The Maternal-Infant Mortality Review (MIMR) has coordinated ongoing retrospective committee reviews of all Alaskan-resident infant deaths since 1992. At the time of the current study, the MIMR committee had reviewed 891 infant deaths; 95% of the deaths occurred from 1992 through 2004 (99% of Alaskan infant deaths occurring from 1992 through 2002, 81% of deaths occurring in 2003, and 24% of deaths occurring in 2004). Case files comprised birth and death certificates, maternal and infant medical records, autopsy reports, first responder reports, and any existing child protective services reports or public health nurse home interviews. At review sessions, three or four medical experts reviewed each file and came to a consensus decision regarding what they believed were the most probable and contributing causes of death.
We included two categories of deaths in the current analysis: (1
) MIMR committee or death certificate report of SIDS or asphyxia as the cause of death and (2
) any case with a report that death occurred during sleep, regardless of assigned cause. We did not include deaths not yet reviewed by the committee because these usually did not have complete information available. For deaths that met our criteria, we used a standard data abstraction tool to collect further details from the MIMR case files. For those that occurred in association with bed sharing, we ascertained the presence of five risk factors:
- Position: Infant was put to sleep prone (or found prone only if no information was available on position in which infant was put to sleep).
- Non-caregiver: Infant was sleeping with someone who was not the primary person responsible for the infant's care at the time of death, including those sleeping with a caregiver and other people.
- Maternal tobacco use: Either prenatal or postnatal cigarette or chew use by the mother was reported in any of the records available.
- Impairment of bed-sharing partner: Infant was sleeping with any person impaired by alcohol, tobacco, or other drug use at the time of death.
- Surface: Infant was sleeping on a sofa or water bed.
To compare demographic characteristics of infant deaths while bed sharing with all infant deaths, we used a dataset of linked birth and death certificate records for all Alaskan infant deaths from 1992 to 2004, provided by the Alaska Bureau of Vital Statistics.
PRAMS is an ongoing population-based randomized surveillance system that samples approximately 18% of Alaskan mothers of newborns two to six months after a live birth. The Centers for Disease Control and Prevention (CDC) designed the PRAMS sampling, implementation, and weighting methodologies, which are described elsewhere.20,21
From 1996 through 2003, there were approximately 79,912 Alaskan births; 14,777 (18.5%) were sampled for PRAMS, and 11,837 mothers responded to a mail or phone survey (response rate = 80.1%). The mean infant age at the time the mother responded was 3.2 months (range: 2.2–9.9). We compared weighted proportions of selected demographic characteristics of survey respondents with all births in Alaska on an annual basis to ensure the sample was representative. For example, survey respondents compared with all births in 2003 were similar on examined characteristics, including maternal education <12 years (14.2% vs. 14.3%), Alaska Native race (25.0% vs. 24.4%), unmarried status (35.2% vs. 34.3%), and low birthweight birth (5.1% vs. 5.1%).
During birth years 1996–1999, Alaska PRAMS asked women, “How often does your new baby sleep in the same bed with you?” During 2000–2004, the question was modified to, “How often does your new baby sleep in the same bed with you or anyone else?” We identified frequent bed sharing by responses of always or almost always, and infrequent bed sharing by responses of sometimes, rarely, or never (rarely was not a response option during 1996–1999). We linked PRAMS data to birth certificates for 10,860 respondents who had non-missing data for the question about bed-sharing habits. This question was designed to be skipped by women whose infants were not living with them at the time they responded to the survey. Among respondents with missing data on bed sharing (n
=703), 16.4% reported that their infant had died before they completed the survey. We analyzed PRAMS data using SPSS® Version 16.022
to adjust for the survey design.
We used PRAMS data to examine background population prevalences of certain risk factors of interest among all births and among women whose infants frequently bed share. Risk factors on PRAMS included maternal cigarette smoking habits currently and during the last three months of pregnancy, prenatal smokeless tobacco or chew use, prenatal marijuana use, number of alcoholic drinks in an average week during the three months before pregnancy and currently, and the position in which the infant was most often laid down to sleep. We combined information on prenatal or current smoking on PRAMS with smoking reported on the birth certificate to create a single smoking risk factor category, and combined PRAMS or birth certificate reports of prenatal smokeless tobacco use to create a single chew risk factor category. To measure the prevalence of women who may have consumed some alcohol every day, we assigned women who drank seven or more drinks per week into the category of daily drinker.
The MIMR records and PRAMS data used for this study were collected by ongoing surveillance programs of ADPH. An employee with legal access to these data for the purpose of developing public health recommendations conducted all analyses. Thus, no institutional review board (IRB) approval or review was sought or obtained for this study. Alaska PRAMS is reviewed annually by IRBs at CDC and the University of Alaska Anchorage.