METHODS

National birth and death registrations for England and Wales, 1993–98, provided denominator data. All deaths with an underlying cause of death certified as “sudden death, cause unknown”, International Classification of Disease, 9th revision (ICD) three digit code number 798 were included.

Total numbers of live births and cases of SIDS in birthweight specific groups were obtained from routinely published national statistics.⁹ Birthweight groups used for the analyses were <1500 g, 1500–2499 g, 2500–2999 g, 3000–3499 g, ≥3500 g, and “birth weight not stated”.

The ONS provided copies of the non-confidential section of all death registrations in twins and higher order multiple births in England and Wales 1993–98 with coding of the causes of death. The death registrations also provided information on the age at death, birth weight, and whether twins were of like or unlike sex. National birth and death registrations do not allow twins to be classified as MZ or DZ, only as of like or unlike sex. As unlike sex twins must be DZ but like sex twins comprise both DZ and MZ twins, the risk of like and unlike sex twins was compared. ONS also provided data on the registered birth weight of the surviving co-twin. Denominator numbers of twins of like and unlike sex were obtained from routinely published annual registrations.¹⁰

The number of SIDS in singletons was obtained by subtracting the number of SIDS in multiple births from the total number of SIDS. The annual numbers shown relate to the year of birth, not year of death. For example, an infant born in 1993, who dies in 1994 will be included in the numerator SIDS for 1993. Similarly, infants born in 1998 but who died with SIDS in 1999 were included in the numerator for 1998. Deaths owing to SIDS that occurred after 1 year of age were excluded from the analyses.

RESULTS

There were 2349 SIDS cases among infants born from 1993 to 98; of these deaths, 2248 occurred in infancy—that is, aged less than 1 year. The remaining 101 (96 singletons and five twins) SIDS occurred when the children were aged 1–4 years and were excluded from the analyses.

DISCUSSION

The first official definition of SIDS was “the sudden death of any infant or young child which is unexpected by history and in whom a thorough necropsy examination fails to show an adequate cause of death”.¹² This definition means that a diagnosis of SIDS is one made by exclusion, and the thoroughness of the postmortem examination may be crucial to eliminating a recognisable adequate cause of death. In England and Wales, it was not until 1971 that “sudden death, cause unknown” was accepted as a natural registrable cause of death by the Registrar General. This acceptance of SIDS as a registrable entity led, in the decade of the 1970s, to an increase in numbers so registered which was largely an artefact owing to transference from respiratory causes of death. There has been a continuous downward trend in SIDS mortality following the earlier artefactual rise. The trend was intensified following the national “back to sleep” campaign in 1991, and we have now shown that SIDS mortality continues to fall significantly in both singletons and twins in the period 1993–98. The thoroughness of postmortem examination is likely to vary throughout the country. As the study is based on nationally registered SIDS, it could be inferred that the downward trend was, in part, an artefact of the improvement in the quality of postmortem examination, thereby reducing the number of deaths by exclusion. Any improvement in the thoroughness of postmortem examination may be counteracted by the falling numbers of postmortem examinations.¹³ However, the decline in SIDS mortality has been of such magnitude to make this an unlikely explanation; the alternate explanation of a real improvement is more probable.

An objective of the study was to test the hypothesis that zygosity may influence the risk of death as a result of SIDS. The failure to observe any effect of zygosity must be interpreted with caution because like or unlike sex of twins is only a partial marker of zygosity. All unlike sex twins are dizygous, but only about 50% of like sex twins are monozygous.

Several studies have found that twins have a two- to fourfold increased risk of SIDS compared with singletons.^2,3,5,14,15 There are at least two components to this increased risk. Preterm delivery and low birth weight is associated with increased rates of SIDS, and a much higher proportion of twins compared with singletons are born preterm or of low birth weight.^2,5,16 Birth weight specific twin/singleton comparisons of risk have produced conflicting results. The 1968 Californian live birth cohort study found a higher risk in twins for all three birthweight specific groups (≤2000 g, 2001–2500 g, and >2500 g) examined.⁶ A more recent United States national study found a higher, but not statistically significant, SIDS risk in singletons compared with twins for the birth weight groups 500–1499 g and 1500–2499 g, but a lower risk for the birth weight group ≥2500 g.⁵ Our results are in accord with the United States study. A possible explanation is that birth weight is a marker for gestational age and, for a given birth weight group, twins are gestationally more mature than twins. The mechanism whereby gestational maturity exerts an influence may be through the immunocompetence of the developing fetus. A newborn’s ability to produce IgG is low, and transplacental transfer of maternal IgG during the second and third trimesters is the basis of passive protection in early neonatal life and could influence the probability of the infant succumbing to SIDS.^17,18 This presupposes that some SIDS deaths have an infectious aetiology.

Our observation that heavier twins (birth weight ≥3000 g) are at greater risk of SIDS than singletons is unusual and has not been reported previously. Within a population preventive perspective of SIDS mortality, the attributable risk associated with large twins is small because they are relatively few of them. However, the high relative risk may be of aetiological significance. It seems unlikely that immunological competence is important in these cases. Possible factors such as maternal diabetes and the greater propensity of large twins to suffer perinatal injury are conjectural. The question of why large twins are at high risk is intriguing. However, the increased risk of SIDS in twins of birth weight ≥3000 g is based on only 16 observed cases when 5.5 were expected. Nevertheless, this is unlikely to be a result of chance and merits specific confidential enquiry into individual cases.