Modelling maternal and perinatal risk factors to predict poorly controlled childhood asthma

Abstract

Asthma is the most common non-communicable pulmonary condition, affecting prepubertal boys more often than girls. This study explored how maternal and perinatal risk factors are linked to poorly controlled childhood asthma in a sex dependent manner. This single centre study was performed at a metropolitan teaching hospital in Western Sydney, Australia, using electronical obstetric records from 2000 to 2017 and electronical pediatric records from 2007 to 2018. The data of 1694 children with complete entries were retrospectively analysed. Risk factors for multiple hospital admission for asthma were selected by backward-eliminated Poisson regression modelling. Selection stability of these parameters was independently confirmed using approximated exhaustive search. Sex-specific regression models indicated that most notably parity (RR[95%CI] for parity = 3; 1.85[1.22–2.81]), birth length z-score (1.45[1.23–1.70]) and birth weight z-score (0.77[0.65–0.90]) contributed to multiple asthma admissions in girls, while boys were affected most prominently by maternal BMI (e.g. BMI 35–39.9; 1.92[1.38–2.67]) and threatened preterm labor (1.68[1.10–2.58]). Allergic status was a risk factors for both boys and girls (1.47[1.18–1.83] and 1.46[1.13–1.89]). Applying ROC analysis, the predictive modelling of risk factors for hospital admissions showed an incremental increase with an AUC of 0.84 and 0.75 for girls and boys respectively for >3 hospital admissions. Multiple hospital admissions for asthma are associated with maternal and perinatal risk factors in a sex and birth order dependent manner. Hence, prospective risk stratification studies aiming to improve childhood asthma control are warranted to test the clinical utility of these parameters. Furthermore, the influence of the early in utero environment on male-female differences in other communicable and non-communicable respiratory conditions should be considered.