A Prospective Longitudinal Cohort to Investigate the Effects of Early Life Giardiasis on Growth and All Cause Diarrhea

PATIENTS AND METHODS

We conducted a prospective longitudinal birth cohort in an urban neighborhood of Dhaka, Bangladesh. 629 children were enrolled within 72 hours of birth from January 2008 to December 2012. We analyzed only children that completed at least 2 years of follow-up and limited our analysis to the first 2 years of life. Previous analysis of the first 147 children enrolled in this cohort has been published and included analysis of birth anthropometry as a predictor of subsequent giardiasis throughout the first year of life. No association was found [23].

The study was conducted in the urban slum of Mirpur, Dhaka. The majority of construction is mud-brick, and open sewers flow throughout the neighborhood. Our participants tended to come from the lower socioeconomic strata of Mirpur due to the area in which recruitment occurred and the location of our study clinic.

Subjects were followed by field research assistants who visited the home of each subject 3 times per week. Through these visits, data on duration of exclusive and partial breastfeeding was achieved. Close monitoring for diarrheal illness was also a priority. In the event of a diarrheal episode (defined as ≥3 unformed stools in a 24-hour window and separated from a previous diarrheal episode by at least 3 days), stool was collected by the field research assistant. Children also had nondiarrheal “surveillance” stool collected monthly and their anthropometric indices measured every 3 months. Field research assistants were trained in pediatric anthropometry measurement. Z-scores were calculated using World Health Organization growth curves. Information on socioeconomic status and education was collected via survey at enrollment. Evaluation of Giardia in stool samples was done via enzyme-linked immunosorbent assay (ELISA) (TechLab, Blacksburg, Virginia).

We constructed 2 multivariable linear regression models with length-for-age Z score (LAZ) and weight-for-age Z score (WAZ) at 2 years of age as the outcomes. Predictors were chosen based on previous literature around growth faltering in low-income children but included both total number of Giardia positive surveillance stools in the child's 2 years of life as well as presence of at least 1 Giardia positive surveillance stool in the first 6 months of life. Duration of exclusive breastfeeding (defined as a diet of only breast milk without formula, animal milk, water, or any other liquid/solid consumable) was also included. As all children were still partially breastfed at 2 years of age, this variable was not included in any analyses. Based on the results of these regressions, we constructed predictive models of any Giardia variable that was statistically significant. Predictors of giardiasis were chosen based on sanitation practices and socioeconomic factors. We utilized conditional backward stepwise selection to isolate significant predictors. A Hosmer-Lemeshow statistic was produced for our final model to assure stability [24]. A nonsignificant χ² statistic was produced assuring goodness of fit.

For our secondary analysis of Giardia's effect on the risk of an all-cause diarrheal episode we used generalized estimating equations to fit a binary logistic regression model, where the outcome was the presence of at least 1 diarrheal episode in each month of life, and predictors included risk factors for diarrheal disease as well as the presence of Giardia in surveillance stool (a time-varying predictor).

For statistical analysis, only subjects with complete data sets for all variables in a given analysis were used for that particular model. GraphPad Prism version 5.01 (GraphPad Software, Inc., La Jolla, California) was used for survival analyses. SAS version 9.4 (SAS Institute Inc., Cary, North Carolina) was used for all other statistical computation.

This study was approved by the Ethics and Research Review Committees at The International Centre for Diarrhoeal Disease Research, Bangladesh, and by the Institutional Review Board at the University of Virginia.

RESULTS

629 children were enrolled in the cohort. 445 children completed 2 years of the study and had a surveillance stool collected for all 24 months. Enrollment characteristics did not differ between the original cohort and the 445 patients completing the study (Table 1). Children in our cohort had an average of 3.6 Giardia positive monthly surveillance stools throughout the first 2 years of life. The median time to first Giardia positive surveillance stool was 17 months. For second and third positive surveillance stools the median survival time was 1 and 2 months after initial infection, respectively (Figure 1). 7% of children had a Giardia positive surveillance stool in the first 6 months of life. By 2 years of age 74% of our cohort had at least 1 Giardia positive surveillance stool. 64% had at least 2 positive monthly stools, and 55% had at least 3.

354 children had complete data sets for anthropometry and were used for LAZ and WAZ regressions. Children in this cohort had an average decline in LAZ score of 1.35 from enrollment to 2 years of age. LAZ at enrollment, income, maternal education, and early life giardiasis were statistically significant predictors of LAZ at 2 years of age. A 1-unit increase in LAZ at enrollment increased LAZ at 2 years by 0.40 (confidence interval [CI], .26–.54; P value <.0001). For every 1000 taka of family income LAZ at 2 years increased by 0.08 (CI, .05–.11; P value <.0001). Presence of any formal education in mothers increased LAZ at 2 years by 0.28 (CI, .07–.49; P value .009). Presence of a least 1 Giardia positive surveillance stool in the first 6 months of life decreased LAZ at 2 years of age by 0.4 (CI, −.80 to −.001; P value .05), whereas the total number of Giardia positive surveillance stools in the child's first 2 years of life was not a statistically significant predictor. WAZ at enrollment, length of exclusive breastfeeding, and gender were not statistically significant predictors of LAZ at 2 years.

We conducted a similar regression for change in WAZ from enrollment to 2 years of age. Children in our cohort had an average decline in WAZ of 0.5 from enrollment to 2 years of age. LAZ at enrollment, WAZ at enrollment, income, and maternal education were predictors of WAZ at 2 years. For a 1-unit increase in LAZ at enrollment, WAZ at 2 years increased by 0.17 (CI, .03–.30; P value .02). A 1-unit increase in enrollment WAZ increased WAZ at 2 years by 0.22 (CI, .07–.37; P value .003). 1000 taka of family income equated to a 0.05 increase (CI, .02–.08; P value .003), and presence of any formal maternal education resulted in a 0.36 increase (CI, .16–.56; P value .0005). Notably, neither presence of giardiasis in the first 6 months of life nor total number of Giardia positive surveillance stools in the child's life were statistically significant predictors of WAZ at 2 years of age. Length of exclusive breastfeeding and gender were also not statistically significant (Table 2).

Complete data on 391 patients at 6 months of age were available to investigate risk factors for early life Giardia (defined as at least 1 Giardia positive surveillance stool in the first 6 months of life). Predictors entered into our original model included sex, income, presence of any maternal education, months of exclusive breastfeeding (limited to the first 6 months), presence of an animal in the home, presence of an open drain/sewer outside the bedroom, presence of an open drain/sewer outside the kitchen, routine covering of drinking water vessels, routine hand washing by primary caregiver prior to feeding the child, use of piped drinking water, use of a tube well for drinking water, presence of a toilet or septic tank, use of a pit latrine, and use of open defecation. After selection, only months of exclusive breastfeeding (OR: 0.72, CI, .58–.90; P value .004) and caregiver hand washing prior to feeding the child (OR: 0.33, CI, .11–.97; P value .04) remained as statistically significant predictors early life Giardia (data not shown). A Hosmer-Lemeshow statistic was produced for our final model with nonsignificant χ² statistic of 12.25 (P value .09).

Finally, we investigated the association between Giardia carriage and all-cause diarrhea. Variables included in our model to predict presence of at least 1 diarrheal episode in a given month included any maternal education, income, exclusive breastfeeding, WAZ at enrollment, LAZ at enrollment, sex, and presence of Giardia in surveillance stool. 274 children had complete data and were analyzed with this model. In these children, 1164 of the total 6576 child-months (17.70%) were positive for a diarrheal episode with an average of 6.07 episodes of diarrhea per child over the first 2 years of life. Of those diarrhea positive child-months, 36.25% (422 child-months) had Giardia detected in the surveillance stool. Only exclusive breastfeeding was significantly associated with decreased diarrhea (P value <.001). No variables in our model were associated with increased diarrhea. Importantly, presence of Giardia in the child's surveillance stool was neither protective nor a risk factor for that child having an acute diarrheal episode.

We then repeated this analysis but separated presence of Giardia into 2 separate covariates; first incidence of Giardia detected in the surveillance stool and subsequent incidence of Giardia detected in the surveillance stool. This did not change the model with exclusive breastfeeding remaining significant (P value <.001) and neither of the Giardia covariates being associated with diarrhea (data not shown).

DISCUSSION

Our study demonstrated that giardiasis in the first 6 months of life but not the total number of Giardia positive surveillance stools over the first 2 years of life, was inversely associated with linear growth. Longitudinal analysis of growth in low-income countries has shown the first 6 months of life to be a crucial window that determines either healthy or diminished height gain [21, 25]. Giardia has been shown to disrupt intestinal tight junctions in vitro and increased intestinal permeability early in life is associated with growth stunting [23, 26]. Giardiasis in weanling mice disrupts the villus:crypt architecture in the presence of chronic inflammation and impairs growth [27]. Furthermore, although some Giardia strains are known to cause intestinal inflammation, it has been shown that the levels of inflammation decrease with subsequent infections [28, 29]. This implies that there is acquired protection against severity of giardiasis but not from reinfection, which may be the reason having a first Giardia infection in early life influences growth more than prolonged infection.

Our findings are significant as they potentially explain the discrepancy in the literature investigating Giardia's effect on growth. Cross-sectional studies that identified associations between Giardia and decreased length likely represented identification of children who had predisposing risk factors to giardiasis and may have had early exposure. Likewise, studies that found no such association may have not discriminated between early giardiasis and later infection.

We showed hand washing practices and exclusive breastfeeding were inversely associated with early life giardiasis, which is consistent with several studies showing similar protection [30, 31]. This suggests that caretakers are transmitting Giardia from the environment or fecal-oral route on their hands. Daycare outbreaks of Giardia in the United States support this concept, thus suggesting that epidemic and endemic routes of transmission may be the same [32, 33]. There is also literature that suggests that breast milk protects against symptomatic Giardia via secretory IgA although offering no protection against colonization or reinfection [34]. Contradictory work shows anti-Giardia IgA offers no protection against diarrheal or asymptomatic infection nor occurrence of repeat infections [35]. This may be explained by work showing breast milk to be protective against only the first infection of Giardia but not chronic colonization [36]. Given our findings that early life giardiasis was a risk factor for growth stunting, it seems that maintenance of exclusive breastfeeding through 6 months of age and good hand hygiene may be crucial interventions in protection from Giardia associated stunting.

Our results also indicate that presence of Giardia carriage is neither a risk factor for acute diarrheal illness nor is it protective. This corroborates the 2 longitudinal studies in the literature that found no change in incidence of acute diarrhea during Giardia positive periods [14, 17]. The hypothesis that Giardia may be protective against other enteropathogens comes from studies comparing diarrheal and nondiarrheal surveillance stools that found lower incidence of Giardia in diarrheal stools. However, we postulate an alternative explanation that colonizing Giardia is being washed out of feces during acute diarrheal episodes from other pathogens and thus does not necessarily imply Giardia is protective.

This study has several important limitations. First, although ELISA has both high sensitivity and specificity, it may be inferior to polymerase chain reaction (PCR), and thus we may have missed low-level Giardia colonization. Second, we did not differentiate Giardia assemblages. It has been shown that although assemblage B is more prevalent in Bangladesh, assemblage A had a stronger association with diarrhea [37]. Associations between assemblages A and B with symptoms are inconsistent across both experimental and field studies [16]. Thus, it may be that a particular assemblage is more or less associated with our outcomes of interest and that the signal is lost when the summative burden of all Giardia is investigated. Third, the confidence interval for Giardia's association with poor linear growth is large, making accurate estimation of the effect size difficult. This is likely due to the low number of children who had early life giardiasis. A higher sample size would have made this estimation more accurate. Lastly, information regarding coinfections, intestinal microbiota, and markers of permeability and intestinal inflammation was beyond the scope of the present study but likely is very relevant to the primary outcomes of growth and diarrhea and may interact with asymptomatic giardiasis.

The study has particular strengths that are unique to other outcome-based studies of pediatric endemic giardiasis. First, intensive home-based surveillance ensured that diarrheal surveillance and breastfeeding data were accurate. Second, while ELISA may be inferior to PCR, it is superior to microscopy, which many of the older studies of giardiasis relied upon. Third, the longitudinal design incorporating surveillance and diarrheal stools allowed us to investigate the incidence of diarrhea during periods of known Giardia positivity. Lastly, our study had the benefit of continued surveillance in a birth cohort allowing us to detect early life Giardia and analyze this as a separate risk factor.

CONCLUSIONS

Giardia was common in our cohort. Using frequent sampling in a prospective pediatric cohort enrolled at birth, we have helped clarify important considerations in resolving the endemic pediatric giardiasis conundrum. For the first time, we show that infection with Giardia within the first 6 months of life was an independent risk factor for linear growth failure but not poor weight gain, whereas total number of Giardia positive monthly surveillance stools had no effect on growth. Presence of Giardia had no effect on the likelihood of a child developing acute diarrhea. In this regard, efforts to prevent Giardia-associated growth sequelae may have the greatest impact during the first 6 months of life. Our data support that these measures focus on introduction of hand-washing practices and exclusive breastfeeding. Whether screening and treatment of giardiasis in early life could salvage later growth delays is an important question for future study. Further investigation into the effects of specific Giardia assemblages and interactions with co-pathogens may provide further understanding of the subacute effects of this ubiquitous but enigmatic pathogen.