Statistics from Altmetric.com
Editor,—The demonstration by Blairet al of an association between poor postnatal growth and an increased risk of sudden infant death syndrome (SIDS),1 is a useful addition to our understanding of the aetiology of this condition. It is unfortunate that the conclusion in the abstract that “Poor postnatal weight gain was independently associated with an increased risk of SIDS and could be identified at the routine six week assessment” goes beyond the data presented.
It can be estimated from the data in this study that the overall risk of SIDS was 0.77/1000 live births. The risk in babies with birth weights greater than the 15th centile, the group in whom the relation with postnatal growth was detected, was 0.68/1000. Given the reported odds ratio of 1.75 associated with being in the slowest growing 16% who might be identified at six weeks, the data suggest that the absolute risk of SIDS among this group would be about 1.1/1000. Even a programme targeted at infants below the 2nd centile for growth at six weeks, would identify a group whose absolute risk of SIDS was about 4.2/1000—that is, for every infant who might benefit from the intervention, there would be 240 who would not. Even if it were accepted that this level of risk was sufficient to trigger an intervention, the nature of the intervention remains unclear. To the best of our knowledge, with the exception of the “Back to Sleep” campaign, there is no convincing evidence of the effectiveness of any intervention aimed at preventing SIDS. None of the intervention programmes described in the accompanying commentary2 have been evaluated in appropriately controlled studies.
The discussion in the paper is rather more circumspect than the conclusion in the abstract, but it is the latter which, reinforced by the accompanying commentary, is likely to have disproportionate impact on those readers who do not read the whole paper. In the commentary Dr Carpenter acknowledges that the low rates of SIDS make such interventions difficult to justify but then suggests that “Targeting such infants for weight monitoring at home . . . will also identify numerous other problems at a remedial stage”. This is a suggestion for which there is currently little evidence.
There are clear dangers in recommending population interventions on the basis of relative risks without taking account of absolute risks or of the need for interventions to be of proven effectiveness.
Drs Blair, Fleming, and Platt comment:
Poor postnatal weight gain was a significant factor in the multivariate analysis despite controlling for low birth weight, prematurity, neonatal problems, poor socioeconomic status, and many other potential confounding factors, and remained significant when further highly predictive covariates of SIDS such as infants put down prone, infants found with head covered, and tobacco exposure were added to the model. In this sense postnatal weight gain was independently associated with an increased risk of SIDS. We also found the difference in weight gain between the SIDS and control infants measured from birthweight to the last known weight was equally apparent if measured from birth to the six week assessment. We therefore stand by our conclusion in the abstract that poor postnatal weight gain “was independently associated with an increased risk of SIDS” and also that poor postnatal weight gain “could be identified at the routine six week assessment”.
We find it difficult to understand how this conclusion advocates an intervention campaign on the basis of these two statements and disagree with Logan and colleagues that our discussion on this matter was circumspect. We point out clearly that poor weight gain itself is not a sensitive marker and that it “should be seen as a thread in a web of factors that render an infant vulnerable to SIDS and is both a consequence of adverse health and social conditions”.
We agree that absolute risks must be used for targeted prevention campaigns but do not advocate such a campaign based on our solitary finding. Preliminary analysis of risk scoring on the first two years of our dataset, tested on the third year show that 42% of SIDS families can be identified from 8% of the population using just a few prenatal factors. Incorporating postnatal factors such as weight gain may improve the specificity and sensitivity of such a scoring system.
To our knowledge the “Back to Sleep” campaign initiated in October 1991 has not been evaluated in an appropriately controlled study yet the SIDS rate in England and Wales more than halved from 1.7 deaths per 1000 live births in 1990 to 0.77 deaths in 1992.1-1 Findings from our study have helped build on the advice regarding safe sleeping practices; parents are now advised not only to avoid placing infants in the prone position but also the side position, bed sharing under certain circumstances, sharing a sofa to sleep, and to avoid head covering by placing the feet of the infant at the foot of the cot. In 1998 the SIDS rate fell a further 25% in England and Wales to 0.45 deaths per 1000 livebirths.1-2 There is no direct evidence that these recommendations and fall in SIDS rates are linked although the dramatic fall in the prevalence of parents placing their infant in the prone position to sleep (57% before the campaign1-3 to 3% after the campaign1-4) suggest many parents take up such recommendations.
A targeted population intervention would perhaps be inappropriate given the reduced number of SIDS deaths in this country but risk scoring systems could be used to identify and study “high risk” families to both increase our understanding of the risks associated with the infant sleeping environment and improve the advice we give parents in the hope that the number of SIDS deaths reduces still further.