bearing blog

My initial judgment of the article that I summarized below, purporting to show a connection between cosleeping and SIDS, is that the research methods appear sound (although I want to run it by an epidemiologist friend of mine).    The conclusions that are based on the research also appear sound, with some caveats.

The FOX News story captured most of the details, much better than I usually expect from science reporting.

Worth pointing out:

• The researchers recommend that all babies sleep in the parents’ room.   If parents smoke, putting baby to sleep in a separate room increases the risk of SIDS.
• The study did NOT show that it was risky to cosleep with a baby older than 11 weeks.
• 5 of the 46 SIDS babies reported as having shared a bed in the last sleep were actually found dead in a crib or "moses basket."
• Bed sharing here is defined as having shared the bed on a single occasion—in the case of SIDS babies, during the "last sleep" before being found dead, and in the case of control babies, in the night before the interview.  Only 13 babies in the SIDS group and 13 babies in the control group regularly slept in the parental bed.  The rest—110 babies in the SIDS group and 240 babies in the control group—regularly slept somewhere else.

I think the most important point for new parents is the last.  When  "continuum-concept" adherents or "attachment-parenting"  proponents (the best-known of which is Dr. William Sears)  speak of co-sleeping, we are referring to  a baby’s sleeping every night in the parental bed, with parents, from birth. 

The question "Should we co-sleep, or not?" means, for most of us, "Should my baby always sleep in my bed, or should he never sleep in my bed?"  It does not mean, "Should my baby sleep in my bed tonight?"   

Yet the latter is really the question that the study purports to answer.

Those of us who regularly co-sleep are really only interested in twenty-six of the studied babies, thirteen who died and thirteen who did not; or, perhaps, thirty-nine of the studied babies, thirteen who died and their twenty-six associated controls.   Is this a large enough sample to evaluate the risk of regularly co-sleeping?  Were there confounding factors?

One missing piece of information is necessary to gain perspective.  This site claims that the rate of SIDS in Scotland is about 0.7 per 1000 births (70 per 100,000).  52,537 babies were born in Scotland in 2001.  The 123 SIDS deaths were spread out over 53 months. Let us suppose for the sake of argument that the percentages shown by the study are accurate measures of the population, that the control group is representative of the whole population with respect to sleeping practices, and that the birth rate was flat at 52,537/year (big assumptions—this is just for the sake of argument).  Then, for every 100,000 babies born in Scotland:

– 4,943 reportedly regularly sleep with parents.  95,057 don’t.

– 70 die of SIDS.  99,930 don’t.

– 7 babies (70 x percent of all SIDS babies who sleep with parents) regularly sleep with their parents and die of SIDS.  4,936 babies regularly sleep with their parents and don’t die of SIDS.

– 63 babies do not regularly sleep with their parents and die of SIDS.  94,994 babies do not regularly sleep with their parents and do not die of SIDS. 

– Chance of dying from SIDS if baby regularly sleeps with parents:  7/4,943=0.14% (14 in 10,000).  Chance of dying from SIDS if baby does not regularly sleep with parents:  63/99,930=0.063% (6 in 10,000).   

Perspective:  Odds that the baby will not die of SIDS if you co-sleep regularly:  99.86%.  Putting the baby in a crib every night improves those odds to 99.94%.   

We co-sleepers do it because we perceive benefits:  more sleep, easier breastfeeding, and a closer relationship with our kids.  Is the slight improvement in the odds worth the loss of those benefits?  It’s a question that the report did not cover.

Let’s extrapolate more wildly from the tiny data set of regular cosleepers, just to see what happens on a per-night basis.   If the extrapolation were, by some amazing coincidence, representative of the population, this would allow us to answer separately questions relevant to "Should I take my baby into my bed tonight?" and "Should I always cosleep or should I never cosleep?"

– On a given night, 19,011 babies per 100,000 births are in bed with their parents.  Assume that all those who normally sleep with their parents are among them. (This may not be valid.)  Then, at most, on a given night, 14,068 babies who do not normally sleep with their parents are in bed with them. 

Out of 100,000 births, then:  26 babies are in their parents’ bed at some time during a sleep session in which they die.  (Only 23 of them actually die in the bed.)  It is reasonable to suppose that 7 of these are the ones that died who regularly slept in the parents’ bed.  So 19 babies who do not regularly sleep in their parents’ bed die during a sleep session that takes place at least partly in their parents’ bed. 

123 deaths over 1590 days:  0.08 SIDS deaths on a given night over the course of the study in Scotland.  If it is true that the SIDS rate is 70/100,000 = 123/175,714 over the course of the study, then on a given night in Scotland,   0.08/175,714 babies die, or 46 per hundred million (not that there are that many babies in Scotland). 

SO.  Out of a theoretical one hundred million babies in Scotland:

On a given night, 14,068,000 who don’t usually sleep with their parents are in bed with their parents.  4,943,000 are in bed with their parents as usual.  80,988,000 are, as usual, not in bed with their parents.  On that night, 46 babies die of SIDS.  Twelve were babies who don’t usually sleep with their parents, but that night, were in bed with their parents.  Five were babies who were in bed with their parents as usual.  The other 29 were not in bed with their parents. 

Risk of a baby who always sleeps with you dying of SIDS tonight:  10 in 10 million.

Risk of a baby who doesn’t usually sleep with you dying of SIDS tonight if you don’t put him in your bed:  4 in 10 million. 

Risk of a baby who doesn’t usually sleep with you dying of SIDS tonight if you do put him in your bed:  9 in 10 million.

So yes, the very tiny risk of death is highest for regular co-sleepers and lowest for regularly separate sleepers—though the data set is perhaps too small to make this conclusion.

Last week I blogged about a new study purporting to show that there is a positive correlation between bed-sharing and SIDS.  I promised to post more after I had a chance to read the article (Tappin et al., J Pediatrics 147, 32-37 (July 2005)).

Here’s my executive summary of the article.

Over about four and a half years, the researchers obtained reports of all sudden unexplained infant deaths, implied to be throughout Scotland.  They then attempted to interview the parents within 28 days.  They eliminated from the group any infant for whom a cause of death could be identified and any infant whose parents they were unable to interview.  That left 123 infants who died of SIDS in Scotland and whose parents were interviewed afterward. 

The control group was obtained by selecting, for each infant in the SIDS group, the baby born immediately after and the baby born immediately before that baby in the same maternity ward.  They attempted to interview each set of parents in the control group within 28 days of the associated "SIDS baby’s" death.  Those that were interviewed successfully numbered 263. 

For the SIDS group "last sleep" in the questionnaire means the last sleep period before being found dead.  For the control group "last sleep" means "last night."

The questionnaire asked about socioeconomic status, smoking, "In which room was your baby for most of the last sleep?", and "What was your baby’s normal sleep place (day and night)?" If the baby shared a sleep surface during the last sleep, the parents were asked whether it was on a bed, couch, or chair; for how long; with whom; how close to others in the bed; and whether between people or between a person and an edge of the surface.  They did not ask about alcohol because they did not have confidence that people would answer accurately.

Here are the reported data:

Normally slept in parental bed:  11% (13/123) of SIDS babies, 5% (13/263) of controls.

Shared surface during last sleep:  52% (64/123) of SIDS babies, 20% (53/263) of controls.

    – Shared a couch:   11% (14/123) of SIDS babies, 1% (2/263) of controls.
    – Shared a chair:    2% (2/123) of SIDS babies, no controls.
    – Shared a bed:   37% (46/123) of SIDS babies,  19% (50/263) of controls.

(N.B.  Of those 46 SIDS babies who shared a bed "during" their last sleep, 5 were recorded as found dead somewhere else:  2 in a crib/cot, 3 in a Moses basket.  So:

    – Were sharing a bed at time of death:  33% (41/123) of SIDS babies.)

    – Sharing a crib with a twin:  2% (2/123) of SIDS babies, 1% (1/263) of controls.

In the same room as parents, not sharing a sleep surface:  36% (44/123) of SIDS babies, 63% (167/263) of controls. 

In a separate room from parents, not sharing a sleep surface:  12% (15/123) of SIDS babies, 19% (43/263) of controls.

Another detail:  Of the 46 SIDS babies who shared a bed during their last sleep, 30 were not breastfed or had been weaned from the breast by the time of the last sleep. This includes all the infants who actually died in the bed.  All of the 16 breast-fed, co-sleeping SIDS babies were under 11 weeks of age.  The number of control infants who were not breastfeeding is not reported explicitly in the article.  Nor is the number of SIDS babies who didn’t bed share and were not breastfeeding.

After doing the statistical analysis, the researchers arrived at the following results:

– Putting the baby to sleep in a separate room *is* measurably riskier than putting the baby to sleep in the same room *if* the parents are smokers.  It is not measurably riskier if the parents are nonsmokers.

– Sharing a bed with parents at some time during the night, especially between the two parents, is measurably riskier than not sharing a bed when the child is younger than 11 weeks.

– The risk is still measurable even if the young (<11 weeks) infant is breastfed and even if the young infant’s mother does not smoke. 

– Sharing a bed with parents at some time during the night is NOT measurably riskier than not sharing a bed when the child is older than 11 weeks. 

  But because of the "wide confidence intervals," the researchers decline to conclude that it is safe for a child older than 11 weeks to co-sleep.  I cannot evaluate the statistics well enough to comment on this, for example, the width of the confidence intervals here compared to the width of those supporting their other conclusions.

– Sharing a couch is measurably "associated with an increased risk."

– Socioeconomic status has no effect except insofar as it affects smoking, birth weight, maternal age, and sleep location.

– The strong effect of sleeping between two parents is possibly due to the known effect of sleeping in proximity to duvets and pillows.  A previous study showed that 81% of infants bedsharing at death were covered by a duvet.

– Why was no connection found before?  If there is no connection between bedsharing and SIDS for infants older than 11 weeks, then lumping all ages together as was done in some other studies might have obscured a connection between bedsharing and SIDS for infants younger than 11 weeks.  Also, recently more people have been bedsharing, so the population available for study is greater and thus better able to demonstrate a connection.

The researchers recommend that all babies sleep in a crib in the parents’ room.