ABSTRACT.
The burden of flaviviral infections, especially dengue and Zika, is high in the Americas. Malnutrition affects the risk and response to infections, but the role of diet on flaviviral infection risk is uncertain. The objective of this study was to investigate the relations between dietary patterns adherence and anti-flavivirus IgG seroconversion in children during a Zika epidemic in a dengue-endemic area of Colombia. In 2015–2016, we followed 424 anti-flavivirus IgG seronegative children aged 2 to 12 years for 1 year. Baseline data included children’s sociodemographic, anthropometric, and dietary information collected through a 38-item food frequency questionnaire (FFQ). IgG testing was repeated at the end of follow-up. The primary exposure was adherence to each of four dietary patterns (animal foods, traditional, ultraprocessed foods, and prudent) that were identified from the FFQ through principal component analysis. Secondary exposures were intake frequencies of foods contributing to relevant patterns. We estimated risk of seroconversion by quartiles of adherence scores and compared them using relative risks (RR) and 95% CI from Poisson regression adjusted for sex, age, and socioeconomic status indicators. Seroconversion risk was 32.1%. Adherence to the traditional pattern was positively related to seroconversion. RR comparing fourth versus first quartiles of adherence was 1.52 (95% CI: 1.04–2.21; P trend = 0.02). Of the most representative foods in this pattern, potato and sugarcane water intake frequencies were related to increased seroconversion risk. In conclusion, adherence to a traditional foods pattern, including potatoes and sugarcane water, was positively associated with anti-flavivirus IgG seroconversion.
INTRODUCTION
Vector-borne flaviviruses, including dengue (DENV), Zika (ZIKV), and yellow fever have a wide global distribution and infect an estimated 500 million people annually.1,2 Since the turn of the century, DENV cases reported worldwide have grown from ∼500,000 in 2000 to 5.2 million in 2019.3 In the Americas, the burden of flaviviruses is increasing due to the reemergence of endemic infections such as DENV paired with the introduction of diseases previously limited to other regions of the world, including ZIKV in 2015. DENV infection can progress to potentially fatal forms, and ZIKV can lead to Guillain-Barré syndrome in adults and microcephaly in newborns.4 Clinical infections and progression to severe disease are more likely in infants and children than they are in adults.5 Because no specific treatments against these diseases are available, the identification of modifiable risk factors is a high research priority.
The nutritional status could modulate the risk and response to flavivirus infections. Both micronutrient status and long-term protein and energy balance have been associated with outcomes following DENV infection in epidemiologic studies among children.6 Both underweight7 and overweight8 have been related to increased risk of adverse dengue-related outcomes, whereas specific macronutrient biomarkers including certain long chain n-6 fatty acids, fatty acids from dairy,9 and some aminoacids10–12 have been associated with decreased risk of severe dengue disease. Also, micronutrients including vitamins D13 and E,14 as well as zinc15 have been related to clinical severity of infection. Some of these studies have been cross-sectional, which prevents establishing the temporality of the associations. In addition, little is known on the potential role of dietary patterns (groups of foods frequently consumed together) on flaviviral infection risk or outcomes. Some studies have linked adherence to dietary patterns and risk of adverse outcomes of other viral infections, but their associations with risk of flaviviral infection are unknown. Studying the role of diet from this perspective facilitates the formulation of hypotheses for future testing on the potential effects of the specific foods represented in a given pattern and of the nutrients those foods contribute to the diet. We conducted a prospective investigation of Colombian children to explore the associations of data-driven (a posteriori) dietary patterns on the risk of flaviviral antibody seroconversion as a proxy for the incidence of infection.
METHODS
Study design and population.
We conducted a longitudinal investigation using data from a cohort study designed to identify optimal dengue vaccination strategies. The study was conducted in Piedecuesta, a municipality within the metropolitan area of the city of Bucaramanga, in northeast Colombia. Bucaramanga is located at an average altitude of 1,005 m (range 600–3,600) above sea level, with mean temperature 24°C, relative humidity 72%, and an estimated population of 152,448 in 2015. Recruitment procedures have been described elsewhere.16 In brief, eligible participants lived in seven neighborhoods where the town’s schools and daycare centers were located, and in five neighborhoods with the highest participation rates in a prior investigation on dengue. We aimed to recruit 2,000 children aged 2 to 15 years and 400 adults (≥ 18 years). Before recruitment, the study team provided educational materials about DENV infection and the upcoming study in the target neighborhoods, and trained research assistants visited households door-to-door, inviting families to participate. Enrollment was offered to all eligible children aged 2 to 15 years plus one adult within the household, typically the child’s mother, until the target sample size proportional to each neighborhood’s population was reached. A total of 2,038 children and 408 adults were recruited between June and October 2015.
This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects were approved by the Universidad Industrial de Santander Ethics Committee; the University of Michigan Institutional Review Board approved the use of data from the study (ethics no. HUM00148922). Written informed consent was obtained from all the children’s parents or primary caregivers, and verbal assent to participate was confirmed with children aged ≥ 7 years.
Baseline information.
At the enrollment visit, we conducted a standardized interview with the household adult recruited, to obtain sociodemographic and health status information. We inquired on the participants age, education level, family income, home ownership, housing characteristics, number of residents, and the household socioeconomic status classification according to the local government’s rating for public services fees. Anthropometry was performed according to standardized procedures using calibrated instruments. Height was measured without shoes to the nearest centimeter using a Seca 213 portable stadiometer (Seca, Hanover, MD), weight was measured in light clothing and without shoes to the nearest 100 g using a Kenwell EF-432-BW digital scale (Bedecol, Cali, Colombia), and waist circumference was measured at the midpoint between the lowest rib and the iliac crest to the nearest centimeter using a inextensible Seca 201 measuring tape. At the end of the recruitment visit, research assistants obtained blood samples by antecubital venipuncture; 5 mL were collected in a tube without anticoagulant for separation of serum.
Dietary assessment.
Between October 2015 and March 2016, trained dieticians administered a 38-item food frequency questionnaire (FFQ) to a random sample of 454 mothers of children aged 2 to 12 years, to assess the children’s usual food intake during the previous year. The FFQ was designed for the dietary assessment of Colombian school-age children in a comparable urban setting.17 The food list was based on the most commonly consumed foods in this population as identified by the Colombian National Nutrition Survey.18 All major sources of energy in this age group (frequently consumed foods with high carbohydrate, fat, or protein content) were included. The FFQ had nine frequency response options: four to five times per day, two to three times per day, once per day, five to six times per week, two to four times per week, once per week, one to three times per month, less than once per month, or never. Dieticians described a reference portion in natural units (e.g., one glass of milk or one egg) or standard measures for commonly consumed portions in this population and inquired about children’s frequency of intake. Intake frequencies were transformed into weights using once per day as the reference unit. We estimated energy intakes by multiplying the consumption frequency weights of each food by the energy contents of the specific portion using composition values from the U.S. Department Agriculture’s Standard Reference food composition database, supplemented with data from manufacturers and published reports (Food Processor software; http://www.esha.com) as well as the food composition table of Colombian foods available from the Colombian Institute of Family Welfare.19 In previous investigations, we found strong correlations between intake of animal food sources per the FFQ and plasma vitamin B12 concentrations, which lends support to the questionnaire’s validity.17
Follow-up.
The study team actively followed study participants for one year through regular phone calls and home visits. An annual seroprevalence evaluation was carried out at a home visit one year after each participant’s date of enrollment. At this visit, research assistants obtained a blood sample for quantification of flaviviral antibodies following the same protocols as at the baseline assessment.
Laboratory methods.
Blood samples were placed in insulated containers precooled at 2 to 8°C and transported to the study laboratory (Parque Tecnológico Guatiguará, Bucaramanga, Colombia) on the same day of collection. Serum was separated by centrifugation at 3500 × g for 10 minutes and aliquoted. Serum anti-flavivirus IgG was quantified using the indirect ELISA Panbio kit. The test’s sensitivity is 96%, and the specificity ranges from 91% to 100%. Both baseline and follow-up samples underwent identical transportation, storage, and analytic procedures.
Statistical analysis.
Of the 454 children with dietary assessment, 30 were seropositive at baseline and were excluded from analyses. Thus, the analytic sample consisted of 424 seronegative children.
The study outcome was anti-flavivirus seroconversion risk (percent of children who seroconverted during the year), defined conventionally as a Panbio test result > 11 units at the 1-year follow-up assessment. The primary exposure was adherence to dietary patterns per the FFQ. We used the a posteriori (“data-driven”) approach20 to identify patterns in which foods and food groups intake frequencies cluster together in this population and named the clusters arbitrarily according to their approximate resemblance to known patterns in this setting.17 Responses to the 38-item FFQ were input to a principal component analysis (PCA). An orthogonal transformation was used to rotate the factors obtained to achieve a simpler structure and facilitate interpretability. To determine the number of factors to retain, we considered eigenvalues > 1, the Scree test, and interpretability. We multiplied the standardized intake frequency weights for each food or food group by the factor score coefficients and the sum of these products was the score for each derived factor. Four dietary patterns were identified (Supplemental Table 1): “animal food sources” (e.g., red meat, milk, cheese, yogurt), “traditional” (e.g., potatoes, legumes, sugarcane water, chicken organs), “ultraprocessed foods” (e.g., candy, soda, ice cream, potato chips), and “prudent” (e.g., spinach/chard, uncanned fish or seafood). Secondary exposures were intake frequencies of individual foods contributing to patterns that were related to the outcome. Covariates included child’s sex, age, height- and body mass index (BMI)-for-age z-scores according to the WHO growth reference for children and adolescents,21 maternal education, housing characteristics, socioeconomic status, and regular use of bed net and insect repellant. These variables were categorized as presented in Table 1.
Table 1.
Risk of flavivirus IgG antibody seroconversion according to baseline sociodemographic characteristics in children from Bucaramanga, Colombia
Characteristic | n | Risk (%) | Relative risk (95% CI)* |
---|---|---|---|
Sex | |||
Female | 208 | 27.4 | 1.00 |
Male | 216 | 36.6 | 1.33 (1.01–1.77) |
P† | 0.045 | ||
Age, years | |||
2 to < 5 | 25 | 32.0 | 1.03 (0.57–1.86) |
5 to < 11 | 328 | 31.1 | 1.00 |
11–12 | 71 | 36.6 | 1.18 (0.83–1.66) |
P, trend‡ | 0.47 | ||
Height-for-age z-score | |||
< –2 | 16 | 37.5 | 1.15 (0.57–2.32) |
–2 to < –1 | 86 | 32.6 | 0.97 (0.66–1.42) |
–1 to < 0 | 152 | 31.6 | 1.00 |
0 to < 1 | 126 | 31.7 | 0.98 (0.66–1.45) |
≥ 1 | 44 | 31.8 | 0.98 (0.58–1.66) |
P, trend | 0.77 | ||
BMI-for-age z-score | |||
< –2 | 18 | 27.8 | 1.14 (0.46–2.21) |
–2 to < –1 | 45 | 24.4 | 1.05 (0.58–1.92) |
–1 to < 0 | 113 | 25.7 | 1.00 |
0 to < 1 | 127 | 41.7 | 1.71 (0.98–2.97) |
1 to < 2 | 82 | 26.8 | 1.10 (0.59–2.05) |
≥ 2 | 39 | 41.0 | 1.68 (0.89–3.17) |
P, trend | 0.11 | ||
Maternal education | |||
Primary or less | 113 | 31.0 | 1.00 |
Secondary | 281 | 34.2 | 1.10 (0.80–1.52) |
University | 29 | 17.2 | 0.56 (0.24–1.29) |
P, trend | 0.55 | ||
Home ownership | |||
No | 343 | 33.8 | 1.00 |
Yes | 81 | 24.7 | 0.73 (0.49–1.10) |
P | 0.13 | ||
Housing type | |||
House or apartment | 281 | 32.0 | 1.00 |
One room or multifamily | 143 | 32.2 | 1.00 (0.75–1.35) |
P | 0.98 | ||
People per room | |||
1 | 34 | 26.5 | 0.79 (0.44–1.42) |
2 | 276 | 33.3 | 1.00 |
3 | 86 | 27.9 | 0.84 (0.57–1.22) |
> 3 | 28 | 39.3 | 1.18 (0.72–1.92) |
P, trend | 0.72 | ||
Socioeconomic status§ | |||
1 | 15 | 33.3 | 1.15 (0.57–2.32) |
2 | 211 | 31.3 | 1.00 |
3 | 181 | 34.3 | 0.97 (0.66–1.42) |
4 | 17 | 17.7 | 0.98 (0.66–1.45) |
P, trend | 0.85 | ||
Regular use of bed net | |||
No | 399 | 31.6 | 1.00 |
Yes | 25 | 40.0 | 1.27 (0.77–2.09) |
P | 0.36 | ||
Regular use of repellent | |||
No | 401 | 31.7 | 1.00 |
Yes | 22 | 40.9 | 1.29 (0.77–2.18) |
P | 0.34 |
From generalized estimating equations (GEE) with the Poisson distribution. IgG seroconversion was the dichotomous outcome and indicator variables for each characteristic were predictors. Robust estimates of the variance were specified in all models.
χ2 statistic.
Wald test for a variable representing ordinal categories of the characteristic introduced into a GEE model as a continuous covariate.
According to the local government classification for public services fees.
We first compared the risk of anti-flavivirus IgG seroconversion between categories of baseline characteristics with the use of relative risks with 95% confidence intervals (CI), to identify independent predictors of the outcome. Next, we examined sociodemographic characteristics in relation to quartiles of adherence to each of the four identified dietary patterns using means ± SD and proportions for continuous and dichotomous characteristics, respectively. Tests for linear trend by quartiles of adherence were estimated from linear or Poisson regression models with each characteristic as the outcome and a variable representing quartiles of adherence introduced as a continuous predictor.
Finally, we compared the risk of anti-flavivirus IgG seroconversion between quartiles of adherence to each dietary pattern. We estimated adjusted relative risks and 95% CI from generalized estimating equations with the Poisson distribution in which IgG seroconversion was the dichotomous outcome and predictors included indicator variables for quartiles of adherence to each pattern. Potential confounders were independent predictors of outcome or variables that could be related to the exposure without being its consequence. These encompassed child’s sex, age, height-for-age z-score, maternal education, home ownership, housing type, number of people per room, and household socioeconomic status. In supplemental analysis, we evaluated the associations between intake frequency of specific food items with sociodemographic variables and risk of seroconversion, adjusting for total energy intake in addition to the covariates listed above. Robust estimates of variance were specified in all models. Analyses were conducted with use of Statistical Analysis Software version 9.4 (SAS Institute, Cary, NC).
RESULTS
Mean age of participants was 8.1 ± 2.3 years; 51% were boys. Overall seroconversion risk was 32.1%. Seroconversion was positively associated with male sex (Table 1).
Adherence to an animal foods dietary pattern was inversely related to age and number of people per room and positively associated with maternal education and socioeconomic status (Table 2). Adherence to the traditional foods pattern was inversely associated with height- and BMI-for-age, house size and socioeconomic status and positively associated with the number of people per room. Adherence to the ultraprocessed foods pattern was positively associated with age and inversely with height- and BMI-for-age, maternal education, home ownership, and house size. Adherence to the prudent dietary pattern was not significantly associated with the characteristics examined.
Table 2.
Correlates of adherence to dietary patterns in children from Bucaramanga, Colombia
Correlates* | Quartile (Q) of adherence to dietary pattern | P, trend† | |||
---|---|---|---|---|---|
Q1 | Q2 | Q3 | Q4 | ||
Animal food sources (n) | 103 | 109 | 107 | 105 | |
Male sex (%) | 57.3 | 50.5 | 52.3 | 43.8 | 0.08 |
Age, years | 8.5 ± 2.4 | 7.9 ± 2.2 | 8.5 ± 2.4 | 7.5 ± 2.1 | 0.009 |
Height-for-age z-score | −0.48 ± 0.87 | −0.20 ± 1.07 | 0.00 ± 0.91 | −0.27 ± 1.07 | 0.10 |
BMI-for-age z-score | 0.15 ± 1.10 | 0.32 ± 1.27 | 0.32 ± 1.44 | 0.35 ± 1.29 | 0.26 |
Maternal education is primary or less (%) | 28.2 | 35.2 | 23.4 | 20.0 | 0.06 |
Home ownership (%) | 15.5 | 23.9 | 25.2 | 11.4 | 0.51 |
Housing is one bedroom/multifamily (%) | 35.9 | 33.9 | 29.0 | 36.2 | 0.84 |
People per room | 2.2 ± 1.2 | 2.0 ± 0.9 | 1.9 ± 0.6 | 1.7 ± 0.6 | <0.0001 |
Socioeconomic status | 2.3 ± 0.6 | 2.5 ± 0.6 | 2.5 ± 0.7 | 2.6 ± 0.7 | 0.002 |
Regular bed-net use (%) | 3.9 | 4.6 | 6.5 | 8.6 | 0.12 |
Regular repellent use (%) | 2.9 | 7.4 | 0.9 | 9.5 | 0.18 |
Traditional (n) | 107 | 107 | 104 | 106 | |
Male sex (%) | 50.5 | 50.5 | 59.6 | 37.7 | 0.24 |
Age, years | 8.0 ± 2.3 | 8.0 ± 2.4 | 8.2 ± 2.2 | 8.2 ± 2.4 | 0.55 |
Height-for-age z-score | −0.04 ± 0.98 | −0.25 ± 0.97 | −0.16 ± 1.03 | −0.50 ± 0.96 | 0.0006 |
BMI-for-age z-score | 0.44 ± 1.33 | 0.41 ± 1.12 | 0.27 ± 1.38 | 0.02 ± 1.25 | 0.01 |
Maternal education is primary or less (%) | 20.8 | 36.5 | 25.0 | 24.5 | 0.99 |
Home ownership (%) | 24.3 | 14.0 | 19.2 | 18.9 | 0.51 |
Housing is one bedroom/multifamily (%) | 21.5 | 34.6 | 36.5 | 42.5 | 0.002 |
People per room | 1.9 ± 0.9 | 1.8 ± 0.6 | 2.0 ± 0.8 | 2.2 ± 1.0 | 0.001 |
Socioeconomic status | 2.6 ± 0.7 | 2.5 ± 0.6 | 2.5 ± 0.7 | 2.3 ± 0.6 | 0.002 |
Regular bed-net use (%) | 2.8 | 5.6 | 8.7 | 6.6 | 0.16 |
Regular repellent use (%) | 7.5 | 3.8 | 5.8 | 3.8 | 0.34 |
Ultraprocessed foods (n) | 105 | 101 | 109 | 109 | |
Male sex (%) | 52.4 | 49.5 | 52.3 | 49.5 | 0.79 |
Age, y | 8.1 ± 2.2 | 7.5 ± 2.1 | 8.0 ± 2.3 | 8.7 ± 2.5 | 0.01 |
Height-for-age z-score | −0.14 ± 0.96 | −0.20 ± 1.04 | −0.18 ± 0.97 | −0.42 ± 1.00 | 0.04 |
BMI-for-age z-score | 0.52 ± 1.31 | 0.20 ± 1.07 | 0.35 ± 1.26 | 0.09 ± 1.43 | 0.04 |
Maternal education is primary or less (%) | 18.1 | 28.7 | 22.2 | 37.6 | 0.006 |
Home ownership (%) | 28.6 | 16.8 | 22.0 | 9.2 | 0.002 |
Housing is one bedroom/multifamily (%) | 25.7 | 30.7 | 35.8 | 42.2 | 0.008 |
People per room | 2.0 ± 0.8 | 1.9 ± 0.8 | 1.8 ± 0.9 | 2.1 ± 1.0 | 0.46 |
Socioeconomic status | 2.5 ± 0.6 | 2.5 ± 0.6 | 2.4 ± 0.7 | 2.5 ± 0.7 | 0.82 |
Regular bed-net use (%) | 3.8 | 7.9 | 5.5 | 6.4 | 0.59 |
Regular repellent use (%) | 2.9 | 6.9 | 6.4 | 4.6 | 0.62 |
Prudent (n) | 105 | 106 | 105 | 108 | |
Male sex (%) | 55.2 | 50.0 | 47.6 | 50.9 | 0.49 |
Age, y | 8.3 ± 2.4 | 8.1 ± 2.2 | 8.0 ± 2.3 | 7.9 ± 2.4 | 0.18 |
Height-for-age z-score | −0.24 ± 1.02 | −0.01 ± 1.03 | −0.35 ± 0.96 | −0.34 ± 0.95 | 0.12 |
BMI-for-age z-score | 0.28 ± 1.34 | 0.45 ± 1.37 | 0.17 ± 1.15 | 0.25 ± 1.26 | 0.56 |
Maternal education is primary or less (%) | 33.3 | 25.7 | 21.9 | 25.9 | 0.18 |
Home ownership (%) | 16.2 | 19.8 | 24.8 | 15.7 | 0.85 |
Housing is one bedroom/multifamily (%) | 36.2 | 32.1 | 37.1 | 29.6 | 0.47 |
People per room | 2.0 ± 0.9 | 1.9 ± 0.8 | 2.1 ± 1.1 | 1.8 ± 0.6 | 0.21 |
Socioeconomic status | 2.4 ± 0.6 | 2.5 ± 0.6 | 2.5 ± 0.7 | 2.5 ± 0.6 | 0.67 |
Regular bed-net use (%) | 5.7 | 6.6 | 2.9 | 8.3 | 0.68 |
Regular repellent use (%) | 4.8 | 4.8 | 3.8 | 7.4 | 0.46 |
Values are mean ± SD, unless otherwise noted.
For dichotomous correlates, Cochran–Armitage χ2 test. For continuous correlates, Wald test for an ordinal variable representing the median for each quartile of adherence, introduced into a linear regression model as a continuous predictor.
We next examined the associations between adherence to each dietary pattern and risk of seroconversion (Table 3). Adherence to the traditional foods pattern was positively, nonlinearly associated with seroconversion. Risk of seroconversion (%) in adherence quartiles 1 through 4 was 28.0, 23.4, 33.7, and 43.4, respectively. Compared with children at the lowest quartile of adherence, the risk of seroconversion in those at the highest quartile was 1.52 times higher (95% CI: 1.04–2.21; P = 0.03), after adjustment. Of the most representative foods in this pattern, potato and sugarcane water intake frequencies were related to increased seroconversion risk (Supplemental Table 2). Intake frequency of both food items was inversely related to socioeconomic status indicators (Supplemental Table 3). Adherence to the animal foods, ultraprocessed foods, or prudent patterns was not related to seroconversion.
Table 3.
Adherence to dietary patterns and risk of flavivirus IgG antibody seroconversion in children from Bucaramanga, Colombia
IgG seroconversion | Quartile (Q) of adherence to dietary pattern | P, trend* | |||
---|---|---|---|---|---|
Q1 | Q2 | Q3 | Q4 | ||
Animal food sources | |||||
n | 103 | 109 | 107 | 105 | |
Risk (%) | 36.9 | 23.9 | 36.5 | 31.4 | |
Unadjusted RR (95% CI) | 1.00 | 0.65 (0.42–0.98) | 0.99 (0.69–1.41) | 0.85 (0.58–1.24) | 0.82 |
Adjusted RR (95% CI)† | 1.00 | 0.70 (0.46–1.08) | 1.01 (0.70–1.46) | 0.92 (0.62–1.38) | 0.91 |
Traditional | |||||
n | 107 | 107 | 104 | 106 | |
Risk (%) | 28.0 | 23.4 | 33.7 | 43.4 | |
Unadjusted RR (95% CI) | 1.00 | 0.83 (0.53–1.32) | 1.20 (0.80–1.80) | 1.55 (1.07–2.25) | 0.006 |
Adjusted RR (95% CI) | 1.00 | 0.83 (0.52–1.30) | 1.22 (0.82–1.83) | 1.52 (1.04–2.21) | 0.02 |
Ultraprocessed foods | |||||
n | 105 | 101 | 109 | 109 | |
Risk (%) | 32.4 | 25.7 | 39.5 | 30.3 | |
Unadjusted RR (95% CI) | 1.00 | 0.80 (0.52–1.22) | 1.22 (0.85–1.75) | 0.94 (0.63–1.39) | 0.89 |
Adjusted RR (95% CI) | 1.00 | 0.81 (0.53–1.24) | 1.21 (0.85–1.74) | 0.87 (0.58–1.30) | 0.82 |
Prudent | |||||
n | 105 | 106 | 105 | 108 | |
Risk (%) | 33.3 | 35.9 | 29.5 | 29.6 | |
Unadjusted RR (95% CI) | 1.00 | 1.08 (0.74–1.56) | 0.89 (0.59–1.32) | 0.89 (0.60–1.32) | 0.43 |
Adjusted RR (95% CI) | 1.00 | 1.11 (0.77–1.60) | 0.93 (0.62–1.39) | 0.91 (0.62–1.35) | 0.47 |
RR = relative risk.
Wald test for an ordinal variable representing the median for each quartile of adherence, introduced into the model as a continuous predictor.
From generalized estimating equations with the Poisson distribution. IgG seroconversion was the dichotomous outcome, and predictors included indicator variables for quartiles of adherence to each pattern plus child’s sex, age, height-for-age z-score, maternal education (one indicator for primary), home ownership, housing type, number of people per room, and household socioeconomic status. Robust estimates of the variance were specified in all models.
DISCUSSION
In this longitudinal study of Colombian children, adherence to a “traditional” dietary pattern was associated with anti-flavivirus IgG seroconversion during 1 year of follow-up, independent of demographic, anthropometric, and socioeconomic characteristics. Among the foods in this pattern, potatoes and sugarcane water intake frequencies were positively related to seroconversion risk.
Previous studies of the role of diet on flaviviral infection outcomes focused on anthropometry as a long-term indicator of energy balance and on biomarkers of nutrient intake. To date, most available evidence on the relations between nutritional factors and arboviral infections has centered on the role of the host’s nutritional status on the risk of progression to severe forms of DENV disease. Our approach to define exposure in terms of adherence to dietary patterns is a novel one in the study of flaviviral diseases. Previous studies have found that adherence to dietary patterns characterized by high intake of refined grains and/or snacking foods are related to increased risk or adverse outcomes of other infections such as human papilloma virus, Helicobacter pylori, and HIV.22–24 However, no prior investigations addressed the potential role of dietary patterns on arboviral infections.
IgG antibody seroconversion could result from infection by any flavivirus present in the region, which include DENV, ZIKV, and yellow fever. There has not been an outbreak of yellow fever in the study area since 1923,25 and participants did not receive a yellow fever vaccine during the follow-up period; hence, yellow fever is an unlikely explanation for seroconversion. ZIKV was first identified in the region in late October 2015,26 coinciding with the recruitment and follow-up periods, whereas incidence of DENV infection at the time of the study was similar to that in previous years.27 Hence, seroconversion likely represented new infections with ZIKV or DENV. Host factors, including diet, could modulate the risk of arboviral infection according to antibody seroconversion at different levels. First, these factors could increase the host attractiveness to infective vectors. Mosquitoes locate feeding sources through olfaction; nutritional characteristics including body size, and intake of specific items including beer and bananas28–31 have been related to host attractiveness to mosquito bites, possibly through skin secretion of compounds related to odor. Whether frequent intake of some foods in the traditional pattern, including potatoes and sugarcane water, affect attractiveness to the vectors is an intriguing possibility. Second, host factors could modulate the infectiousness of a bite through immunological or mechanical means. Adherence to specific dietary patterns could be related to the amount and distribution of subcutaneous fat which in turn could affect the number and location of blood capillaries from which a mosquito would feed. Third, because not all infective mosquito bites result in detectable infection, dietary factors might affect viral replication directly or through immunological and inflammatory responses.32,33 Finally, these factors could influence the body’s ability to generate and maintain the antibodies used to define seroconversion. Noncausal explanations of the findings are also plausible. We noted strong associations between indicators of the socioeconomic status and both adherence to the traditional pattern and intake of potatoes and sugarcane water, the pattern’s foods most strongly related to seroconversion. There were also associations of these socioeconomic status indicators (e.g., people per room) with seroconversion, although statistical power was limited. Hence, socioeconomic status could be a confounder of the diet-seroconversion association. Although the analyses were controlled for these socioeconomic factors, it is not possible to rule out residual confounding. Also, high intake of low nutrient density foods including potatoes and sugarcane water could indicate low intake of high-micronutrient-rich foods that might contribute to prevent infection.
This study had several strengths. The longitudinal design precludes reverse causation bias (i.e., that infection causes adherence to a given dietary pattern instead of the opposite) and differential misclassification of the outcome (i.e., that the proportion of children erroneously classified as seroconverting or not differs between quartiles of adherence to a given pattern). Furthermore, the use of an objective measure of infection instead of self-report prevents recall bias. Follow-up was complete, which minimizes selection bias (i.e., that the associations observed in a smaller subset of followed participants differ from those who would have been seen if all had been followed). The dietary patterns identified through PCA of the FFQ were comparable to those previously found in a similar pediatric population,17 which supports the external validity of the dietary questionnaire. In addition, the associations between adherence to the patterns identified and sociodemographic characteristics were in expected directions; this suggests that the internal validity of the FFQ was adequate.
Some limitations are also worth noting. First, as with any observational study, inferring causation would require strong assumptions that may have not been met. For example, residual or unmeasured confounding cannot be ruled out. Lack of variability in some covariates including socioeconomic status could have hindered our ability to control fully for confounding. Also, the association between adherence to the traditional pattern and seroconversion could be due to chance. Second, the FFQ is subject to measurement error; although this is likely nondifferential with respect to the outcome, it could attenuate linear trends. Third, dietary patterns identified a posteriori through a data-driven approach may be difficult to interpret because not all foods with high loading factors within a pattern may align with known or expected configurations for such pattern. Fourth, incidence of clinical infections was low, which prevented us from examining the associations of diet with symptomatic events. In addition, the outcome measure used, anti-flavivirus IgG, is unspecific to a single arboviral infection.
In conclusion, adherence to a traditional foods dietary pattern was positively associated with anti-flavivirus IgG seroconversion in children. The association was primarily driven by intake of potatoes and sugarcane water. Future research on the role of diet on clinical outcomes and immunological responses to arboviral infections could entail examining associations with specific nutrients in foods that contribute to relevant patterns.
Supplemental Material
Note: Supplemental materials appear at www.ajtmh.org.
REFERENCES
- 1. Bhatt S. et al. , 2013. The global distribution and burden of dengue. Nature 496: 504–507. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Gaythorpe KA, Hamlet A, Jean K, Garkauskas Ramos D, Cibrelus L, Garske T, Ferguson N, 2021. The global burden of yellow fever. eLife 10: 10.7554/eLife.64670. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. World Health Organization , 2022. Dengue and severe dengue. Available at: https://www.who.int/news-room/fact-sheets/detail/dengue-and-severe-dengue. Accessed April 14, 2023.
- 4. Baud D, Gubler DJ, Schaub B, Lanteri MC, Musso D, 2017. An update on Zika virus infection. Lancet 390: 2099–2109. [DOI] [PubMed] [Google Scholar]
- 5. Hammond SN. et al. , 2005. Differences in dengue severity in infants, children, and adults in a 3-year hospital-based study in Nicaragua. Am J Trop Med Hyg 73: 1063–1070. [PubMed] [Google Scholar]
- 6. Villamor E, Villar LA, Humphries DL, Scott ME, Vermund SH. Nutrition and Infectious Diseases: Shifting the Clinical Paradigm. Cham, Switzerland: Humana Press, 283–296. [Google Scholar]
- 7. Kalayanarooj S, Nimmannitya S, 2005. Is dengue severity related to nutritional status? Southeast Asian J Trop Med Public Health 36: 378–384. [PubMed] [Google Scholar]
- 8. Zulkipli MS, Dahlui M, Jamil N, Peramalah D, Wai HVC, Bulgiba A, Rampal S, 2018. The association between obesity and dengue severity among pediatric patients: a systematic review and meta-analysis. PLoS Negl Trop Dis 12: e0006263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Villamor E, Villar LA, Lozano-Parra A, Herrera VM, Herran OF, 2018. Serum fatty acids and progression from dengue fever to dengue haemorrhagic fever/dengue shock syndrome. Br J Nutr 120: 787–796. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Voge NV. et al. , 2016. Metabolomics-based discovery of small molecule biomarkers in serum associated with dengue virus infections and disease outcomes. PLoS Negl Trop Dis 10: e0004449. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. El-Bacha T, Struchiner CJ, Cordeiro MT, Almeida FCL, Marques ET, Jr, Da Poian AT, 2016. 1H nuclear magnetic resonance metabolomics of plasma unveils liver dysfunction in dengue patients. J Virol 90: 7429–7443. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Cui L, Pang J, Lee YH, Ooi EE, Ong CN, Leo YS, Tannenbaum SR, 2018. Serum metabolome changes in adult patients with severe dengue in the critical and recovery phases of dengue infection. PLoS Negl Trop Dis 12: e0006217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Villamor E, Villar LA, Lozano A, Herrera VM, Herran OF, 2017. Vitamin D serostatus and dengue fever progression to dengue hemorrhagic fever/dengue shock syndrome. Epidemiol Infect 145: 2961–2970. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Vaish A, Verma S, Agarwal A, Gupta L, Gutch M, 2012. Effect of vitamin E on thrombocytopenia in dengue fever. Ann Trop Med Public Health 5: 282–285. [Google Scholar]
- 15. Rerksuppaphol L, Rerksuppaphol S, 2019. Zinc deficiency in children with Dengue viral infection. Pediatr Rep 11: 7386. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Barry MR , Villar LA Herran OF Lozano-Parra A Estupinan MI Herrera VM Villamor E , 2021. Seropositivity and history of hospitalisation for dengue in relation to anthropometric indices among Colombian children and adults. Epidemiol Infect 149: e58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Villamor E, Mora-Plazas M, Forero Y, Lopez-Arana S, Baylin A, 2008. Vitamin B-12 status is associated with socioeconomic level and adherence to an animal food dietary pattern in Colombian school children. J Nutr 138: 1391–1398. [DOI] [PubMed] [Google Scholar]
- 18. Instituto Colombiano de Bienestar Familiar , 2005. Encuesta Nacional de la Situación Nutricional en Colombia, 2005. https://www.icbf.gov.co/sites/default/files/libro_2005.pdf. Accessed April 14, 2023.
- 19. Instituto Colombiano de Bienestar Familiar , 2005. Tabla de Composición de Alimentos Colombianos. https://www.icbf.gov.co/system/files/tcac_web.pdf. Accessed April 14, 2023.
- 20. Zhao J, Li Z, Gao Q, Zhao H, Chen S, Huang L, Wang W, Wang T, 2021. A review of statistical methods for dietary pattern analysis. Nutr J 20: 37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. de Onis M, Onyango AW, Borghi E, Siyam A, Nishida C, Siekmann J, 2007. Development of a WHO growth reference for school-aged children and adolescents. Bull World Health Organ 85: 660–667. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Barchitta M, Maugeri A, Quattrocchi A, Agrifoglio O, Scalisi A, Agodi A, 2018. The association of dietary patterns with high-risk human papillomavirus infection and cervical cancer: a cross-sectional study in Italy. Nutrients 10: 469. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Shu L, Zheng PF, Zhang XY, Feng YL, 2019. Dietary patterns and Helicobacter pylori infection in a group of Chinese adults ages between 45 and 59 years old: an observational study. Medicine (Baltimore) 98: e14113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Hendricks KM, Mwamburi DM, Newby PK, Wanke CA, 2008. Dietary patterns and health and nutrition outcomes in men living with HIV infection. Am J Clin Nutr 88: 1584–1592. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Groot H, 1980. The reinvasion of Colombia by Aedes aegypti: aspects to remember. Am J Trop Med Hyg 29: 330–338. [DOI] [PubMed] [Google Scholar]
- 26. Instituto Nacional de Salud , 2015. Boletin Epidemiologico Semanal: Semana Epidemiologica Numero 42. http://www.ins.gov.co/buscador-eventos/BoletinEpidemiologico/2015%20Boletin%20epidemiologico%20semana%2042.pdf. Accessed April 14, 2023.
- 27. Instituto Nacional de Salud , 2022. Protocolo de Vigilancia en Salud Pública de Dengue. https://www.ins.gov.co/buscador-eventos/Lineamientos/Pro_Dengue.pdf. Accessed April 14, 2023.
- 28. Port GR, Boreham PFL, Bryan JH, 1980. The relationship of host size to feeding by mosquitoes of the Anopheles gambiae Giles complex (Diptera: Culicidae). Bull Entomol Res 70: 133–144. [Google Scholar]
- 29. Shirai O, Tsuda T, Kitagawa S, Naitoh K, Seki T, Kamimura K, Morohashi M, 2002. Alcohol ingestion stimulates mosquito attraction. J Am Mosq Control Assoc 18: 91–96. [PubMed] [Google Scholar]
- 30. Lefevre T, Gouagna LC, Dabire KR, Elguero E, Fontenille D, Renaud F, Costantini C, Thomas F, 2010. Beer consumption increases human attractiveness to malaria mosquitoes. PLoS One 5: e9546. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Paskewitz S, Irwin P, Konwinski N, Larson S, 2018. Impact of consumption of bananas on attraction of Anopheles stephensi to humans. Insects 9: 129. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Beck MA, Handy J, Levander OA, 2004. Host nutritional status: the neglected virulence factor. Trends Microbiol 12: 417–423. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Broome CS, McArdle F, Kyle JA, Andrews F, Lowe NM, Hart CA, Arthur JR, Jackson MJ, 2004. An increase in selenium intake improves immune function and poliovirus handling in adults with marginal selenium status. Am J Clin Nutr 80: 154–162. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.