Abstract
Objective. Prone sleeping is a strong risk factor for sudden infant death syndrome (SIDS). We investigated whether the combined effect of prone sleeping position and prenatal risk factors further increased the SIDS risk.
Methods. In the Nordic Epidemiological SIDS Study, parents of SIDS victims in Denmark, Norway, and Sweden completed a questionnaire on potential risk factors for SIDS. Forensic pathologists verified the SIDS diagnosis. Four controls of the same gender, age, and place of birth were selected. This matched case–control study, which included 244 SIDS cases and 869 controls from 1992 to 1995, was analyzed by conditional logistic regression.
Results. Odds ratios (ORs) for prone and side sleeping compared with supine sleeping for the last sleep were 13.9 (95% confidence interval 8.2–24) and 3.5 (2.1–5.7). Infants 13 to 24 weeks old had particularly high risk in prone and side sleeping, at 28.5 (7.9–107) and 5.9 (1.6–22). OR for prone sleeping was higher in girls, at 30.4 (11–88), than in boys, 10.3 (5.5–19). We found strong combined effects of sleeping position and prenatal risk factors (more than multiplicative). The OR for prone and side sleeping was increased for infants with birth weight <2500 g, at 83 (25–276) and 36.6 (13–107); for preterm infants, at 48.8 (19–128) and 40.5 (14–115); and for intrauterine growth retarded, at 38.8 (14–108) and 9.6 (4.3–22), compared with supine position in infants without these prenatal factors. The combined effect of nonsupine positions and intrauterine growth retarded was highest among 13- to 24-week-old infants. Effects of combined presence of nonsupine sleeping positions and each of the factors of smoking in pregnancy, young maternal age, higher parity, low level of maternal education, and single motherhood were more than additive. Attributable fractions in the population for prone and side sleeping were 18.5% and 26.0%.
Conclusions. Both prone and side sleeping increased the risk of SIDS. The risk was increased further in low birth weight infants, preterm infants, and infants at the age of 13 to 24 weeks, suggesting that SIDS may be triggered by nonsupine sleeping in infants with prenatal risk factors during a vulnerable period of postnatal development.
- birth weight
- fetal growth retardation
- gestational age
- infant mortality
- infant
- premature
- prone position
- sudden infant death
The Nordic countries experienced a dramatic increase in sudden infant death syndrome (SIDS) during the 1980s, concurrent with an increase in total postperinatal deaths.1-3 The highest incidence was observed in Norway in 1988, with 2.69 SIDS per 1000 infants surviving the first week of life.3-5 At that time, knowledge of modifiable risk factors was limited, and in 1990 a multicenter case–control study was initiated by SIDS investigators in Denmark, Norway, and Sweden in cooperation with parental SIDS organizations. However, reports from the Netherlands, England, New Zealand, and Norway strongly suggested that prone sleeping position was associated with SIDS,6-9 and recommendations to avoid prone sleeping were given in Norway in 1990,9 in Denmark in 1991,5 and in Sweden in 1992.10 In Norway, these recommendations were followed by an abrupt drop in the prevalence of prone sleeping and in the SIDS rate.11 Rates were also reduced in Denmark5 and Sweden.
Still, the role of sleeping position in the etiology of SIDS is far from clarified. Studies support the hypothesis that SIDS infants in general have experienced suboptimal intrauterine conditions.12-18 SIDS deaths peak at ∼3 months of age, and most deaths occur within the first 6 months of life.19SIDS is associated with postnatal events before death, such as thermal stress, infection, and prone sleeping.6-720-21These findings have led to the triple risk hypothesis, ie, that a sudden infant death may occur when a predisposed infant, in an unstable period of homeostatic control, is exposed to triggering factors.22
In the present article from the Nordic Epidemiological SIDS Study, we investigated the combined effects of prone sleeping position and prenatal risk factors on the risk of SIDS.
METHODS
A consecutive multicenter case–control study was conducted from September 1, 1992 to August 31, 1995, with cases recruited among all infants dying suddenly and unexpectedly between 7 and 364 days in Denmark, Norway, and Sweden. Forensic pathologists verified the SIDS diagnosis according to standardized criteria of the Forensic Group of the Nordic SIDS Study,23 based on Beckwith's definition of 1970: “The sudden death of any infant or young child, which is unexpected by history, and in which a thorough postmortem examination fails to demonstrate an adequate cause of death.”24 For each case, regional coordinators selected six live control infants of the same gender and born at the same maternity ward 14 days after the index case. A delay period of 14 days was expected to ensure that controls had an age similar to the SIDS baby when parents completed the questionnaire. Parents of the first four of these selected infants were invited to participate. If a family was unwilling to participate, another family of one of the two remaining infants was invited. Some coordinators initially invited the parents of all six infants.
Parents of SIDS victims and control infants were contacted by the regional project coordinators shortly after the death of the infant and asked to complete a questionnaire. They were asked about the usual sleeping position immediately after homecoming from the maternity hospital the last week before death or interview as well as about the last time the infant was placed to sleep. Altogether, the parents were asked 272 questions, including those about anthropometric measures; complications in pregnancy and labor; smoking before, during, and after pregnancy; and postnatal factors such as clothing, bedding, and infection. Information on parents' age, level of education, and social situation was also collected. The study was approved by the regional ethics committees in all three countries.
Differences in infants' age distribution were evaluated with the Mann–Whitney test.25 SIDS risk among exposed infants was compared with SIDS risk among unexposed infants and estimated with odds ratio (OR) with 95% confidence interval (CI) in a matched case–control analysis by a conditional logistic regression analysis.26 We used stratified Cox regression models to fit the conditional logistic regression model, because these two methods give the same likelihood functions and thus the same estimates.27 In the Cox regression analyses, survival time was set to 1 for all cases and controls, the case was the event, and each case–control unit was the separate strata in the analysis. The statistical package SPSS for Windows 6.0 (SPSS Inc, Chicago, IL) was used.28
First, we analyzed the effect of prone and side positions within each stratum of the matching variables. Second, the combined effect between sleeping positions and prenatal factors was explored in the following way.26 Three categories of sleeping position and two categories of a prenatal factor were combined into a new variable with six categories: prone with a prenatal risk factor, side with a prenatal risk factor, supine with a prenatal risk factor, prone without a prenatal risk factor, side without a prenatal risk factor, and supine without a prenatal risk factor. All were compared with the sixth category. OR for these cross-classified variables was estimated using the conditional logistic regression model and reported with 95% CI. Cross-classified variables of position and birth weight were also analyzed within three strata of infants' age. Combined effects of sleeping position and maternal characteristics, such as maternal age, birth order, education level, and marital status, were also explored. We tested whether the combined effects of two factors exceeded additivity,26,29 using conditional logistic regression with additive relative risk models available in EGRET (Statistics and Epidemiology Research Cooperation, Seattle, WA),30 because the standard conditional logistic regression model implies multiplicative effect between two factors.26
Third, potential confounders of the association between sleeping position and SIDS were evaluated in the conditional logistic regression analysis. Finally, attributable fraction in the population of the prone and side sleeping positions was estimated with the formula, attributable fraction = % exposed among cases × (OR − 1)/OR.31 We assumed that nonsupine sleeping positions were causally related to SIDS and that the distribution of sleeping position among the controls reflected the distribution in the normal infant population.
The following variables were analyzed: position placed to sleep immediately after homecoming from the maternity hospital; usual position placed to sleep during the last week; position placed to sleep for the last time; position when infant was found dead for case subjects or after the last sleep for controls; country; year of birth; gender; age; birth weight; gestational age (<37 weeks, ≥37 weeks); preeclampsia; incidence of hypertension and urinary tract infection in pregnancy; illness during the first week of life (listlessness, irritability, screaming, poor sucking, excessive vomiting); maternal smoking in pregnancy; maternal age; birth order; maternal education; and marital status.
The effect of birth weight was evaluated in two ways. First, birth weight was categorized as <2500 g and ≥2500 g. However, the cut-off point of 2500 g may not be very useful to identify intrauterine growth retarded (IUGR) infants.32 In general, birth weight is not normally distributed and may be described as two underlying distributions: a predominant (normal) distribution and a residual distribution consisting of primarily preterm births.33 In previous studies, we have shown that SIDS infants have a lower mean birth weight than other infants, because their birth weight distribution is shifted downward and also contains a higher proportion of preterm births.18 In the present study, we compared the observed birth weight distributions for SIDS infants and controls with the estimated birth weight distributions for SIDS infants and controls, assuming that each distribution contained two underlying distributions (predominant and residual distribution), using the computer software Peak Fit (Jandel Scientific, San Rafael, CA).34 Therefore, IUGR infants were identified as term births with birth weight below the 10th percentile, calculated from the controls' birth weight distribution.
RESULTS
According to official statistics, 380 sudden and unexplained infant deaths occurred in Denmark (N = 90), Norway (N = 113), and Sweden (N = 177) during the study period. Among these, regional coordinators were able to contact the parents of 306 infants, 268 (87.6%) of whom responded. Among the 268 infants, 22 cases without controls and 2 cases later classified as non-SIDS by pathologists were excluded, leaving 244 SIDS cases for the study. Parents of 1292 matched controls were invited, 1004 (77.7%) of whom responded. Among the 1004 infants, 24 controls with a nonresponding case, 105 controls without a case, and 6 controls with a matching case reclassified as non-SIDS were excluded, leaving 869 controls for the study. The age-at-death distribution for SIDS and the age distribution for controls had similar shapes, but the controls were older (P < .01). The distributions were unimodal and skewed to the right, and SIDS deaths showed a peak at 9 to 12 weeks, whereas the peak age of controls was 12 to 15 weeks. The mean ages for SIDS infants and controls were 16.1 and 21.4 weeks, and the median ages 13.0 and 18.0 weeks.
“Sleeping position immediately after homecoming,” “usual sleeping position,” and “sleeping position the last time” demonstrated that all nonsupine positions were associated with increased risk of SIDS (Table 1). Compared with that for supine sleeping, the ORs for placing the infant either prone or on the side for the last sleep were 13.9 (95% CI 8.2–24) and 3.5 (2.1–5.7). The ORs for found prone and found on the side compared with found supine were 14.1 (8.8–23) and 2.0 (1.2–3.6). For cases placed to sleep in the side position, only 34.2% were found on the side, whereas 41.1% were found prone and 24.7% were found supine. For controls placed in the side position, 46.9% were found on the side, 6.8% were found prone, and 46.3% were found on the back.
Sleeping Positions and the Risk of SIDS
Combining data on “usual sleeping position” and “sleeping position the last time,” usual prone sleeping was associated with increased SIDS risk only for infants also placed prone the last time (Table 2). Among the 211 controls usually placed prone, 72% were placed prone the last time, and among 124 SIDS infants usually placed prone, 90% were placed prone the last time. Additionally, prone the last time increased the risk even more when the usual position had been nonprone. Similar results were observed for position placed after homecoming from the maternity hospital and position placed the last time (Table 2). In subsequent analyses, we therefore studied the position placed to sleep the last time.
Risk of SIDS by Sleeping Position Placed Usually and the Last Time, and by Sleeping Position Placed Immediately After Homecoming From the Maternity Hospital and the Last Time
The effect of sleeping position the last time was stratified by the matching variables country, year of birth, and gender and also by infant's age divided into tertiles (Table3). Denmark had the highest OR for prone sleeping, whereas the effects were lower and similar in Norway and Sweden. Throughout the study period, the SIDS risk in prone and side positions increased. The risk in prone position was higher for girls than for boys, a consistent finding in the three countries. Among controls, more boys were placed prone than girls, 23.4% and 13.4%, respectively (P < .01), whereas among SIDS infants, 56.1% boys and 51.8% girls slept prone before death (P = .68). The risks in prone or side sleeping were highest when the infant was 13 to 24 weeks old, whereas the risk was intermediate during the first 12 weeks, and lowest after 6 months of age.
The Effect of Sleeping Position Placed to Sleep the Last Time in SIDS, Stratified by Each Matching Variable, Country, Year of Birth, and Gender, and by Infant's Age
The birth weight distribution for SIDS infants was both shifted downward and contained a higher proportion of births in the lower tail compared with the controls (Figure, part A). The observed mean birth weight for SIDS infants was 3173 g (standard deviation [SD] 780) compared with 3599 g (SD 535) for controls. The observed mean birth weights for SIDS babies and controls when excluding preterm births were 3428 g (SD 562) and 3641 g (SD 489), and were similar to the estimated mean birth weights in the predominant distributions of SIDS infants and controls (Fig 1, B) at 3423 g (SD 636) and 3630 g (SD 465), respectively. The estimated residual distributions, primarily corresponding to preterm births, were 12.5% for SIDS infants and 3.0% for controls. The higher proportion of residual births (preterm births) among SIDS infants only partially explained the lower observed mean birth weight and the broader SD of SIDS infants' birth weights compared with that of controls, meaning that SIDS babies also were IUGR. To quantify the risk of SIDS associated with IUGR, term births below the 10th percentile (2967 g) were selected.
A, Observed birth weight distributions for SIDS infants and controls (smoothed). From the Nordic Epidemiological SIDS Study, 1992–1995. B, Estimated birth weight distributions assuming two underlying distributions for both SIDS infants [predominant (a) and residual (b)] and controls.
Prenatal and maternal risk factors for SIDS (Table4), such as low birth weight, preterm birth, IUGR, preeclampsia, illness during the first week of life, smoking in pregnancy, young maternal age, higher birth order, lower level of maternal education, and single motherhood, were all significantly associated with an increased risk of SIDS, whereas hypertension and urinary tract infection during pregnancy were not, although the ORs also were elevated for these conditions.
Prenatal and Maternal Risk Factors for SIDS
We then examined the effect of the combined presence of nonsupine sleeping and a prenatal risk factor (Table5). Compared with supine sleeping and birth weight ≥2500 g, the OR for supine sleeping and birth weight <2500 g was 2.7 (1.6 to 4.5), the OR for prone sleeping and birth weight ≥2500 g was 11.6 (6.8 to 20), and the OR for prone sleeping and birth weight <2500 g was 83.2 (25 to 276). The OR resulting from a combination of exposure to prone sleeping and birth weight <2500g was far greater than the sum of the risk from each exposure alone. Likewise, side sleeping and birth weight <2500 g increased the SIDS risk dramatically. Also, the combined effect of nonsupine sleeping and the other prenatal or maternal factors, ie, preterm birth, IUGR, any sign of illness the first week of life, maternal smoking during pregnancy, young age of mother, lower level of maternal education, higher birth order, and single motherhood, carried very high SIDS risks, whereas preeclampsia, hypertension, and urinary tract infection did not increase the risk.
Combined Effect of Position Placed to Sleep the Last Time and Prenatal and Maternal Risk Factors, and the Risk of SIDS
When combining the effect of nonsupine position and birth weight below the 10th percentile within strata of infants' age, we found that the SIDS risk was most pronounced among 13- to 24-week-old infants (Table6).
SIDS: Combined Effect of Position Placed to Sleep the Last Time and Birth Weight Within Three Categories of Infant's Age
The combined effects of prone or side sleeping position and the prenatal or maternal risk factors were unchanged after adjustment for each of the potential confounders such as low birth weight, preterm birth, maternal smoking, maternal age, birth order, and maternal education.
Finally, we calculated that 73.7% (50.3 + 23.4) of all SIDS deaths could have been avoided if all infants had slept in the supine position. Elimination of maternal smoking in pregnancy alone could have reduced the number of SIDS deaths by 46.7%. Had preterm birth and low birth weight been eliminated, only 15.5% and 16.2% of SIDS deaths could have been avoided.
DISCUSSION
We confirm that the prone sleeping position is a strong and independent risk factor for SIDS. We also show that the side sleeping position is less safe than the supine position. The combination of nonsupine sleeping and low birth weight, premature birth, or IUGR was associated with a particularly high risk of death. At between 13 and 24 weeks of age, the combined effect of nonsupine sleeping and lower birth weight carried the highest risk for SIDS.
The current study was carried out after strong warnings against the prone position were issued to the public in Norway in January 1990,9 Denmark in December 1991 and 1994,5and Sweden during spring 1992 and spring 1994.10 The prevalence of prone sleeping has been recorded in various surveys including the Nordic Epidemiological SIDS Study. In Norway, prone sleeping was reduced from 49% in 1989 to 28% in 199111and to 3.8% in 1994 to 1995 among controls in the present study; in Denmark, prevalences were 52% in 1991,35 between 4.1% and 10%36 in 1992, and 3.3% in 1994 to 1995; and in Sweden, 42% in 1991 to 1992, 34% in 1993, and between 17%37 and 19% in 1994 to 1995. Corresponding to the surveys of infant sleeping position, SIDS rates in Norway were 2.4 per 1000 in 1989, 1.3 in 1991, and 0.6 in 1995; in Denmark, 1.6 in 1990, 1.2 in 1992, and 0.2 in 1995; and in Sweden, 1.1 in 1991, 0.7 in 1993, and 0.4 in 1995 (1995 rates are preliminary in Denmark and Sweden).38-40
Prone sleeping position is an established risk factor for SIDS, with reported relative risks ranging from 3.5 to 9.6 for last time sleep.7-841-44 In the present study, we found an even stronger association between prone sleeping and SIDS; the OR was 13.9. One explanation for the high OR of prone sleeping in the Nordic study could be that the group of parents who failed to follow the recommendations had additional behavioral risk factors for SIDS as well.
Although parents of case subjects may report differently from parents of control subjects, differential recall has been shown to have little impact on the validity of the study results,45-46particularly if the prevalence of exposure is frequent45 or the time between exposure and interview is short,47 as in our study.
An increased risk of SIDS for sleeping in a side position has been reported previously from New Zealand and England, with an OR of 2.0.48,42 It is possible that both parents and health professionals have been hesitant to place babies on their back and have chosen the intermediate side position instead. Among the 309 controls placed on their side the last time, less than half were found sidewise after sleep, confirming previous reports that the side position is unstable during infancy.6,43,49-51 The finding that SIDS babies were more often found prone than were the control subjects after being placed on the side for sleep suggests that ending up in a prone position may represent the accident leading to SIDS. The first interventions did not stress that parents should place the dependent arm forward to prevent the baby from rolling over. Devices to maintain the side position have never been recommended and are probably rarely used.
In the present study, being placed in the prone position for the last sleep was associated with the highest OR, regardless of whether the usual position had been prone or nonprone, whereas being placed usually prone was not associated with increased SIDS risk for infants placed nonprone the last time. The effect of being placed usually prone merely reflects that most of these infants also slept prone the last time. The infants who were usually placed nonprone, but had been placed prone the last time, carried the highest SIDS risk. Infectious symptoms were not more common among these “turned around” infants. One may speculate that these infants may have slept restlessly in the usually nonprone position before the night they died and died in deep sleep after the parents turned them around to the belly.
Prone sleeping may be dangerous because infants have more episodes of quiet sleep and sleep for longer periods with fewer arousals.52,53 Infants placed prone have a decreased ability to lose heat,53-55 which may be enhanced by illness, excessive wrapping, fever, or a heated room. If the infant is sleeping in a hyperthermic situation, the response to a hypercarbic or hypoxic episode may be altered.56 The baby susceptible to SIDS may respond inadequately to a critical, although normal situation. Such arousal deficits or physiologic aberrations may be acquired in utero and enhanced at a certain postnatal age.56,57
Infants may be particularly vulnerable to the prone sleeping position at a certain age. Studies performed in The Netherlands and England have reported respectively the highest risk during 17 to 20 weeks41 and in ≥10 weeks, 7 whereas in Tasmania, infants <11 weeks of age were most at risk.20 In the present study, the highest risk was observed among infants 13 to 24 weeks old. Interestingly, this vulnerable period was observed several weeks after the peak age of SIDS occurrence, 9 to 12 weeks. We suggest three possible explanations. First, controls were older than cases, because local coordinators spent more time recruiting controls than anticipated. However, this age difference may not explain the high risk at 13 to 24 weeks, because the percentage of prone sleeping among controls was similar in the young and the old age groups. Second, infants at 13 to 24 weeks are more mobile than young infants, but may lack the motor ability to get themselves out of a dangerous position, such as face-down on soft bedding or out of a position with the head covered by bedding.42-43 However,in the present study, more SIDS infants at ≥25 weeks were entangled or covered, and no age difference was found in controls. Third, infants may have a vulnerable period in maturity of their respiratory control, sleep–wake development, and thermoregulation,57-59 which may correspond to an increased risk in the prone position. Preterm babies probably have a delayed vulnerable period, because preterm SIDS babies die later than term babies,60 also confirmed in our study. The mean ages for preterm and term SIDS babies were 19.6 and 15.0 weeks, respectively (P < .01).
Suboptimal intrauterine conditions may strongly impact postnatal development,61,62 and disturbances in postnatal maturation may be related to a vulnerability for SIDS when positioned prone.62 In the present study, we found strong combined effects of nonsupine sleeping and either birth weight <2500 g, preterm birth, IUGR, maternal smoking in pregnancy, or any illness during the first week of life. Additionally, we demonstrated that infants with birth weight less than the 10th percentile who were placed in the prone or side sleeping position had the highest SIDS risk at 13 to 24 weeks. Although the numbers were small, these finding may support the triple risk hypothesis, namely, that a sudden infant death may occur when a susceptible infant, at a critical developmental stage of homeostatic control, is exposed to exogenous triggers (such as prone sleeping).22
The risk of SIDS in the prone position was higher for girls than for boys. Among the controls, 23.4% of the boys and 13.4% of the girls were placed prone. This gender difference is consistent with surveys from Holland63 and Sweden.37 In contrast, no difference between the sexes was found for the SIDS babies. We speculate whether infant girls may be more susceptible to the prone position compared with infant boys, a hypothesis that warrants future sex-specific physiologic studies on infants. Alternatively, care takers may respond differently to girls and boys, reflected by the choice of sleeping position. Symptoms before death, possibly associated with SIDS, may be expressed differently in girls and boys.
The effects of prone and side sleeping were stronger in infants of mothers <25 years of age, mothers with <10 years of education, single mothers, or mothers with more than one child. Except for mothers with more than one child, these mothers did not place their baby prone more frequently. The joint effect of maternal risk factors and nonsupine positions probably reflects underlying risk factors associated with maternal risk factors such as infant care practice. Conceivably, mothers with more than one child have placed the subsequent infants in the same sleeping position as they did before the campaign.
The present study may be subject to nonresponder bias. Among those invited, 12% of the SIDS parents and 28% of the control parents did not participate. We did not have information on the nonresponders. Therefore, we compared demographic characteristics of our SIDS cases in Norway with the complete SIDS data base in the population-based Medical Birth Registry of Norway,4,38 and demographics of our controls in Norway with the total population of live births at risk. The percentages of young mothers, first births, single mothers, preterm, and birth weight <2500 g were not significantly different between our SIDS cases and cases in the SIDS database of the Medical Birth Registry of Norway. In comparing our controls with the population at risk, only the percentage of first births differed. There were slightly fewer first births among controls compared with the population at risk. Similar comparisons are not yet available for Denmark and Sweden. However, the demographics were not significantly different in SIDS cases among the three countries or in controls between countries. We believe that nonresponder bias was not a problem in our study. In certain regions, it was difficult to recruit local project coordinators. The 74 SIDS parents who we were unable to contact resided primarily in the middle region of Norway and the southern region of Sweden. These geographic regions do not experience excess infant mortality.64
In conclusion, the prone sleeping position is dangerous for infants. In addition, the side position carries a higher risk for SIDS than the supine position, probably because of its instability. Side sleeping is common, which means that avoidance may prevent future SIDS. Nonsupine sleeping is particularly dangerous to the low birth weight and premature infant as well as to the 13- to 24-week-old infant. If premature infants are placed in a nonsupine position in neonatal units, they should be turned to the supine position several days before discharge to reassure the parents before taking the baby home. The high risk of SIDS associated with prone and side sleeping positions suggests a need to monitor sleeping position in the infant population so that proper actions can be taken to ensure supine sleeping.
ACKNOWLEDGMENTS
This research was funded by the Nordic Council, Københaven, Denmark; the Norwegian Research Council, Oslo, Norway; the Norwegian Parental SIDS Organization, Oslo, Norway; and the University of Bergen, Bergen, Norway.
We thank the project coordinator P Schreuder, data consultant S Nilssen, the staff at the Medical Birth Registry of Norway, and the regional coordinators in Denmark, including The Danish midwives in cooperation with K. Helweg-Larsen (København), K. Kock (Odense), M. Gregersen (Århus); in Norway, including L. Stoltenberg, R. Lindemann, J. Grøgaard, A. Whitelaw (Oslo); A. Ernø (Nordbyhagen); S.-H. Anderssen (Fredrikstad); R. Palat (Elverum); H. Farstad, P. Christensen (Lillehammer); H. Bævre, K. Iversen (Gjøvik); A.W. Søyland, K. Brekke (Drammen); R. Solberg (Tønsberg); S. Slinde (Porsgrunn); Å. Lærdal (Stavanger); G. Stangeland (Kristiansand S); P. Scheel (Arendal); T. Markestad, B. Skadberg (Bergen); K. Sydnes (Haugesund); H. Thomassen (Ålesund); A. Alme (Volda); A.-M. Edsberg (Kristiansund N); O.S. Haga (Nordfjordeid); J. Moldestad (Førde); S. Slørdahl (Trondheim); S. Børsting (Levanger); J. Holt (Bodø); C. Møller (Tromsø); H. Dramsdahl (Harstad); H.P. Fundingsrud (Hammerfest); and in Sweden, including B. Ericsson (Danderyd); J. Milerad (Stockholm); T. Norsted (Uppsala); N. Skanke (Eskilstuna); I. Leijon (Linköping); A. Minkova (Norrköping); N.O. Jonsson (Jönköping); B. Larsson (Växjo); M. Aldman (Västervik); I. Fosdal (Karlskrona); J. Landehag (Karlskrona); B. Selander (Kristianstad); T. Forsberg (Ängelsholm); N.W. Svenningsen (Lund); P. Henriksson (Helsingborg); J. Ottosson (Malmö); D. Andersson (Halmstad); R. Olegård (Mölndal);Ö.M. Lichtenstein (Uddevalla); B. Alm (Göteborg); G. Norvenius (Hisings Backa); G. Wennergren (Göteborg); L. Hammaren (Borås); L. Inganäs (Trollhättan); U. Selstam (Trollhättan); K. Rex (Skövde); C. Lindgren (Karlstad); L. Ekholm (Örebro); Ö.B. Malmström (Västerås); B. Eckerberg (Falun); K. Hedberg (Gävle); R. Sidenvall (Hudiksvall); S. Mjönes (Sundsvall); I. Axelsson (Östersund); S. Håkansson (Umeå); M. Jurvanen (Boden); C. Nilsson (Gällivare); A. Lind (Kalmar).
We thank Drs Robert Lipton, Grethe S. Tell, and Stein E. Vollset for valuable comments on previous versions of the manuscript.
Footnotes
- Received November 7, 1996.
- Accepted February 17, 1997.
Reprint requests to (N.Ø.) Medical Birth Registry of Norway, University of Bergen, Armauer Hansen Building, N-5021 Bergen, Norway.
- SIDS =
- sudden infant death syndrome •
- OR =
- odds ratio •
- CI =
- confidence interval •
- IUGR =
- intrauterine growth retarded/retardation •
- SD =
- standard deviation
REFERENCES
- Copyright © 1997 American Academy of Pediatrics