Using latent variable modelling to identify etiological heterogeneity in preterm delivery

Abstract

Aims

To identify a subgroup of mothers at high risk of preterm delivery, defined by empirical classes of multimorbidity and recurrence across three consecutive births.

Methods

The data were extracted from the perinatal data collection (PDC) of all inpatient live births (n = 435 912) occurring in the Australian state of Queensland between January 2009 and December 2015. Within this data, a total of 7714 primiparous mothers delivered three consecutive singleton live births (total births = 23 142), and comprise the sample for all analyses.

Results

The LCA indicated a four-class solution fit the data best at each time point, including (i) a ‘normative’ or healthy class with little morbidity (including >80% of the sample at each birth); (ii) a preterm, high morbidity class (<2% of the sample); (ii) a delivery morbidity class (4–8% of the sample); and (iii) preterm, low morbidity class (5–6% of the sample). Each group exhibited unique and consistent associations with maternal and pregnancy-related factors across births. After accounting for these factors, the high morbidity class and preterm, low morbidity class strongly predicted these same classes across consecutive births, and from birth 1 to birth 3 (second-order transition).

Conclusions

A small but highly morbid class of neonatal deliveries emerged, exhibiting strong continuity across consecutive births (odds ratios >10), independent of a range of maternal and pregnancy-related factors. This group of women, if subject to further investigation, could provide valuable insight into the aetiology of prematurity and associated morbidity, perhaps providing information to improve birth outcomes among all women.