Choline, DHA, and Diarrheal Disease Associated with Growth Faltering in a Case-Control Study.

Background: Children with recurrent infectious diarrhea are susceptible to growth faltering. DHA and choline may play a role in this relationship due to their involvement in lipid metabolism, gut immunity, and inflammatory pathways.
Objectives: This study aimed to characterize the contributions made by DHA and choline status and enteric damage in young children in the association between diarrheal illness and child growth.
Methods: A longitudinal case-control study was conducted among children aged 6-36 mo (N = 195) in Cap-Haitien, Haiti. Mother-child dyads were recruited from community health posts and outpatient clinics. Cases were defined as children experiencing acute diarrhea within the last 3 d and matched to healthy controls. Child anthropometry, dietary intake, and blood and stool samples were collected at baseline and follow-up. Plasma DHA, choline, and betaine were determined by LC-MS/MS methods (n = 49) and intestinal fatty acid-binding protein (I-FABP) by ELISA (n = 183). Multivariate regression models were applied with mediation analyses to examine associations and adjust for confounding factors.
Results: At baseline, mean plasma DHA concentrations (1.03 µg/mL; 95% CI: 0.91, 1.15) were not significantly different between cases and controls, nor was there a difference in mean plasma choline concentrations (4.5 µg/mL; 95% CI: 3.8, 5.1). Mean plasma I-FABP concentrations were significantly higher at follow-up in cases (3.34; 95% CI: 3.28, 3.40) than controls (3.20; 95% CI: 3.13, 3.27; P = 0.002). In adjusted multilinear regression models, higher plasma DHA concentrations at follow-up were associated with a negative change in weight-age z score (P = 0.016), and follow-up I-FABP was inversely associated with height-age z score (P = 0.035). No interaction or mediation effects were found. Conclusions: I-FABP concentrations were significantly higher in cases as compared with controls at follow-up, suggesting ongoing enteric damage and increased risk for malnutrition. Plasma DHA and I-FABP may have a role in childhood growth outcomes.

Introduction

Diarrheal disease and undernutrition are 2 predominant causes of morbidity and mortality in children living in low- and middle-income countries (LMICs) (1–3). A nationally representative study conducted in Haiti found that approximately 38% of infants aged 6–11 mo experienced a diarrheal illness within the past 2 wk (3). Undernutrition in the form of stunting is also widely prevalent among young children in Haiti, affecting 21.9% (3). Repeated diarrheal disease often exacerbates undernutrition and may magnify underlying energy and nutritional deficits (4). This unforgiving cycle of repeated diarrheal disease and compounding undernutrition has the potential to result in a condition known as environmental enteric dysfunction (EED), a subclinical disorder of the small intestine characterized by villous blunting and crypt hyperplasia (5, 6). The presence of EED is a postulated etiology of stunted growth in children living in LMICs (5, 6). Thus, the potential for public health nutrition interventions that alleviate these diarrheal complications warrants investigation to improve health and mitigate disease (7).

Interventions comprising nutrient-dense foods can significantly affect child growth and development (8). One randomized controlled trial conducted in Ecuador targeted eggs as an abundant source of choline to children early in the complementary feeding period (ages 6–9 mo), a phase known to have the highest risks for growth faltering and diarrheal disease (9). Findings showed significant reductions in stunting and increased concentrations of DHA and choline status for children in the egg group (10, 11). Choline is an essential nutrient for neurotransmitter synthesis, lipid transport, and 1-carbon metabolism. Choline deficiencies have been shown to alter specific immune system processes resulting in increased inflammatory responses within the intestine (12). Similarly, evidence supports the role of DHA, an ω-3 long-chain PUFA, as an anti-inflammatory mediator of the immune system (13, 14). Coinciding with this function, DHA is vital for linear growth and brain and eye development (15–18). A deeper understanding of how these nutrients influence diarrhea, enteric health, and child growth may be critical in LMIC contexts.

A longitudinal case-control study was conducted to gather preliminary data on these complex interactions across nutrition, diarrhea, and EED factors (). We hypothesized that plasma DHA, choline, and enteric health would mediate childhood growth outcomes in those with diarrheal illness.

FIGURE 1

Conceptual diagram and mediating mechanisms involved in growth outcomes. HAZ, height-age z score; I-FABP, intestinal fatty acid–binding protein; WAZ, weight-age z score; WHZ, weight-height z score.

Methods

Study design and participants

We conducted a case-control study with a 1-mo longitudinal outcome assessment in Cap-Haitien, Haiti. Participants were recruited from Hôpital Universitaire Justinien (HUJ), community clinics (Fort St Michel and La Fossette), and associated community health posts. Inclusion criteria were children 6–36 mo of age and caregivers >18 y old. Those who could not provide consent or required immediate medical care were excluded. Children with symptoms suggestive of coronavirus 2019 (COVID-19) infection were referred to the pediatrician at HUJ. Cases were chosen on the basis of acute diarrhea 3 d before enrollment. Acute diarrhea was defined as ≥3 liquid or semiliquid stools in 24 h by self-report (19). Those who fit the study inclusion criteria but had no history of acute diarrhea were assigned to the control group.

Enrollment occurred between December 2020 and May 2021. Additional measures were taken to improve the safety of the staff and study participants during the COVID-19 pandemic. These included specific COVID-19 symptomatic questionnaires at each visit time point and the provision of face masks for everyone >3 y old. Staff were trained on proper hand-washing techniques. Hand sanitizer was provided for all staff, caregivers, and children. Data and specimen collection stations were thoroughly disinfected between participants. Enumerators, nurses, and the phlebotomist were all supervised by the on-site field coordinator to ensure that all these measures were followed.

Ethical review

Study approvals were obtained from the Washington University Institutional Review Board and the National Bioethics Committee in Haiti (Comité National de Bioéthique). Written informed consent was obtained from all caregivers. Caregivers were also asked for permission to use their data in future research. Refusal did not affect their eligibility for the study. Caregivers were not paid to participate but were compensated for travel costs to the study sites. Participants did not directly benefit from the study; however, we hope that the results will help children with diarrheal disease in the future.

Data were anonymized after the initial visit by using different identification numbers at enrollment. All data collected on paper forms were kept in a locked cabinet and were not accessible to members outside the research team. Electronic data were password protected and deidentified. Our field coordinator, a trained registered nurse, will inform participants in small groups of the results.

Survey management

Trained nurses completed baseline and follow-up assessments with the child and caregiver in their native Creole language (Supplemental Table 1). In addition, written surveys were conducted and entered into the REDCap tools (Research Electronic Data Capture) housed at Washington University (20, 21). Double data entry was completed on 10% of participants selected at random to assess for interrater reliability with a goal of >90% agreement. Inconsistencies, if found, were reviewed and corrected.

Measures

Outcomes

Anthropometric measurements were obtained following the WHO standardized reference manual (22, 23). Weight (Seca model 874 digital scale) and length (ShorrBoard stadiometer) were measured twice. Measurements were repeated a third time when differences were >0.1 kg or >0.7 cm, and the 2 closest were averaged. We subtracted 0.7 cm from the length for children >2 y old when standing height was not measured (24). The WHO Anthro Survey Analyser Software (version 3.2.2) was used to calculate height-age z score (HAZ), weight-age z score (WAZ), and weight-height z score (WHZ) (23, 25).

Exposures and sample collection

Plasma samples were collected in lithium heparin-containing BD Vacutainers and placed on ice. After centrifugation at 1100–1300 × g for 20 min, aliquots were stored at –20 °C (26).

A random subset of cases and controls was selected for nutrient analysis, excluding hemolyzed samples. Unfortunately, product cost changes limited our initial planned analysis. Paired plasma choline, DHA, and betaine were measured using modified LC-MS/MS methods at the Washington University Metabolomics Facility (27). The betaine:choline ratio was used to measure potential flux into 1-carbon metabolism and away from lipid synthesis as choline was metabolized to betaine. Higher betaine:choline ratios suggest an increase in 1-carbon metabolism (28). When this occurs, it limits the amount of choline used for phosphatidylcholine synthesis, and previous studies suggested that this may have a role in growth faltering (29).

Stool samples were collected at the time of the visit. If the child did not or was unable to provide a stool sample during the study visit, caregivers were given a clean collection cup containing a sterile spoon and instructed to return with a sample. Equal amounts of stool and stabilization solution (Monarch DNA/RNA Preservation Reagent, T2011L; New England Biolabs) were mixed and stored at –20 °C.

All samples were transported on ice from HUJ to the Kuhlmann Laboratory at Washington University in St Louis via the Haitian National Laboratory. Upon arrival, samples were placed in the –80 °C freezer.

To evaluate for intestinal damage, we detected plasma intestinal fatty acid–binding protein (I-FABP) in duplicate. I-FABP was detected using an ELISA (Human FABP2 DuoSet; R&D Systems) according to the manufacturer's protocol, substituting BSA for normal goat serum. I-FABP is a marker of enteric damage since it is highly expressed in the small intestinal enterocytes and immediately released into the circulation during acute insults (30). To evaluate intestinal inflammation, we detected fecal calprotectin (FC), a major protein found in the cytosol of monocytes and neutrophils (31). FC was determined in duplicate on preserved stool samples using ELISA (Hycult Biotech) according to the manufacturer's protocol; a separate proprietary buffer (K 0001.C.100; Immunodiagnostic) provided optimized results. Both markers have been associated with EED (6).

Covariates

Age was determined using the difference between the date of birth and the date of the initial encounter. The following information was collected at baseline: socioeconomics, demographics, drinking water source, sanitation (drinking water, toilet type, number of people using the toilet, and waste disposal), and hygiene (hand-washing techniques). Maternal level of education, occupation, home ownership, house composition, and household income were utilized to evaluate socioeconomic status. Child health and morbidity questionnaires inquired about any history of asthma, congenital anomalies or developmental delays, a 2-wk symptom recall, and vaccination status. Dietary intake was evaluated by breastfeeding history, vitamin and mineral supplementation, and a 24-h FFQ, previously validated in other studies in Cap-Haitien and catchment communities (32). Based on the FFQ, foods were categorized into groups: breastmilk; milk, cheese, yogurt; bread, rice, pasta, potatoes, cereal; roots and tubers; beans and legumes; red meat, chicken; eggs; seafood; dark leafy green vegetables; all other vegetables; fruits; oil and butter; and bonbon sel (thin flour cracker), juice, soda, and other sweets. Animal source foods (ASFs) included red meat, chicken, eggs, seafood, and milk products. Self-reported dietary recalls are inherently subject to social desirability bias (33, 34). To temper possible social desirability bias, caregivers were interviewed by local Haitian women trained to minimize expressing any judgments or opinions. Nevertheless, bias may still be present and should be considered when interpreting results.

Statistical analysis

All continuous variables were assessed for normal distribution and outliers using histograms, scatter plots, and box plots. Those with nonnormal distributions were log10 transformed, which included I-FABP and FC. Baseline continuous variables were assessed for significant differences using independent samples t test or Mann-Whitney U test when nonparametric tests were indicated. A paired samples t test was used to compare continuous baseline variables with those from follow-up. Continuous variables were compared by Pearson correlations or Spearman ρ when nonparametric tests were indicated. Categorical variables were evaluated using χ2 analysis. For longitudinal analyses, we included the difference in WAZ/underweight (WAZ < –2) and WHZ because of the increased probability that weight could change in 1 mo. For follow-up, only HAZ/stunting (HAZ < –2) was examined as the linear growth outcome due to the low probability of change in 1 mo (22). Difference-in-difference changes were determined by subtracting baseline values from follow-up and comparing changes across groups. This method was employed for WAZ, WHZ, and plasma and stool biomarkers.

Two sets of multivariate models were tested: 1) logistic regression to examine the association between the nutritional biomarkers and underweight and stunting at 1-mo follow-up and 2) linear regression models for primary exposures on anthropometric z scores. Using the logistic regression model allowed us to assess for trends in the nutritional biomarkers concerning the smaller sample size. In the first analysis, the crude OR was determined and adjusted for by variables where the association was potentially evident (P < 0.15). Then, models were assessed by omnibus tests of model coefficients (significant P < 0.05), Nagelkerke R2 for variability, and goodness of fit by Hosmer and Lemeshow. The Benjamini and Hochberg procedure was used to control for a maximum false discovery rate of 0.05 (35).

In the multilinear regression analysis, we examined the primary exposure and z score outcomes with the coefficients for all covariates to provide insight into the varying contributions of factors. Regression modeling diagnostics were applied to test the underlying assumptions, examine collinearity, and optimize model fit. Stepwise regression analysis was done for age, sex, socioeconomic and demographic factors, WASH variables (water, hygiene, and sanitation), and dietary intake variables. Covariates were retained in final models when P < 0.15 in association with growth outcomes (follow-up HAZ, change in WAZ, change in WHZ), with few exceptions. The child's last dietary intake before the venous draw was assessed, and the time of blood draw was obtained, which was adjusted for in analyses where significant interactions were found. Analyses were repeated with and without the inclusion of outliers. Outliers tested for differences in analyses included I-FABP concentrations >15,000 pg/mL and <200 pg/mL, and difference in I-FABP >5000 pg/mL or < –5000 pg/mL was noted. Our main exploratory variables (enteric biomarkers, DHA, choline, and betaine) were tested in models as interaction terms and using structural equation modeling. Type I error was set to be 2-sided and at a value of 0.05. All analyses were completed using SPSS software (version 27.0; IBM).

Results

Participant enrollment and baseline characteristics

A total of 195 children were enrolled in the study (case: n = 96; control: n = 99), and 136 children returned for the follow-up assessment (). The difference between the number of participants lost to follow-up among cases and controls was nonsignificant. At baseline, our cases and controls appeared similar (). However, symptoms associated with acute gastroenteritis, such as vomiting and poor appetite, were significantly higher in the cases compared with controls. WASH practices and dietary intake of breastmilk, fruits, vegetables, and ASF were comparable between groups. Intake of bonbon sel, equivalent to a thin cracker made of white flour, was reported in 43% of all participants and was a significant component of the diet.

FIGURE 2

Flow diagram illustrating progression and biomarker analyses in cases and controls. Specific sample size per variable is shown. Children who met the inclusion and exclusion criteria were assessed for ≥3 liquid stools over 24 h in the last 3 d. Those who met this definition of diarrhea were assigned to the cases, and those who did not were considered controls. I-FABP, intestinal fatty acid–binding protein.

TABLE 1

Baseline characteristics by group

	Control (n = 99)	Case (n = 96)	P Value
Child
Age,2 mo	18.8 ± 8.3	17.1 ± 7.5	0.114
Sex: female	56.9	43.1	0.075
Dietary intake
Currently breastfeeding	47.5	52.5	0.243
Times breastfeed, 24 h	14.4 ± 5.8	13.3 ± 6.3	0.182
Bonbon sel	42.7	46.2	0.645
Animal source foods	48.7	51.3	0.478
Eggs	15	12	0.478
Fish	4.1	11.7	0.129
Morbidities, 14-d recall
Vomiting	14.6	31.3	0.0293
Suppressed appetite	34	54.2	0.0053
Nasal congestion/rhinorrhea	68.7	63.5	0.448
Cough	51.5	56.3	0.594
Difficulty breathing/wheezing	15.2	15.6	0.507
Rash	23.7	19.8	0.509
Fever, >38.0 °C	2.1	5.3	0.248
Days with diarrhea2	4.7 ± 3.7	2.8 ± 1.6	0.0133
Vaccinations received
Polio	100	97.3	0.117
Rotavirus	91.5	91.9	0.437
Typhoid	38.2	47.1	0.095
Maternal
Maternal age,2 y	29 ± 6.7	25.7 ± 5.7	0.086
Completion of secondary school and higher	50.6	47.9	0.7
Household
Household occupancy	5.9 ± 2.1	6.0 ± 2.2	0.784
Drinking bottled water	86.9	89.6	0.557
Electricity	22.4	46.9	0.0013
Material floor is rock or dirt	9.2	22.9	0.0233
Flush toilet	11.2	3.1	0.15
Number of people using toilet[2,4]	2.8 ± 1.3	2.7 ± 1.1	0.886

Values are presented as mean ± SD and percentages. T test and χ2 test were used to assess for statistical significance unless otherwise specified.

Mann-Whitney U test.

P < 0.05.

Control, n = 42; case, n = 34.

Child anthropometry

There were no significant differences in anthropometric z scores or undernutrition indicators between groups at baseline, nor were there significant differences in growth when the difference-in-difference analysis was applied between cases and controls ().

TABLE 2

Anthropometric data compared between cases and controls at baseline and 1-mo follow-up

	Baseline			Follow-up			Change from Baseline
	Control (n = 97)	Case (n = 96)	P Value	Control (n = 69)	Case (n = 67)	P Value	Control (n = 69)	Case (n = 67)	P Value
HAZ	–1.10 ± 1.23	–1.14 ± 1.31	0.78	–1.18 ± 1.28	–1.28 ± 1.15	0.63	–0.05 ± 0.69	–0.03 ± 0.54	0.9
WAZ	–0.87 ± 1.33	–1.00 ± 1.16	0.47	–0.81 ± 1.17	–0.98 ± 1.03	0.38	0.03 ± 0.51	0.08 ± 0.53	0.61
WHZ	–0.37 ± 1.23	–0.53 ± 1.01	0.33	–0.26 ± 0.97	–0.46 ± 0.98	0.22	0.01 ± 0.78	0.05 ± 0.71	0.77
Stunted	20.4	25	0.44	21.7	25.4	0.61	−1.3	−0.4	0.34
Underweight	21.4	17.7	0.51	15.9	14.9	0.87	5.5	2.8	0.84
Wasted	9.2	10.4	0.77	4.3	4.5	0.97	4.9	5.9	0.79

Values are presented as mean ± SD and percentages. HAZ, height-age z score; WAZ, weight-age z score; WHZ, weight-height z score.

Biomarkers of nutrition and enteric damage

Plasma DHA and choline concentrations were nonsignificant when compared between cases and controls (). However, plasma concentrations of I-FABP were significantly higher in cases than in controls at follow-up.

TABLE 3

Biomarkers of nutrition, enteric damage, and enteric inflammation compared between cases and controls at baseline and 1-mo follow-up

	Baseline					Follow-up
	Controls		Cases			Controls		Cases
	n	Mean (95% CI)	n	Mean (95% CI)	P Value2	n	Mean (95% CI)	n	Mean (95% CI)	P Value3
Plasma DHA, µg/mL	25	0.95 (0.79, 1.10)	24	1.11 (0.91, 1.31)	0.19	25	0.64 (0.52, 0.76)	24	0.82 (0.65, 0.99)	0.17
Plasma choline, µg/mL	25	4.02 (3.39, 4.66)	24	4.94 (3.80, 6.07)	0.27	25	2.53 (2.21, 2.85)	24	3.58 (2.70, 4.46)	0.14
Plasma betaine, µg/mL	25	8.02 (6.70, 9.24)	24	7.29 (6.15, 8.44)	0.40	25	7.65 (6.62, 8.67)	24	6.88 (5.74, 8.02)	0.22
Betaine:choline ratio	25	2.35 (1.88, 2.81)	24	1.63 (1.00, 2.27)	0.19	25	3.18 (2.74, 3.62)	24	2.46 (1.86, 3.05)	0.0494
Log I-FABP, pg/mL	93	3.15 (3.10, 3.20)	90	3.22 (3.10, 3.38)	0.13	62	3.20 (3.13, 3.27)	62	3.34 (3.28, 3.40)	0.0024
Log FC, ng/mL	21	3.43 (2.91, 3.94)	22	3.39 (2.91, 3.87)	0.77

Statistical significance was determined by Student t test and was considered for P < 0.05, except where indicated. FC, fecal calprotectin; I-FABP, intestinal fatty acid–binding protein.

Statistical significance was determined by Mann-Whitney U test and was considered for P < 0.05 for plasma choline and betaine:choline ratio at baseline.

Statistical significance was determined by Mann-Whitney U test and was considered for P < 0.05 for plasma DHA, plasma choline, and plasma betaine at follow-up.

P <  0.05.

Paired analyses of the entire cohort showed mean plasma DHA concentrations at baseline (1.03 µg/mL; 95% CI: 0.91, 1.15) to be significantly higher than those at follow-up [0.73 µg/mL; 95% CI: 0.63, 0.83; t(48) = 4.9, P < 0.001]. Similar findings were also noted with plasma choline concentrations, with mean baseline values (4.5 µg/mL; 95% CI: 3.8, 5.1) significantly higher than those at follow-up [3.0 µg/mL; 95% CI: 2.6, 3.5; t(48) = 5.97, P < 0.001]. These findings were consistent between the case and control groups, with the difference-in-difference changes of plasma choline and DHA nonsignificant when compared between groups. In contrast, the mean betaine:choline ratio at baseline (2.0; 95% CI: 1.7, 2.3) was significantly lower than the ratio at follow-up [2.8; 95% CI: 2.5, 3.2; t(48) = –5.6, P < 0.001]. The difference-in-difference changes of plasma betaine and the betaine:choline ratio were nonsignificant between groups.

After log transformation, plasma I-FABP concentrations at baseline were significantly lower than those at follow-up [t(59) = 3.3, P = 0.002] in the cases but not the controls. The difference-in-difference changes for plasma I-FABP were nonsignificant in cases as compared with controls.

Analyses of plasma I-FABP concentrations in association with age showed that children in the control group had I-FABP concentrations that inversely correlated with age at follow-up (Spearman correlation r = –3.25, P = 0.009). Decreasing I-FABP concentrations with age were not appreciated in the case group, where plasma I-FABP concentrations remained high regardless of age (r = 0.008, P = 0.95).

Nutritional biomarkers in association with underweight and stunting

In the first set of multivariate models, plasma biomarkers of nutrition were assessed after adjustment for maternal education, maternal age ≤19 y, ASF intake at baseline, and time of blood draw to determine the association with underweight status at follow-up (). Adjusted analyses did confirm that covariates—including breastfeeding, maternal education, adolescent mother, ASF intake at baseline, and time of blood draw—were significantly associated with underweight status. A similar multivariate analysis was done to assess the relationship of nutritional biomarkers with stunting at follow-up after adjusting for age, breastfeeding at baseline, and time of blood draw. Plasma DHA, choline, and betaine concentrations were nonsignificant in association with stunting. In addition, the goodness of fit using Hosmer and Lemeshow was nonsignificant in both models.

TABLE 4

Association between nutritional biomarkers and underweight or stunted at the 1-mo follow-up

	Underweight			Stunted
	Crude OR (95% CI)	Adjusted OR (95% CI)2	Adjusted P Value	Crude OR (95% CI)	Adjusted OR (95% CI)3	Adjusted P Value
Baseline
Plasma DHA, µg/mL	0.29 (0.03, 2.61)	0.08 (0.001, 4.16)	0.207	0.64 (0.14, 3.40)	0.69 (0.11, 4.47)	0.840
Plasma choline, µg/mL	0.70 (0.39, 1.26)	0.43 (0.14, 1.31)	0.233	1.10 (0.83, 1.45)	1.16 (0.85, 1.58)	0.513
Betaine:choline ratio	1.96 (0.90, 4.26)	3.81 (0.91, 16.0)	0.136	1.43 (0.76, 2.69)	1.40 (0.72, 2.74)	0.644
Follow-up
Plasma DHA, µg/mL	0.12 (0.01, 2.65)	0.09 (0.002, 3.27)	0.282	3.32 (0.54, 20.6)	8.68 (0.80, 94.2)	0.456
Plasma choline, µg/mL	0.25 (0.05, 1.37)	0.10 (0.01, 0.90)	0.120	1.24 (0.85, 1.80)	1.48 (0.93, 2.34)	0.288
Betaine:choline ratio	2.19 (1.02, 4.70)	5.65 (1.34, 23.8)	0.108	1.09 (0.65, 1.84)	0.95 (0.51, 1.77)	0.949

OR and 95% CI were determined by logistic regression and adjusted P values by the Benjamini-Hochberg procedure.

Adjusted for maternal education, adolescent mother, intake of animal source foods at baseline, and time of blood draw.

Adjusted for age, breastfeeding at baseline, and time of blood draw.

Determinates of growth faltering

The second multivariate model found that sociodemographic determinants contributing to growth faltering included age, source of drinking water, days with diarrhea, and intake of ASF (). Case-control group assignment did not significantly contribute to growth faltering (P > 0.15). After adjusting for multiple comparisons, plasma DHA concentrations at follow-up were significantly associated with growth outcomes. I-FABP concentrations at follow-up were inversely related to HAZ (P = 0.035). Factors and biomarkers identified as significant in regression models were nonsignificant in mediation analysis with structural equation modeling.

TABLE 5

Multilinear regression model for predicting change in growth parameters

	Height-Age Z Score2		Weight-Age Z Score		Weight-Length Z Score
	β (SE)	P Value	β (SE)	P Value	β (SE)	P Value
Child age, mo	–0.05 (0.02)	0.0083	0.02 (0.02)	0.371
Baseline nutrition
Breastfeeding			0.29 (0.30)	0.402
ASFs	0.85 (0.32)	0.013
Egg intake					0.35 (0.17)	0.058
Bonbon sel					–0.22 (0.09)	0.0263
Morbidities
Vomiting, baseline					0.37 (0.17)	0.057
Days with diarrhea, baseline	0.10 (0.05)	0.056			–0.01 (0.01)	0.076
Days with diarrhea, follow-up					0.06 (0.06)	0.331
Household
Drinking bottled water	–1.34 (0.53)	0.0163
Drinking public pump water					–1.42 (0.25)	<0.0013
Number using toilet			0.37 (0.09)	0.0243
Biomarkers
DHA, follow-up			–1.97 (0.40)	0.0163
I-FABP, pg/mL, follow-up	–1.37 (0.63)	0.0353
I-FABP, pg/mL, difference					7.25e−5 (0.00)	0.055
Adjusted R2	0.28	0.0013	0.85	0.0383	0.73	0.0013

Values are coefficient β (SE) for the difference in growth parameters from baseline to the 1-mo follow-up unless otherwise specified. ASF, animal source food; I-FABP, intestinal fatty acid–binding protein.

Height-age z score at 1-mo follow-up.

P <  0.05.

Discussion

In this longitudinal case-control study, we examined the relationship between plasma DHA and choline concentrations on growth faltering in children with and without diarrheal illness. Findings showed that plasma DHA and I-FABP concentrations were associated with growth outcomes in young Haitian children after adjusting for confounding factors. Moreover, plasma I-FABP was significantly higher in those with diarrheal illness and remained elevated at the follow-up time point. This study contributes to the limited evidence base for these nutrients and their role in enteric health, diarrhea, and child growth. In our view, it suggests a crucial need for ongoing investigation into the potential impacts of integrated nutrition and diarrhea prevention interventions.

DHA status of young children may indicate different processes at the nexus of nutrition and infection. Our study found that children in Haiti had lower plasma DHA concentrations (mean: 1.03 µg/mL; 95% CI: 0.91, 1.15) than healthy children of similar ages from other contexts (36). These concentrations matched the plasma lipid profiles of severely malnourished Pakistani children (37). However, the pathophysiology for low plasma DHA concentrations in children with undernutrition and diarrheal illness is still unclear. Plasma DHA has been shown to be influenced by diet (10), though nutrient losses in the stool may occur particularly in those with underlying enteropathy (6). Our study showed that higher concentrations of plasma DHA at follow-up were negatively associated with a change in WAZ but not stunting or HAZ. These findings diverge somewhat from other studies showing an association with stunting (18), though this may be explained by the shorter follow-up period, lower intake of foods containing DHA, and high prevalence of diarrhea in our sample.

Plasma DHA concentrations were nonsignificantly higher in cases than controls. Although this did not align with our original hypothesis, the increase in the systemic inflammatory response in children with acute illness may justify these results. In a cross-sectional study of 1609 children aged 6–23 mo with moderate acute malnutrition, researchers noted that whole blood DHA concentrations were negatively correlated with C-reactive protein, which is generally elevated during systemic inflammation. Conversely, they noted a positive association between DHA concentrations in whole blood and the presence of diarrhea (38). Since plasma DHA concentrations fluctuate more than whole blood DHA concentrations, it is plausible that plasma DHA may increase to attenuate the inflammatory reactions in response to infection (39–41).

Plasma choline concentrations and the association with growth outcomes differed from a study by Semba et al. (29), which reported that HAZ positively correlated with serum choline concentrations and negatively with betaine and betaine:choline ratio. Nevertheless, we found that the betaine:choline ratio was significantly lower in cases as compared with controls at the follow-up visit, which may be explained by choline homeostasis pathways, where the oxidation of choline to betaine is reduced in response to a choline-deficient diet to ensure that choline is available for lipid metabolism (28). To our knowledge, no studies have evaluated changes in plasma choline in the setting of acute diarrheal illness in young children, thus warranting more extensive trials for further investigation.

Malnutrition and recurrent diarrheal illness are associated with morphologic and functional changes within the intestine (42). In previous studies, plasma I-FABP was linked to EED and growth faltering and, as such, can be a valuable, sensitive, and immediate marker for intestinal damage (30, 43). Regarding age-based differences in I-FABP concentrations in the control group, our findings mirror those of other studies, where I-FABP concentrations decrease after 12 mo of age (44). Persistent elevation of I-FABP in cases suggests ongoing intestinal damage in these malnourished children after an intestinal insult. Additionally, I-FABP concentrations were, on average, higher in children 3–12 mo of age living in LMICs when compared with healthy US children and adults (44, 45). This suggests that subclinical intestinal damage may be ubiquitous in LMICs and enteric recovery may be delayed.

Evidence for the relationship between I-FABP and child growth is equivocal in the literature. We showed that plasma I-FABP at follow-up was negatively associated with HAZ at follow-up in multivariate linear regression modeling. Similar results were observed in Peruvian children 5–12 mo of age, where I-FABP concentrations at baseline were significantly higher in children with stunting (46). However, a study on Zimbabwean infants showed inconsistent results regarding I-FABP, noting that plasma I-FABP concentrations were associated with increased length-age z score velocity in children 1–3 mo old but finding the opposite in children 6–12 mo old (47). Although findings have been mixed in previous studies, our results support the use of plasma I-FAPB concentrations in assessing growth outcomes and their potential utility in combination with other biomarkers.

In concert with our primary focus, this study explored other potential determinants of young child growth in the Haitian context, revealing some key factors: breastfeeding frequency, dietary intake of ASFs, quality of drinking water, and the duration of diarrhea. Given the complex interactions of infection, bowel function and integrity, and nutritional status, it is paramount to assess these covariates as they are likely confounders in the association between plasma biomarkers and growth outcomes. Specifically, drinking water from a public pump resulted in an adverse change in WHZ. Poor access to water is common in Haiti, where 31% of people have access only to water contaminated with Escherichia coli (48, 49). Consumption of contaminated drinking water has been associated with poor growth and EED, aligning with our results (50).

Some limitations were present in this study. First, as a case-control observational study, we were limited in the conclusions that we could draw from the findings. However, measures were taken to reduce bias and confounding in regression modeling and the longitudinal design. The number of samples assayed for plasma DHA, choline, and betaine was also reduced due to budget changes during COVID-19. Despite this, we examined trends in those with acute diarrhea and the relationship between intestinal damage and nutritional biomarkers on growth within a community setting in Cap-Haitien, Haiti. This study served as preliminary data for a large randomized controlled trial, Grandi Byen, ongoing to test the efficacy of a parenting and nutrition intervention on a full range of outcomes, including biomarkers assessed here (51).

The findings from this case-control study showed high plasma I-FABP in both groups with significantly higher concentrations in those with acute diarrheal illness, evincing the rampant malnutrition and enteric damage present in young Haitian children. This evidence also suggests a role for DHA in association with growth outcomes, and the overall low concentrations across the sample population point to likely dietary deficiencies. Our results provide formative data highlighting the absolute necessity of these underappreciated nutrients in childhood diarrheal illness and malnutrition.

Click here for additional data file.