Association of environmental markers with childhood type 1 diabetes mellitus revealed by a long questionnaire on early life exposures and lifestyle in a case-control study

Abstract

Background: The incidence of childhood type 1 diabetes (T1D) incidence is rising in many countries, supposedly because of changing environmental factors, which are yet largely unknown. The purpose of the study was to unravel environmental markers associated with T1D.

Methods: Cases were children with T1D from the French Isis-Diab cohort. Controls were schoolmates or friends of the patients. Parents were asked to fill a 845-item questionnaire investigating the child's environment before diagnosis. The analysis took into account the matching between cases and controls. A second analysis used propensity score methods.

Results: We found a negative association of several lifestyle variables, gastroenteritis episodes, dental hygiene, hazelnut cocoa spread consumption, wasp and bee stings with T1D, consumption of vegetables from a farm and death of a pet by old age.

Conclusions: The found statistical association of new environmental markers with T1D calls for replication in other cohorts and investigation of new environmental areas.

Trial registration: Clinical-Trial.gov NCT02212522 . Registered August 6, 2014.

Keywords: Case–control; Data-driven; Environment; Epidemiology; Type 1 diabetes.

Fig. 1

Flowchart of the samples definition. Missing value criterion is verified if at least half the questionnaire was filled. Delay refers to the time between diagnosis and questionnaire reception. Participants have made the primary school mistake if they answered that they went to primary school even though their reference age is smaller than 5.5 years. The two samples on which analyses were performed are in the bottom right corner

Fig. 2

Volcano plot for the matched analysis. The x-axis shows the effect size with protective factors on the left and risk factors on the right. The y-axis indicates the significance. The higher line indicates the Bonferroni threshold while the lower line shows the more lenient threshold for 5 % of false discovery rate. The unlabeled variables above the FDR threshold are from most significant to least: week-ends with other children, taste for sugar as a baby, death of a pet from old age, vegetables from farm, home-made delicatessen, stings (mainly wasps and bees), siblings before birth, friend’s pool, plane, fresh exotic fruits, vegetables from a rural market during pregnancy

Fig. 3

Volcano plot for the propensity analysis. The x-axis shows the effect size with protective factors on the left and risk factors on the right. The y-axis indicates the significance. The horizontal line indicates the Bonferroni threshold. The unlabeled variables above the threshold are from most significant to least: fruits from a farm or a family garden during childhood, stings, diarrhea, diarrhea during winter, contact with cats in the neighborhood, pet shop, swimming pool during pregnancy and death of a pet of old age

Fig. 4

Comparison of the results of the two analysis. -log10 (p-value) of the two analysis plotted against each other. The most associated variables in both analysis are in the top right corner. The Bonferroni threshold for the propensity analysis is the vertical line. The false discovery rate threshold for the matched analysis is the horizontal line. A more lenient statistical control is used for the matched analysis as it is less prone to bias. All variables passing both thresholds are labeled