Development and validation of a multivariable model for predicting prolonged length of stay following caesarean...

Postpartum length of stay is an important metric of recovery following delivery. Predicting prolonged hospital stay could be useful for postpartum care, facilitate patient counselling, allow targeted interventions for modifiable risk factors and support management of maternal bed capacity. Our aim was to develop and internally validate a predictive model for prolonged length of postpartum stay (≥90^th percentile) following caesarean delivery (CD), with the secondary aim to elucidate factors influencing postpartum length of stay.

Following ethics approval in 107 centres in the UK, we conducted a prospective, multicentre study. Eligible patients were enrolled and baseline demographic, anaesthetic, obstetric and medical data were collected on day 1 postpartum and followed by telephone between day 28 and 32 postpartum, with data on length of stay, patient reported outcome measures, recovery, complications and readmission to hospital.

Data from 1164 patients who underwent CD were included. A total of 119 patients had a prolonged (≥90^th centile) length of stay (≥102 hours). The receiver operator characteristic curve for a prolonged length of stay under a lasso regularised logistic regression model had an area under the curve of 0.7808, with Obstetric Quality of Recovery (ObsQoR) score, neonatal intensive care admission, gestational age and urgency of CD the most important variables.

Using prospectively collected data from a large and diverse national cohort, we developed and validated a model to predict prolonged length of stay following CD in the UK. Further studies are required to determine if targeted interventions can help reduce prolonged length of stay.