Article Text
PDF
Abstract
Objectives Children from immigrant families may be at risk for iron deficiency (ID) due to differences in pre-migration and post-migration exposures. Our objectives were to determine whether there is an association between family immigrant status and iron stores and to evaluate whether known dietary, environmental or biological determinants of low iron status influence this relationship.
Design This was a cross-sectional study of healthy urban preschool children (12–72 months) recruited from seven primary care practices in Toronto. Laboratory assessment of serum ferritin and haemoglobin and standardised parent-completed surveys were completed between 2008 and 2013 during routine health maintenance visits. Multiple regression analyses were used to evaluate the association between family immigrant status and serum ferritin, ID (ferritin <14 μg/L) and iron deficiency anaemia (IDA) (ferritin <14 μg/L and haemoglobin ≤110 g/L).
Results Of 2614 children included in the analysis, 47.6% had immigrant family status. The median serum ferritin was 30 μg/L and 10.4% of all children had ID and 1.9% had IDA. After adjusting for maternal ethnicity and education, age, sex, income quintile, cow's milk intake, breastfeeding duration and bottle use, there were no significant associations between immigrant status and ferritin, ID or IDA. Significant predictors of low iron status included age, sex, cow's milk intake and breastfeeding duration.
Conclusions We found no association between family immigrant status and iron status after including clinically important covariates in the models. These data suggest immigrant children may not need enhanced screening for iron status or targeted interventions for iron supplementation.
- General Paediatrics
- Race and Health
- Evidence Based Medicine
- Nutrition
- Paediatric Practice
Statistics from Altmetric.com
What is already known on this topic?
Prevention, identification and treatment of iron deficiency (ID) are important for growth and development. Certain dietary and sociodemographic factors are risk factors for ID.
What this study adds?
We found no association between family immigrant status and iron status in English-speaking, preschool-aged children followed in primary care practices.
Introduction
Iron is an essential micronutrient for optimising developmental trajectories in early childhood. Iron deficiency (ID) is the most common micronutrient deficiency in the world with prevalence rates of 5–8% in Canadian urban ‘low-risk’ children.1–3 Early identification of ID is important for mitigating the potentially irreversible deleterious effects of low iron stores on long-term neurodevelopment and behaviour.4
Preschool age children are a high-risk group for ID and iron deficiency anaemia (IDA) because they undergo a rapid phase of growth, leading to a depletion of stores, with relatively low iron bioavailability in the foods they eat. Known risk factors for low iron stores include preterm delivery, low birth weight, longer total duration of breast feeding, low socioeconomic status, excessive cow's milk intake after 1 year of age, obesity and chronic illness.4–12
Children from immigrant families, particularly those from non-industrialised regions where the proportion of children with anaemia is over 50%,13 may be at high risk for ID due to a greater burden of acute and chronic infectious diseases, difficulty accessing health services before and after migration and cultural differences in dietary practices.14–18 However, there is little data on immigration as an independent risk factor for ID. The role of ethnicity in predicting ID has also not been well studied. Immigration and ethnicity may contribute to a number of environmental, behavioural and biological differences that affect iron intake and metabolism.
The primary objective of this study was to determine whether preschool-aged children from immigrant families have lower serum ferritin levels than children from non-immigrant families. Our secondary objectives were to evaluate if there is an association between family immigrant status and ID and IDA, as well as to evaluate whether ethnicity and other known dietary, environmental or biological determinants of low iron status influence this relationship.
Methods
Study design and participants
This was a cross-sectional observational study of urban children living in Toronto, Canada, between the ages of 12 and 72 months whose families could communicate in English. Of the Canadian foreign-born population, 93.5% can converse in English and/or French,19 Canada's two official languages. Healthy children were recruited though TARGet Kids!, a primary care practice-based research network in Toronto between December 2008 and February 2013. TARGet Kids! is a collaborative network between University of Toronto child health outcomes researchers and primary care physicians designed to collect health information to advance knowledge in paediatric primary care in an evidence-based manner. A complete description of the TARGet Kids! cohort has been published.20
Data acquisition
Participants were recruited during scheduled health maintenance visits by trained research personnel embedded in seven participating paediatric and family medicine practices. Data were collected on standardised parent-completed surveys based on the Canadian Community Health Survey.21 Data collected included sociodemographic and dietary information and anthropometric measures. Venous blood sampling for serum ferritin and haemoglobin was obtained at the clinic visit and the results were electronically stored as previously reported by the TARGet Kids! collaboration.6 ,7 ,20
Children were excluded if they had a chronic disease (other than asthma and high functioning autism), acute inflammation (C reactive protein (CRP) >10 mg/L) which may falsely elevate serum ferritin values,4 severe developmental delay or the family could not communicate in English.
Variables
The primary exposure was family immigrant status, divided categorically into non-immigrants, immigrants from non-industrialised regions (based on Unicef's World Regions)22 and immigrants from industrialised regions. Immigrant status was determined from the response to the questions “Where were your child's biological parents born?” and “Where was your child born?” A child was considered to be from an immigrant family if either the parent or the child was not born in Canada. Parental immigrant status has been shown to be associated with child health outcomes.23 ,24 Immigrants were classified into the ‘non-industrialised immigrant’ group if one parent or the child was from a non-industrialised country.
The main outcome measure was serum ferritin level (µg/L). Secondary outcome measures were the presence of ID (ferritin <14 µg/L and haemoglobin >110 g/L) and IDA (ferritin <14 µg/L and haemoglobin ≤110 g/L).
A number of clinically plausible biological and environmental variables that might influence iron stores were included in the analysis. Covariates considered for our models were maternal ethnicity, maternal education, after-tax income, child's age, sex, body mass index (BMI), current daily cow's milk intake, current bottle use and duration of breast feeding. Maternal recall has been shown to be a valid and reliable estimate of breastfeeding duration.25 Maternal ethnicity was captured by the question “What were the cultural origins of your child's ancestors (an ancestor is usually more distant than a grandparent)?” Only maternal ethnicity was included in this analysis as it has been previously validated.26 Ethnicity responses were collapsed and categorised as European, Asian and other. The top ethnic origins of first-generation immigrants in Canada are Chinese and East Indian.19 Weight was measured using a precision digital scale (±0.025%; SECA, Germany) and standing height was measured using a stadiometer (SECA). BMI z-scores were calculated using WHO growth standards.27
Statistical analysis
Descriptive statistics were performed for the main independent variable, outcome variables and covariates. From previous TARGet Kids! studies, we knew ferritin was not normally distributed and was therefore log-transformed to obtain a linear relationship between the main predictor and primary outcome and then back-transformed after analysis for easier interpretability.7 Univariate linear regression was used to determine the unadjusted association between family immigrant status and ferritin as well as with each putative exposure and ferritin. A multivariable linear regression model was developed with serum ferritin as the primary outcome variable with family immigrant status as the main exposure variable. This model was adjusted for the aforementioned covariates which were all specified a priori and included in the final model.
Secondary analyses, which were more exploratory in nature, aimed to test the association between family immigrant status and ID as well as IDA across various demographic and clinical scenarios, as described. Univariate logistic regression was used to determine the unadjusted association between family immigrant status and iron status (ID and IDA) as well as each putative exposure and iron status. Univariable screening with p values of <0.25 identified covariates to be included in the final model. Two multivariable logistic regression models were then developed with ID and IDA as the outcome variables and immigrant status as the exposure variable adjusting for identified covariates from univariable screening. p Values ≤0.05 were considered significant. Because this study was observational in nature, missing data were expected. Multiple imputation methods using predictive mean matching were used to analyse the data. As imputed estimates were consistent with complete case analysis, the latter are presented. Analyses were carried out using SAS, V.9.4. All parents of participating children consented to participation.
Results
In this study, 2697 children of ages 12 and 72 months, meeting study criteria were identified and had venous sampling. Previous TARGet Kids! work has found those who opt to have blood sampled are similar in demographics to those who do not.7 ,20 Two thousand six hundred and fourteen children were included in the main analysis after excluding those with a CRP of >10 mg/L. In testing assumptions of linear regression, studentised residuals were used to assess for influential outliers. Three BMI z-scores were influential and not clinically plausible and were counted as missing. All other data were kept in the analysis. Of the 2614 children with serum ferritin values, 2512 had data on family immigrant status.
Patient characteristics are presented in table 1. The participant median age was 36.7 months (IQR 18.4–50.0 months) and 53% were male. Of the children, 48.5% were from non-immigrant families, while 29.9% were from immigrant families from non-industrialised countries and 17.7% from immigrant families from industrialised countries. Within the immigrant families, 92.4% of children were born in Canada to immigrant parent(s). Ethnicity was predominantly European (67.8%). The median serum ferritin was 30 µg/L (IQR 19–39). Of the children, 10.4% and 1.9% had ID and IDA, respectively. Figure 1 shows the serum ferritin and haemoglobin levels by family immigration status.
Population characteristics
Serum ferritin and haemoglobin of TARGet Kids! cohort by family immigrant status.
For our primary analysis, univariate linear regression revealed no association between the three immigrant groups and serum ferritin (table 2). All per cent changes in predicted serum ferritin reflect changes relative to the sample population median serum ferritin. Asian ethnicity (compared with Europeans) resulted in a 9.0% increase in serum ferritin (p=0.004). For every month increase in age, serum ferritin was 0.5% higher. For every cup increase in daily milk intake, serum ferritin was between 9.2% and 16.2% lower compared with no milk intake (table 2). For a change in income category from $15 000 to $29 999 compared with over $80 000, there was a 24.1% increase in serum ferritin. For every additional month of breast feeding, there was a 0.5% reduction in serum ferritin and not using a bottle resulted in a 5.7% increase in serum ferritin. No other variables were statistically significant in terms of their association with serum ferritin.
Unadjusted and adjusted linear regression models of ferritin and family immigrant status
In the multivariable linear regression model, we found no association between immigrant groups and serum ferritin levels (table 2). Statistically significant covariates associated with increased serum ferritin included older age (p<0.001), male sex (p=0.02), income $15 000–$29 999 (p=0.002), no or low cow's milk intake (p<0.001), shorter breastfeeding duration (<0.001) and no current bottle use (p=0.02). BMI z-score and maternal education were not associated with serum ferritin.
In the multivariable logistic regression model, we did not demonstrate differences in the odds of ID between immigrant groups (table 3). Ethnicity and bottle use also did not demonstrate independent effects on the presence of ID. Covariates that appear to increase the odds of ID include younger age (OR 0.96, 95% CI 0.95 to 0.97), lowest income group (OR 3.42, 95% CI 1.10 to 10.65), longer breastfeeding duration (OR 1.04, 95% CI 1.02 to 1.06) and higher cow's milk intake (OR 1.82, 95% CI 1.13 to 4.22) for three cups of milk per day versus none. Female sex was associated with a decreased odds of ID (OR 0.73, 95% CI 0.54 to 0.99).
Unadjusted and adjusted logistic regression analyses of ID and immigrant status and other covariates
Using a multivariable logistic regression model, we found no association between immigrant groups and IDA, or for the covariates sex, BMI z-score, income group or milk intake and IDA (table 4). ‘Other’ ethnicity had a 2.79-fold higher odds of IDA compared with the European group. As with the other iron status models, breastfeeding duration showed a 7% higher odds of IDA for every additional month of breast feeding and 6% lower odds of IDA for each increase in month of age of the child. Maternal postsecondary education reduced the odds of IDA by 61%.
Unadjusted and adjusted logistic regression analyses of IDA and immigrant status and other covariates
Discussion
We did not identify an association between family immigrant status and iron status, including serum ferritin, ID and IDA, in a cohort of preschool-aged children, 1–6 years of age, living in Toronto, Canada, followed in primary care practices. When biologically and clinically important covariates related to iron intake and synthesis were included in our adjusted model, the observed relationship between exposure groups did not change. Of the children studied, 10.4% and 1.9% had ID and IDA, respectively, which are similar to reported rates from other urban studies in developed nations.1–3 ,28 ,29
Traditionally, children from immigrant families may be considered a vulnerable population due to challenges faced when migrating to a new environment, differences in adherence to Canadian guidelines for recommended dietary intake, fewer social and material resources and more difficulties accessing the healthcare system compared with children born to Canadian-born parents.14 ,30 Cultural differences in dietary practices may also put immigrants at different risks (either protective or harmful) for nutritional deficiencies.18
Conversely, children from immigrant families may experience better health outcomes compared with their Canadian-born counterparts due to selection and settlement policies when immigrating to Canada that bias health outcomes. Support from family or the community to which they migrate can also improve their health. Better health outcomes in adult immigrants have been observed and termed ‘the healthy immigrant effect’.31 ,32 This effect is less well documented in children though studies show better neonatal mortality and low birthweight outcomes and better rates of immunisations in children born to immigrant mothers, despite facing more disadvantages.23 ,31 ,33 These latter observations and theories may, in part, explain the lack of association found in the current study between immigrant status and iron status.
Understanding immigration as an exposure is important due to the high frequency of immigration in much of urban North America, including Toronto. There is a paucity of existing evidence to guide practitioners in nutritional counselling and screening for young children when it comes to immigrant health. Conventional thinking about the vulnerability of immigrants may lead practitioners to perform enhanced screening for growth and nutrition or offer nutritional supplements.14 ,30 ,34 Such practices are not evidenced informed and, based on the findings of this study, may not be necessary for monitoring iron status for some immigrant groups, in particular those with access to primary care.30 Moreover, a recent study of preventative interventions for primary care practitioners identified ID as one of 21 screening recommendations for which there is a lack of evidence.35 We found no other studies in the literature comparing the iron status of preschool-aged immigrants with non-immigrants and only one study suggesting American Hispanic children has worse iron status compared with white non-Hispanic children.9 Studies on pregnant women suggest immigrant women have higher rates of ID compared with native-born women.36 ,37
Major strengths of this study are that it provides a large sample size with a wide range of measured clinical, environmental and biological variables that may affect iron status. The reported income and education levels of our cohort are in line with population-based measures of Canadian family incomes and Canada's relatively educated immigrant population.38 ,39 We included an ethnically diverse sample from one of the most multicultural cities in the world and a large immigrant population, which improves the generalisability of the results.39 Questions used in this study about feeding behaviours have been validated across a number of cultural/ethnic groups in Canada.40
There are a number of important limitations to the study. First, there may be residual confounding from unknown and unmeasured covariates, such as recency of immigration. Given English-speaking immigrants were the focus of this study, excluding those who could not communicate in English may have underestimated some potential effect of immigration. However, 93.5% of Canadian foreign born can converse in either English or French,19 and only 100 of the initial 17 173 patients assessed for study eligibility were excluded on the basis of a language barrier.20 ,32 Because children were recruited from primary care practices with access to a regular care provider and high rates of postsecondary education, arguably, the most vulnerable patients were not included in this study. Therefore, these results may not be applicable to immigrants without a primary care provider. Children were required to give blood samples for study inclusion. Parents of such children may have more health-seeking behaviours and consequent better health compared with the general population. However, rates of ID and IDA in our study were comparable with rates reported from other large urban studies in developed countries.1–3 ,28 ,29
Finally, given that many of the answers to the questions were self-reported, they were subject to recall error. Other studies have shown maternal recall to be a valid estimate for certain nutrition behaviours, including estimating breastfeeding duration.25
Conclusions
We found no association between family immigrant status and serum ferritin, ID and IDA. These data suggest that English-speaking immigrants in primary care practices may not need enhanced screening for ID or targeted interventions to implement iron supplementation to reduce the risk of ID. Ethnicity also did not appear to independently explain differences in serum ferritin or ID, but may have an independent effect in contributing to IDA. Other clinical factors such as milk intake, breastfeeding duration and prolonged bottle use are more strongly associated with ID, which may better guide practitioners as to who should be considered for further evaluation of ID.
Acknowledgments
The authors thank all of the participating families and practitioners, and the paediatric and family medicine practices that are currently involved in the TARGet Kids! research network. They also thank the TARGet Kids! Collaboration Steering Committee (Tony Barozzino, Brian Chisamore, Mark Feldman and Moshe Ipp); the research team (Kathleen Abreo, Dharma Dalwadi, Kanthi Kavikondala, Tarandeep Malhi, Antoinetta Pugliese, Megan Smith, Laurie Thompson); Christopher Allen, Yang Chen, Gerald Lebovic, Magda Melo and Patricia Nguyen at the Applied Health Research Centre and Azar Azad at the Mount Sinai Services Central Laboratory.
References
Footnotes
Collaborators TARGet Kids! Collaboration: Co-Leads: Patricia C. Parkin, Catherine S. Birken, Jonathon L. Maguire; Scientific Advisory: Colin Macarthur, Muhammad Mamdani; Scientific Committee: Kawsari Abdullah, Laura N. Anderson, Imaan Bayoumi, Cornelia M. Borkhoff, Sarah Carsley, Yang Chen, Mikael Katz-Lavigne, Kanthi Kavikondala, Christine Koroshegyi, Christine Kowal, Grace Jieun Lee, Dalah Mason, Jessica Omand, Navindra Persaud, Meta van den Heuvel, Weeda Zabih; Site Investigators: Jillian Baker, Tony Barozzino, Joey Bonifacio, Douglas Campbell, Sohail Cheema, Brian Chisamore, Karoon Danayan, Paul Das, Mary Beth Derocher, Anh Do, Michael Dorey, Sloane Freeman, Keewai Fung, Charlie Guiang, Curtis Handford, Hailey Hatch, Sheila Jacobson, Tara Kiran, Holly Knowles, Bruce Kwok, Sheila Lakhoo, Margarita Lam-Antoniades, Eddy Lau, Fok-Han Leung, Jennifer Loo, Sarah Mahmoud, Rosemary Moodie, Julia Morinis, Sharon Naymark, Patricia Neelands, James Owen, Michael Peer, Marty Perlmutar, Navindra Persaud, Andrew Pinto, Michelle Porepa, Nasreen Ramji, Noor Ramji, Alana Rosenthal, Janet Saunderson, Rahul Saxena, Michael Sgro, Susan Shepherd, Barbara Smiltnieks, Carolyn Taylor, Thea Weisdors, Sheila Wijayasinghe, Peter Wong, Ethel Ying and Elizabeth Young.
Contributors NRS conceptualised and designed the study, carried out the analysis, interpreted the results, drafted the initial manuscript, revised the manuscript and approved the final manuscript as submitted. PCP conceptualised and designed the study, interpreted the results, reviewed and revised the manuscript and approved the final manuscript as submitted. JLM and CSB conceptualised and designed the study, reviewed and revised the manuscript and approved the final manuscript as submitted. CMB conceptualised and designed the study, carried out the analysis, interpreted the results, reviewed and revised the manuscript and approved the final manuscript as submitted. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
Funding NRS was supported by a Restracomp Award through The Research Institute at The Hospital for Sick Children. The Pediatric Outcomes Research Team is supported by a grant from The Hospital for Sick Children Foundation. Support for the TARGet Kids! programme was provided by the Canadian Institute for Health Research (CIHR) Institute of Human Development, Child and Youth Health (IHDCYH), the CIHR Institute of Nutrition, Metabolism and Diabetes and the St. Michel's Hospital Foundation. Funding agencies had no role in the design, collection, analyses or interpretation of the results of this study.
Competing interests None declared.
Ethics approval Research Ethics Board at the Hospital for Sick Children.
Provenance and peer review Not commissioned; externally peer reviewed.