How often are late preterm births the result of non-evidence based practices: analysis from a retrospective cohort study at two tertiary referral centres in a nationalised healthcare system




To determine the proportion, characteristics, and predictors of late preterm birth (LPTB) in relation to evidence-based (EB) and non-evidence based (NEB) indications.


Retrospective cohort study.


Two Canadian tertiary referral centres.


All live singleton LPTBs over 1 year from 2010 to 2011, excluding major congenital anomalies.


Indications for LPTB were classified a priori as EB (i.e. based on practice guidelines or on evidence from randomised controlled trials) or NEB. Data were abstracted from maternal antenatal and labour records. Univariate analyses were completed using Fischer's exact, Pearson's chi-square, or analysis of variance (anova) F–tests. Logistic regression included gestation at birth, delivery provider, previous stillbirth, previous caesarean section, corticosteroid administration, and previous preterm birth as predictors for NEB LPTB.

Main outcome measures

The proportion, characteristics, and predictors of women with NEB versus EB LPTBs.


Of 524 LPTBs, 25.2% (n = 132) were NEB. Logistic regression revealed that NEB LPTBs were less likely if patients were delivered by their own doctor or their doctor's practice partner (OR 0.53, 95% CI 0.34–0.83). However, NEB LPTBs were more likely in women who had experienced a previous stillbirth (OR 2.57, 95% CI 1.20–5.49).


Approximately one-quarter of LPTBs are NEB. Further research is needed to see if a review of the indications for LPTB, and subsequent reduction in NEB LPTBs, translates into improved neonatal outcomes and cost savings.