MCHB/EPI Miami Conference — December 7 - 9, 2005

Measurement of Gestational Age: Challenges to Research and Surveillance

SHARON KIRMEYER: This presentation developed by myself and my colleague Joyce Martin concentrates on the birth weight distribution of a subset of very preterm infants that is the infants which were born at 28 to 31 weeks of gestational age. In the following slides we will concentrate on two themes. One theme concerns the changes in the birth weight distribution over the past decade for the 28 to 31 week births and the other theme concerns the differentiation between birth weight distributions by race, by Hispanic origin and by other characteristics.

This topic is important for three reasons. Firstly, that short gestational periods have adverse outcomes for infant mortality and for infant and child morbidity hence preterm births particularly need good surveillance. The questions have arisen regarding the quality of gestational age dated for preterm births due to issues of misreporting, misclassification, fetal pathologies and obstetric procedures. There is particular concern regarding the pronounced and some would even say the infamous curve that contain the heavier than expected births at 28 to 31 weeks of gestation.

The data used for the survey come from the birth certificates in the NCHS vital registration system. The three main variables used are those of gestational aged primarily based on date of last menstrual period, the infant's birth weight in grams and the race and Hispanic origin of mother. We have subset the data to contain births at 28 to 31 weeks of gestation for singletons only and for U.S. residents and finally the time period of interest span 1990 to 2002. We used as end points three years of data aggregated in order to have sufficient numbers of births to work with.

This is the birth weight distribution for births at 28 to 31 weeks of gestational age in 1990 to 1992. As you can see the curve is definitely bimodal. We wanted to measure or divide the curve into two portions and particularly to have a portion that we would call heavier than expected at any birth weight distribution for 28 to 31 weeks. So we used the (inaudible) at 2,300 grams to divide the distribution into two parts and we found this low point, 2,300 grams to be constant over the years and across groups. I hope you can see this division point. In 1990 to '92 the second curve contained 39 percent of the births. Ten years later a significant change has taken place. By 2000, 2002 the second curve has dropped to contain one-third or 33 percent of the births. Over the period there has been a change in population composition of these young preterms. There was a substantial decrease as you can see from the two bars in the total number of births at 28 to 31 weeks. There's been a major decrease in the number of non-Hispanic blacks and that's quite notable. Percentage wise there's been an increase of Hispanic births and non-Hispanic whites remain the largest groups. Now, the significance of this change in composition really concerns the change for non-Hispanic blacks. There have been changes that we will see for non-Hispanic blacks, however, as proportionately they have become smaller, their weight is less toward the overall outcomes.

This slide shows in some simplest terms, side by side, the birth weight distributions for three race or Hispanic origin groups. In 1990 to '92 the non-Hispanic whites had the lowest second curve. The non-Hispanic blacks the intermediate and Hispanics the largest. Here is the key finding. Over the decade we find that the second curve has declined in each group presented here. Hispanics maintained the largest second curve and non-Hispanic blacks had the greatest decline in the second curve.

This table provides some data points for the changes just seen. In 1990 nearly half the Hispanic births were in the second curve and Hispanics had the smallest percent change over the decade. The big change as was shown graphically was for non-Hispanic blacks. It was such a large change that non-Hispanic blacks and non-Hispanic whites then had essentially the same percentage point, 30 percent, or 30 out of 100 births in the second curve in the year 2000, 2002. We will now look at some shifts in the initiation of prenatal care and how it affects the second curve.

These bar graphs show the trimester when prenatal care began by years and race or Hispanic origin. Essentially almost all infants in each group received some prenatal care that's represented by the purple segment. By 2000 and 2002 the important change, which occurred, was the one that occurred to the large and equal increase for non-Hispanic blacks and Hispanics who began their prenatal care in the first trimester. In 2000, 2002 2/3 of these women in these groups initiated care in the first trimester of their pregnancy.

Well, we now ask whether or not the timing of prenatal care makes any difference on birth weight distributions. Here we look at changing birth weight distributions by the months when the care was initiated. If prenatal care is begun very early in the first or second months then you see a small second curve. If prenatal care is begun in the third month of pregnancy the curve looks like the general overall distribution by birth weight. If prenatal care is begun in the second trimester we now see that the second curve is larger than the first. And for the few women who started their prenatal care in the seventh or eighth month then that second curve is twice the size of the first curve. We add one more variable to what we were just looking at.

In this graph we add race and Hispanic origin to the data just shown. The X-axis would be the month prenatal care began. The Y-axis would be the percent of births in the second curve and we have representation for the three race Hispanic origin groups. This graph shows that there is a strong positive relationship between the month prenatal care began and the percent of births falling in the second curve. In other words the later that care begins the greater is the probability for an infant to fall into the second curve. If we were to fit trend lines to any of these three lines we would find that month by month for each month later that care is begun the curve would be five percentage points higher. However, there is as you can see a definite differential by group. For the same time that prenatal care is begun on average the Hispanic population is 10 points higher than the non-Hispanic whites and the non-Hispanic whites are three points higher than the non-Hispanic blacks. Besides prenatal care there are a couple other factors that are strongly associated with the second curve. As maternal age increases the second curve decreases and somewhat paradoxically as birth order goes up the second curve increases.

Now, we want to untangle this paradox of maternal age and birth order. So we have mapped independently the first birth from the second and higher birth and we find that regardless as maternal age goes up the proportion of the second hump or curve goes down but at each point the first birth has a lower percent in the second curve. And here are three other associations. One would be maternal education. The more years of schooling that the mother has had the smaller is the percent in the second curve.

This graph is concerned with method of delivery. In 2000 and 2002 almost half of the births that were delivered vaginally were in the second curve but only one out of six caesarian deliveries were. There's an even stronger distinction in curves for multiples and singletons. One in three singletons as we know from an earlier slide would be in the second curve but now that we are able in this slide to add the multiples we find that only one in 20 multiple births are in the second curve. In summary, over the past decade there was a significant decline in percent in the second birth weight curve of the 28 to 31 week infants. This decline is seen for each race or Hispanic origin group. The largest decline was for non-Hispanic blacks. The smallest for Hispanics. And the proportions for non-Hispanic blacks and non-Hispanic whites in the second curve converge to the same new low level of 30 out of 100 births. And regardless of race or Hispanic origin of mother we found that the larger percent of births in the second curve was associated with six factors, the later initiation of prenatal care, the younger maternal age, higher birth order, lower educational attainment of the mother, vaginal deliveries in singleton births and also in reference to the prenatal care there is a quandary to explore. There was a major and equal increase in prenatal care for both non-Hispanic blacks and whites but regardless of when prenatal care began significantly more Hispanics were in the second curve than were non-Hispanic blacks and if you have the hand out, if you want to contact myself or Joyce here would be our emails. Thank you.