Sudden unexpected death in infancy and socioeconomic status: a systematic review
- 1School of Postgraduate Medical Education, University of Warwick, Coventry, UK
- 2Peninsula Medical School, University of Exeter, UK
- Correspondence to: Professor N Spencer School of Postgraduate Medical Education, University of Warwick, Warwick, Coventry CV4 7AL, UK;
- Accepted 28 October 2003
This paper aimed to systematically review observational studies documenting the relation between sudden unexpected death in infancy and socioeconomic status. A search of two electronic databases (Medline 1966 to November 2002; Embase 1981 to November 2002) yielded 52 case-control or cohort studies meeting the inclusion criteria. An increased risk of sudden unexpected death in infancy was reported in 51 studies and 32 of 33 studies reporting graded measures of socioeconomic status showed a dose-response relation of sudden death with socioeconomic status. Of the 10 studies in which adjustment was made for maternal smoking, socioeconomic status retained an independent effect on infant death in nine. The effect of socioeconomic status was also independent of birth weight in 10 of 11 studies and independent of sleeping position in two. The included studies reported a significant association of socioeconomic status with sudden unexpected death in infancy with risk of infant death increasing with greater exposure to adverse social circumstances. The findings support a significant role for adverse social circumstances in the pathways to sudden unexpected death in infancy.
Despite recent changes in the epidemiology of sudden unexpected death in infancy,1 it remains the most significant cause of infant death in developed countries. The term sudden unexpected death in infancy (SUDI) is used in this study in preference to the more commonly used sudden infant death syndrome (SIDS). Firstly, SUDI incorporates all cases of SIDS, defined as the sudden death of an infant under the age of 1 year that remains unexplained after thorough case investigation including performance of a complete necropsy, examination of the death scene, and review of the clinical history,2 while allowing for inclusion of unexpected deaths not fully complying with this strict definition but having common associations and risk factors.3 Secondly, many of the studies included in this review predated the adoption of a strict definition for SIDS and few would have had systems in place to ensure the thorough case investigation required by the definition. For these reasons, we have adopted a similar approach to that used by the UK Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI).3
A link between low socioeconomic status and SUDI has been noted in much of the literature.4,5 Socioeconomic status represents a constellation of factors reflecting social position and social circumstances including income, occupation, education, and ownership of resources such as housing. For the purposes of this study, we have included marital status and maternal age that are not strictly socioeconomic status variables but have been consistently shown to be strongly associated with measures of social status especially in studies of pregnancy and infant outcome.
The relation of low socioeconomic status and SUDI is important in that aetiological hypotheses need to be consistent with the observed social patterns. It has been suggested that smoking “accounts for” the socioeconomic differences in risk of SUDI.6,7 It is well reported that smoking patterns by social class vary between countries and, over time, within countries.8 Other independent risk factors for sudden unexpected infant death, such as birth weight and sleeping position,1 are also socially patterned. It seems probable that socioeconomic status rather than being a direct, proximal cause of sudden infant death exerts its effect through mediating variables such as smoking, birth weight, and sleeping position. This study provides an opportunity to examine the links between socioeconomic status and SUDI between countries and over time and contribute to a better understanding of the nature of the relation.
A systematic search was undertaken to identify case-control and cohort studies in which the relation between socioeconomic status and SUDI was examined.
Case-control and cohort studies of SUDI or SIDS that included author defined useable relative risks or odds ratios associated with measures of socioeconomic status and met methodological criteria.
For case-control studies
(a) selection of an appropriate control group—that is, controls from the same population, selected with minimal bias and of similar age
(b) the same method of ascertainment of exposure data for cases and controls
(c) analysis took account of matching where appropriate
(d) exposure data were available for >80% of both cases and controls
The association of sudden unexpected death in infancy with social disadvantage was demonstrated in 51 of 52 case-control and cohort studies published since 1965.
The association was independent of maternal smoking in 9 of 10 studies.
Social disadvantage is important to an understanding of the epidemiology of sudden unexpected infant death and should not be dismissed as an unmodifiable variable.
For cohort studies
(a) no evidence of biased case ascertainment by exposure status—only studies in which cases were ascertained from whole populations and not selected sources such as specialist hospitals were included
Socioeconomic status measures
Measures of socioeconomic status were included whether measured at the individual or area level. These were: social class (registrar general’s social class in UK studies), occupation, income, deprivation (as measured by deprivation indices), housing tenure, overcrowding, maternal education, marital status, and maternal age.
Two electronic databases (Medline 1966–2002; Embase 1981–2002) were searched for studies of the outcome of interest (search terms included: SIDS; sudden infant death syndrome; sudden unexpected death in infancy; postneonatal mortality; cot death; crib death) and the exposure of interest (search terms included: social class; occupation; maternal age; maternal education; marital status; socioeconomic status; income; poverty; deprivation usually represented by area based deprivation indices based on census data; overcrowding; social factors; risk factors) with appropriate truncations and mis-spellings. A citation search was undertaken for identified studies in the Science Citation Index. Additional socioeconomic status measures were included in studies identified using the above list of measures. They included: parental education; economy of the home, house repair, housing density, cohabitation, and social disadvantage. A secondary search using these terms yielded 34 abstracts and produced no new studies.
Secondary search of the bibliographies of the papers identified by the electronic search was undertaken. Where gaps in data were identified, authors were asked to provide supplementary data. Titles and abstracts of studies identified by electronic searches were examined for possible relevance and those that might meet the inclusion criteria retrieved. The full papers were read by both reviewers (NJS and SL) independently to determine whether they met the inclusion criteria. Disagreements were resolved by consensus.
Each of the studies identified was classified according to whether an increased risk of SUDI was reported to be associated with each measure of socioeconomic status studied. In addition, studies that reported graded measures of socioeconomic status were examined for the presence of a gradient or trend. To further explore the debate on the relation between socioeconomic status and SUDI, the results of studies that attempted to control for potential confounding of the effect of socioeconomic status on SUDI were included in a separate table (table 3). Smoking in pregnancy, birth weight, and sleeping position are socially patterned and also important risk factors for sudden infant death. Studies that adjusted for these variables were included in table 3.
The titles and abstracts of 380 studies identified by the search strategy were examined. A total of 110 published papers and abstracts potentially meeting the inclusion criteria were reviewed. Sixty discrete studies, fulfilling the inclusion criteria, were identified with publication dates from 1966 to December 2002 reporting data on infants born between 1956 and 1998. Eight studies were excluded as they failed to meet methodological criteria (table 1).
Fifty two studies were included in the review (table 2). The studies were conducted in 16 countries, 15 in developed countries and one in a developing country (Brazil). Most of the studies were conducted in the USA and the UK. Thirty were case-control studies and 22 cohort studies.
Only 2 of the 52 studies failed to show a statistically significant relation between sudden unexpected death and at least one measure of socioeconomic status. In one of these studies, the results suggested an association with low social class but failed to reach conventional levels of statistical significance. Six of the remaining 50 studies reported a positive relation between low socioeconomic status and sudden unexpected death for all socioeconomic status measures used. No studies reported an increased risk with any measure of higher socioeconomic status. Thirty three studies reported graded measures of socioeconomic status. Thirty two of these showed a significant gradient for at least one socioeconomic status measure and the remaining study showed a trend but failed to reach conventional levels of statistical significance.
Sixteen studies reported odds ratios associated with measures of socioeconomic status after adjustment for the main socially patterned risk factors, smoking, birth weight, and sleeping position. Maternal smoking during and/or after pregnancy was adjusted for in 10 studies (table 3). In 8 of these 10 studies, the 95% confidence intervals for at least one socioeconomic status measure excluded one. One study51 did not report 95% confidence intervals but p values indicate that maternal education remained significant after adjustment. Birth weight was adjusted for in 11 studies and sleeping position in two studies. Socioeconomic status measures remained independently associated with sudden unexpected infant death in all but one of the studies adjusting for birth weight and in both studies adjusting for sleeping position.
Methods for the systematic review of observational studies are less well defined than for randomised control trials. Meta-analysis of observational data poses particular problems related to heterogeneity between populations and in measures used. None the less, the principles of systematic reviews, an exhaustive search for primary studies and clear, pre-defined inclusion criteria, seem to offer an approach to minimising bias in the review of observational data. In this review no attempt was made to combine data statistically but rather to examine whether a consistent relation was found between social factors and sudden unexpected infant death over time and between countries. The range of countries and populations studied, the 42 year period over which the studies were conducted, and the wide range of socioeconomic status measured used would have made combined statistics difficult to interpret and potentially misleading.
Preventive programmes should address the social circumstances of families with young infants in addition to promoting parental behaviour change
Fifty two studies, undertaken in 16 countries and including over 10 000 sudden unexpected infant deaths during the period 1956 to 1998, were included in this systematic review. A broad definition of sudden unexpected infant death was used to take account of the changing definition of sudden unexpected death over the period studied. The results show an increased risk of sudden unexpected infant death associated with low socioeconomic status, measured by a range of indicators, which is consistent over time and between countries. These findings suggest that socioeconomic factors have an important role in the pathways leading to SUDI. Smoking is known to be strongly correlated to SIDS70 and it has been suggested that it “accounts for” the social gradient.
The changing social pattern of women’s smoking in developed countries over the 40 year period covered by these studies, the finding that in 10 studies socioeconomic status measures retained statistical significance after adjustment for maternal smoking (see table 3) and the variation in smoking patterns between countries8 makes it unlikely that smoking “accounts for” all the effects of socioeconomic factors in sudden infant death.6,7
Socioeconomic status is not a discrete variable that can be said to “cause” sudden unexpected infant death. It is likely to act, as it does in relation to mortality throughout the life course,71 as a distal determinant exerting its influence through a range of variables including low birth weight,72 smoking,73 overwrapping,74 and prone sleeping75 all of which have been shown to have a social gradient in the same direction as sudden infant death. These risk factors are likely to be on the pathway from socioeconomic status to sudden infant death. As table 3 shows, socioeconomic status exerts an effect on sudden infant death independent of these major risk factors and it is possible that other socially patterned variables are mediating this independent effect.
Further work is required to elucidate the mechanisms by which socioeconomic factors influence sudden infant death. However, this study shows that the search for possible proximal causes of sudden infant death must be consistent with observed social patterns.
It has been suggested that the socioeconomic status of families with young children is “immutable”62 and “unmodifiable”58 justifying exclusive concentration on proximal risk factors in the prevention of sudden infant death. However, the proportion of UK children living in poverty (by the EU definition of income below half the average after housing costs) rose from 9% in 1979 to 34% in 1995–6.76 These trends can be traced directly to government policy and could be reversed. If, as the results of this study suggest, socioeconomic factors act as distal determinants of SUDI, preventive programmes need to address the social circumstances into which infants are born as well as the health related behaviours of their parents.