Access to prenatal care in North Carolina: a comparative study of women with some prenatal care and women with no prenatal care

              DATA AND METHODS
The Access to Prenatal Care Survey (APCS) was a
statewide frequency-matched design in which infants
whose mother had no prenatal care, identified from
their 1985 birth certificates, were matched to women
with less than adequate prenatal care based upon the
Kessner Index of prenatal care. The Kessner Index
categorizes infants on the basis of the timing and
frequency of prenatal care and gestational age of the
infant at birth. Frequency ofvisits is classified by the
American College of Obstetrics and Gynecology
standard schedule of visits. These categories are
Inadequate care, Intermediate care, and Adequate
care. Women with adequate prenatal care must have
initiated care in their first trimester and received a
sufficient number of visits for the gestational age of
their infants. Women receiving less-than-adequate
care would have begun their care after the first
trimester or received less than the standard number
of visits for gestational age of their infants.
The choice of less-than-adequate care as the
standard ofcomparison was based upon both practical
and empirical criteria. While the ideal of public
health is to ensure access to adequate prenatal care
for all women, an interim step is to reduce the
number of women who receive no prenatal care. In
the short run a more realistic goal is to push this
group of women into the some-care category.
Empirical data also support the importance of some
care on outcome variables such as birthweight and
prematurity. Ryan et al. (1980) showed Aat the
greatest improvements in newborn health are between
those infants with no care and those with some care.
No-care infants were frequency matched to some-care
infants on the basis of four match variables:
mother's age, in 5-year age groups starting with 10-
14 and ending with 30 and over; mother's education,
in categories less than high school, high school, and
greater than high school; race, with white and
nonwhite groups; and region of the state, using the
four Department of Human Resources regions.
Frequency matching sets sample size criteria within
each category of the match variables. The goal of
frequency matching is to set the percentage distribu-tion
of the some-care women to that of the no-care
women within each cell ofthe 5X3X2X4 matrix of
match variables. In other words, the sample ofsome-care
women chosen resembled, with respect to
sociodemographic and geographic criteria, the no-care
women.
A sample of 1,200 1985 birth certificates was
chosen; 600 some-care births and 600 no-care births
from the automated birth registry file maintained by
the State Center for Health Statistics. Lists composed
of mother's name, child's name, father's name, and
address were compiled by county of residence of the
mother. Interviewers assigned by each county health
department participating in the APCS were trained
in data collection techniques and provided with lists
of interviewees.
Data collection began in April, 1986 and was
completed in early July. A total of 709 completed
interviews were collected from the 1 ,200 respondents
initially selected for the study. Based upon compari-sons
of 354 no-care respondents and 355 care
respondents interviewed, no discernible response
biases were found, with the age-race-education
characteristics of the entire sample not significandy
different from that of the actual respondents. Com-parisons
did, however, reveal discrepancies between
birth certificate entries of no prenatal care and self-report
of respondents of their prenatal care. The
original file of 709 was reduced by 146 respondents
with discrepancies in care status. This report is
limited to these 563 observations.
The primary methodological tool used here to
distinguish between no-care women and some-care
women after controlling for sociodemographic and
geographic confounders was logistic regression for
the analysis of categorical dependent variables. Each
logistic regression model must include match variables
as predictors as well as other variables of interest.
Logistic regression provides a determination of the
effect of predictor variables such as marital status,
employment, income, and attitudes related to primary
care on the odds of receiving some prenatal care.
Effiect parameters take the form of integers between
and infinity with numbers less than 1 indicating a
reduction in the odds of receiving care and numbers
greater than 1 indicating an increase in the chances of
having some pretuital care.
THE PROCESS OF SEEKING PRENATAL
CARE
The decision to use prenatal care services is a
complex interplay of attitudinal factors prevalent at
the time of pregnancy and biological conditions that
change with gestation, worked out within the context
of the family, the broader social group, and what
Levine et al. (1969) describe as the culture of
medicine, the health care deUvery system. Prenatal