Introduction

Soil-transmitted helminths infect approximately 1.5 billion people worldwide annually [1], contributing to more than five million disability-adjusted life years (DALYs)[2]. Symptoms of these infections include diarrhea and abdominal pain, and infection can have long-term consequences on development and nutritional status [3]. The term ‘soil-transmitted helminths’ refers to several species of parasitic worms, the most common of which are roundworms, Ascaris lumbricoides, whipworms, Trichuris trichiura, and hookworms, Ancylostoma duodenale and Necator americanus [4]. Pregnant women and children are especially vulnerable to hookworm infection, a known risk factor for iron deficiency anemia, since these populations tend to have the lowest iron stores [5,6]. During their adult worm-stage, helminths live in the intestines of infected individuals and transmit their eggs within human feces. In areas with inadequate sanitation and plumbing, human feces may contaminate the soil which is ingested from improperly cleaned or prepared vegetables, contaminated water sources, or direct ingestion by children [7]. Hookworm species infect individuals through penetration of the skin when it is exposed to larvae from hatched eggs in the soil [8]. Developing countries, particularly in Sub-Saharan Africa (SSA), are at higher risk of soil-transmitted helminth infection due to conditions of poverty, poor sanitation, limited resources, and an abundance of the population living in rural areas [9,10].

Prevalence of soil-transmitted helminth infection during pregnancy (SHIP) in SSA is estimated to range from 11–31%, according to previous studies [10–12]. Often, infected individuals in SSA countries display polyparasitism, meaning they are simultaneously infected with more than one species of helminth [13,14]. SHIP is known to have negative consequences on birth and maternal outcomes, such as pre-term birth, low birth weight, and maternal anemia [15,16]. The World Health Organization (WHO) recommends preventive chemotherapy in the form of a single dose of either 400mg of albendazole or 500mg of mebendazole for pregnant women during the first trimester in areas of the world with a high baseline prevalence of helminth infection (20%) and anemia (40%) in pregnant women [17].

Several studies have shown the detrimental impacts of soil-transmitted helminth infection during childhood on child growth and neurocognitive development [18,19]; however, little research has focused on how exposure to these infections during pregnancy may impact child neurocognitive development. Two studies from a birth-cohort in Uganda have shown that prenatal exposure to helminth infection did negatively impact early cognitive functioning in 15-month old infants [20]; however, anti-helminthic treatment did not reverse or impact cognitive development in these offspring in early or late childhood [21]. Evidence from some studies have shown that infections such as maternal Human Immunodeficiency Virus (HIV) and malaria during pregnancy remain associated with poor neurocognition in children even at 16 years of age [22,23], suggesting that infections during pregnancy can have long-term consequences on child development. A recent study in Benin found associations between hookworm infection during pregnancy and poor cognitive and gross motor scores, as measured by the Mullen Scales of Early Learning (MSEL), in children at one year of age [24]. However, we found no literature that explored these associations in school-aged children. Additionally, there is a growing body of literature on the consequences of prenatal infection on behavioral development in offspring [25–27]. We found no evidence of studies investigating how maternal parasitic infection, specifically SHIP, may impact long-term behavioral development in children.

Hookworm infection during pregnancy has been shown to increase systemic inflammation and more specifically increase erythrocyte sedimentation rate (ESR), a general marker of inflammation [28,29]. Rodent models have demonstrated that maternal infection leads to inflammatory immune responses that can alter brain development of the fetus and have long-term neurodevelopment consequences [30,31]. Existing evidence suggests that systemic maternal infection and subsequent inflammation response disrupts placental vasculogenesis and angiogenesis, which construct the materno-fetal interface during pregnancy, and in turn negatively impacts fetal growth and development [32]. We were unable to identify previous studies investigating the role of the maternal immunoinflammatory response from SHIP on long-term child neurocognitive development.

Therefore, the main objectives of our study were to investigate the associations between SHIP and neurocognitive and behavioral development outcomes in a population of children in Benin prospectively at six years of age. Infections from hookworms were analyzed separately from other helminth infections due to the significant susceptibility in this population to these infections and their subsequent consequences on iron stores. A secondary objective was to test the indirect effects of SHIP exposure on development through potential mediating factors identified in the literature to explore previously hypothesized mechanisms. Our hypothesis was that SHIP would remain associated to impaired neurocognitive and behavioral development in this prospective cohort at six years of age, and that this relationship would be mediated by hypothesized factors such as inflammation.

Methods

Results

Of the 863 singletons born alive to mothers in the MiPPAD clinical trial, 487 (61%) of eligible children were assessed at 6 years of age and included in our analyses (Fig 1). Four children with severe neurological and mental disorders were excluded from analyses, as well as two children who presented with fever (temperature ≥ 37.5°C) at the time of neurodevelopment assessments.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Population flowchart from second trimester of pregnancy until child follow-up at six years of age.

https://doi.org/10.1371/journal.pntd.0009260.g001

Table 1 shows the comparison of maternal and infant demographic characteristics at birth and maternal exposure to helminths during pregnancy between children included and excluded in analyses. Mother-child pairs included in our analyses were of higher maternal age, more likely to be multigravida, and of higher parental education levels.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographic and exposure characteristics of mothers and children included in analyses compared to those lost to follow-up.

https://doi.org/10.1371/journal.pntd.0009260.t001

Prevalence of helminth infection in pregnant women during follow-up was 12.6% at the 1^st ANV, 8.5% at the 2^nd ANV, and 1.2% at delivery (Table 2). The majority of infections in pregnancy were not chronic, with 15% of infected women having more than one infection over the course of her pregnancy. The proportion of infections at each visit caused by hookworm species was 77.9%, 85%, and 100%, respectively. The prevalence of helminth infection in children at six years was 44.6%; 63.7% of infections due solely to hookworms, 27.4% due solely to roundworm, 8.2% due to polyparasitism of hookworm and roundworm species, and <1% due to tapeworm.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Description of infection in women during pregnancy and in children at six years.

https://doi.org/10.1371/journal.pntd.0009260.t002

Crude and adjusted multiple linear regression models tested the associations between helminth infection and hookworm density in women during pregnancy and neurocognitive development outcomes in their children at six years of age (Table 3). Adjusted models were controlled for maternal age, education, pre-pregnancy BMI, family possession score, HOME score, child sex, age at the time of assessment, and malaria incidence during pregnancy. Helminth infection diagnosed in pregnant women at each ANV and overall helminth infection and hookworm density during pregnancy were not found to be associated with neurocognitive development, motor function, or attention/hyperactivity disorder in children at six years of age measured by the KABC-II, BOT-2, and TOVA tests, respectively. Helminth infection in children at the time of neurocognitive assessments was also not found to significantly impair test scores.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Multiple linear regression models between SHIP and child neurocognitive development scores at six years.

https://doi.org/10.1371/journal.pntd.0009260.t003

Table 4 illustrates that SHIP was not associated with a higher total SDQ score in crude or adjusted models. However, SHIP was associated with impaired emotional development as measured by the SDQ internalizing score. Helminth infection at the 1^st ANV was associated with higher internalizing scores 0.82 (95% CI 0.02, 1.62) in the crude model. Helminth infection at the 2^nd ANV/delivery was associated with higher internalizing scores in both crude 1.13 (95% CI 0.21, 2.05) and adjusted 1.07 (95% CI 0.15, 2.00) models. Helminth infection at any point in pregnancy was also associated with higher internalizing scores in crude 0.78 (95% CI 0.78, 1.45) and adjusted 0.79 (95% CI 0.12, 1.46) models. Infection did not seem to significantly impair behavioral development measured by the SDQ externalizing scores. Hookworm density during pregnancy and helminth infection in children at the time of assessment were not found to be associated with SDQ scores.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Multiple linear regression models between SHIP and child behavioral and emotional development scores at six years.

https://doi.org/10.1371/journal.pntd.0009260.t004

Mediation analyses using SEM methods investigated potential mechanisms behind the significant association seen between SHIP at least once during pregnancy and impaired emotional development as measured by the SDQ internalizing score (Fig 2). Models were adjusted for potential confounding factors used in multiple linear regression and coefficients with 95% confidence intervals are shown. Residuals of mediator and outcome variables are represented by ε_1–5. Analyses did not reveal significant mediation through the indirect effects of each mediator: inflammation during pregnancy -0.002 (95% CI -0.03, 0.02), child weight-for-age-z-score 0.02 (95% CI -0.03, 0.06), gestational age at delivery 0.01 (95% CI -0.06, 0.09), or birth weight -0.004 (95% CI -0.03, 0.02). The total effect of SHIP on impaired emotional development was statistically significant 0.78 (95% CI 0.08, 1.47), however less than 3% of the total effects could be explained by the indirect effects of potential mediators tested.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Mediation analysis between SHIP at least once during pregnancy and impaired emotional development in children at six years¹.

https://doi.org/10.1371/journal.pntd.0009260.g002

After multiple imputation of missing outcomes, co-variables, and children lost to follow-up between birth and six years of age, helminth infections at 1^st ANV, 2^nd ANV/delivery, and at least once during pregnancy were significantly associated to higher SDQ internalizing scores after adjusting for confounding factors (Table 5). Hookworm infections at the 2nd ANV/delivery and at least once during pregnancy were associated with higher internalizing scores in crude and adjusted models. Blood lead level in children, known to impair long-term neurocognitive development [41,42], was considered in models in sensitivity analyses. Neither blood lead level measured in children at one year nor at six years had a significant impact on the association between SHIP and child development. Hospitalization of children between one and six years was not found to significantly impact associations between SHIP and child development. Analyses between SHIP and MSEL scores at one year were rerun and confirmed previous findings that hookworm infection in pregnancy was associated with impaired cognitive scores in this population.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Crude and adjusted coefficients (95% CI) from multiple linear regression models after imputation of missing variables and children lost to follow-up.

https://doi.org/10.1371/journal.pntd.0009260.t005

Discussion

This study found that helminth infection during pregnancy was not associated with impaired cognitive ability, gross and fine motor skills, or hyperactivity disorder in children at six years of age. Our analyses did reveal that helminth infection was associated with impaired behavioral development, specifically emotional development in children, after controlling for risk factors. The observed association was not found to be mediated by inflammation during pregnancy, child weight-for-age z-score, or poor birth outcomes. Multiple imputation of missing variables and children lost to follow-up since birth confirmed that both helminth infection and hookworm infection during pregnancy were associated with impaired emotional development.

Our study is the first to our knowledge to investigate how helminth infection measured at different points during pregnancy may impact long-term child neurocognitive development, and is the first study to investigate how these infections during pregnancy impact behavioral development. There is an overall lack of existing literature investigating this subject because of the difficulty in extensive follow-up of SSA cohorts with advanced laboratory and neurocognitive assessment testing. A strength of this study is the prospective nature of the cohort of mothers and children followed from pregnancy until six years post-partum, which included several biological measurements during pregnancy and child development assessments during the follow-up period. The construct validity of the KABC-II was verified in a pilot study within this population, and the SDQ has been used in several studies taking place in SSA countries [36,43,44]. A quality control of the SDQ test was completed at the Global Health Assessment center in Uganda before its use in this study. On-site investigators were trained by study Principal Investigators (PIs) MB and FBL before administering neurocognitive and behavioral assessments. Another strength of this study was the analysis of potential mediators such as gestational age at birth, birth weight, child growth, and inflammation during pregnancy.

Limitations of this study include the small sample size and insufficient power of analyses to show a meaningful difference in KABC-II scores between comparison groups. Given the mean and standard deviation of the KABC-II, a difference of 2.6 points as observed in our results would correspond to an effect size of 0.21; therefore, our study was underpowered to be able to show such a difference in our analyses. Although all investigators administering assessments and questionnaires received the same training and a universal oral translation of instructions was agreed upon, there is still the potential for measurement error. Also, the auto-evaluation of child behavior by parents using the SDQ could have introduced self-reporting bias into our study. However, this is often a bias from child behavioral assessments as many tools rely on observations from parents or teachers. The BOT-2 and TOVA assessments were not validated for use in this population; however, they have been validated for use in other SSA countries [37]. Another limitation of this study is the use of a multitude of statistical tests to explore associations between SHIP and development outcomes, thus the possibility that significant results seen among SDQ internalizing scores could be due to chance cannot be ruled out. Children lost to follow-up between birth and six years were also found to have differing sociodemographic characteristics, however there was no difference in exposure to helminths during pregnancy between these two groups. While this study did take into account several potential confounding and mediating factors, there is still the possibility that some factors were overlooked. For example, conditions of extreme poverty and low social support provided for parents could influence SHIP and neurocognitive development of children and were not included in analyses. Finally, the MiPPAD clinical trial and the APEC nested cohort study supplied helminth prophylaxis to infected women after diagnosis, thus potentially under-estimating exposure to helminths during pregnancy in our population.

A previous study involving 635 Beninese infants found that children with mothers infected by helminths at the 1^st ANV and with mothers who had at least one helminth infection during pregnancy had significantly lower mean gross motor scores and cognitive ability than children with mothers who were not infected [24]. We sought to test whether these children still had lower neurocognitive test scores prospectively at six years of age, however our analyses revealed that helminth infection during pregnancy no longer had an impact on long-term child neurocognitive development. Within sensitivity analyses, we reran the same linear regression analyses conducted in this population of children assessed at one year [24] within the population of 487 children included in this paper. We found the same significant associations between SHIP and cognitive development in these children at one year as those found in the previous publication, thus ruling out the possibility that selection bias is responsible for the lack of significant association between SHIP and long-term neurocognitive development at six years (S1 Table). Helminth infection in childhood has been linked to impaired child development; however, results remain equivocal [19,21]. Additional analyses were conducted within this population to test for associations between helminth infection in children at the time of assessment at six years and impaired development scores; however, no significant associations were revealed. Due to its insignificance in univariate analyses, and the fact that helminth infection in pregnancy and in childhood may share common risk factors, authors did not further evaluate childhood helminth infection in mediation analyses. Within our population, 45% of children were diagnosed with helminth infection at six years, despite the national recommendation in Benin to treat children with anti-helminthic prophylaxis every three months beginning at nine months of age. The high prevalence of hookworm infection in children, as opposed to Ascaris, could be attributed to water, sanitation, and hygiene (WASH) interventions that have been shown to significantly impact Ascaris infection, but have less impact on hookworm infection due to differing modes of transmission of these helminths [45]. This paper provides evidence that these national guidelines are often not respected due to poor governance of public health systems in low-income countries like Benin.

The importance of timing of maternal infection during pregnancy on risk of impaired birth and child outcomes has been widely questioned. Conclusions from previous studies seem inconsistent, with some studies citing no significant difference in child health outcomes based on infection timing during pregnancy [46,47] and others showing that various infections during the second trimester of pregnancy, as opposed to the third trimester, led to more severe neurocognitive and behavioral impairments in children [27,48,49]. Due to the limited existing research on the impact of SHIP on child neurocognitive development, we were unable to form concrete hypotheses for whether or not timing of infection during pregnancy would have a greater impact on neurocognitive scores in children at six years. While results from analyses at one year showed a trend that earlier helminth and hookworm infections (diagnosed in the 2^nd trimester) were associated with lower cognitive and motor scores, we do not see the continuation of this trend in our analyses at six years. Further research within prospective cohorts should investigate the role of the timing of SHIP on child health and development outcomes to make clearer conclusions.

Additionally, our study investigated how SHIP could impact child behavioral development measured by the parent-reported SDQ. The hyperactivity and conduct problem scores are combined to form an externalizing score, indicating external behavioral issues, and the peer and emotional problem scores are used to generate an internalizing score to indicate overall emotional issues. We found that SHIP was associated with higher internalizing scores in children, indicating more emotional development problems in children. Several previous studies have found associations between maternal infection during pregnancy and increased risk of emotional and behavioral problems in children through disruption of blood vessel formation between placenta and fetus that have long-term implications on development [26,32,50]. Maternal infection in pregnancy activates the mothers’ inflammatory immune response; several studies have identified this increase in inflammation as the mechanism that leads to increased risk of behavioral problems and disorders in offspring [51–53]. Rodent models have also shown that increased maternal inflammation during pregnancy can directly increase brain inflammation in the fetus and thus alter fetal brain development [25,52]. One previous study specifically found that elevated markers of inflammation in pregnant women were associated with more internalizing and externalizing symptoms in their offspring later in childhood [54]. Ours is the first study to investigate plausible mechanisms of the association between maternal intestinal worm infection and externalizing and internalizing symptoms in school-aged offspring.

Mediation analysis tested the indirect effects of potential mediators in the association between SHIP and impaired emotional development in children measured by the SDQ. SHIP is a known risk factor for birth outcomes such as pre-term birth [48] and low birth weight [55]. Helminths have also been attributed to stunted growth and wasting in children [2,56]. We examined how these mediators, along with inflammation in pregnancy measured by the presence of CRP, could indirectly affect the significant relationship seen between SHIP and impaired emotional development. However, our analyses did not reveal any significant indirect effects from these mediators. Inflammation within mothers in our population could be the result of several different factors, such as malaria or other maternal infection [57] or increased maternal BMI [58]. Therefore, it is difficult to ascertain whether the cause of inflammation within this population is due primarily to helminth infection in pregnancy. Additionally, CRP was the only marker used to measure inflammation during pregnancy; previous studies have indicated eosinophils, IgE, and other cytokines could be more specific inflammatory markers of helminth infection [28]. Other serum cytokines known to be activated in the inflammatory immune response to helminth infection could be studied in future research to evaluate their role as mediators in this relationship. Sensitivity analyses were conducted to test mediation between hookworm infection and impaired emotional development through inflammation, since hookworm infection in particular has been previously linked to systemic inflammation [28], and similarly no significant results were found. Other potential mediating factors such as maternal anemia and child malnutrition were not considered in this paper. Malnutrition was not included due to the fact that it could be both a potential confounding factor and a mediating factor, and because there are several causes of malnutrition in low-income populations such as this. There are also several causes of anemia in at risk populations, and mechanisms between anemia and long-term child development are still unclear. A separate study exploring this complicated and multi-factorial relationship is currently underway within this population, and therefore was not expanded upon for the purposes of this paper.

This study does not confirm hypotheses that SHIP impacts long-term child neurocognitive development at six years. However, it provides evidence that these infections can influence behavioral development during childhood and beyond. The clinical manifestation of impaired neurocognitive development in SSA populations according to Western-based assessment scores is widely unknown, and studies with comparisons of symptoms within SSA populations are needed. Maternal infection by helminths remains a significant public health challenge in many areas of the world, and efforts to control these infections and understand the full scope of their burden on mother and child health are needed. Additionally, the high prevalence of helminth infection in six-year-old children in this study shows that national prevention guidelines for helminth infection may not be adequately followed and implementation of these guidelines must be strengthened to reduce the burden of these infections in school-aged children. Future research should be done to confirm our findings and further investigate how these parasitic infections can impair child development and the role that maternal inflammation may have on this association.

Supporting information

Acknowledgments

We would like to thank the on-site researchers and public health nurses in Allada, Benin for their collaboration on all studies mentioned in this paper.