Introduction Prematurity is a leading cause of neonatal morbidity and mortality. Tests are available to help predict the likelihood of pre-term labour (PTL), although optimal protocols remain uncertain. We assessed the changing pattern of testing in English maternity units.

Methods 163 maternity units were surveyed online in Sept/Oct 2011, and again in Sept/Oct 2012. In 2012, non-responders were followed up by telephone contact. The overall response rate improved from 32.5% (54 units) in 2011 to 73% (119 units) in 2012. Data were analysed quantitatively using contingency tables, and spatially using Geomapping software.

Results In 2012, 87% (CI; 80–92%) of units used biochemical testing to predict PTL, a significant (p < 0.05) increase from 2011 (76%, CI; 63–85%). For units where data were available for both years, 33% altered their method of PTL testing between 2011–2012, with 40% of these initiating biochemical testing. 14 units did not test for pre-term labour (11%, CI; 7–18%). The most commonly cited barriers to testing were cost and inexperience of operators, each cited by 16% of units (CI; 10–24%). On the basis of test results, 94% (C1; 87–97) of units gave steroids, but only 77% (CI; 67–84) discharged home and 82% (CI; 73–88%) arranged in utero transfer.

Conclusions Our results suggest a heterogeneous pattern of test utilisation. The high proportion of units changing methods within a year implies confusion regarding optimal strategies for PTL prediction. There is an urgent need for further research and clearer guidance in this area. Heterogeneity in protocols could lead to suboptimal allocation of valuable neonatal network resources.