Risk Factors for Sudden Infant Death Syndrome Following the Prevention Campaign in New Zealand: A Prospective Study
Objectives. To identify the risk factors for sudden infant death syndrome (SIDS) following a national campaign to prevent SIDS.
Methods. For 2 years (October 1, 1991 through September 30, 1993) data were collected by community child health nurses on all infants born in New Zealand at initial contact and at 2 months.
Results. There were 232 SIDS cases in the postneonatal age group (2.0/1000 live births) and these were compared with 1200 randomly selected control subjects. Information was available for 127 cases (54.7%) and 922 (76.8%) of controls.
The previously identified modifiable risk factors were examined. The prevalence of prone sleeping position of the infant was very low (0.7% at initial contact and 3.0% at 2 months), but was still associated with an increased risk of SIDS. In addition, the side sleeping position was also found to have an increased risk of SIDS compared with the supine sleeping position (at 2 months: adjusted odds ratio (OR) = 6.57; 95% confidence interval (CI) = 1.71, 25.23). Maternal smoking was found to be the major risk factor for SIDS. Bed sharing was also associated with an increased risk of SIDS. There was an interaction between maternal smoking and bed sharing on the risk of SIDS. Compared with infants not exposed to either bed sharing or maternal smoking, the adjusted OR for infants of mothers who smoked was 5.01 (95% CI = 2.01, 12.46) for bed sharing at the initial contact and 5.02 (95% CI = 1.05, 24.05) for bed sharing at 2 months. In this study breastfeeding was not associated with a statistically significant reduction in the risk of SIDS.
The other risk factors for SIDS identified were: unmarried mother, leaving school at a younger age, young mother, greater number of previous pregnancies, late attendance for antenatal care, smoking in pregnancy, male infant, Maori ethnicity, low birth weight, and shorter gestation.
Conclusions. After adjustment for potential confounders, prone and side sleeping positions, maternal smoking, and the joint exposure to bed sharing and maternal smoking were associated with statistically significant increased risk of SIDS. A change from the side to the supine sleeping position could result in a substantial reduction in SIDS. Maternal smoking is common in New Zealand and with the reduction in the prevalence of prone sleeping position is now the major risk factor in this country. However, smoking behavior has been difficult to change. Bed sharing is also a major factor but appears only to be a risk to infants of mothers who smoke. Addressing bed sharing among mothers who smoke could reduce SIDS by at least one third. Breastfeeding did not appear to offer a statistically significant reduction in SIDS risk after adjustment of potential confounders, but as breastfeeding rates are comparatively good in New Zealand, this result should be interpreted with caution as the power of this study to detect a benefit is small.
- sudden infant death syndrome
- prospective study
- case-cohort study
- risk factors
- sleep position
- maternal smoking
- bed sharing
M;-2qortality from sudden infant death syndrome (SIDS, or cot death) dropped dramatically in New Zealand in 1990 (1989: 4.1/1000 live births; 1990: 2.9/1000). This reduction was not attributable to diagnostic transfer as deaths from all causes in the postneonatal period also decreased (1989/1000; 1990: 4.2/1000).1Since then SIDS and postneonatal mortality has declined further (1993: 2.1/1000 and 3.5/1000, respectively).2 This reduction in mortality has been attributed to the decline in numbers of infants placed prone to sleep.3
The New Zealand Cot Death Study, a large 3-year (November 1, 1987 to October 31, 1990) nationwide case-control study, identified an increased risk of SIDS with prone sleeping position, maternal smoking, and not being breastfed.4,5 A national cot death prevention programme was formally launched in New Zealand in early 1991,6 although the prevalence of infants sleeping prone had begun to decline at the end of 1989.7 The campaign initially recommended that infants should be placed on their side or back, that infants should not be exposed to tobacco smoke either during pregnancy or for the first year of life, and mothers should breastfeed if at all possible. At the end of 1991 parents were also advised not to sleep with their infants. Further analysis of the New Zealand Cot Death Study found that the increased risk of SIDS from bed sharing was substantial if the infant had been exposed to maternal smoking.8,9
We have previously reported changes in SIDS and total postneonatal mortality.1 This current study was established to monitor changes in the prevalence of the modifiable risk factors targeted by the prevention campaign and other risk factors and to ascertain if the risk factors or their magnitude changed as the prevalence of other risk factors changed. This article reports the risk factors for SIDS following a prevention programme that has substantially reduced the number of infants sleeping prone.
In outline, the study design was a case-cohort. Data were recorded routinely on every infant by Plunket nurses (community child health nurses) for all live births between October 1, 1991 and September 30, 1993. The data from SIDS cases plus a random sample of control infants were extracted and examined in the usual case-control method.
Data on infant care practices were recorded by the nurse at two points in time, at the first contact (initial) and at approximately 2 months of age. These data were recorded in the nursing record. The dates of the visits were also recorded. Information on the following infant care practices in the previous 24 hours were recorded at initial contact and at 2 months: main type of infant feeding, maternal and paternal smoking, the positions infants were placed to sleep, and whether the mother fell asleep with the infant in bed.
Additional data recorded at the initial contact included (a) sociodemographic variables: marital status, age mother left school, age of mother, self-reported ethnicity of mother and father; (b) pregnancy factors: number of previous pregnancies, when antenatal care began (months) and maternal smoking in pregnancy; (c) infant variables: infant's date of birth, reported ethnicity of infant, sex, birth weight and gestation; and (d) the type of infant feeding at discharge from the obstetric hospital. At this contact the nurse explained the study and the mother signed a consent to participate in the study, which was printed in the nursing record.
All deaths registered by the New Zealand Health Information Service as attributable to SIDS in the postneonatal age group (dying after 28 completed days and within the first year of life) form the cases of this study. This takes into account the findings at autopsy, if performed. Although autopsy was not an essential component of the definition, but we have previously reported that almost all (98%) of deaths classified as SIDS have had an autopsy.5
Although data were prospectively collected on all births in the 2-year period, it was not economic or logistically feasible to process all the data. Accordingly, a sample of infants was obtained that was representative of all births within New Zealand. The method for sampling was: (a) a date of birth was randomly selected from all the days in the study period; (b) an obstetric hospital was randomly chosen in proportion to the number of births; (c) in the obstetric hospital with multiple births on nominated date of birth random numbers were used to select a particular infant from among those born on that day; and (d) a direction variable, which indicates to either go forward or back in looking for a birth in the situation where the hospital did not have one on the nominated day, was also randomly chosen. This results in a sample that is representative of all infants born in New Zealand.
The data from SIDS cases and the selected controls were transcribed from the nursing record to a data sheet only if consent had been signed prospectively. The data sheet was then sent to the study center.
The number of live births for years 1992 and 1993 was 59 266 and 58 866, respectively. The distribution of the mothers' age and marital status and the infants' birth weight and gestation for all live births in New Zealand in 1993 were obtained from official publications.2,10 The expected distribution of the infants' ethnicity and maternal smoking in pregnancy were estimated from a representative sample of all live births in New Zealand in a previous study.5
Although this was a case-cohort study it was analyzed in the usual case-control method. This results in confidence intervals (CIs) that are slightly conservative.11 Relative risks were estimated by calculation of odds ratios (OR). The univariate ORs have CIs calculated by the method of Cornfield, except for when the cell numbers were small and the exact method was used.12 The multivariate ORs are obtained from unconditional logistic regression modelling as are their CIs.
Population-attributable risks (PARs) were calculated to estimate the proportion of deaths explained by exposure to particular risk factors.13
Ethical permission for this study was obtained from the University of Auckland Human Subjects Research Ethics Committee, the Plunket Ethics Committee, and the ethics committees in each region.
In the 2-year cohort there were 429 (3/1000) postneonatal deaths (Table 1). There were 232 SIDS deaths (2/1000) in the postneonatal age group (Table 1). In addition there were 26 SIDS deaths in the neonatal age group and 2 deaths after the first year of life. For this case-control study only SIDS deaths occurring in the postneonatal age group were considered. A total of 1200 controls were selected. There was a significant difference in the proportion of data sheets returned for cases and controls (cases: 127 (54.7%) returned; controls: 922 (76.8%) returned; χ2 = 48.8, df = 1, P< .001). The reasons data sheets were not returned were: families who refused to participate (cases: 15.9%; controls: 10.8%); not supervised by Plunket nurses (cases: 7.8%; controls: 4.9%); and other reasons, such as: families moved without leaving forwarding address; families not invited to participate in the study; and records lost (cases: 21.6%; controls: 7.5%). Control participants in the study were compared with all live births in New Zealand. Mothers in the study were more likely to be married (71.0% vs 63.3%, respectively, χ2 = 17.6, df = 1, P < .001) and not to have smoked in pregnancy (75.1% vs 69.0%, χ2 = 10.3, df = 1, P < .001), but their ages did not differ (χ2 = 5.4,df = 3, P = .145). Infants participating in the study did not differ from all live births in regards to birth weight (χ2 = 2.2, df = 3, P = .527), gestation (χ2 = 0.8,df = 1, P = .366) and ethnicity (χ2 = 2.6, df = 2, P = .276).
The ages of the cases and controls were not significantly different at the initial visit (cases: mean = 2.6 weeks, standard deviation [SD] = 1.7; controls: mean = 2.4 weeks, SD = 1.2) and at the 2 months contact (cases: mean = 9.1 weeks, SD = 1.9; controls: mean = 9.2 weeks, SD = 2.2).
Univariate relative risks of SIDS associated with variables are shown in Tables 2-6. For variables related to socioeconomic status, infants of mothers in de facto relationships or single had an increased risk of SIDS compared with infants of married mothers; the risk of SIDS was inversely related to the age of mother and the age she left school (Table 2).
For variables related to the pregnancy, there was a positive association between the risk of SIDS and the number of previous pregnancies, smoking in pregnancy, and late attendance at antenatal clinic (after 3 months of pregnancy) (Table3).
With regard to the infants, male infants had an increased risk of SIDS compared with female infants; Maori infants were more likely to die from SIDS than non-Maori infants; birth weight and gestation were inversely associated with the risk of SIDS (Table4).
For factors after birth, the three measures of breastfeeding (exclusive breastfeeding at discharge from the obstetric hospital, any breastfeeding at initial contact, and any breastfeeding at 2 months) were not associated with a statistically significant reduction in risk of SIDS; the risk of SIDS became greater with increasing levels of maternal smoking in the previous 24 hours; the risk of SIDS increased if the father smoked; the prone sleeping position in the previous 24 hours and the side sleeping position (at 2 months) increased the risk of SIDS compared with the supine position; and falling asleep with the infant in bed was associated with an increased risk of SIDS compared with infants not bed sharing (Table 5).
We have previously shown an interaction between bed sharing and maternal smoking.8 In this study there was a substantial increase in the risk of SIDS where the mother both smoked and fell asleep with the infant in the same bed compared with infants not exposed to either risk factor (Table 6). Infants exposed to maternal smoking but not bed sharing had an increased risk of SIDS; however, there was no statistically significant increase in risk of SIDS from bed sharing if the mother was a nonsmoker.
The effect of smoking by either parent was examined (Table7). The relative risk of SIDS for infants of parents who both smoked was substantial (OR = 10.27; 95% CI = 6.17, 17.08) compared with infants whose parents did not smoke. The risk was increased when only the mother smoked, but there was no significant increased risk if only the father smoked.
The modifiable risk factors (sleep position, breastfeeding, maternal smoking, and bed sharing) were examined after controlling for potential confounders (Table 8). The prone and side sleeping positions were significantly associated with an increased risk compared with the supine sleeping position. Bed sharing was a significant risk factor for infants of mothers who smoked, but not for infants of mothers who did not smoke.
PAR provides an estimate of the proportion of cases that can be attributed to the risk factor, assuming there is a causal relationship between the putative risk factor and the death. The PAR estimates for selected variables are shown in Table 9. The estimates are similar for the initial and 2-month contacts. A substantial reduction in SIDS might be achieved if infants were placed supine instead of on their side. Almost 50% of SIDS cases can be attributed to maternal smoking only and a third to the joint exposure to bed sharing and maternal smoking. In contrast, improving breastfeeding rates further will have little impact on SIDS mortality in New Zealand.
Ninety-two percent of newborn infants are seen by Plunket nurses in New Zealand (unpublished data), and 77% of all children born in the 2-year cohort participated in the study. However, the response rate of SIDS cases was significantly lower than that of controls. Indeed refusal to participate in the study or lack of supervision by community child health nurses were significantly associated with an increased risk of SIDS. A previous report from the Christchurch region of the study also found that nonparticipants were more likely to smoke than participants.14 In this national study those who did participate differed from all live births in New Zealand. Mothers in participating families were more likely to be married and not to have smoked in pregnancy than expected. However, mothers' age and the infants' birth weight, gestation, and ethnicity did not differ. This means that caution is required to extrapolate the results to the whole community. However, as measurements were made before the categorization into case and control, there is no differential selection bias; hence, the results are valid within this sample.
New Zealand was the first country to formally launch a national SIDS prevention campaign, although earlier work from Holland had suggested that SIDS mortality might be reduced by the recommendation to avoid placing the infant prone to sleep.15,16 This campaign resulted in a rapid decline in the prevalence of the prone position and an associated decrease in SIDS mortality.1,7
Recall bias is a potential problem with retrospective studies. In view of the publicity surrounding the prevention campaign, recall bias might be expected to be a major problem. Collecting data prospectively, as in this study, avoids this problem. This report is the first national study of risk factors for SIDS following a national SIDS prevention campaign. The previously identified major sociodemographic, pregnancy, and infant risk factors were significantly associated with an increased risk of SIDS.
Despite the fact that the prevalence of prone sleeping position has fallen to a low level (0.7% and 3.0% at initial and 2 months, respectively), it continues to be a significant risk factor for SIDS. We previously calculated that prone sleeping position was causally associated with 52% of SIDS cases in New Zealand in 1987/88.4 We now estimate that prone sleeping position is associated with 10% of SIDS cases. This indicates the importance of maintaining the advice on infants' sleeping position in the national SIDS education program.
Most infants are now placed to sleep on their sides in New Zealand. We have previously reported that the risk of SIDS is greater in the side sleeping position compared with the supine position; however, the result was not statistically significant after controlling for potential confounders.5 This study found a statistically significant increased risk for SIDS with the side sleeping position after adjustment. A similar finding has recently been reported from the United Kingdom.17 A further substantial reduction in SIDS may occur with a change to the supine sleeping position. It now seems appropriate to advise the back sleeping position rather than the side or back sleeping positions.
We previously reported that breastfeeding was associated with a reduced risk of SIDS.18 Although breastfeeding was not found to be protective in this study, this result should be interpreted with caution as breastfeeding rates are high in New Zealand and this study might lack the power to detect a benefit. Furthermore, in countries with a low breastfeeding rate the potential benefit on SIDS mortality will be much greater.
Bed sharing was again identified as a risk factor for SIDS. There are now 10 studies that have examined the association between bed sharing and SIDS.5,17,19-25 As our data were recorded prospectively, recall bias cannot be an explanation for this result. We have previously reported that the increased risk of SIDS with bed sharing was predominantly for infants of mothers who smoked.8,9 This study and two other studies confirm that finding.17,25 No statistically significant increased risk with bed sharing was seen for infants of mothers who did not smoke. As the bed sharing measure referred to a practice that occurred previously (sometimes months before), we cannot exclude the possibility that bed sharing on the night of death is a risk factor. Mothers of infants should be advised that bed sharing substantially increases the risk of SIDS if she smokes. For mothers who do not smoke the evidence is inconclusive; however, there is no evidence that bed sharing in any group is protective.
PAR calculations for the joint exposure to bed sharing and smoking indicates that approximately 30% of SIDS cases might be prevented if either bed sharing or maternal smoking was avoided. This proportion may be even greater than 30% as a higher proportion of nonparticipants in the study are likely to have been bed sharing smokers.
There have been many studies showing an increased risk of SIDS associated with maternal smoking and we have argued that maternal smoking is causally related to SIDS.26 We and others have shown fathers who smoke also increase the risk of SIDS, however, this increased risk is predominantly for infants of mothers who smoke.27-29 Maternal smoking either alone or in combination with bed sharing is the major risk factor for SIDS in New Zealand. Smoking behavior has proved difficult to change. Indeed the prevalence of smoking in pregnancy does not appear to have changed in New Zealand.30 This suggests the need for other strategies, including enforcing laws that prohibit the sale of tobacco to minors, minimum package size, restricting smoking in public places, plain packaging, stronger health warnings, and banning advertising and sport sponsorship.
Although SIDS mortality rates have declined in many countries with the reduction in the prevalence of the prone sleeping position,1,15,31-34 this study indicates that further substantial reductions in SIDS mortality might be achieved with a change to back sleeping position, a reduction in maternal and paternal smoking, and a reduction in the joint exposure to bed sharing and maternal smoking.
This study was funded by the Cot Death Association and the Public Health Commission. They also funded Mrs C. Everard and Mr Clements. Mr Stewart is funded by the Health Research Council of New Zealand.
Mrs C. Everard coordinated the study. We sincerely thank the Plunket nurses who collected the data. We thank Dr D. Geddis, Dr R. Scragg, Ms N. Taylor, Ms A. Counsell, and Ms D. Saunders who helped develop the study. We also thank the staff of the New Zealand Health Information Service for assistance in the production and provision of infant mortality data.
- Received December 6, 1996.
- Accepted April 15, 1997.
- Address correspondence to: Dr Edwin A. Mitchell, Department of Paediatrics, School of Medicine, University of Auckland, Private Bag 92019, Auckland, New Zealand.
Reprints not available.
- SIDS =
- sudden infant death syndrome •
- OR =
- odds ratio •
- PAR =
- population-attributable risk •
- CI =
- confidence interval •
- SD =
- standard deviation
- Mitchell EA,
- Brunt JM,
- Everard C
- ↵New Zealand Health Information Service. Fetal and Infant Deaths, 1993. Wellington, New Zealand: Ministry of Health; 1996
- Mitchell EA,
- Ford RPK,
- Taylor BJ,
- et al.
- Mitchell EA,
- Taylor BJ,
- Ford RPK,
- et al.
- Mitchell EA,
- Tonkin S
- Scragg R,
- Mitchell EA,
- Taylor BJ,
- et al.
- ↵Department of Statistics. Demographic Trends, 1993. Wellington, New Zealand: Department of Statistics; 1995
- Whittemore AS
- de Jonge GA,
- Burgmeijer RJF,
- Engelberts AC,
- Hoogenboezem J,
- Kostense PJ,
- Sprij AJ
- Fleming PJ,
- Blair PS,
- Bacon C,
- et al.
- Ford RPK,
- Taylor BJ,
- Mitchell EA,
- et al.
- ↵Carpenter RG. Sudden and unexpected deaths in infancy (cot death). In: Camps FE, Carpenter RG, eds. Sudden and Unexpected Deaths in Infancy (Cot Death). Bristol, England: John Wright; 1972:7–15
- Luke JL
- Lee NNY,
- Chan YF,
- Davies DP,
- Lau E,
- Yip DCP
- ↵Mitchell EA, Thompson JMD. Co-sleeping increases the risk of sudden infant death syndrome, but sleeping in the parent's bedroom lowers it. In: Rognum TO, ed. Sudden Infant Death Syndrome. New Trends for the Nineties. Oslo, Norway: Scandinavian University Press; 1995:266–269; chap 50
- Klonoff-Cohen H,
- Edelstein SL
- ↵Willinger M, Hoffman HJ, Wu K-T, Gloekner CK, Hillman LS. Sleep environment: the NICHD SIDS Cooperative Epidemiological Study (Abstract). Presented at the Fourth SIDS International Conference; Washington, DC; June 23–26, 1996
- ↵Mitchell EA. Smoking: the next major and modifiable risk factor. In: Rognum TO, ed. Sudden Infant Death Syndrome. New Trends for the Nineties. Oslo, Norway: Scandinavian University Press; 1995;114–118: chap 21
- ↵Department of Health. Report of the Chief Medical officer's expert Group on the Sleeping Position of Infants and Cot Death. London, England: Her Majesty's Stationery Office; 1993
- Mitchell EA,
- Ford RPK,
- Stewart AW,
- et al.
- Blair PS,
- Fleming PJ,
- Bensley D,
- et al.
- ↵Mitchell EA. Monitoring of the New Zealand SIDS Prevention Programme: Final Report to the Public Health Commission. Auckland, New Zealand: Auckland Uniservices; 1994
- Gilbert R
- Copyright © 1997 American Academy of Pediatrics