Collection Mémoires et thèses électroniques
Accueil À propos Nous joindre

Annexe 3 – Impact d’une déficience néonatal en vitamine A sur l’incidence d’infections respiratoires chez les enfants du Nunavik

Table des matières

Cameron C., Dallaire F., Vézina C., Muckle G., Bruneau S., Ayotte P., Dewailly E., 2004. Vitamin A deficiency in cord blood and its impact on acute respiratory infections among preschool Inuit children. En preparation.

Objectif: Évaluer si la concentration de vitamine A dans le sang de cordon ombilical est associée à l’incidence d’infections respiratoires aiguës chez les enfants inuits d’âge préscolaire au Nunavik. Devis d’étude: Les dossiers médicaux de 305 enfants ont été revus pour une période couvrant les cinq premières années de vie. L’association entre la vitamine A dans le sang de cordon ombilical et l’incidence d’otites moyennes aiguës (OMA), d’infections des voies respiratoires supérieures (IVRS) et inférieures (IVRI) et d’hospitilasation pour une IVRI a été évaluée à l’aide de la régression de Poisson. Résultats : Comparativement aux enfants ayant une concentration de vitamine A ≥ 20 µg/dl, les rapport de taux d’incidence ajustée (RT) pour les enfants avec < 20 µg/dl s’étendaient de 1,06 à 1,62 pour les OMA, de 1,12 à 1,34 pour les IVRI et de 1,09 à 1,43 pour les hospitalisations pour une IVRI. La plupart de RT étaient statisitquement significatifs pour les OMA et les IVRI, mais pas pour les hospitalisations pour une IVRI. Aucune association n’a été détectée pour les IVRS. Conclusion : Une déficience néonatale en vitamine A serait un facteur de risque significatif pour l’incidence d’OMA et d’IVRI dans cette population.

Objective: To assess if vitamin A concentration in umbilical cord blood is associated with incidence and severity of respiratory infections in preschool Inuit children from Nunavik. Study design: The medical charts of 305 children were reviewed from 0 to 5 years of age. The association between vitamin A concentration in umbilical cord plasma and the incidence rates of acute otitis media (AOM), upper and lower respiratory tract infections (URTIs and LRTIs) and hospitalization rates for LRTIs was evaluated using Poisson regression. Results : Compared to children with vitamin A concentration ≥  20 µg/dl, adjusted rate ratios (RR) for children below 20 µg/dl ranged between 1.06-1.62 for AOM, 1.12-1.34 for LRTIs, and 1.09-1.43 for hospitalization for LRTIs. Most RRs were statistically significant for AOM and LRTIs, but not for hospitalization for LRTIs. No association was found for URTIs. Conclusion : Neonatal vitamin A deficiency appears to be a significant risk factors for AOM and LRTIs in this population.

Vitamin A plays a major role in growth, development and vision (Sommer 1997, West et al. 1989). Vitamin A deficiency is well known to impair resistance to infection, especially in early age (Bhaskaram 2002, Kapil & Bhavna 2002, Ross 1996, Ross & Stephensen 1996, Semba 1994, West et al. 1989) and many authors have identified increased rates or augmented severity of infection in vitamin A deficient children (Basu et al. 2003, Bloem et al. 1990, Dudley et al. 1997, Pinnock et al. 1986, Sommer 1990, Sommer et al. 1984). Insufficient vitamin A intake has typically been described in developing countries. However, our group recently showed that a significant proportion of Inuit infants from Nunavik (Northern Québec, Canada) were born with low blood levels of vitamin A (Dallaire et al. 2003).

The Nunavik region is located in the northernmost part of the province of Québec. Around 9 600 Inuit inhabit 14 Inuit communities spread out on the Coast line of the Hudson Bay, the Hudson Strait, and the Ungava Bay. Children from Nunavik are burdened by a high rate of acute infections (unpublished results), a situation similar to other Inuit communities throughout the world (Banerji et al. 2001, Davidson et al. 1994, Koch et al. 2002, Wainwright 1996).

This study is the second phase of a previous study that evaluated vitamin A concentration in cord blood of newborns from three regions of the province of Québec (Dallaire et al. 2003). Before considering the implementation of a supplementation program, it was important to evaluate the potential impact of low vitamin A concentration at birth on infection incidence in the first years of life. Thus, in this study, we tested the hypothesis that low vitamin A concentration at birth was associated with higher incidence rate of acute respiratory infections, and with rates of admission for lower respiratory tract infections in the first five years of life.

Participants of this study were originally recruited for evaluation of prenatal exposure to food chain contaminants, results to be published separately.

We attempted to review the medical charts of all children included in our previous study for the first 5 years of age. We made a list of all the medical chart numbers available in our database and worked with the staff of the communities’ health centers to locate the charts. Then, copies of the charts were sent to our research center to be reviewed by five 2 nd - and 3 rd -year trained medical students using a standardized questionnaire. For every diagnosis of infection noted in the charts, we recorded the date of diagnosis, whether antibiotics were prescribed, and whether the child was hospitalized. When the child was hospitalized, we recorded the main reason of hospitalization, whether concurrent illnesses were present, the date of hospitalization, the length of stay, and whether the child was transferred to another hospital. For the present study, only acute respiratory infections were targeted. Four categories were created: upper respiratory tract infections (URTIs), lower respiratory tract infections (LRTIs), hospitalization for lower respiratory tract infections, and acute otitis media (AOM). AOM was analyzed separately from URTIs because AOM is a significant, well-recognized problem in Nunavik. The URTIs category included streptococcal pharyngitis and tonsillitis, acute upper respiratory tract infection not otherwise specified (NOS), acute rhinitis, head cold, nasopharyngitis, pharyngitis, coryza, sinusitis, tonsillitis, laryngitis, tracheitis, croup, and influenza. The LRTIs category included acute bronchitis and bronchiolites, acute lower respiratory infection NOS, chest infection NOS, laryngotracheobronchitis, tracheobronchitis, bacterial and viral pneumonia, bronchopneumonia, influenzal pneumonia, and pneumonitis. For ear infections, only acute otitis media were included. Otitis media with effusion, chronic otitis media and glue ears were excluded.

Two infectious episodes affecting the same anatomic site were considered separate if there was at least fifteen days between the two diagnosis and if it was not specified in the chart that the second episode was related to the first.

Poisson regression was used to evaluate the associations between the vitamin A concentration and the infection incidence rates. The main dependant variables were the number of diagnosed episodes of infection during the first five years of life, and the main independent variable was vitamin A concentration in cord blood. Two regression models were constructed: one in which vitamin A was treated in categories [< 10µg/dl, 10-14µg/dl, 15-19µg/dl and ≥ 20µg/dl (reference)], and one in which it was treated in continuous. The continuous model yielded a single RR corresponding to the relative increase in rate for each 5 µg/dl increase in the concentration of vitamin A.

Adjustment for confounding factors was done using multiple regression (Poisson regression). Potential confounding factors were tested in the model one by one, but only those influencing the incidence rate ratios (RRs) by more than 5% were included in the final model. The variables initially excluded were retested one by one in the final model to ensure that their exclusion did not influence the results. The variables included in the final multivariate model were sex and birth weight (dichotomous, < 3000 g or ≥ 3000 g). Gestational age, smoking during pregnancy, contaminant exposure, and reviewer of the chart were excluded. Some potential confounding factors were available only for children included in the 5-year follow-up subgroup. These factors were breastfeeding duration, crowding, and socioeconomic status. None of these variables influenced the association in a significant manner and they were also excluded from the final model.

We used SPSS Data Entry Builder 2.0 for data entry (Chicago, Illinois, United States) and SAS 8.02 (Cary, NC, United States) for database management and statistical analyses. A p -value < 0.05 was considered significant.

Table 11.3 shows the association between acute respiratory infections and vitamin A concentration in cord blood. For AOM, compared to children in the ≥  20 µg/dl group, children in vitamin A groups of < 10 µg/dl, 10-14 µg/dl, and 15-19 µg/dl had RRs of 1.63, 1.25, and 1.08, respectively. Statistical significance was reached for children with < 10 µg/dl ( < 0.0001), and for those with 10-14 µg/dl ( p  < 0.05). A statistically significant negative association (p<0.0001) was also observed in the continuous model (RRs = 0.88 for each 5 µg/dl increase of vitamin A concentration, p  < 0.0001). Similar results were observed in the adjusted model.

AOM = acute otitis media, URTIs = upper respiratory tract infections, LRTIs = lower respiratory tract infections, CI = confidence interval.

* Statistically significant ( p  < 0.05).

** Statistically significant ( p  < 0.0001).

For URTIs, there was no significant association except for the 15-19 µg/dl group in the adjusted model (RR = 1.16). For LRTIs, compared to children in the ≥  20 µg/dl group, children in vitamin A groups of < 10 µg/dl, 10-14 µg/dl, and 15-19 µg/dl had RRs of 1.17, 1.31, and 1.34, respectively. Statistical significance was reached for children with 10-14 µg/dl ( < 0.05), and for those with 15-19 µg/dl ( p  < 0.05). The dose-response relationship that was apparent for AOM in the categorical was not observed for LRTIs. Nevertheless, the continuous model yielded a statistically significant negative association (RRs = 0.92 for each 5 µg/dl increase of vitamin A concentration, p  < 0.05). Results were similar in the adjusted model, but statistical significance was lost in the adjusted continuous model.

For hospitalization for LRTIs, children in lower vitamin A category all had RRs above 1.0, but statistical significance was reached only for children 10-14 µg/dl in the unadjusted model (RR = 1.66, p  < 0.05). The continuous model showed a negative association (RRs = 0.81 for each 5 µg/dl increase of vitamin A concentration, p  < 0.05), but the statistical significance was lost in the adjusted model.

This study was undertaken in order to document the association between vitamin A concentration in cord blood and acute respiratory infections during the first five years of life in Inuit children. A statistically significant association between lower vitamin A concentration at birth and higher incidence rate was found for AOM and LRTIs, but not for URTIs and hospitalization for LRTIs. For AOM, the association showed a dose-response pattern.

Reports of placebo-controlled prospective studies on supplementation are controversial as some authors found that vitamin A supplementation decreased respiratory infections incidence rates (Pinnock et al. 1986, Sempertegui et al. 1999) while some did not (Barreto et al. 1994, Basu et al. 2003, Biswas et al. 1994, Kartasasmita et al. 1995, Venkatarao et al. 1996). In observational prospective studies, vitamin A deficiency was associated with rate of respiratory infections (Pandey & Chakraborty 1996, Sommer et al. 1984) but not with rate of AOM (Durand et al. 1997). In a cross-sectional, follow-up and interventional trial study performed in Thailand with a similar deficient population, Bloem et al. (1990) found a dose-response relationship between respiratory diseases incidence rates in children and mild vitamin A deficiency.

Our categorical model yielded an apparent dose-response relationship for AOM, but not for LRTIs and hospitalization. Nevertheless, for LRTIs, rate ratios for children in the 10-14 µg/dl and 15-19 µg/dl categories were significant. Furthermore, the continuous model for LRTIs was significant in the unadjusted model, and borderline significant in the adjusted model. These results suggest two possibilities. First, there could be an increased incidence of LRTIs in children with < 20 µg/dl, but we lacked statistical power to identify a clear dose-response pattern in the categorical model. Second, there could exist at cut-off vitamin A value below which the susceptibility to LRTIs is increased without further dose-response relationship. Further studies with a greater number of participants are needed to resolve this issue. In the previous phase of this study, we argued that the cut-off concentration for vitamin A deficiency in umbilical cord blood was 10 µg/dl (Dallaire et al. 2003). The results of the present study, however, show that although no specific clinical effect of vitamin A deficiency, such as night blindness, were apparent above 10 µg/dl, a possible adverse health effect could be present for children below 20 µg/dl. However, because of the unique socioeconomic situation of the Inuit children, it would be hard to infer from our results that such an effect could be present in non-Inuit populations.

It has been discussed by some authors that the effect of a deficient level of vitamin A could better be observed on the severity of infectious episodes rather than on the incidence rate (Roy et al. 1997). Indeed, an association between vitamin A level and severity has been demonstrated in some studies (Basu et al. 2003, Julien et al. 1999), but not all (Kartasasmita et al. 1995). To assess the impact of deficient vitamin A level at birth on the severity of infections in our population, we examined the incidence of hospitalization for LRTIs. Unfortunately, we lacked statistical power and no significant association was found. However, associations were positive and the effect-size in both the categorical and the continuous models was greater when only LRTIs that led to an admission were considered, compared to the total rate of LRTIs. This suggests that not only the rate of LRTIs is increased in children with lower vitamin A levels, but that these episodes were also more severe. Further studies with a greater number of subjects are needed to clarify this issue.

In the present study, a medical chart review was used to evaluate incidence rates of infection. There is only one health center in each community included in this study and participants always go to that health center when they seek medical care. Copies of consultations done elsewhere are also routinely requested to complete local medical charts. We are therefore confident that we have reviewed the majority of outpatient visits. However, parent’s decision of seeking medical attention is related to many cultural factors, which in turn could be associated with dietaries habits. A bias could therefore be introduced if the propensity to seek medical attention was associated with low or high vitamin A consumption. If this bias was present in our data, it is likely that it would be insignificant when severe symptoms were present, symptoms for which every parent would go to the clinic. It would also not be present for hospitalization, as the decision is not one of the parents. We cannot, however, exclude the possibility of a biased association for AOM or URTIs.

Nutrition status is an important factor that can modulate the immune system. Our preliminary analyses allowed us to find that socioeconomic status and crowding, two factors that could be related to malnutrition, did not influence the association between vitamin A and infection rate. However, we found that vitamin A concentration was correlated with omega-3 fatty acids, which means that a greater maternal fish consumption – and most likely a more traditional diet – was associated with a greater vitamin A intake (results not shown). It also means that a lower vitamin A concentration could be associated with a diet composed of a greater proportion of imported food, a diet that is often less-nutritive that the traditional inuit diet (Blanchet et al. 2000). Therefore, we cannot exclude that part of the association between vitamin A and infection rate could be due to other nutriments deficiencies, or to a generally less healthy diet. We did however exclude the potential confounding effect of organochlorines exposure because the association found between prenatal organochlorines exposure and infections in this population was completely independent from the association between neonatal vitamin A and infections (data not shown).

This study underlines a possible link between low vitamin A concentration at birth and acute respiratory infections in Inuit children from Nunavik. Because children from this population are burdened by a high incidence of AOM and LRTIs compared to other North-American populations, the identification of a preventable risk factor such as vitamin A deficiency is of paramount importance. Together with the first phase of this study, these results indicate that a carefully planned supplementation program should be considered for both pregnant women and newborns of Nunavik.

We are grateful to the Nunavik population for their participation in this research. We thank Marie-Lise Mercier, Mélanie Gaudreault, Catherine Lalonde, Élisabeth Leblanc, and Valérie Marchand for medical charts review, and Patsy Tulugak and Mary Nulukie for help with charts retrieval and copying. We are indebted to Daria Pereg for her valuable inputs during the preparation of this manuscript.

© Frédéric Dallaire, 2006