• Comm Child Health
• Data Collection
• Epidemiology
• General Paediatrics
• Mortality

The recognition of risk factors for unexpected and unexplained infants deaths (sudden infant death syndrome, (SIDS))—most importantly the risk associated with the prone sleeping position—and the implementation of risk reduction public health campaigns (eg, the ‘Back to Sleep’ campaigns in the UK in 1991 and the USA in 1994) was followed in many countries by a significant fall in the number of such deaths. In England and Wales, the overall post-neonatal mortality (PNM) rates fell in parallel with the recorded fall in SIDS, and continued to fall over the next decade, providing evidence that the fall in SIDS was not solely a consequence of diagnostic transfer.1 ,2

International comparisons of SIDS rates have been hampered by the wide variation in quality of investigation after unexpected infant deaths—particularly with regard to death-scene investigations and autopsy rates and procedures.

A further impediment to effective international comparisons has been the way in which unexpected and unexplained deaths are categorised in different jurisdictions. In the UK, the increasing use since the late 1990s of the term ‘unascertained’ rather than SIDS by some pathologists has led to further uncertainty, as these terms may translate into different International classification of Diseases, Version 10 (ICD-10) codes.3

Further uncertainty arises from the concept of how much pathology is required to fully explain an unexpected death; for instance, practising paediatricians will commonly see children who survive an illness despite radiological evidence of severe pneumonia. In many such children the apparent severity of the radiological findings is greater than is seen in some infants who die unexpectedly and whose death may deliberately or inadvertently be attributed by the pathologist to pneumonia. Even if the pathologist does not consider the death was caused by pneumonia, under the rules of ICD-10 any mention of pneumonia as a contributory factor, unless the main cause of death is given as SIDS (ICD-10 code R95; which is not regarded as ‘ill-defined’), means that the death will be categorised as due to pneumonia, as this takes precedence over any of the other codes (eg, R99: ‘other ill-defined and unspecified causes of mortality’), which may be used if ‘unascertained’ (which ICD-10 regards as ‘ill-defined’) is written on the cause of death certificate.

The standardised approach to investigation of unexpected infant deaths introduced in England as part of the Children Act 2004 has led to greater multi-agency involvement and arguably higher standards of investigation, though the Child Death Review process is commonly conducted in relative isolation from the Inquest, where the final recorded ‘cause’ of deaths is decided. Some paediatricians attending child death review meetings do not challenge the label of ‘unascertained’ applied by the pathologist.4 The Kennedy Report5 recommended that, while it is acceptable for the pathologist to label a death as ‘unascertained pending further investigation’, this label should be replaced by ‘SIDS’ at the review meeting unless another causal condition has been identified.

The study by the multi-national group led by Taylor3 to investigate the international variance in certification of unexpected infant deaths is to be warmly welcomed as it shows wide discrepancies in the use of different ICD-10 codes for cause of death in different countries, and aggregating all of these categorisations leads to a much clearer picture of just how many infants are dying unexpectedly in each country.

Prospective, population-based studies of unexpected infant deaths have shown that, while many remain unexplained even after intensive investigation, some are shown to be caused by recognisable acute illness with clear evidence of the pathological process leading to death.2

National and international studies based on death certification cannot identify those deaths which occurred suddenly and unexpectedly but were caused by identified pathology as these will be correctly classified under the identified underlying cause (eg, infections, metabolic disorders, cardiac conditions). Death certificate data are dependent on the approach to categorisation of unexpected deaths adopted by the person or people responsible for the categorisation and do not have any intrinsic way of identifying those deaths that were sudden or unexpected.

Taylor et al show that in Australia, New Zealand and the USA 25%, 43% and 15%, respectively, of unexpected infant deaths are attributed to ICD-10 code W78 (‘accidental suffocation and strangulation in bed’), while this category is much less commonly used in Canada (5%), England and Wales (4%) or Germany (1%). These differences are unlikely to represent real differences in the ‘causes’ of such deaths (in which convincing evidence of suffocation is uncommon), even after careful scene investigations.

This aggregation of possible ICD codes that might denote unexpected and unexplained infant deaths which the authors have lumped together as ‘sudden unexpected deaths in infancy’ (SUDI) gives a different picture of how patterns of infant death have changed over the past decade in the countries included in this study to that which emerges from consideration of those deaths attributed to SIDS alone. It shows that, while Japan has from published death certification data been considered to be a country in which few infants die as SIDS, the total number included within the aggregated ‘SUDI’ codes suggests that slightly more infants per thousand live births die suddenly and unexpectedly in Japan than in Australia, Germany, Canada or England and Wales.

It is also clear from this approach that, while SUDI rates have continued to fall in some countries over the first decade of this century, in others, notably the USA and New Zealand, there has been little change.

The very low SIDS rates reported from the Netherlands raises the question as to whether such deaths are occurring but are given a different label, or whether the SUDI rate in the Netherlands is genuinely very low. The study by Taylor shows that the overall SUDI rate in the Netherlands is indeed very low, but so too is the total PNM rate.

The number of babies dying between 1 month and 1 year of age from causes other than SUDI (the non-SUDI PNM rate) in the eight countries included in this study ranged from 1.31/1000 live births in the USA to 0.70 in Japan, with the rate for the Netherlands close to the middle at 0.88. This suggests that, while the overall SUDI rate in the Netherlands is indeed very low this is probably not as a result of diagnostic transfer to other identified causes. However it also shows that, whatever is responsible for the very low SUDI rate in the Netherlands it is not a factor that has a major protective effect against other causes of post-neonatal infant mortality.

Conversely the relatively high SUDI rates in both the USA and New Zealand are accompanied by non-SUDI PNM rates that are the highest in this series of eight developed countries.

These data may allow further investigation of factors contributing to higher or lower SUDI rates—either together with or in the absence of effects on other identifiable causes of death in infants.

Further detailed studies of how infant care practices and other potentially contributory factors vary within and between countries with differing SUDI rates may help shed more light on these continuing discrepancies than isolated local studies of potential contributory (and avoidable) factors.