A time of change: iron deficiency and depression among Mexican adolescent girls

Previous studies have also reported an association between iron deficiency and depressive symptoms in children and adolescents. Nonetheless, there is limited information on the prevalence of iron deficiency and depression among adolescent girls in Mexico, and the potential link between iron deficiency and depression is not completely elucidated. Therefore, this thesis aimed to investigate the interrelationships between nutritional exposures and health outcomes among adolescent girls in Mexico, focusing on iron deficiency and depression.

In chapter 2, we investigated the time trends in age at menarche (AAM) during the 20th century in Mexico, and we tested the association between age at menarche and the risk of non-communicable diseases in adult life, using data from adult women from the Mexican National Health Survey. Linear and log-binomial regression were used for nutritional and disease outcomes, while Welch-ANOVA was used to test for a time trend. AAM decreased from 13.6 y in 1920s to 12.6 y in 1980s. AAM was negatively associated with weight and positively with height. AAM was associated with diabetes and hypercholesterolemia, but not with hypertension, breast cancer or arthritis. W e concluded that AAM decreased significantly during the 20th century in Mexico, and earlier AAM was positively associated with chronic disease in adulthood.

Chapter 3,  explored the existing dietary patterns among Mexican adolescents and examined their association with nutritional status. This chapter used data from adolescents aged 12–19 years from the National Survey of Health and Nutrition (ENSANUT-2006). Principal component analysis was used to derive the dietary patterns (DPs). Associations between DP and nutritional status were determined by prevalence ratios.  We identified four DPs: nontraditional and breakfast-type, Western, plant-based, and protein-rich. The prevalence of overweight and obesity was higher in adolescents who scored high on the Western pattern or the plant-based pattern. The Western pattern was also positively associated with anemia in adolescent girls while the nontraditional breakfast-type pattern was inversely associated with anemia in younger adolescents and in girls. The Western pattern was simultaneously associated with overweight–obesity, and anemia. In the context of the double burden of malnutrition, dietary advice must consider malnutrition in all its forms.

Chapters 4-6, result from a cross-sectional study conducted in Santa Catarina and Monterrey, Northern Mexico, from September 2018 to January 2019. We collected data from 408 girls aged 12-20 years from public schools. Exclusion criteria were: diagnosis of systemic disease that may affect iron status, history of major surgery in the last month, and regular use of medication (except contraceptives). To assess depressive symptoms, we used the 6-item Kutcher Adolescent Depression Scale (6-KADS). Blood samples were collected for subsequent assessment of iron indicators and inflammatory markers while fecal samples were collected to evaluate gut microbiota composition. In addition, anthropometrics and other sociodemographic variables were also collected.

Chapter 4 studies the subtypes of depressive symptoms and their relation with iron status, body weight, and pubertal onset. Latent class analysis (LCA) was used to identify and characterize groups of girls based on depressive symptoms. LCA yielded three classes of depressive symptoms; 44.4% of the adolescents were “unlikely to be depressed,” 41.5% were “likely to be depressed,” and 14.1% were “highly likely to be depressed.” Our analyses demonstrated that iron-deficient girls had greater odds of being “likely depressed” (OR=2.01, 95% CI 1.01-3.00) or “highly likely depressed (OR=2.80, 95% CI 1.76-3.84). Linear regression analyses revealed that lower hemoglobin concentrations and higher body weight increased the probability of being “likely depressed.” This study shows that iron-deficient adolescent girls are more likely to suffer from depressive symptoms and that lower concentrations of hemoglobin and higher body weight increase the probability of experiencing depressive symptoms.

Chapter 5 investigated if gut microbiota composition in girls with depressive symptoms differs from that of girls without depressive symptoms. Fecal samples were collected from 139 adolescent girls. The relative quantification of bacterial taxa was done using 16S ribosomal RNA gene amplicon sequencing. Beta-diversity revealed no significant differences in bacterial composition between participants within the different subgroups of   symptoms. Phyla and genera showed no nominal statistical differences between subgroups. However, several other variables (BAZ, ferritin, CRP, and AGP) that may be associated with depression seem to explain slightly the variation in the microbiota composition and suggest a potential link between iron metabolism, inflammation, and depression.

Chapter 6 examines the existing methods to adjust biomarkers of iron status for inflammation and proposes an alternative method to account for inflammation when assessing iron status. For that purpose, inflammatory marker concentrations CRP and AGP were transformed in several ways (via logarithmic, fractional polynomial, cubic root, and inverse hyperbolic sine transformation) to achieve homoscedasticity, to improve model fit, and to normalize the distribution. The balance between model fit and model simplicity on SF and sTfR was tested with Akaike and Bayesian Information Criteria (AIC/BIC) and visual inspection of the residual plots. The best model for SF in our dataset, included CRP transformed by inverse hyperbolic sine but did not include AGP, whereas sTfR did not benefit from any adjustment. The alternative approaches to transform CRP led to an increase by up to 5.0 percentual points in the estimation of iron deficiency. On the other hand, transforming CRP and AGP did not modify the estimation of iron-deficient erythropoiesis. In conclusion, the estimation of iron deficiency can be statistically improved by using alternative methods to transform the markers of inflammation. More research is needed to determine the reproducibility of these alternative approaches to adjust iron status indicators in settings with varying degrees of inflammation.

Finally, chapter 7 summarizes the main findings and integrates them to fill in some of the research gaps mentioned in this introductory chapter. Overall, the research in this thesis illustrates a decrease in AAM in Mexican women. Furthermore, in our secondary and primary data analysis, we show that the double burden of malnutrition persists in Mexican adolescents and that iron deficiency, anemia, and overweight and obesity are associated with the severity of depressive symptom.