Association between going-to-sleep position (last available position, within last 2 weeks) and stillbirth in comparison to left lateral going-to-sleep position

• High quality evidence from an IPD meta-analysis of 5 case control studies (N=3108) found an important association between supine going-to-sleep position and stillbirth: aOR 2.63 (95% CI 1.72 to 4.04).
• Low quality evidence from an IPD meta-analysis of 5 case control studies (N=3108) found no important association between right sided going-to-sleep position and stillbirth: aOR 1.04 (95% CI 0.83 to 1.31).
• Very low quality evidence from an IPD meta-analysis of 5 case control studies (N=3108) found no important association between prone going-to-sleep position and stillbirth: aOR 0.63 (95% CI 0.12 to 3.25).
• Very low quality evidence from an IPD meta-analysis of 5 case control studies (N=3108) found no important association between variable side going-to-sleep position and stillbirth: aOR 0.97 (95% CI 0.70 to 1.35).
• Very low quality evidence from an IPD meta-analysis of 5 case control studies (N=3108) found no important association between propped up going-to-sleep position and stillbirth: aOR 1.30 (95% CI 0.68 to 2.49).
• Moderate quality evidence from an IPD meta-analysis of 5 case control studies (N=3108) found an important association between not remembering going-to-sleep position and stillbirth: aOR 2.26 (95% CI 1.48 to 3.46).

Association between going-to-sleep position (last available position, within last 2 weeks) and small for gestational age in comparison to left lateral going-to-sleep position

• High quality evidence from an IPD meta-analysis of 4 case control studies (N=1760) found an important association between supine going-to-sleep position and babies being born small for gestational age: aOR 3.23 (95% CI 1.37 to 7.59).
• Very low quality evidence from an IPD meta-analysis of 4 case control studies (N=1760) found no important association between right sided going-to-sleep position and babies being born small for gestational age: aOR 1.05 (95% CI 0.58 to 1.90).
• Very low quality evidence from an IPD meta-analysis of 4 case control studies (N=1760) found no important association between other going-to-sleep position and babies being born small for gestational age: aOR 1.14 (95% CI 0.62 to 2.09).

Association between going-to-sleep position on last night and stillbirth in comparison to left lateral going-to-sleep position on last night

• Very low quality evidence from 1 phase 2 prospective population-based case-control study (N=1024) examined whether there was an association between prone going-to-sleep position on last night and stillbirth compared to left lateral going-to-sleep position on last night. However, no association was found in this study: aOR 1.01 (95%CI 0.13 to 7.81)
• Very low quality evidence from 3 phase 2 case-control studies (N=2122) - 2 non-nested prospective population-based studies and 1 retrospective nested study in an online cohort of known size - examined whether there was an association between right-lateral going-to-sleep position on last night and stillbirth compared to left lateral going-to-sleep position on last night. However, none of the 3 studies found any such association: aOR 0.67 (95%CI 0.44 to 1.02); aOR 1.11 (95%CI 0.70 to 1.77); aOR 1.74 (95%CI 0.98 to 3.01)
• Very low quality evidence from 2 phase 2 case-control studies - 1 non-nested prospective population-based study and 1 retrospective nested study in an online cohort of known size (N=1657) - examined whether there was an association between sitting/propped going-to-sleep position on last night and stillbirth compared to left lateral going-to-sleep position on last night. However, none of the studies found any such association: aOR 0.44 (95%CI 0.13 to 1.49); aOR 0.71(95%CI 0.22 to 2.30).
• Very low quality evidence from 3 phase 2 case-control studies (N=2122) - 2 non-nested prospective population-based studies and 1 retrospective nested study in an online cohort of known size - examined whether there was an association between supine going-to-sleep position on last night and stillbirth compared to left lateral going-to-sleep position on last night. Two studies found that supine going-to-sleep position was associated with an increase in stillbirth compared to left lateral going-to-sleep position (aOR 2.31 [95%CI 1.04 to 5.11]; aOR 2.54 [95%CI 1.04 to 6.18]) whereas no association was found in another study: aOR 1.05 (95%CI 0.32 to 3.50). However this latter study, which was powered to detect an association between supine going-to-sleep position and stillbirth assuming a 20% exposure frequency, reported an exposure frequency of only 2.3%.
• Very low quality evidence from 2 phase 2 case-control studies - 1 non-nested prospective population-based study and 1 retrospective nested study in an online cohort of known size (N=1657) - examined whether there was an association between variable-lateral going-to-sleep position on last night and stillbirth compared to left lateral going-to-sleep position on last night. However, none of the studies found any such association: aOR 0.93 (95%CI 0.51 to 1.69); aOR 0.75 (95%CI 0.34 to 1.64).

Association between going-to-sleep position on last month of pregnancy and late stillbirth in comparison to left lateral going-to-sleep position

• Very low quality evidence from 2 phase 2 case-control studies (N=928) - 1 non-nested prospective population-based study and 1 retrospective nested study in an online cohort of known size - examined whether there was an association between right-lateral going-to-sleep position on last month and stillbirth compared to left lateral going-to-sleep position on last month. However, none of the studies found any such association: aOR 1.1 (95%CI 0.43 to 2.6); aOR 1.14 (95%CI 0.70 to 1.85).
• Very low quality evidence from 1 phase 2 case control study – a retrospective nested study in an online cohort of known size (N=633) - examined whether there was an association between sitting/propped going-to-sleep position on last month and stillbirth compared to left lateral going-to-sleep position on last month. However, no association was found in this study: aOR 1.20 (95%CI 0.39 to 3.68).
• Very low quality evidence from 2 case-control studies in (N=928) - 1 phase 1 non-nested prospective population-based study and 1 phase 2 retrospective nested study in an online cohort of known size - examined whether there was an association between supine going-to-sleep position on last month and stillbirth compared to left lateral going-to-sleep position on last month. One study found that supine going-to-sleep position was associated with an increase in stillbirth compared to left lateral going-to-sleep position: aOR 6.26 (95%CI 1.2 to 34.00) whereas no association was found in another study: aOR 0.37 (95%CI 0.04 to 3.12).
• Very low quality evidence from 1 phase 2 retrospective nested case control study in an online cohort of known size (N=633) examined whether there was an association between variable-lateral going-to-sleep position on last month and stillbirth compared to left lateral going-to-sleep position on last month. However, no association was found in this study: aOR 0.87 (95%CI 0.48 to 1.55).

Association between going-to-sleep position on last night or last week of pregnancy and pre-term stillbirth (28 to 36 weeks gestation) in comparison to left lateral going-to-sleep position on last night or last week

Association between going-to-sleep position on last night or last week of pregnancy and term stillbirth (≥37 weeks gestation) in comparison to left going-to-sleep position on last night or last week

The quality of the evidence

The quality of each primary study was appraised using the QUIPS checklist. The main reasons for downgrading the quality of studies were recall bias due to the time delay between recruitment and interview, and selection bias due to limited participation from all the eligible women. The overall quality of evidence for each outcome was assessed using a modified GRADE framework for prognostic reviews and ranged from very low to low. Although a phase 3 prospective cohort study is the best study design to establish the causal relationship between a prognostic factor and an outcome, the frequency of stillbirth is low and such a study is therefore not practically feasible to conduct. Therefore, no restrictions were initially placed on the appropriate types of study considered for this review with phase 1 and 2 studies initially rated as providing a moderate and high quality of evidence, respectively. The main reasons for downgrading the overall evidence for all sleep positions was risk of bias associated with each of the contributing studies (recall bias, selection bias), inconsistency (variation in effect estimates across studies), indirectness (sample not clearly representative of target population), imprecision (effect estimate has wide 95% confidence intervals) and publication bias (due to the small number of early phase studies reporting outcomes).

The IPD meta-analysis quality was appraised using the ROBIS checklist and both IPD meta-analyses were at low risk of bias. The individual outcomes within the IPD meta-analyses ranged in quality rating from high (for example supine sleeping position and its association with stillbirth) to very low. The IPD meta-analyses outcomes were typically downgraded due to imprecision in their estimates and publication bias (due to the small number of early phase primary studies reporting outcomes).

Benefits and harms

Overall, the evidence suggests that there may be an increased risk of stillbirth after 28+0 weeks and babies being born small for gestational age (SGA) associated with going to sleep on one’s back (in other words in the supine position) compared to going to sleep on one’s left-hand side – in both the primary studies and the IPD meta-analysis. The best estimate, from the IPD meta-analysis was that sleeping in a supine position approximately doubled the odds of stillbirth and trebled the odds of babies being born SGA.

While the quality of the evidence from the primary studies ranged depending on the timing and precise outcomes considered, the quality of the evidence from the IPD, particularly for the evidence around supine sleeping position, was relatively high. The evidence still may not be considered to be definitive evidence of causality between going to sleep position and stillbirth or SGA as the IPD meta-analysis does not overcome issues with the study design of the primary studies. However the committee agreed it was of sufficient quality to advise women to try to avoid going to sleep on their back after 28 weeks and inform women of the likely link with stillbirth, alongside a caveat that the evidence is uncertain. The committee chose to specifically highlight stillbirth as this is a more concerning outcome than babies being born SGA and they agreed that including SGA in the recommendations made the advice less clear.

The evidence also suggests that the risk of going to sleep in any other position is the same as that of going to sleep on one’s left side (that is, there is not an increased likelihood of stillbirth associated with any other going-to-sleep position). The committee agreed that, as there has been some concern that there is such an association, it was important to highlight that the evidence does not support the view that there is an increased likelihood of stillbirth associated with going to sleep on one’s right side irrespective of the time of reporting compared to going to sleep on one’s left side. However the evidence for other sleep positions was generally an absence of definitive association (with serious imprecision) as opposed to definitive evidence of no association, therefore the committee did not include this information in the recommendations.

The committee discussed that the studies looked at the association between going to sleep position and outcomes because due to the nature of the observational studies relying on women’s recall, it would not be possible to study the association between the woman’s sleeping position and outcomes. However, the going to sleep position should be considered a proxy for sleeping position. Sleeping position is perhaps best controlled by controlling the going to sleep position. In addition, pillows or other props could be used to aid that the position stays when sleeping. This review did not assess the effectiveness of any interventions to modify sleeping position but in the committee agreed, based on their knowledge and experience, to recommend advising women to consider using for example pillows so that they can maintain their position when sleeping.

The committee noted that there may be a psychological impact of informing pregnant women of the potential link between sleeping on one’s back and stillbirth or SGA and did not want to cause undue anxiety. The committee also noted the relatively low incidence of stillbirth (1 in every 244 births in the UK according to 2018 Office for National Statistics [ONS] data).

However on balance they agreed that the evidence was strong enough that women should be advised about the risk.