Having just completed a course in statistics last week for my graduate work,
I thought it would be interesting to look critically at the AAP article on
Suffocation Deaths. I would like to share some interesting observation that
my still limited knowledge revealed:

1. We can all agree that the safety of the sleeping environment is
important. This article does a good job of highlighting some of the most
dangerous practices, such as bedding involving plastics or proximity to
cords. The category of greatest concern to me is the overlain category
since it feeds the fears of families who would want to bring their baby into
bed. This is the category that I am focusing on in order to refute the
conclusion that co-sleeping is unsafe.  (While the article only mentions bed
sharing, most of us will interpret that as co-sleeping.)

2. This experiment is a quasi-experiment; it does not have independent or
dependent variables that are controlled or manipulated.
It is also non-parametric.  It does not have to
meet the assumptions of normality, random sampling or homogeneity of
variance, (which are considered to be scientific prerequisites for the
validity of the analysis). A quasi-experiment simply observes a naturally
occurring phenomenon and measures it.

3. The only statistical analysis in this study is the Chi-square test that
analyzes the
"relationship among products, patterns and age groups." In this case, I
believe that the
study is measuring the increase in occurrence across 3 time frames.
It does not give us any information about the significance of the
occurrences.  The Chi-square test adds only an illusion of credibility to
this study, which is actually not a statistical analysis at all.  This does
not support a conclusion about co-sleeping.

4. In the category of overlying, there
is no statistical analysis AT ALL. The only facts that are reported are in
terms of percentages. The percentages that are chosen to be included are
done so in a way that inflates their significance. "Seventy percent" (of the
total overlain deaths) or
"more than half" (of the total overlain deaths) sensationalize and I feel
mis-represent the actual numbers
which are proportionally lower with respect to the total population in the
study, N=2178. The overlain category is n=180, and of this number, only 5%
of the total deaths (102) occurred in a bed with an adult. Compare this to
the way that wedging is represented. Wedging accounts for 879 deaths, but
it is referred to in the body of the text in subcategories of wedging, such
as 22% or 32%. These figures have less impact.
The conclusion that babies should not co-sleep is not founded statistical
analysis at all. The overlain in bed category accounts for 5%
of the total deaths. All the rest are from babies sleeping unsupervised in
inappropriate conditions.

5. Other problems with the overlain discussion include a very old reference
from 1970 about co-sleeping inviting disaster. A 30-year-old reference that
is an opinion is inappropriate. The text further contradicts itself and the
AAP position on co-sleeping and SIDS, specifically the AAP McKenna articles
listed in the bibliography. These articles are current, 1997, and conclude
that "bed sharing might be protective to infants at risks for SIDS" or
"bedsharing promotes breastfeeding that is thought to be protective of
SIDS".

6. I would also like to point out several factors that are unaccounted for
in the overlain category. I believe that drug use plays a role in the
increased occurrence of the overlain category over time. During the 3 time
spans that the study is using, the Cesarean rate has at least tripled
(including repeat Cesareans in all categories). Cesareans require some
postpartum medication, and for some women, this is for a protracted period
of time during the first three months. This is the period of time that most
of the overlain deaths occurred. It seems plausible that some of these
occurrences are actually iatrogenic results of the Cesarean. Additionally,
drug use and alcohol use has increased as well. We all know that
co-sleeping is inappropriate in all circumstances that involve drug use. I
suggest that drug use has inflated the incidence of overlain deaths, not the
increase in co-sleeping which, in biological/evolutionary terms, has always
existed.

7. All the figures are from the CPSC. This means that the data was
collected by an outside agency, and the reliability of the data is dependant
on the accuracy of the reporting that was done for reasons other than this
study. It is always a good idea to collect data specifically for the
purpose of the analysis that is intended to insure accuracy of collection.

8. The article does a good job of highlighting the dangers of unsafe sleep
environments, such as the presence of plastics or cord, or unsupervised
sleeping, including wedgings in cribs, a more common occurrence, 295 as
opposed to 102 totals. By these figures, cribs are still more dangerous.

9. The only conclusion that can be reached in this study is that parents
need to be meticulous and vigilant about sleep environments. Bed-sharing,
if meant as unsupervised use of an adult bed for an infant is different from
co-sleeping. This article says nothing about co-sleeping that is
statistically supported.

Maybe we can use this article to help parents understand the hazards of
unsafe sleep environments. But it is clear upon inspection that this
article does not support any conclusions about co-sleeping.

Joanne McCrory

             ***********************************************
The LACTNET mailing list is powered by L-Soft's renowned
LISTSERV(R) list management software together with L-Soft's LSMTP(TM)
mailer for lightning fast mail delivery. For more information, go to:
http://www.lsoft.com/LISTSERV-powered.html