Population

This study was designed as a case-control study. The cases were patients diagnosed with Type 1 DM who were aged 4–14 years and followed in the Endocrine and Metabolic Diseases polyclinic of the Ege University Pediatrics Hospital, between January and March 2020. The controls were children who came to the General Pediatric Outpatient Clinic of the same hospital, with the same age range and were not diagnosed with any chronic disease including Type 1 DM. General Pediatrics outpatient clinic admissions were due to newly developing acute conditions [the majority of the admissions were upper respiratory tract infection (65%), vitamin D deficiency (5%), and healthy child follow-up (3%)], and 85–90% were first visits to the hospital; 10–15% were invited for a follow-up one month later, so the follow-up children were also of the same age. If they have a chronic condition, they go to pediatric specialty clinics and start follow-up in those clinics.

The sample size was calculated as a minimum of 210 children, each group comprising 105 children, using the t-test group in the G*Power program for two groups, with an effect size of 0.5 (a medium anticipated difference between two groups for a scale-type variable), a margin of error of 0.05, and a power of 95%. A total of 252 volunteers were planned to be included in the study by adding 20% to this sample size to account for possible non-response, and the recruitment stopped with a total of 248 volunteers. The case group included 122 children, and the control group had 126, meeting the minimum required sample size. In the data collection process (before reaching the sample size), a total of 11 parents who did not want to participate in the study in the general pediatric outpatient clinic were excluded. In the control group, children who presented to the outpatient clinic with an acute condition but had an underlying chronic disease were also excluded from the study. Also, seven parents did not want to participate in the study because they did not want to share their information, parents of two children thought they would not remember the study questions, and two parents did not want to participate in the outpatient clinic. The response rates for all eligible cases and controls admitted during the data-collecting period were 96% and 91%, respectively.

Questionnaire

The study questionnaire, structured by the researchers in light of the literature, was applied face-to-face. The study was explained to the children included in the study and their parents, and the questions were asked. The questions relating to the characteristics of both cases and controls (sex, age, body mass index, form of delivery, birth order, and birth interval), for cases only (duration of Type 1 DM and age at diagnosis, HbA1c levels at diagnosis were examined retrospectively from the patient file), and both groups’ maternal history of GDM and their family characteristics (family history of Type 1 DM, education level, residence), were included.

Smoking status was ascertained during pregnancy and lactation, and in the household. To evaluate a child’s smoking exposure before diagnosis of Type 1 DM, parents were asked (yes/no) about the status of smoking near the child before diagnosis, and those who answered ‘yes’ were asked for the duration of smoking and the number of cigarettes smoked in the same closed environment with the child at home. Maternal smoking status during pregnancy and lactation was asked with a response of yes/no.

As for anthropometric measures, in calculating the percentile values, reference values for Turkish children were used¹⁵. Body mass index (kg/m²) was used to assess childhood obesity. In the study, children’s BMI percentiles were classified according to the percentiles of the CDC growth chart¹⁶. Children with a BMI <5th percentile were considered underweight, 5th–84th percentile normal weight, 85th–94th percentile overweight, and ≥95th percentile obese.

Measurement of SHS

Mothers were individually asked whether they smoked during pregnancy, and if so, they were further queried about their daily cigarette consumption during that time. The same set of questions was reiterated for the lactational period. Importantly, the case group (children with Type 1 DM) and the control group (healthy children) were subjected to identical inquiries during the maternal and lactational periods. Subsequently, comparisons were performed using the data collected from these specific time frames.

For the household smoking variable, parents of controls were asked about their smoking behavior when they were included in the study. However, for the cases, the inquiry occurred before the diagnosis of Type 1 DM to explore the association with SHS. After collecting data, we compared smoking-related variables between the two groups: for controls during their participation in the study and for cases before the Type 1 DM diagnosis period. Notably, we specifically questioned parents about the number of daily cigarettes smoked at home to examine children’s passive smoking status. The study refers to smoking within the home as household smoking. The exposure resulting from this household smoking is defined as secondhand smoke (SHS). Researchers also examined children’s SHS exposure during three different periods: during pregnancy, during lactation, and the at-home period.

The child’s exposure to cigarettes before the diagnosis of Type 1 DM was calculated by asking the number of cigarettes smoked per day by the parents who smoked before the diagnosis of their children.

Statistical analyses

The data were analyzed using the PASW Statistics Version 18.0 software platform¹⁷. The descriptive variables of cases and controls were compared with two-tailed tests: Student’s t-test (continuous variables), Mann Whitney U test (when a non-parametric test was necessary), and chi-squared test (categorical variables). The smoking status of the parents in the case group before the diagnosis and the number of cigarettes they smoked daily, were compared with McNemar’s test and t-test for dependent groups. To be able to measure the relationship between smoking and the odds of having Type 1 DM disease, the parent’s smoking status before the diagnosis date was questioned. In the case group, the parent’s smoking status and cigarettes smoked per day before and after diagnosis (current, during data collection) were also compared.

Multivariable analysis with the enter method evaluated adjusted odds ratios (AOR) and 95% confidence intervals (CIs) of other confounders. Logistic regression analyses were conducted with the adjusted variables based on biological plausibility (age and sex). Multivariable analyses were performed with sensitivity analyses (Supplementary file Table 1) with models with probable confounders both biological variables (sex, age) and variables that were significant in univariate analysis (birth interval, family history, residence) and other probable confounders (education status) for maternal and lactational smoking status and smoking at home. In all analyses, p<0.05 was accepted as a statistical significance level.

Ethics

Written permission was obtained from Ege University Children’s Hospital, Department of Pediatrics, in order to conduct the research in the relevant institution. Approval was obtained from the Ethics Committee for Clinical Research, Ege University, Faculty of Medicine (decision number 10059, dated 9 January 2020).

Sensitivity analyses

Additional adjustment for early life variables did not modify the identified associations, and the results remained even when missing covariate data, as shown by a complete case analysis with and without covariate adjustment. Combined data on maternal smoking during pregnancy and lactation did not show higher odds of being exposed to maternal smoking in children with Type 1 DM. In all models, children with Type 1 DM had higher odds of being exposed to an increased number of cigarettes smoked by parents at home (Supplementary file Table 1). We also performed a post hoc analysis which showed that we achieved an effect size of 0.28 and a power of 0.59.

Parental smoking at home

Studies investigating the relationship between passive smoking and risk of Type 1 DM, are few and the possible mechanisms have not been clarified^8,20-23. Regarding the mechanism of passive smoking promoting diabetes metabolism, some studies suggest that nicotine affects the function of islet cells and insulin²⁴. Moreover, nicotine promotes the degradation of insulin receptor substrate-1 (IRS-1) by activating AMP-activated protein kinase α2 in adipocytes to increase lipolysis in adipose tissue²⁵. Through these mechanisms, passive smoking disrupts carbohydrate metabolism and promotes the development of abnormal glucose metabolism¹⁹.

We found increasing odds of exposure to a higher number of cigarettes smoked at home by parents among children with Type 1 DM. Therefore, the increase in risk might be dose-dependent. Further studies should focus on the relationship between Type 1 DM and the dose of SHS besides evaluating SHS smoking status only. A 10-year follow-up prospective study investigating the relationship between passive smoking and Type 2 DM showed that women who never smoked and women who were exposed to SHS from spouses had a higher risk of diabetes compared to women who were not exposed. In addition, passive smoking at home, estimated from the number of cigarettes smoked per day by the spouse in the study, was associated with a higher likelihood of developing diabetes²⁶. Another study evaluating the association between smoking and metabolic parameters in patients with Type 1 DM showed that a significant proportion of patients were active or passive smokers, and suggested a relationship between smoking and adverse metabolic profile²².

The United States National Health and Nutrition Examination Survey (NHANES 1999–2010) demonstrated that SHS is correlated with obesity and worse glycemic parameters²⁷. Our study did not compare SHS and glycemic parameters, except HbA1c. We observed that children with Type 1 DM whose parents smoked before diagnosis had slightly higher HbA1c levels at the time of diagnosis compared to non-smokers. However, this result was not statistically significant, but it may be evaluated more efficiently in studies with higher numbers of participants in which the level of cotinine could be measured.

Maternal smoking during pregnancy and lactation periods

We found no association between Type 1 DM and neither maternal smoking during pregnancy or lactation periods. Nevertheless, several studies showed that maternal smoking during pregnancy has a triggering effect²⁸, no effect²⁹, and even a protective effect^30-32 on the risk of Type 1 DM in the offspring. Surprisingly, prospective studies suggest a reduced risk of Type 1 DM in the offspring of mothers who smoked during pregnancy^7,33. According to an animal study, smoking may help prevent the development of autoimmune diabetes by preserving pancreatic insulin content and encouraging anti-inflammatory mechanisms³⁴. This mechanism is thought to be related via the ‘nicotinic anti-inflammatory pathway’³⁵. However, the mechanisms are not fully elucidated, and human studies are inconclusive. On the other hand, these results should be interpreted cautiously as smoking status was self-reported, as in our study, and there was a large amount of missing data in certain studies³⁰. Moreover, the selection of the control group in our study from a tertiary hospital may have caused a difference (and bias) in the results.

Similarly, a meta-analysis study including 23 studies showed that maternal smoking reduces the risk of Type 1 DM in the offspring. However, only one of the included studies³⁶ measured the cord blood cotinine level, which indicated prenatal smoking exposure. In addition, the bias in the research due to the heterogeneity of the studies is underlined³⁷.

Limitations

There are some limitations in the study. First, passive smoking exposure in children was not measured with biological samples, and it was estimated from the self-report of parents. Therefore, it may have been underreported. Second, the reporting of SHS exposure is subject to recall bias. In addition, the selection of the control group from a tertiary hospital may have created a bias in the representation of this group. At the same time, it enabled a similar education distribution in both parents of the two groups.

Given that a significant portion of the control group visited the clinic due to upper respiratory tract infection, and considering the established correlation between passive smoking and upper respiratory tract infection, it is prudent to approach the study results with due caution. However, it is essential to acknowledge that selecting the control group exclusively from a tertiary hospital introduces potential bias in the study. Third, there may be response bias due to parents in both groups, particularly among the cases, who may have misrepresented their smoking status during pregnancy and lactation out of guilt. Fourth, we questioned maternal smoking during pregnancy and the lactation period, but we did not directly ask about paternal smoking in these two periods. Therefore, the results may not accurately reflect pregnancy or lactational exposure to smoking. Finally, those exposed to SHS at home may have many other risk factors for DM or factors that could lead to DM (reduced access to healthcare, lower level of education, socioeconomic status, etc.) that could affect the interpretation of the results.

Future studies with larger populations would increase the power of the study. They would enable the comparison of paternal smoking status besides the maternal smoking status, duration, the number of cigarettes smoked near the children per day, and in which place in the house they smoked (balcony or living room, etc.). Although we performed a sensitivity analysis, residual confounding factors may have been due to unforeseen variables and/or missing data.

The health-related effects of passive smoke are known to be linked with temperature, humidity, ventilation, depth of breathing, and distance from the smoker⁶. Another limitation is that these variables could not be evaluated in our study.