia.

7Risk factors for the first detection of Giardia in surveillance stool samples were identified using pooled logistic regression to estimate hazard ratios (HRs) and adjusting for site and a restricted quadratic spline for age. Variables in the multivariable model were included on the basis of statistical significance, model fit by the quasi-likelihood information criterion, covariance between factors, and variability of factors within sites for site-specific models. Comparing by the Akaike information criterion (AIC) to models with linear week of the year, seasonality was assessed by modeling Giardia detection with linear, quadratic, and cubic terms for the week of the year (w), and the terms sin(2πw/52), cos(2πw/52), sin(4πw/52), and cos(4πw/52). We used Poisson regression to evaluate associations between zinc and vitamin A status with Giardia detection in surveillance stool samples and adjusted for previous Giardia detection and potential confounders included in the multivariable risk factor model. We estimated the effect of Giardia detection on subsequent diarrheal rates using pooled logistic regression with general estimating equations (GEEs) and robust variance to account for correlation between outcomes within children and adjusted for the same confounders and illness symptoms during the exposure periods. We estimated the effect of Giardia in all stools on gut biomarker concentrations using multivariable linear regression with GEEs and adjusted for stool consistency and presence of the 2 other pathogens of highest prevalence, enteroaggregative E coli (EAEC) and Campylobacter. Last, we estimated the effect of Giardia detection in surveillance stools on WAZ and LAZ attainment at 2 years of age using multivariable linear regression with GEEs. Confounders, listed in the table footnotes, included baseline sociodemographic characteristics associated with Giardia detection identified above and EAEC and Campylobacter stool positivity. Data from SAV were excluded from zinc-related analyses and data from PKN were excluded from length-related analyses because of measurement quality concerns at those sites. For analyses limited to surveillance stool samples, results (not shown) were consistent when we repeated analyses with diarrheal stool samples.

8Of 34916 stool samples (27092 surveillance and 7824 diarrheal) tested for Giardia by an EIA, 33796 (96.8%) were also tested for Giardia by wet-prep microscopy. Compared to EIA, the sensitivity of microscopy was 46.2%, and its specificity was 99.3%. Giardia positivity by microscopy was 21% less likely if the stool was watery or liquid than if it was soft or formed (risk ratio [RR], 0.79 [95% confidence interval (CI), 0.68–0.90]), but we found no association between stool consistency and EIA results (RR, 0.94 [95% CI, 0.86–1.03]).

9Among 2089 children with at least 1 tested stool, the overall Giardia prevalence according to the EIA in stool samples was 14.7% (n = 5135). Giardia was detected at least once in two-thirds (n = 1178) of the 1741 children followed to 2 years of age (range, 37.7% [BRF] to 96.4% [PKN]). The overall median times to Giardia detection, which varied according to site, were 18.0 and 20.0 months for surveillance and diarrheal stool samples, respectively (Figure 1).

10The incidence of Giardia-positive diarrhea was 40.4 cases per 100 person-years. However, measures of Giardia-related diarrhea incidence decreased by approximately 60%, 75%, and more than 80% when we required the previous stool sample to have tested negative for Giardia, the current stool sample to have no detection of diarrhea-associated pathogens, and the current stool sample to have no detection of any other pathogens, respectively (Table 1).

11The overall prevalence of Giardia detected in surveillance stool samples was 13.6% (Table 1). However, the prevalence decreased by more than half (6.0%) when we required the previous surveillance stool to have tested negative for Giardia. Repeated Giardia detections in surveillance stool samples occurred in 838 (40.1%) children (Supplementary Figure 1). The prevalence of persistence was less than 5% before 6 months of age in all except the PKN site but increased to 31.8% overall in the second year of life.

12Giardia detection increased with age over the first 2 years of life; a 1-month increase in age was associated with an 11% increase in the risk of Giardia detection in surveillance stool samples (RR, 1.11 [95% CI, 1.10–1.12]). The percentage of days in the previous month that the child was exclusively breastfed was a strongly protective factor against first Giardia detection (Table 2). Socioeconomic factors, including increased socioeconomic score (Water, Assets, Maternal Education, Income score), household income, and older maternal age, were also protective. Metronidazole exposure in the previous 15 days was associated with a 31% relative decrease (95% CI, 16–44) in Giardia detection in surveillance stool samples, but we found no association with exposure more than 15 days earlier or with exposure to any other antibiotics.

13Multiple hygiene and environmental risk factors were associated with Giardia detection. Hand-washing, treatment of drinking water, and increased water access were protective, whereas the presence of siblings was a strong risk factor. Associations with having a dirt floor and owning chickens indicate the importance of environmental exposure to Giardia (Table 2). The distribution of environmental factors differed according to site, and although risk factor trends were generally consistent, there were site-to-site variations in the magnitude and even the direction of associations in some cases (Supplementary Figure 2).

14Giardia was positively correlated with Campylobacter detection (Pearson correlation coefficient [PCC], 0.15; P <.0001) but not with the detection of EAEC (PCC, −0.02; P =.0001) or viruses (PCC, −0.00; P =.7) in all stool samples, which suggests that the routes of transmission and/or age-susceptibility patterns are similar to those of Campylobacter.

15We found a significant increase in first Giardia detections in surveillance stool samples in July through September and a smaller peak in March/April in the south Asian sites (BGD, INV, NEB, and PKN) (Supplementary Figure 3). Giardia seasonality was variable at the other sites, with peaks in December/January in the BRF and PEL sites and a small peak in March/April in the TZH and SAV sites. In contrast, we found no evidence of seasonality when we included all Giardia detections. No association between site-specific mean temperature or rainfall and Giardia positivity was found. The seasonality of Giardia detection in diarrheal stool samples matched the seasonality of all-cause diarrhea, which suggests that many Giardia-positive diarrheal episodes were caused by other pathogens (not shown).

16Higher plasma zinc and retinol concentrations at 7 months of age were associated with decreased subsequent Giardia detection (Supplementary Table 1). A combined 1 standard deviation greater zinc and retinol concentration was associated with an adjusted 22% (95% CI, 2%–37%) lower Giardia-detection rate in surveillance stool samples from 8 to 24 months of age. A higher retinol concentration at 15 months of age was also associated with an approximate 10% decrease in the subsequent Giardia-detection rate. There were no associations between zinc status and Giardia detection at 15 months of age and no associations at either time period between zinc or vitamin A status and incidence of Giardia-positive diarrhea.

17Giardia detection in surveillance stool samples in the period between vitamin A status measurements at 7 and 15 months was associated with an adjusted −1.58 mg/dL (95% CI, −2.82 to −0.34 mg/dL) change in retinol concentration over that time period. There were no associations between Giardia and change in zinc concentration.

18Giardia detection in surveillance stool samples was not associated with short-term diarrheal risk (adjusted RR for diarrhea in the following 30 days, 1.07 [95% CI, 0.96–1.19]). In addition, the apparent negative association between Giardia detection and diarrhea previously reported (RR adjusted for age and site, 0.90 [95% CI, 0.86–0.95]) might be explained by treatment of 37% of all diarrhea episodes with metronidazole, such that Giardia might have been cleared before the diarrheal stool was collected. When we adjusted for recent metronidazole exposure, the association between Giardia detection and diarrhea moved toward the null (RR, 0.95 [95% CI, 0.90–1.00]).

19Giardia detection was not associated with subsequent diarrheal rates in any except the PKN site (Supplementary Table 2). Giardia persistence before 6 months of age was common at the PKN site (17.4%) and was associated with an adjusted 28% relative decrease (95% CI, 11–41]) in diarrheal rates from 6 to 24 months of age. Giardia persistence in the first year of life at the PKN site (47.7%) was also associated with an adjusted decrease in subsequent diarrheal rates (adjusted incidence rate ratio, 0.81 [95% CI, 0.64–1.03]). The protective effect was driven largely by protection against subsequent diarrhea in which enteropathogenic bacteria were detected.

20The presence of Giardia in stool samples was associated with an elevated marker of increased intestinal permeability, LMZ, across sites (Table 3), as indicated by an average increase in lactulose (z-score difference, 0.12 [95% CI, −0.00 to 0.24]) and a decrease in mannitol (z-score difference, −0.15 [95% CI, −0.26 to −0.04]). We found no significant differences in the associations across ages, although the magnitude was greatest in the second year of life (adjusted LMZ difference at 15 months, 0.25 [95% CI, 0.10–0.40]). Among markers of inflammation, Giardia was also associated with a decrease in NEO concentration (Table 3) but not consistently with MPO, ALA, or AGP.

21Giardia detection was associated with reduced weight and length attainment at 2 years of age (Table 4). Compared with those with low Giardia exposure (the 10th percentile of Giardia positivity in surveillance stool samples over the first 2 years of life), children with high exposure (the 90th percentile of Giardia positivity) had an adjusted −0.12 LAZ decrement (95% CI, −0.25 to 0.01) and −0.11 WAZ decrement (95% CI, −0.23 to −0.00) at 2 years of age. Giardia persistence in the first 6 months of life was statistically significantly associated with more than double that decrement in both weight and length at 24 months of age (Table 4).

22The prospective MAL-ED birth-cohort study provided a unique opportunity to investigate the complex relationships between Giardia exposure, micronutrient status, intestinal permeability, diarrhea, and growth. Giardia detection was common at all sites and increased in frequency through the second year of life. Major reductions in Giardia diarrhea incidence when more specific definitions were used suggest that true Giardia-caused cases of diarrhea are difficult to identify in settings of high endemicity. In the absence of molecular typing, evidence of visual clustering of Giardia detections within children and seasonal patterns limited to first detections suggest that repeated detections can often represent persistent infections. We identified both hygiene and environmental risk factors for Giardia infection and confirmed that the fecal-oral and waterborne routes both might be important modes of transmission. Host factors, including zinc and vitamin A deficiencies, might also contribute to Giardia susceptibility, because children with higher micronutrient concentrations had less subsequent Giardia detection. This relationship might be bidirectional, because Giardia detection from 7 to 15 months of age was also associated with a decrease in retinol concentration during that period.

23In contrast to data from previous reports, our longitudinal data did not suggest that Giardia infection was protective against diarrhea across the sites. The protective effect of Giardia on subsequent acute diarrheal risk was limited to early exposures at the PKN site, which might be explained by the uniquely high detection rate in the first year of life at this site, environment-specific unidentified biological susceptibility factors, Giardia strain variability, and coinfection factors. Because the diagnostic EIA was insensitive to stool consistency, dilution of Giardia in diarrheal stools does not explain previously reported inverse associations with diarrhea. In contrast, clearance of Giardia in diarrheal stools by metronidazole, the most common antibiotic used for diarrhea treatment in MAL-ED (17% of all episodes were treated with metronidazole), might contribute to this inverse association.

24Even in the absence of diarrheal symptoms, Giardia infection, especially early persistent infection, was associated with reduced weight and height attainment at 2 years. This finding was reported recently from an independent cohort study in Bangladesh. This early impact might be the result of a critical period of susceptibility in which the infant gut and intestinal microbiota are developing. The increased lactulose/mannitol excretion ratio associated with Giardia suggests a mechanism through increased intestinal permeability and malabsorption, components of environmental enteropathy. Giardia was not associated with increased markers of intestinal inflammation (ALA, MPO, or NEO), which suggests that Giardia might disrupt epithelial cells through pathways different from those of chronic immune activation typical of environmental enteropathy. Given the divergence between diarrhea and growth outcomes associated with Giardia infection, diarrhea-independent mechanisms are likely responsible for the growth impact such that physiologic insult occurs without necessary manifestation of diarrhea.

25We found that the subset of children with early and persistent Giardia infection experienced a significant disease burden caused by Giardia. However, no association was observed between Giardia detection and growth in the comprehensive risk factor models from MAL-ED that compared high and low levels of Giardia exposures both defined by no detections in the first 6 months (MAL-ED Network Investigators, unpublished data). Although other determinants might influence growth at later ages more strongly, targeting Giardia early in life might confer substantial benefit.

26This study was limited by fewer surveillance stool samples tested in the second year of life when Giardia was most common. In addition, although more sensitive than microscopy, the EIA has a lower sensitivity than nucleic acid–based detection methods. We also did not use Assemblage typing, which might have helped distinguish persistent infection from reinfection and explain variability in clinical outcomes. Last, antihelminthic medications that are active against Giardia (eg, albendazole or mebendazole) were not recorded, although they might have been used at the individual level and potentially through mass treatment campaigns. Despite these limitations, the breadth of data available, including markers of gut inflammation and permeability, enabled robust analysis of the effects and potential mechanisms of Giardia infection.

27The MAL-ED study enabled a comprehensive description of Giardia epidemiology in children early in life across 8 diverse sites. In general, interventions to limit exposure might reduce Giardia burden better than treatment, because metronidazole only transiently reduced detection. Interventions to reduce early exposure (eg, through sustained exclusive breastfeeding) might have the greatest effect on long-term outcomes, given increased the susceptibility of children to the effects of Giardia during their first few months of life.

28Supplementary materials are available at the Journal of the Pediatric Infectious Diseases Society online.