Design: prospective observational cohort with accelerometer-measured body position
Population: 84 women at 32-36 weeks attending an Australian public antenatal clinic, 556 nights of data
Primary outcome: customised birthweight centile (GROW)
Effect: no relationship with broad supine time (R² = 0.0027, p = 0.64) or low-tilt supine time (R² = 0.0029, p = 0.63)

Eighty-four women between 32 and 36 weeks wore a tri-axial accelerometer around the abdomen for seven consecutive nights, giving 556 nights of position data measured in degrees of roll. Birthweights were converted to customised centiles using GROW software.

Women spent a median of 65.9 minutes a night in broad supine position - anywhere from fully supine to 45 degrees of tilt - and only 9.1 minutes in true low-tilt supine, under 15 degrees. Neither measure showed any relationship with customised birthweight centile (broad supine R² = 0.0027, p = 0.64; low-tilt R² = 0.0029, p = 0.63), and neither did two purpose-built risk indices weighting time by degree of tilt. Birthweight, gestational age at delivery and fetal growth trajectory were all null as well.

The practical value is in the descriptive finding as much as the negative one. The advice given in clinic - do not sleep on your back - is aimed at a position women in this cohort held for about nine minutes a night. If supine going-to-sleep position really is a risk factor for late stillbirth, and the observational data for that remain reasonably consistent, this cohort suggests the mechanism is not chronic impairment of fetal growth. The authors are candid that a study large enough to settle it may not be feasible given how rare stillbirth is.

Keep giving left-lateral going-to-sleep advice - this does not contradict the stillbirth data it was based on
Reassure a woman who wakes on her back: there is no evidence here of an effect on growth
Do not use maternal sleep position as a reason to order additional growth scans
Note the sample was 84 women in one Australian clinic - small, and not a population in which stillbirth could be counted

Why it matters

It separates the advice we give from the mechanism we assume is behind it.

The statistics, in plain English

An R² of 0.0027 means supine time explained about a quarter of one percent of the variation in birthweight centile - statistically and clinically nothing. But 84 women is a small cohort, so this rules out a large effect on growth rather than a small one, and a null result in a study this size is weak evidence of absence for anything subtle.