E L S E V I E R Early Human Development 45 (1996) 119-131

Early Human Development

The 'new' risk factors for SIDS: is there an association with the ethnic and place of birth

differences in incidence in Victoria, Australia?

ag< Anne Potter ' , Judith Lumley b, Lyndsey Watson a

~'Centre Jor the Study q[' Mothers" and Children "s Health, La Trobe University, 463 Cardigan Street, Carlton, Victoria 3053, Australia

b . Natzomzl Perinatal Epidemiology Unit, Radcliffe Infirmary.". Oxford, OX2 6HE, UK

Received 12 June 1995; revised 30 November 1995; accepted 5 December 1995

Abstract

Home interviews and assessments of infant development were carried out in a project examining the family environments of four groups of women and their infants with different risks of sudden infant death syndrome (SIDS): a reference group of 200 Australian-born (Anglo-Celtic) women (SIDS incidence, 2.04/1000), 101 women born in Asia (0.97/1000) and 56 women born in Southern Europe (0.58/1000) whose infants have a very low risk of SIDS and 102 women who chose to give birth at home whose infants have a high risk of SlDS (8.9/1000). As these differences are not explained by the classic social and perinatal risk factors, it was possible that they might be attributable to the 'new' risk factors: prone sleeping position, not fully breast feeding, exposure to cigarette smoke and bed sharing. Analysis of the data did not show this expected association: there were no significant differences between the groups in the use of the prone position; although only two Asian-born women smoked, infant exposure to cigarette smoke was similar in the other three groups; choice of infant feeding method did not fit the expected pattern - - homebirth infants were fully breast fed almost exclusively while this was so for only about 50% of infants in both low risk groups; more than 50% of homebirth families slept with their infants, but bed sharing was also significantly more common in the Asian-born group than in the reference group.

Ke3words: Sudden infant death (SIDS); Ethnic differences; Smoking; Bed sharing; Prone sleeping; Breast feeding

Corresponding author. Tel.: + 6 l 3 93481211: fax: +61 3 93481129.

0378-3782/96/$15.00 © 1996 Elsevier Science Ireland Ltd. All rights reserved PII S 0 3 7 8 - 3 7 8 2 ( 9 6 ) 0 1 7 2 6 - 4

120 A. Potter et al. / Early Human Development 45 (1996) 119-131

I. Introduction

In Victoria, record linkage of death certificates and the birth records of infants who died suddenly and unexpectedly during the postneonatal period in 1985-1989 revealed 601 cases with a diagnosis of unexplained sudden infant death, The incidence of sudden infant death syndrome (SIDS) was 2.04/1000 in infants of Australian-born mothers. In infants whose mothers had been born in Southern Europe, the relative risk of SIDS was 0.28 (95% confidence interval (CI) 0.15, 0.55) and 0.48 (95% CI 0.29, 0.78) in infants whose mothers had been born in Asia [1]. Southern European was defined as Greece, Italy, Malta, Spain, Portugal and the former Yugoslavia. Asia was defined according to the Australian Bureau of Statistics categories, with the exclusion of the Middle East.

Record linkage for the years 1985-1987 identified several cases whose births had been planned out-of-hospital births. The relative risk of SIDS in this group was 3.63 (95% CI 1.62, 8.10) [2]. (In Victoria planned out-of-hospital births during the period 1983-1992 accounted for 0.2-0.3% of all births [3].) These significant ethnic and place of birth differences are not explained by the classical social and perinatal risk factors.

Over the past decade, one strand of SIDS research has focused on the care of infants; on parental behaviour and its effect on the infant's vulnerability to SIDS [4-6] and on the part cultural and ethnic influences may play in determining the characteristics of the setting in which the infant is raised [7]. The study described in this paper was designed to explore the micro-environments of the four groups of infants within a subsequent 3-year period. It was not a study of SIDS mortality.

While the study was underway, 'new' risk factors for SIDS, (sleeping in the prone position, overheating, co-sleeping, not fully breast fed and exposing infants to cigarette smoke) were defined in cohort and case-control studies elsewhere [3,4,6]. These aspects of infant care are now the basis of advice given to new mothers by health professionals and in public education campaigns throughout the western world. In New Zealand, where Maori infants are reported to be 3.81 times more likely to die of SIDS than non-Maori infants, the increased risk is said to be almost entirely attributable to higher rates of maternal smoking, bottle feeding, prone sleep position and co-sleeping [5]. Thus the first priority for analysis was to determine whether or not ethnic and place of birth differences in SIDS rates in Victoria were associated with the distribution of these 'new' risk factors.

2. Methods

The protocols of the Tasmanian Infant Health Study, a prospective cohort study of the causes of sudden infant death syndrome [4], were used in the collection of data at the time of recruitment and at the home interview 4 weeks after birth. A pilot study, in which six women were recruited and interviewed, was conducted at the end of 1990.

A. Potter et al. / Early Human Development 45 (1996) 119-131 121

The groups which the record linkage had identified as being at either significantly lower risk of SIDS (infants of women born in Southern Europe and Asian countries) or significantly raised risk of SIDS (infants of women having a planned out-of- hospital birth) and a reference group of Australian-born women of Anglo-Celtic background comprised the four groups in the study.

2.1. R e c r u i t m e n t - - h o s p i t a l

A large teaching hospital which services the eastern suburbs of Melbourne, Victoria was chosen for the recruitment of women in the Australian-born (Au), Southern European-born (SE) and Asian-born (As) groups.

Women whose place of birth was Australia, Southern Europe or Asia and who had a reasonable command of English, and whose infants were healthy singletons, born at term, were approached during their postnatal hospital stay. When women agreed to participate they were asked to sign a consent form which was witnessed by a family member, visiting friend or a staff member, An information sheet with the names and telephone numbers of members of the study team and a card indicating when to expect a phone call to arrange the time of a home interview were left with participants. Details of the infants' birth measurements and other perinatal data were recorded. At the time of recruitment, women were asked to complete a questionnaire which covered the educational levels and employment of both parents, family income, type of infant feeding intended and smoking history during pregnancy. Recruitment was carried out on one day each week and all eligible women were approached in both the public and private wards.

Recruitment commenced in January 1991 and continued until April 1993. In an attempt to keep the Australian-born group culturally homogeneous, every endeavour was made to ensure that the parentage of the women in this group was Anglo-Celtic. Even women who were uncertain about their family history were usually able to exclude Southern European or Asian links.

2.2. R e c r u i t m e n t - - h o m e b i r t h s

The recruitment of women in the planned out-of-hospital birth (Hb) group took place concurrently with the help of the doctor and midwives in a GP practice caring for families choosing to birth at home, and an independent homebirth midwife with her own practice. The midwives discussed the project with eligible families late in pregnancy and after a family had indicated a readiness to participate they were given a consent form, information sheet and other postnatal forms. The researcher (AP) was then notified of the mother's name, address and telephone number.

2.3. The h o m e i n t e r v i e w

The home interview was conducted when the infant was between 3 and 5 weeks of

122 A. Potter et al. / Early Human Development 45 (1996) 119-131

age. A structured questionnaire was used; this covered infant feeding, clothing and bedding, infant behaviour and minor illnesses. There were questions also about the mother's wellbeing and the domestic environment, including the smoking behaviour of occupants. Measurements of temperature and humidity were recorded.

2.4. Analysis

All the questionnaires were coded by the researcher (AP) who conducted the interviews. Data were analysed using Epi Info, Version 5.01 [8] and SPSS PC. The analysis and Statcalc components of EPI Info were used for descriptive statistics and statistical tests. Associations are expressed in terms of odds ratios and 95% confidence limits.

3. Results

Our aim was to recruit 200 women in the reference group and 100 in each of the other three. Very few problems were encountered in arranging and carrying out the interviews. It was not difficult to achieve the desired sample size in the Australian- born and Asian-born groups, but fewer than expected births to women in the Southern European categories occurred at the hospital during the period of the project, and of the women in this group who did give birth there, a greater number than expected had insufficient English. At the end of April 1993, the hospital-based sample comprised 200 Australian-born, 56 Southern European-born and 101 Asian-born women. By then, 102 women who had had births at home had been interviewed. This was nearly half (41.3%) of the planned out-of-hospital births which occurred in the Melbourne metropolitan area during the study period. Of the 503 women who were recruited, 43 were not interviewed, (Au 13, SE 2, As 7, Hb 4) and one withdrew from the project after the interview (n = 44, 8.7%). The most common reasons for not wanting an interview were family illness, a change of heart about participation and being 'too busy'.

Difficulties with English language which were confined to the Asian-born group, were not sufficient to compromise the overall quality of the data. Two questions which asked about attitudes to and satisfaction with motherhood were puzzling for a small minority of Asian-born women and were not able to be completed. Seventeen countries were represented in the Asian-born group, with Malaysia, Vietnam, and India providing the largest numbers, (18, 14 and 13, respectively). There were fewer than 10 women l¥om each of Cambodia, Singapore, Sri Lanka, Indonesia, China and Hong Kong and only one or two from Taiwan, Timor, Thailand, Japan, Korea, Laos, and Burma. In the Southern European group, nearly half the women were born in Greece (n 25, 45%), 16 in the former Yugoslavia, 13 in Italy and one in each of Spain and Portugal. The sample of women born overseas is distributed similarly to these groups in Victoria.

A. Potter et al. / Early Human Development 45 (1996) 119-131 123

e ~

©

e~

~g ~ . ~

• ~ ~

124 A. Potter et al. / Earn Human Development 45 (1996) 119-13l

More than three quarters (n = 79, 77%) of the planned out-of-hospital birth group were Australian-born. Of the remainder, the UK was the country of birth for 10, Germany for five and eight other countries were represented.

Table 1 shows some of the social differences between the mothers by region of birth. There are marked differences in parity. The percentage of women cohabiting (marr ied/unmarried) in each group is similar, although the proportion living together but not married differs (Au 14.0%, SE 3.6%, As 3.0%, Hb 19.6%).

A substantial minority of women in each group expected to return to work within the 6 months following the birth (Au 19.0%, SE 28.7%, As 32.4%, Hb 25.0%). However, the differences between the groups are more marked when part-time work and lull-t ime work are considered separately; homebirth mothers were significantly more likely to intend to work part-time than Asian-born mothers (31.4% vs. 1.0%) with the reverse being true for full-time work (7.9% vs. 20.8%).

Table 2 shows some of the characteristics of the infants by group. The lower median birth weight of the infants in the Asian-born group is consistent with the Victorian perinatal data. Six infants had a birth weight less than 2500 g (range, 2340-2465). Three of these were in the Australian-born group with one in each of the other three. The groups show marked differences in the use of the dummy.

There were no significant differences between the groups in the number of infants who had 'an unusual cry ' (Au 6.5%, SE 1.8%, As 5.9%, Hb 6.8%), or who were said to 'cry for no apparent reason' (Au 16.0%, SE 12.5%, As 16.0%, Hb 8.8%). Fewer than half of the infants in all groups had been 'found to be sweaty ' (Au 44.5%, SE 40.2%, As 45.5%, Hb 46.4%). The infants of Southern European-born mothers were more likely to have had a cold, but the majority in all groups had not (Au 78.0%, SE 69.6%, As 80.2%, Hb 77.5%). In those who had suffered from a cold, 'snuffles' was the most common symptom.

3.1. W e w ' risk fac tors

Table 3 shows the distribution of the 'new' risk factors. All infants in the Southern European-born group were usually put to sleep on the side and in the other three groups fewer than 10% were placed in the prone position, with rather more in each of these groups usually sleeping supine (Au 7.5%, As 13.9%, Hb 16.6%).

Table 2 Some characteristics of the infants by group

Aust (n = 200) SE (n 56) Asian (n - 101 ) Hb (n = 102)

Sex (male/female) 97/103 25/31 51/50 62/40 Gestation (median: weeks) 40 39.5 39 40 Birth weight (median: g) 3440 3290 3175 3480 Uses dummy (%) 52.3 45.5 28.7 15.7

A. Potter et al. / Early Human Development 45 (1996) 119-1.31

c

r E e '~

~ ~ ~

125

126 A. Potter et al. / Earl)' Human Development 45 (1996) 119-131

In all but the homebirth group, significantly fewer women were fully breast feeding at the time of the interview than had expressed the intention to do so immediately after the birth. More than a quarter of Southern European- and Asian-born women were using a combination of breast and bottle feeding (SE 26.8%, As 26.7%), but relatively few in the Australian-born and homebirth groups (Au 5.5%, Hb 2.9%). Only one infant in the homebirth group was exclusively formula fed, while this was the type of feeding used by 20.5% of Australian-born mothers, 21.4% of the Southern European-born and 26.7% of the Asian-born.

The great majority of women in all groups were feeding on demand (Au 91.5%, SE 96.4%, As 89.1%, Hb 97.0%), but there were differences in the frequency of feeding, with significantly more of the infants in the homebirth group having more than eight feedings in 24 h (Au 15.5%, SE 16.0%, As 23.8%, Hb 52.0%).

Fewer women were smoking at the time of interview than had smoked at some time during pregnancy, although when compared with the number smoking in the last trimester, there were small increases in the number of Australian-born and Southern European-born women smoking after birth. Only one Asian-born woman smoked during pregnancy (in the first trimester) and another was smoking at the time of the interview. More of the Australian-born smoked in excess of 10 cigarettes a day. Very few mothers ever smoked in the same room as the infant, and this was true also for other members of the household. Smoking by one or more adults, other than the mother, was more common in the homes of the Southern European-born group (Au 32.0%, SE 44.7%, As 23.8%, Hb 24.5%).

The proportion of infants who slept in the same room as the parents was much lower in the Australian-born group than in the other three groups (Au 58.0%, SE 87.5%, As 96.0%, Hb 94.2%). Bed sharing was recorded where an infant had regularly spent more of the night time sleep in the bed of an adult (usually a parent) than in his/her own bed. The original questionnaire did not ask this question specifically, but as it was obvious, right from the start of interviewing, that it was a common practice in families in the Asian-born and homebirth groups (in both these groups there were some infants who had no bed of their own), the interviewer added bed sharing as an option in the question that asked where the infant usually slept.

Recent interest in the risk associated with the combination of bed sharing and maternal smoking [7] prompted us to look at the presence of this combination in the study groups. Twelve women of the 91 who co-slept with their infants were current smokers. The prevalence of co-sleeping with current smoking was 1/101 (1%) of Asian-born women, 1/200 (0.5%) of women born in Australia 1/56 (2%) of women born in Southern Europe and 9/102 (9%) of the home birth group. There were only small differences in the figures for co-sleeping and smoking during pregnancy (Au 0.5% (1/200), SE 3.6% (2/56), Hb 10.8% (11/102)). The Asian-born woman who had smoked during pregnancy was not co-sleeping.

Table 4 gives the odds ratios for prone sleeping; not fully breast feeding; maternal smoking during pregnancy and at interview; smoking of other family members and infant sleeping in the parents' bed with the reference group being infants of Australian-born mothers.

A. Potter et al. / Early Human Development 45 (1996) 119-131

Table 4 Comparison of the 'new' risk factors across the four groups

127

Odds ratio [95% Confidence interval] P-value

Prone sleeping Au" 1.00 SE 0.00 [0.00-2.081 As 2.35 [0.77-7.22] Hb 2.61 [0.89-7.85]

Not fully breast fed Au ~ 1.00 SE 2.65 [1.37-5.10] <0.01 As 3.27 [1.92-5.57] <0.001 Hb 0.12 [0.03-0.33] <0.001

Smoking (pregnancy) Au ~ 1.00 SE 0.62 [0.25-1.41] As 0.03 10.00-0.20] <0.001 Hb 0.75 [0.39-1.40]

Smoking finterview) Au ~ 1.00 SE 1.71 [0.89-3.27] As 0.66 [0.37-1. l 8] Hb 0.69 [0.38-1.22]

Smoking (other family members) Au ~ 1.00 SE 1.71 [0.89-3.271 As 0.66 [0.37 - 1.181 Hb 0.69 [0.38-1.22]

Infant sleeps in parents" bed Au' 1.00 SE 2 . 2 1 10.33-11.73] As 1 3 . 5 2 [4.82-46.32] <0.00 l Hb 49.40 [18.2-164.1] <0.001

~' Reference group.

4. Discussion

The ethnic groupings in this study are the same as those used in the SIDS risk reference paper [1]. To have examined infants by mother's country of birth would have provided unreliable estimates of such a rare phenomenon as SIDS. Malaysia, Vietnam and India comprised 45% of the Asian-born sample and in fact represented 47% of the Asian-born population of women giving birth in Victoria in 1991 [9]. This distribution is markedly different from that of the UK where Asian immigrants are predominantly from India, Pakistan and Bangladesh. Although cultural beliefs and practices differ between Asian countries, the infant care practices of mothers from different parts of Asia are more similar to each other than to those of Australian-born or European-born mothers, (PL Rice, personal communication). In the S1DS risk reference paper the estimated risks were not adjusted for mother's period of residence in Australia, as this information was not recorded in the perinatal data. Because the further subdivision into length of residence groupings in this study would not have allowed meaningful comparisons, we did not collect these data.

128 A. Potter et al. / Early Hunum Development 45 (1996) 119-131

The biases that may have been introduced by the selection process are more likely to be in the direction of minimising the differences between the groups, given the relatively older age of the mothers and the high educational levels across the groups, as well as the necessity for the women in the non-English speaking background groups to have a reasonable command of spoken English. The relatively large number of Asian-born women in the study who had completed tertiary education probably reflects the fact that the recruitment hospital is close to a major university at which they or their families may have been students or staff members. The level of education in this group is not representative of the Asian population in Victoria as a whole. This tendency to be generally older, better educated and to have a reasonable grasp of English makes it more likely that the messages of the 'Reducing the Risks' campaign were understood by women across the groups, nevertheless differences in the distribution of the 'new' risk factors were found.

Most, if not all, parents who participated in the project would have been exposed to information about the risk of the prone position. Even before the study began, the research findings had begun to influence advice given to new mothers about preferred sleep position and only months later, the national campaign, 'Reducing the Risks', was launched. We were interested to find that the pattern of differences between the study groups was similar to that reported by a survey of infant sleeping practices carried out by the Australian Bureau of Statistics in July 1992 [10]. That too found Southern European-born mothers much less likely to use the prone position than either Asian-born or Australian-born mothers (SE 4.1%, As 15.8%, Au 13.2%). Over the period of recruitment, the homebirth midwives were also recommending the use of the side-lying or supine positions, but the homebirth mothers were not as likely as women in the other groups to attend maternal and child health centres, where this advice would have been reinforced. It is obvious that the 'Reducing the Risks' campaign affected the use of the prone sleep position in the study population. Beal and Finch report the prevalence of the prone sleep position in the control populations of Australian case-control studies to have been between 30-40% during the 1980s [ 11 ]. However, we do not know to what extent the use of the prone sleep position varied between the groups pre-campaign, or whether the groups were equally responsive to the campaign messages.

Almost all the infants in the homebirth group (high risk) were fully breast fed at time of interview, as distinct from only roughly half of the infants in both of the low risk groups. This is the opposite finding to that which the 'new' risk factors might have predicted.

A strong association between maternal smoking and SIDS has been established [12]. Although only two women in the Asian-born group smoked, the level of maternal smoking in the Southern European-born group (the other low risk group), was not significantly different from that of the high risk group. More of the Southern European-born families had other members who smoked, however very few families in the study admitted to smoking in the same room as the infant, making it hard to know how much passive smoking these infants were exposed to. Our smoking data may have been affected by the under-reporting that is known to be more likely when the interviewer is a health-related professional.

A. Potter et al. / Earl)' Human Development 45 (1996) 119-131 129

Sharing a bed with an infant has been claimed to be both risky [5] and beneficial [13]. More than half of the homebirth group and a quarter of the Asian-born were co-sleeping, while in the Australian-born and the Southern European-born groups there were only five and three families, respectively. Homebirth families were more likely to be co-sleeping because they thought it was 'more natural', easier to breast feed, and 'better' for the infant, whereas the Asian-born mothers were following traditional practices. Similar differences have been recorded in other studies; for example in Birmingham, in a study comparing the infant-rearing practices of white and Asian mothers. This study found three times as many Asian infants as whites slept with their parents [14]. Scragg et al., in following up the evidence that bed sharing was a risk factor for SIDS in Maoris, estimated that the combitiation of smoking and infant co-sleeping might account for 20% of deaths from SIDS in New Zealand [15]. We found this combination of factors to be more prevalant in the high risk group as compared with the low risk groups, but given the very small numbers involved, would caution about the over-interpretation of this finding.

That the regular use of a dummy might be protective against SIDS was one of the findings of the New Zealand Cot Death Study [16], The prevalence of dummy use reported in that study is considerably lower than our data would suggest for Victoria. The homebirth mothers made significantly less use of the dummy than mothers in the other groups (Table 2). This low usage may be associated with the very high rates of breast feeding of homebirth infants. Despite the similar infant feeding patterns at 1 month in the low risks groups, significantly more of the infants in the Southern European group were given dummies than in the Asian group.

It is possible that during the study period changes in infant care practices as a result of the 'Reducing the Risks' campaign contributed to the lowering of SIDS incidence from 2.13/1000 livebirths in 1989 to 1.05/1000 livebirths in 1991 and 1992 [17]. However, changes in SIDS incidence were significant only in the Australian-born group and the relative risks between the four groups remained similar over time, (during the years 1991-92 they were 1/1000 (Australian-born), 0.84/1000 (Asian- born), 0.00/1000 (Southern European-born) and 3.28/1000 (women giving birth out-of-hospital)), suggesting that factors other than the 'new' risks were influencing the differences between the groups.

5. Conclusion

Given that the ethnic and place of birth differences in the incidence of SIDS in Victoria cannot be explained by social and perinatal risk factors, and that it has been shown that the 'new' risk factors account for most of the ethnic differences in New Zealand, we were interested to see whether the same might be true here. Fewer than 10% of mothers in all groups used the prone position. The very low incidence of smoking in the Asian-born group aside, there were no significant differences between the groups in smoking during pregnancy or at interview. Our data show the reverse of what would be expected if not being fully breast fed were a risk factor. Although more than 50% of the infants in the high risk group regularly slept with their parents,

130 A. Potter et al. / Early Human Development 45 (1996) 119-131

so too did a quar te r of in fan t s in the A s i a n - b o r n group, this b e i n g l 0 t imes the rate o f

co - s l eep ing in the re fe rence group. It s eems that the ' n e w ' r i sk factors do not c lar i fy

the reasons for the d i f fe rences in Victor ia .

Acknowledgments

The au thors wish to t hank B r o n w y n Handf ie ld , Jane Yel land and the m i d w i f e r y

s ta f f at M o n a s h Med ica l Cen t r e for the i r he lp wi th r ec ru i tmen t ; P ro fesso r Ter ry

D w y e r and Dr A n n e - L o u i s e P o n s o n b y , the M e n z i e s Cen t re for P o p u l a t i o n Hea l th

Research , Un ive r s i t y o f T a s m a n i a , w h o k ind ly a l lowed us to use the T a s m a n i a n

Infant Hea l th Study pro toco ls ; the h o m e b i r t h prac t i t ioners w h o gave the i r suppor t and

all the w o m e n w h o gene rous ly a l l ow ed us to in t rude on the i r f ami ly life. The pro jec t

was funded by the Na t iona l S IDS Counc i l of Aus t r a l i a and the Vic to r i an Hea l th

P r o m o t i o n Founda t ion .

References

l l] Kilkenny, M. and Lumley, J. (1994): Ethnic differences in the incidence of the sudden infant death syndrome (SIDS) in Victoria, Australia 1985 1989. Paediatr. Perinat. Epidemiol., 8, 27-40.

[2] Lumley, J. and Sombekke, M. ( 1993): Differences in the incidence of sudden infant death syndrome by place of birth: Victoria, Australia, 1985-1987. Second SIDS International Conference. Perinatol- ogy Press.

[3] Perinatal Data Collection Unit (1994): Births in Victoria 1983-1992, Table 7.1. Perinatal Data Collection Unit, Melbourne.

[4] Dwyer, Y., Ponsonby, A.-L., Newman, N.M. and Gibbons, L.E. (1991): Prospective cohort study of prone sleeping position and sudden infant death syndrome. Lancet, 337, 1244-1247.

[5] Mitchell. E.A., Stewart, A.W., Scragg, R. et al. (1993): Ethnic differences in mortality from sudden infant death syndrome in New Zealand. Br. Med. J., 306, 13-16.

[6] Fleming, EJ., Gilbert, R., Azaz, Y. et al. ( 1990): Interaction between bedding and sleeping position in the sudden infant death syndrome: a population based case-control study. Br. Med. J., 301, 85-89.

[7] Gantley, M., Davies, D.E and Murcott, A. (1993): Sudden infant death syndrome: links with infant care practices. Br. Med. J., 306, 16-20.

[8] Dean, A.G., Dean, J.A., Burton, A.H. and Dicker, R.C. (1990): EP1 lnlo, Version 5: A Word Processing, Database, and Statistics Program for Epidemiology on Microcomputers. USD, Incorpo- rated, Stone Mountain, GA.

[9] Rice, EL., Watson, L.F. and Lumley J. (1994): Women born in Asia. Their obstetric profiles. A Victorian study. In: Asian Mothers, Australian Birth, pp. 15-30. Editor: EL. Rice. Ausmed, Melbourne.

[10] Australian Bureau of Statistics. (1993): Survey of Infant Sleeping Positions. Australia 1992, Table 5. Australian Bureau of Statistics, Canberra.

[ 11] Beal, S.M. and Finch, C.F. ( 1991 ): An overview of retrospective case-control studies investigating the relationship between prone sleeping positions and SIDS. J. Paediatr. Child Health, 27, 334-339.

[12] Nicholl, J. and O'Cathain, A. (1992): Antenatal smoking, postnatal passive smoking and the Sudden Infant Death Syndrome. In: Effects of Smoking on the Fetus, Neonate and Child, pp. 138 149. Editors: D. Poswillo and E. Alberman. Oxford University Press, London.

[13] McKenna, J.J. and Mosko, S. (1993): Evolution and infant sleep: an experimental study of infant-parent co-sleeping and its implications for SIDS. Acta Paediatr. Suppl., 389, 31-36.

A. Potter el al. / Early Human Development 45 (1996) 119-131 131

[14] Farooqi, S., Perry, l.J. and Beevers, D.G. (1993): Ethnic differences in infant-rearing practices and their possible relationship to the incidence of sudden infant death syndrome (SIDS). Paediatr. Perinat. Epidemiol., 7, 245-252.

[15] Scragg, R., Mitchell, E.A., Taylor, B.J., et al. (1993): Bed sharing, smoking and alcohol in the sudden infant death syndrome. Br. Med. J., 307, 1312-1318.

[16] Mitchell, E.A., Taylor, B.J., Ford, R.EK., el al. (1993): Dummies and the sudden infant death syndrome. Arch. Dis. Child., 68, 501-504.

[17] Consultative Council on Obstetric and Paediatric Mortality and Morbidity (1989, 1991, 1992): Annual Reports. Health Department, Victoria.