Arch Dis Child 1999;81:117–119 117

Use of and the risk of sudden death Arch Dis Child: first published as 10.1136/adc.81.2.117 on 1 August 1999. Downloaded from syndrome

E A Mitchell, S M Williams, B J Taylor

Abstract examined sharing because we postulated Background—The use of duvets in infancy that there might be a diVerence in the risk of is not recommended in the UK and SIDS between duvets used for children and Australia because of a reported associ- those used by adults. We also examined ation with sudden infant death syndrome whether the infant was tucked in firmly (SIDS). because we postulated that this might attenuate Aims—To examine the association be- the risk if duvets were a risk factor for SIDS. tween the use of duvets and the risk of SIDS. Methods Methods—A nationwide case control The New Zealand cot death study was a nation- study (393 cases, 1592 controls). The use of wide case control study (1 November 1987 to 31 duvets was assessed by interview with the October 1990). The methods have been re- 78 parent or guardian. ported in detail previously. Four hundred and Results—The use of duvets was associated eighty five SIDS cases in the postneonatal age with an increased risk of SIDS (odds ratio group were compared with 1800 controls, who (OR) = 1.65; 95% confidence interval were a representative sample of all births in the (CI), 1.31 to 2.08); however, after adjust- study areas. For control , a nominated ment for potential confounders there was date and time was used to base interview ques- no increased risk of SIDS (OR = 1.04; 95% tions on, which ensured a similar age distribu- CI, 0.77 to 1.38). Furthermore, subgroup tion to that expected for the cases. analysis did not identify any group in Interviews with parents or guardians were which the use of duvets was associated conducted for 393 (81.0%) cases and 1592 with an increased risk of SIDS. (88.4%) controls. Obstetric records were Conclusions—This study does not support examined in 465 cases (95.9%) and 1762 con- the recommendation to avoid duvets. trols (88.4%). (Arch Dis Child 1999;81:117–119) The use of duvets was obtained by interview from the question: “Was there a or Keywords: sudden infant death syndrome; duvet; quilt; over the baby?”. case control study We calculated univariate odds ratios (OR) http://adc.bmj.com/ using the method of Cornfield for confidence intervals (CI). Logistic regression was used to Some infant care practices are associated with calculate odds ratios adjusted for possible con- an increased risk of sudden infant death founders. These variables were related to 1 syndrome (SIDS). In the UK a recommen- socioeconomic (marital status, occupation, and dation was made in 1993 that duvets () age mother left school), pregnancy (age of 2 should not be used for in infancy. mother at first pregnancy, age of mother at on September 27, 2021 by guest. Protected copyright. Australia followed this lead in 1997. Studies infant’s birth, months’ pregnant when first examining the association between the use of Department of attended antenatal clinic, attendance at ante- duvets and the occurrence of SIDS are natal education classes, and number of previ- Paediatrics, University 3–6 of Auckland, Private conflicting. We reported previously that the ous pregnancies), infant (ethnicity, sex, birth- Bag 92019, Auckland, use of duvets was associated with an increased weight, and gestation), and postnatal factors New Zealand risk of SIDS at the univariate level, but not (age of infant, time of day, season, maternal E A Mitchell after adjustment for potential confounders.4 A smoking, bed sharing, region, infant admission recent study from the UK reported a small, but Department of to a neonatal unit, sleeping position, tucked in Preventive and Social not significant, increased risk of SIDS with tightly, and ). These variables Medicine, University duvets after adjustment, although the authors have been defined previously and the preva- of Otago Medical concluded that the use of duvets should be lence and relative risks published.378 We 4 School, Box 913, strongly discouraged in infancy. An earlier classified age as aged < 13 weeks or > 13 Dunedin, New Zealand report from the Tasmanian case control study S M Williams weeks. found no association between the use of duvets We assessed interactions by comparing the 5 Department of and SIDS, but recently they reported an change of deviance brought about by including a Paediatrics, University increased risk of SIDS with duvets in infants term for the interaction with the ÷2 distribution. of Otago, Box 913, sleeping supine or on the side, particularly Dunedin among infants > 12 weeks old.6 Duvets had no Results B J Taylor adverse eVect if the infant slept prone. The use of duvets was associated with an Correspondence to: We have re-examined our case control study increased risk of SIDS (OR = 1.65; 95% CI, Dr Mitchell. to see whether duvets increase the risk of SIDS 1.31 to 2.08) (table 1); however, after adjust- email: e.mitchell@ in any subgroups. The subgroups of interest ment for potential confounders there was no auckland.ac.nz suggested by the Tasmanian study are sleep increased risk of SIDS (OR = 1.04; 95% CI, Accepted 26 March 1999 position and age of infant. In addition, we 0.77 to 1.38). 118 Mitchell, Williams, Taylor

Table 1 Numbers and odds ratios (OR) for use of duvets natal clinic (0–3 months, 44.1%; 4+ months, Arch Dis Child: first published as 10.1136/adc.81.2.117 on 1 August 1999. Downloaded from 53.2%; ÷2 = 7.4; df = 1; p = 0.007); and (4) Cases Controls attendance at antenatal education classes Useofduvets No % No % Univariate OR (95% CI) Multivariate OR (95% CI) (attended, 43.0%; did not attend, 48.0%; ÷2 = 3.9; df = 1; p = 0.049). Duvet use was not No 162 42.0 856 54.3 1.00 1.00 Yes 224 58.0 719 45.7 1.65 (1.31 to 2.08) 1.04 (0.77 to 1.38) associated with the number of previous preg- nancies. CI, confidence interval. With regard to factors related to the infant, Complete data were available for 364 cases duvet use was associated with ethnicity (Maori, and 1538 controls. The risk of SIDS associated 60.5%; Pacific Island, 51.5%; other, 41.1%; with duvets was examined by including interac- ÷2 = 38.2; df = 2; p = 0.001). Duvet use was tion terms in the model in the following not associated with sex, birthweight, or gesta- subgroups: sleep position, age of infant, bed tion. sharing, and being tucked in firmly. The risk of The use of duvets was associated with the SIDS associated with duvet use did not diVer following postnatal factors: (1) age of infant by sleep position (change in deviance, ÷2 = 3.2; (< 13 weeks, 37.9%; 13–19 weeks, 49.0%; degrees of freedom (df) = 1; p = NS (not 20–25 weeks, 53.0%; 26+ weeks, 56.3%; significant)); age of infant (change in deviance, ÷2 = 35.8; df = 3; p = 0.001); (2) time of day 2 ÷ = 0.03; df = 1; p = NS); bed sharing (00:00–05:59, 48.7%; 06:00–23:59, 43.1%; 2 (change in deviance, ÷ = 0.03; df = 1; ÷2 = 4.9; df = 1; p = 0.026); (3) season p = NS); or being tucked in firmly (change in (January/February, 34.1%; July/August, 2 deviance, ÷ = 1.5; df = 1; p = NS) (table 2). 55.0%; ÷2 = 39.5; df = 5; p = 0.001); (4) The characteristics of families that used maternal smoking (smoker, 51.6%; non- duvets were examined in the control group 2 smoker, 42.9%; ÷ = 10.5; df = 1; p = 0.001); only. With regard to socioeconomic factors, (5) bed sharing (yes, 63.9%; no, 43.5%; duvet use was associated with: (1) marital sta- ÷2 = 24.8; df = 1; p = 0.001); (6) being tucked tus (married, 42.3%; unmarried, 55.1%; 2 2 in firmly (yes, 41.1%; no, 48.7%; ÷ = 8.6; ÷ = 20.0; df = 1; p = 0.001); (2) occupation (high, 38.9%; middle, 46.0%; low, 57.6%; df = 1; p = 0.003); and (7) ever been found ÷2 = 26.6; df = 2; p = 0.001); and (3) age with head covered by bedding (yes, 49.9%; no, 43.7%; ÷2 = 5.1; df = 1; p = 0.024). Although mother left school (< 16 years, 52.5%; 16 2 years, 44.3%; 17+ years, 43.2%; ÷2 = 8.9; duvet use was associated with region (÷ df = 2; p = 0.012). = 13.1; df = 4; p = 0.011), the diVerences were 2 With regard to factors related to pregnancy, not explained by a linear trend (÷ for linear the use of duvets was associated with: (1) age of trends, 0.031; p = 0.859, regions being or- mother at first pregnancy (< 20 years, 60.6%; dered from north to south). Duvets were not 20–24 years, 47.5%; 25+ years, 38.4%; associated with infant admission to a neonatal 2 unit, sleeping position, or breastfeeding. ÷ = 41.3; df = 2; p = 0.001); (2) age of http://adc.bmj.com/ mother at infant’s birth (< 20 years, 58.5%; For cases only we have data for being found 20–24 years, 52.7%; 25–29 years, 42.8%; 30+ with head completely covered by bedding. Of years, 41.2%; ÷2 = 21.0; df = 3; p = 0.001); (3) those cases using a duvet, 18.1% were found months’ pregnant when first attended ante- with head covered compared with 12.6% of

Table 2 Numbers and odds ratios (OR) for use of duvets by selected variables

Cases Controls on September 27, 2021 by guest. Protected copyright.

Useofduvets n% n%Univariate OR (95% CI) Multivariate OR (95% CI)

Infants sleeping prone only No 97 39.3 279 54.2 1.00 1.00 Yes 150 60.7 236 45.8 1.83 (1.33 to 2.52) 1.29 (0.88 to 1.89) Infants sleeping non-prone only No 63 46.3 575 54.6 1.00 1.00 Yes 73 53.7 479 45.4 1.39 (0.96 to 2.02) 0.77 (0.50 to 1.19)

Infants bed sharing last sleep only No 25 27.5 60 36.1 1.00 1.00 Yes 66 72.5 106 63.9 1.49 (0.82 to 2.72) 1.80 (0.92 to 3.52) Infants not bed sharing last sleep only No 137 46.4 793 56.5 1.00 1.00 Yes 158 53.6 610 43.5 1.50 (1.16 to 1.94) 0.98 (0.48 to 2.00)

Infant tucked in firmly No 42 42.0 356 58.9 1.00 1.00 Yes 58 58.0 248 41.1 1.98 (1.26 to 3.11) 1.35 (0.80 to 2.29) Infant not tucked in firmly No 118 41.5 496 51.3 1.00 1.00 Yes 166 58.5 470 48.7 1.48 (1.13 to 1.96) 0.93 (0.66 to 1.30)

Age of infant < 13 weeks No 92 46.0 450 62.1 1.00 1.00 Yes 108 54.0 275 37.9 1.92 (1.38 to 2.67) 1.06 (0.70 to 1.60) 13+ weeks No 70 37.6 406 47.8 1.00 1.00 Yes 116 62.4 444 52.2 1.52 (1.08 to 2.13) 1.01 (0.68 to 1.51) Duvets and the risk of SIDS 119

2

those not using a duvet (÷ = 2.1; df = 1; We have reported previously that firm Arch Dis Child: first published as 10.1136/adc.81.2.117 on 1 August 1999. Downloaded from p = 0.145). tucking reduces the risk of SIDS.3 In our study, duvet use among the control infants was asso- Discussion ciated with not being tucked in firmly. The use of duvets was determined by a single However, the risk of SIDS with the use of question. In New Zealand duvets/quilts are duvets did not diVer with whether or not the very varied in thickness and composition. They infant was firmly tucked in. range from being thick, soft, and made from Our study does not support the recommen- down or synthetic material to being thin dation to avoid duvets. However, further stud- crochet cotton covers. Thus, our results may ies are needed to be more precise about the not apply to all types of duvets. characteristics of the duvet, such as its thermal We found that the use of duvets in the properties and the potential for rebreathing control population was strongly associated with expired air should the face be covered by the winter and bed sharing. It was also strongly duvet. associated with socioeconomic disadvantage. The use of duvets was found to be associated with an increased risk of SIDS at the univariate Other members of the New Zealand cot death study group are DMO Becroft, AW Stewart, RPK Ford, R Scragg, JMD level, but after adjustment for socioeconomic Thompson, DMJ Barry, EM Allen, AP Roberts, and IB Hassall. and other infant care practices there was no This study was funded by the Health Research Council of New significant increased risk. This finding is Zealand (HRCNZ) and the Hawkes Bay Medical Research Foundation. We thank the research interviewers who undertook consistent with the UK study, which found a the interviewers and the pathologists in the study regions for small but non-significant increased risk of participating in the study. Finally, we sincerely thank the parents and families who participated in this study, without whom the SIDS with duvet use after adjustment for study would have been impossible. Mrs C Everard coordinated sleeping environment and socioeconomic the study and Mr J Thompson assisted with data management. 4 Mrs S Williams, Mrs Everard, Mr Thompson, and Mr Stewart factors. were funded by HRCNZ. We were not able to confirm the Tasmanian findings6; indeed, our findings were in the opposite direction to theirs. We found a 1 Fleming PJ, Gilbert R, Azaz Y, et al. Interaction between bedding and sleeping position in the sudden infant death non-significant decreased risk with duvet use in syndrome: a population based case-control study. BMJ infants sleeping supine or on their side 1990;301:85–9. 2 Report of the Chief Medical OYcers’ expert group on the (adjusted OR = 0.77; 95% CI, 0.50 to 1.19). sleeping position of infants and cot death. London: There are two possible explanations. Duvets in Department of Health, HMSO, 1993. 3 Wilson CA, Taylor BJ, Laing RM, Williams SM, Mitchell Australia diVer from those in New Zealand. EA. Clothing and bedding and its relevance to sudden Australian duvets tend to be the same size as infant death syndrome: further results from the New the cot , thus they cannot be tucked in. Zealand cot death study. J Paediatr Child Health 1994;30: 506–12. In contrast, New Zealand duvets are larger 4 Fleming PJ, Blair PS, Bacon C, et al. Environment of infants than the mattress and can be tucked in. during sleep and risk of the sudden infant death syndrome: results of 1993–5 case-control study for confidential Second, the Tasmanian study did not adjust for inquiry into stillbirths and deaths in infancy. BMJ socioeconomic factors. In contrast to the 1996;313:191–5. http://adc.bmj.com/ 5 Ponsonby A-L, Dwyer T, Kasl SV, Cochrane JA. The Tasmanian study, in our study the risk of SIDS Tasmanian SIDS case-control study: univariate and multi- with the use of duvets was not influenced by the variate risk factor analysis. Paediatr Perinatal Epidemiol age of the infant. 1995;9:256–72. 6 Ponsonby A-L, Dwyer T, Couper D, Cochrane J. Associ- In general, bedding incorporating duvets has ation between use of a quilt and sudden infant death greater thermal insulation properties than bed- syndrome: case-control study. BMJ 1998;316:195–6. 9 1 7 Mitchell EA, Scragg R, Stewart AW, et al. Results from the ding not using duvets. We and others have first year of the New Zealand cot death study. NZMedJ shown that the prone sleeping position in- 1991;104:71–6. 8 Mitchell EA, Taylor BJ, Ford RPK, et al. Four modifiable on September 27, 2021 by guest. Protected copyright. creases the risk from excess thermal insulation. and other major risk factors for cot death: the New Zealand Thus, our finding of a non-significant in- study. J Paediatr Child Health 1992;28(suppl 1):S3–8. 9 Williams SM, Taylor BJ, Mitchell EA. Sudden infant death creased risk with the use of duvets when sleep- syndrome: insulation from bedding and clothing and its ing prone is not unexpected. eVect modifiers. Int J Epidemiol 1996;25:366–75.