Background
The CONSORT statement calls for complete data on flow of participants, including all losses and exclusions. Incomplete reporting of flow into trials versus flow through trials is not uncommon. Where complete data exist in obstetric trials, poor recruitment seems a recurring theme.

Aims
To explore difficulties in recruitment and differences between assessed‐but‐not‐recruited and included women to improve future trial participation, using a case study of a recently published randomised trial of outpatient Foley catheter versus inpatient PGE2 gel for cervical ripening.

Materials & Methods
The assessed‐but‐not‐recruited population of an obstetric trial (ACTRN:12609000420246) was prospectively studied for reasons for noninclusion, demographic data and pregnancy outcome. Women assessed‐but‐not‐recruited due to declined consent or obstetrician declined participation were compared to included women. Main outcome measures included demographic and outcome differences associated with trial participation.

Results
Of 468 assessed participants, 220 (47%) were not eligible by exclusion criteria (potential ‘trial factor’ recruitment difficulties), 147 (31%) declined consent (n = 100, ‘participant factor’) or their obstetrician declined participation (n = 47, ‘clinician factor’) and 101 (22%) were included. Declining women were more likely than participants to be parous (24 vs 10%, P < 0.05), induced for nonmedical reasons (18 vs 4%, P < 0.001), privately admitted (31 vs 3%, P < 0.001) and have longer inpatient stay (4.9 vs 4.2 days, P < 0.05). Conclusion The high assessed‐but‐not‐recruited rate highlights important issues with external validity and feasibility when conducting obstetric trials, including recruitment difficulties related to participant, clinician and trial factors. Assessed: recruited ratios and demographic and outcome differences need consideration in planning and interpretation of randomised trials.