Stata ic version 15

Statistical analysis was conducted using STATA/IC version 15. Descriptive statistics were expressed as medians and interquartile ranges (IQR) for continuous variables and as percentages for categorical variables. Comparisons of categorical data were performed with chi-squared test and Fisher's exact test, while continuous data were analyzed with nonparametric Mann–Whitney U test. A p value ≤ 0.01 was considered statistically significant.

All analyses were conducted using STATA IC Version 15 (Stata, College Station, TX, USA). Chi-square tests were used to test for associations between categorical variables. Logistic regression was used to examine associations between potential risk factors and binary outcome variables (physician-diagnosed asthma or the presence of respiratory symptoms). We examined the associations in unadjusted models and stepwise adjustment for a priori selected covariates (minimally adjusted model: age and sex, language of survey response) and fully-adjusted final models (adjusted for age, sex, language of survey response, education level of the parent/caregiver survey respondent, school of enrollment). Statistical significance was defined as p < 0.05.
Children were classified as asthmatic if the child’s parent answered affirmatively to the question “Has your child ever received an official doctor’s diagnosis for asthma?” A child was considered to have bronchitic symptoms if the child’s parent reported at least one of the following: 1) daily cough for three months in a row, 2) congestion or phlegm for at least three months in a row, or 3) bronchitis in the past year [20 (link)]. Health insurance was categorized into public, private, or none. Respiratory medications were categorized as rescue, control, or other [21 ].

Statistical analyses (mean, min, max, standard deviation) were performed with Stata IC version 15 (Stata Corporation, Texas, Usa) and Microsoft Office Excel 2007.

We calculated measures of central tendency for continuous variables and proportions for ordinal and nominal variables. We reported on the overall response rate to the survey and evaluated the distribution of the study variables. Total debt was not normally distributed and was log-transformed to meet the assumptions required to conduct linear regression. We conducted separate univariate linear regression analyses to determine variables associated with our outcomes: log-transformed total debt and quality of life. In the following step, we conducted multivariable linear regression to find variables independently associated with the outcomes. For all other questions, analyses were two-tailed, a p-value < 0.05 was considered significant, and analyses were conducted with Stata I/C version 15, Stata Corp LLC, College Station, TX, USA.