Statistical analysis
In the main model, a 2-level logistic regression was employed to estimate associations of different family history types with childhood asthma. Children were considered as the first level and studied districts were the second level. The reference group was participants who had no family member affected asthma. Odds ratios (ORs) and its 95% confidence intervals (CIs) were calculated after adjusting for covariates. Mediation effects of parents between grandparental asthma and childhood asthma were estimated using R package mediation and 95%CIs were estimated using quasi-Bayesian approximation (100 simulations). All statistical tests were two-tailed. P<0.05 was considered statistically significant in the study. Data analyses were conducted using R version 3.6.3. The computation was fulfilled in the high-performance computer system in the School of Public Health, Sun Yat-sen University.