RT Journal Article
SR Electronic
T1 Recurrence rates for sudden infant death syndrome (SIDS): the importance of risk stratification
JF Archives of Disease in Childhood
JO Arch Dis Child
FD BMJ Publishing Group Ltd and Royal College of Paediatrics and Child Health
SP 936
OP 939
DO 10.1136/adc.2007.121350
VO 93
IS 11
A1 M J Campbell
A1 D Hall
A1 T Stephenson
A1 C Bacon
A1 J Madan
YR 2008
UL http://adc.bmj.com/content/93/11/936.abstract
AB Objective: To investigate the importance of stratification by risk factors in computing the probability of a second death from sudden infant death syndrome (SIDS) in a family.Design: Simulation study.Background: The fact that a baby dies suddenly and unexpectedly means that there is a raised probability that the baby’s family have risk factors associated with SIDS. Thus one cannot consider the risk of a subsequent death to be that of the general population. The Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI) identified three major social risk factors: smoking, age<27 and parity>1, and unemployed/unwaged as major risk factors. It gave estimates of risk for families with different numbers of these risk factors. We investigate whether it is reasonable to assume that, conditional on these risk factors, the risk of a second event is independent of the risk of the first and as a consequence one can square the risks to get the risk of two SIDS in a family. We have used CESDI data to estimate the probability of a second SID in a family under different plausible scenarios of the prevalence of the risk factors. We have applied the model to make predictions in the Care of Next Infant (CONI) study.Results: The model gave plausible predictions. The CONI study observed 18 second SIDS. Our model predicted 14 dealths (95% prediction interval 7 to 21).Conclusion: When considering the risk of a subsequent SIDS in a family one should always take into account the known risk factors. If all risks have been identified, then conditional on these risks, the risk of two events is the product of the individual risks. However, for a given family we cannot quantify the magnitude of the increased risk because of other possible risk factors not accounted for in the model.