Sudden infant death syndrome (SIDS) is defined as “the sudden unexpected death of an infant <1 year of age, with onset of the fatal episode apparently occurring during sleep, that remains unexplained after a thorough investigation, including performance of a complete autopsy and review of the circumstances of death and the clinical history” . Sudden unexpected death in infancy (SUDI) is a broader term, which includes all sudden and unexpected deaths in infancy, unexplained (SIDS) or explained (suffocation, malformations, arrhythmias, etc). The incidence of SIDS peaks between the ages of 1 and 4 months, with 90% cases occurring before 6 months of age. A triple risk model has been proposed for the pathogenesis of SIDS, which requires the convergence of three risks: exogenous stressors (prone sleep, soft bedding etc), critical development period (2-4 months of age), and vulnerable infant (preterm, low birth weight, nicotine exposed etc) .
Prone and side sleep positions have been considered as external stressors for infants, based on observational studies, and supported by numerous physiological studies. These studies have shown that prone sleep position may alter the autonomic control of the infant cardiovascular system, particularly at 2 to 3 months of age, and may result in decreased cerebral oxygenation [3,4]. Prone position was also shown to decrease cardiac output, mean arterial pressure, oxygen saturation, minute ventilation, and arousal responses to various stimuli [5–8].
The initial safe sleep recommendations were published in the early 1990s after a realization in the late 1980s that prone sleep position was strongly associated with SIDS . The launch of safe sleep campaigns was followed by a significant decline in the rates of SIDS and SUDI in the United States and other parts of the world . The American Academy of Pediatrics (AAP) latest 2016 guidelines recommend placing every infant in supine position for every sleep by every caregiver until one year of age . However, some studies have raised concerns about prolonged supine positions causing delayed motor development and deformational plagiocephaly [12,13].
Even several years after “Back to Sleep” campaign, some parents continue to place their newborns and infants in non-supine position, thus putting them in a potentially unsafe environment . Therefore, it is necessary to look systematically into evidence and further strengthen the recommendations for a safe sleeping position for neonates and infants. This review aimed to determine the effect of supine sleep position compared to non-supine sleep position on health outcomes in term healthy newborns and infants (neurodevelopment, plagiocephaly, SIDS, SUDI, and acute life-threatening events (ALTE)).
Randomized controlled trials (RCTs) including cluster randomized trials or quasi-randomized trials in human neonates were eligible for this review. If the number of RCTs was found to be inadequate (<3) or the optimal information size was not met, we included the observational studies (before-after/cohort/case-control/cross-sectional analysed like case-control). The study population considered were term neonates (up to 28 completed days of life). We excluded the studies if most participants (≥50%) either had low birth weight or were preterm neonates. Studies were included if supine sleep position was compared with non-supine (prone or side) sleep position in neonates. Studies that did not report the number of infants in supine sleep position separately were excluded.
The outcomes of interest were: neonatal mortality (all-cause death in the first 28 days of life); SIDS; ALTE (an episode that is characterized by some combination of apnoea, colour change, marked change in muscle tone, choking or gagging) ; positional plagiocephaly (flattening of the skull) and physiological parameters (cerebral regional oxygen saturation, cardiac output/stroke volume etc.).
With inclusion criteria described above, only one of the included studies reported the risk of SIDS in the neonatal period while none reported the risk of neonatal mortality – the two critical outcomes of the review . Therefore, we modified our inclusion criteria to include all studies reporting SIDS for infants (up to 1-year age). We included additional outcomes which were considered critical to this review: SUDI and long-term neurodevelopment (as assessed by standardized/ validated neurodevelopment tools).
We initially searched for the existing systematic reviews. We planned to update the existing reviews depending upon the year of publication: if published in 2019 or 2020, we intended to use the results of the review without updating them; if published before 2019, we planned to update the review using the same search strategy used in the review.
The databases were searched independently by two authors (MP and BB). The search was conducted in the following databases: MEDLINE (1966 onwards) via PubMed, Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library), EMBASE (1988 onwards), and CINAHL (1981 onwards). We conducted the first search by March 31, 2020 and updated it until November 30, 2021. Searches were limited to human studies. There were no language restrictions. Related conference proceedings (like Pediatric Academic Societies (PAS) abstracts) were also be searched for relevant abstracts. Organizations and researchers in the field were contacted, if necessary, for information on unpublished and ongoing trials. Reference lists of all relevant studies were searched. The ClinicalTrials registry (www.clinicaltrials.gov) was searched to identify any ongoing trial. The search strategy is provided in Appendix S1 in Online Supplementary Document.
Data extraction and management
Two authors (MP and BB) extracted data independently using a pilot-tested data collection form to collect information on design, methods, participants, interventions, outcomes, and treatment effects from each included study. We discussed disagreements until we reached a consensus. If data from trial reports were insufficient, we contacted study authors to request further required information or clarifications. Data were extracted from a systematic review (Gilbert 2005) for six studies (full text unavailable) . Additional data were obtained by contacting the review author.
Assessment of risk of bias in included studies
Two authors (MP and BB) independently assessed the methodological quality of the selected studies. Quality assessment was undertaken using the Cochrane Risk of Bias (RoB 2.0) tool for randomized trials and the ROBINS-I tool for observational studies [17,18]. Any disagreements between the review authors were resolved by discussion.
Meta-analysis was performed with user-written programs on Stata 15.1 (StataCorp, College Station, TX, USA). Pooled estimates for categorical outcomes were calculated from the relative risk (RR) or odds ratio (OR) and 95% confidence intervals by the generic inverse variance method. If available, we used the adjusted RR and OR from the studies for pooling the results in the meta-analysis. The studies recorded different infant sleep positions such as last sleep position, usual sleep position, and position last found at the death scene. The last sleep position was used (or usual sleep position, if the former was not reported) for estimating the odds ratios in individual studies, if more than one position was reported. The studies reported sleep positions as supine, prone, and side. We considered prone and side position as non-supine and compared them with supine position as the side position has also been shown to increase the risk of SIDS in a previous systematic review . We examined heterogeneity between study results by inspecting the forest plots and quantifying the impact of heterogeneity using the I2 statistic. If there was no significant heterogeneity (I2<60% or P ≥ 0.1), we pooled the results using the fixed-effect model. If there was significant heterogeneity (I2>60% or P < 0.1), we explored the possible causes of heterogeneity. If there was no obvious clinical heterogeneity, we used the random-effects model for meta-analysis. The possibility of publication bias was evaluated using funnel plots and the Egger and Begg tests. We used GRADEpro software for assigning the certainty of evidence .
We found one systematic review assessing associations between infant sleeping positions and SIDS, which included observational studies published till January 2003 . We performed a further search using a date filter from July 2002 until November 2021 (Figure 1). We included 54 studies, of which 49 studies were included in the quantitative analysis (Table 1).
Figure 1. PRISMA flow chart depicting the selection of studies included in the review.
Table 1. Characteristics of the studies included in the review
BW – birth weight, ECG – electrocardiogram, Gest – gestation, HIC – high-income country, LMIC – lower middle-income country, OR – odds ratio, SIDS – sudden infant death syndrome, SUDI – Sudden Unexpected Death in Infancy, UMIC – upper middle -income country, y – year, mo – month, wk – week
The designs of the included studies were case-control (n = 28), cohort (n = 12), intervention trials (n = 11; including 6 crossover-randomized trials), ecological (n = 2) and survey (n = 1). The interventional trials evaluated physiological parameters in infants in university hospitals at one or several time points (see Online Supplementary Document).
53 studies were conducted in high-income and one in an upper-middle income country (Brazil). All observational studies were based on data from the community, either from national health database, health surveys, or follow-up data from cohort studies.
This review included data from 54 studies involving 474 672 participants, of which 49 studies with 80 974 participants were included in the quantitative analysis. Studies reporting SIDS, SUDI, or ALTE included infants up to 365 days of age, while studies reporting other outcomes involved participants who had their sleep position practices (exposure status) recorded as neonates. Though detailed population characteristics were not available in the case-control studies, available information suggests that 7%-13% of the participants were born before 37 weeks, with higher proportion of preterm infants among SIDS cases. There were no indications that the included infants were not healthy before the occurrence of SIDS, SUDI, or ALTE.
Exposure status/ intervention (sleep position)
In case-control studies, data on sleep position were collected from records of death scene evaluation after SIDS, or through interviews at home visits, postal questionnaires and mail or phone surveys for control infants. The infant sleep positions (recorded in SIDS meta-analysis) were last sleep position (10 studies), usual sleep position (13 studies), and position last found at the death scene (2 studies). One study  did not mention the type of recorded sleep position.
In cohort studies, sleep position practices were prospectively recorded at one or multiple time points during follow-up. One study  used a mailed questionnaire to enquire about the usual infant sleeping position from 400 mothers in a survey. In ecological studies, the authors estimated the effect of sleep position looking at time-trends in SIDS and SUDI rates from national databases.
None of the included studies reported the risk of neonatal mortality. The critical outcomes reported in the included studies were SIDS, SUDI, ALTE, and neurodevelopment outcomes (details in Appendix S2 in Online Supplementary Document).
SIDS was reported by 29 studies, 26 of which could be included in meta-analysis. These studies included SIDS cases reported in various infant death registries, national databases, or health department reports, from 1958 up to 2006. The definitions of SIDS in the studies were consistent with our protocol, recorded during the first year of life. Two studies reported deaths in children from 2 weeks to 2 years  and 2 weeks to 3 years . We included these two studies because the mean age at the time of SIDS was 3-4 months, similar to other included SIDS studies. The controls were selected from hospital records, population databases, or randomly selected from the same geographical area and time as the cases. The control infants were matched for age, region, sex, and/or ethnicity in 17 studies.
Risk of bias in included studies
A summary of the risk of bias assessment for 54 included studies is provided in Appendix S3 in the Online Supplementary Document. Of 48 observational studies, 46 were considered to be at serious risk of bias, mostly due to confounding and misclassification bias.
Effects of interventions
A summary of the included studies’ effects on reported outcomes is shown in Table 2. The results on physiological parameters and studies with qualitative data have been summarized in Appendix S5, S6 in the Online Supplementary Document.
Table 2. Summary of studies: effect of sleep position on various outcomes by type of study design*
ALTE – apparent life-threatening event; FS – favors supine; FN – favors non -supine; FP – favors prone; NE – no effect; SIDS – sudden infant death syndrome; SUDI – sudden unexpected death in infancy
*The numbers in the boxes against FS/FN/FP/NE indicate the number of studies in their respective categories which favored supine/ non-supine/ prone or showed no effect.
†The overall effect is based on subjective assessment of the following factors: results from all studies (including ecological studies and surveys), number of studies, their sample sizes and study quality (risk of bias).
26 observational studies, comprising 59 332 infants, could be pooled together in a meta-analysis for the outcome of SIDS (Figure 2). The meta-analysis favoured supine sleeping position (OR = 0.51, 95% CI = 0.42-0.61; very low certainty evidence) for the prevention of SIDS (0-1 year). Though there was substantial heterogeneity across the studies (I 2 = 64%), the direction of the effects was mostly consistent. There was doubtful asymmetry in the funnel plot (Figure S5 in Online Supplementary Document); however, the Egger and Begg tests for publication bias were not significant (lowest P = 0.79) for small-study effects. When compared to prone and side position separately, supine was protective for SIDS compared to prone (OR = 0.31, 95% CI = 0.21-0.45), but not to side position (OR = 0.80, 95% CI = 0.63-1.02), though the trend was still in favour of supine position (evidence not graded; Figures S6, S7 in the Online Supplementary Document).
Figure 2. Forest plot for comparison: supine vs. non-supine sleep position. Outcome: SIDS (0-1 year).
One study reported the effect of sleep position on SUDI by comparing the last sleep positions in 126 cases of SUDI and 258 controls during 2012-2015 . The unadjusted OR was reported to be 0.39 (95% CI = 0.23-0.65; low certainty evidence) for the prevention of SUDI, thereby favouring supine sleep position.
A case-control study reported on 29 newborns who suffered unexplained SID or severe-ALTE and 90 control newborns, with cases occurring within 24 hours of life in German hospitals . The unadjusted OR was 0.16 (95% CI = 0.03-0.82; very low certainty evidence) for supine sleep position, compared to potentially asphyxiating position (defined in the study as infant lying on mother’s breast/abdomen or near to and facing her).
Another study  using data on 3729 infants, examined the effect of sleep position on hospital admission related to ALTE in their follow-up cohort of infants who slept consistently in prone, side, or supine position at 1, 3, and 6 months of age. The study did not report any difference (OR = 0.23, 95% CI = 0.005-2.04; very low certainty evidence) in the outcome among infants sleeping in supine position (1/1745), compared to non-supine (5/1984).
Another study followed-up a cohort of 14 138 infants delivered during 1991-92 in the United Kingdom to 18 months of age to assess the effect of sleep positioning on infant development . The authors assessed various domains of development using the DDST at 6 and 18 months. The study reported the position effect as odds ratio by dichotomizing the gross motor scores using a -0.5 SD cut-off point after transforming the scores into a mean of zero and a standard deviation (SD) unit of 1. Based on data from 2097 participants, supine position significantly increased the odds of being 0.5 SD below mean on the Gross Motor Scale at 6 months, when compared to prone position (OR = 1.67, 95% CI = 1.22-2.27; moderate certainty evidence) but no difference when compared to side sleep position (OR = 1.02, 95% CI = 0.91-1.15; 8012 infants; low certainty evidence). The effect of the sleep position, however, diminished over time and was no longer significant at 18 months of age. The odds ratios for supine position being 0.5 SD below the mean on the Gross Motor Scale at 18 months were 1.16 (95% CI = 0.96-1.43; low certainty evidence) when compared to prone and 0.89 (95% CI = 0.69-1.16; low certainty evidence) when compared to side sleep position.
Six observational studies reported the effect of sleep position on the occurrence of positional plagiocephaly at various ages (2-7 months; Figure 3). Supine position was associated with increased odds of positional plagiocephaly (OR = 2.77, 95% CI = 2.06-3.72; 1774 infants; I2 = 53.6%; low certainty evidence) compared to non-supine sleep position.
Figure 3. Forest plot for comparison: supine vs. non-supine sleep position. Outcome: Positional plagiocephaly.
Low- to very low-certainty evidence suggested that supine sleep position might result in reduction of SUDI (0-1 year), SIDS (0-1 year), and unexplained SID/s-ALTE (neonatal period). The effect of supine sleep position was uncertain on ALTE-related hospital admissions in the first 6 months of life. Supine sleep position, compared to prone, likely increased the odds of being 0.5 SD below the mean on the Gross Motor Scale at 6 months, but the effect might not persist at 18 months. Supine, when compared to side sleep position, might not result in any difference in gross motor development (6 or 18 months). There might be an increase in the incidence of positional plagiocephaly (at 2-7 months of age) due to supine sleep position, compared to non-supine position. The evidence for most outcomes were of low- to very low-certainty, owing to high risk of bias (in observational studies), heterogeneity, and imprecision (where results were based on single study).
SIDS outcome was affected by significant heterogeneity (I2 = 64%) partly attributable to disparate study designs, geographical locations (13 countries, 5 continents), and study periods (1958 to 2004), based on exploratory subgroup analyses (details in Appendix S4 in Online Supplementary Document). Safe sleep campaigns have been shown to impact SIDS and SUDI rates differently in various parts of the world . Our review did not find an association between sleep position and ALTE-related hospital admissions. ALTE is considered to be a distinct entity from SIDS, and the SIDS prevention interventions have not made any significant impact on the occurrence of ALTEs . The association of unexplained SID/s-ALTE episodes on the first day of life with potentially asphyxiating position was very uncertain, and reassuringly, increasing rates of skin-to-skin care have been temporally associated with decreasing SUDI prevalence in the first 6 days after birth in the US and Massachusetts .
Though development outcomes were evaluated by four included studies, the results could not be meta-analysed together due to variations in reported outcome measures. We presented data from a study by Dewey et al.  because this study had the largest number of participants with objective assessment scores (using Denver Developmental Screening Test) and was judged to be at moderate risk of bias (others were at serious risk of bias). Supine sleep position was found to increase the risk of positional plagiocephaly in our review. Positional plagiocephaly is considered benign and of mainly cosmetic concern, though some studies have associated its severity with poor neurodevelopment. These associations are thought to be marker of developmental risk rather than truly causal .
The findings of our review are in coherence with multiple infant sleep safety guidelines recommending supine sleep position for the prevention of SIDS . One systematic review included 40 case-control studies on the association of infant sleep position and SIDS , 17 of which overlapped with our review. The other studies in the review were excluded either because they reported comparison of only ‘prone vs non-prone’ position without further qualification of non-prone position (16 studies) or their data sets were included in the more recent publications (seven studies) . The results from that review  found increased odds of SIDS with prone and side sleep positions, compared to supine position. A review on factors affecting gross motor development (GMD) concluded that prone sleeping was associated with better GMD at 4 to 10 months but not at a later age (11 to 17 months), similar to our review . Two studies from this review could not be included in our review because they did not record the exposure status (sleep position) in neonatal period.
This review aimed to reinforce the evidence on a safe sleep position for mortality and serious morbidities in term healthy neonates. We followed a rigorous methodology, with an all-inclusive literature search and no language filters. Given the rarity of SIDS and SUDI, we could not find any interventional studies which assessed the effect of intervention (sleep position) on these outcomes. Though a systematic review had looked at studies on interventions to reduce the risk of SIDS, it could only find studies evaluating the effectiveness in changing infant sleep practices, rather than the risk of SIDS itself . We used manually calculated unadjusted unmatched ORs for pooling the results for the main comparison, since matched ORs were not reported separately for non-supine position in the studies. The sleep position was recorded differently across the studies. While last sleep position is the most likely to be related to SIDS risk, the usual and last found positions might be less accurate and therefore were least preferred for meta-analysis.
Low- to very low-certainty evidence suggests that supine sleep position may reduce the risk of SIDS (0-1 year), SUDI (0-1 year), and unexplained SID/severe-ALTE (neonatal period), compared to non-supine position. There may be delay in short-term ‘gross motor’ development (6 months) and increased incidence of positional plagiocephaly (2-7 months) with prolonged supine sleep position, compared to prone, but the evidence suggests that it does not affect the long-term neurodevelopment (18 months). However, most reported outcomes in this review are limited by low- to very low-certainty evidence.
Disclaimer: The authors alone are responsible for the views expressed in this publication and they do not necessarily represent the views, decisions, or policies of the World Health Organization.