Low Birth Weight and the Interval Between Pregnancies
There is general agreement that when women become pregnant too soon after giving birth, the risk of having low birth weight infants is increased. This leads to the idea that if fewer women had short intervals between birth and conception, there would be fewer low birth weight infants and a lower overall low birth weight rate.
Some researchers have also noted that, in some populations, black mothers are more likely to have shorter intervals between birth and conception compared to white mothers (New England Journal of Medicine, January 12, 1995 Volume 332, number 2 : "Prevalence of Low Birth Weight and Preterm Delivery in Relation to the Interval between Pregnancies Among White and Black Women" by Rawlings, Rawlings and Read). In populations where birth intervals are shorter for black mothers, some of the difference in low birth weight rates, between black and white mothers, is accounted for by the shorter birth intervals.
The first purpose of this paper is to present descriptive statistics regarding the intervals between the last live birth and the date of last menstrual period for Florida resident births in 1994. A second purpose is to quantify the decrease in the low weight births that would result if all women postponed conception past the point of increased risk. And a third purpose is to estimate how much of the difference between black and white low birth weight rates is due to differences in the intervals between birth and conception.
Birth records for Florida residents with birth dates in calendar 1994 were used in this analysis. Birth records where the mother had no previous live birth were excluded from the analysis since there are no birth intervals in these cases.
The interval between the prior birth and conception is calculated by using the date of last live birth and the date of last menses, both of which are on the birth record. Records with invalid values for either of these dates were excluded from the analysis. Out of 110,396 records, 9,778 or 8.9% were excluded for invalid dates.
On the birth record there is a month and year of last live birth, but no day. For calculating the interval between the prior birth and conception, the day of the month of the last live birth is set to 15. Using 15 will always be within 16 days of the actual day since in the worst case the actual day may be the 31st of the month which is 15 days after the 15th. Likewise the actual day may be the 1st which would be 14 days earlier than the 15th. In most cases the difference between the actual day and the assumed day will be less than 16 days. There are some records where the day in the date of last menses is missing. In these cases 15 was used for the day in the date of last menses, based on the same rationale used for the date of last live birth.
The number of days between the date of last live birth and the date of last menses was computed and then divided by 30.4375 to convert it to months. The result is the number of months between the date of last live birth and the date of last menses. This is truncated so that, for example, 1 month and 30 days is counted as 1 month and 1 month and 31 days is counted as 2 months.
The number of births and low birth weight births (birth weight less than 2500 grams) were determined for each value of the birth interval months and these data were used to compile the tables and graph used in the analysis. Low birth weight (LBW) was used as the outcome variable in this analysis because LBW is a commonly used outcome variable and the risks associated with birth interval can be compared to other risk factors for LBW such as maternal age, education and marital status.
Results and Discussion
Table 1 shows the number of births, LBW births and LBW percentages by 4 categories of birth interval calculated as explained above. In table 1 it can be seen that risk of LBW is generally higher for the shorter intervals and lower for the longer intervals. For the 0 to 2 month category the LBW percentage is 9.49% while for the 9+ category the LBW percentage is 5.51%.
Tables 2 and 3 are stratified by the race of the mother. The LBW percentage for both black and white mothers decreases fairly evenly across the 4 categories of birth interval but the LBW percentages are consistently higher for black births compared to white births within each category. In table 4 it can be seen that the LBW percentages are significantly higher for black births compared to white births in all birth interval categories. In the Rawlings study, cited above, there were no significant differences between LBW percentages for Black and White births in the shortest and longest birth interval categories. Only in the 2 middle categories, 3 to 5.9 months and 6 to 8.9 months, were the differences between races significant in the Rawlings study (table 3 in the Rawlings paper). In contrast, there were large and significant differences across all birth interval categories in the Florida data (table 4).
It is possible that this pattern is related to differences in the 2 populations regarding poverty status, access and quality of prenatal care. The population studied in the Rawlings article, referenced above, was composed of military families who had fairly equal income levels, access and quality of prenatal care, across racial lines. In contrast, the Florida population is probably not nearly as equal, regarding these characteristics, as the Rawlings population across racial lines. It may be that black women and white women with birth intervals of 9 months or more would have similar LBW percentages if they had similar income and prenatal care levels.
The graph shows the low birth rates for the Florida data for black and white births in relation to the interval calculated as explained above. In an effort to smooth the curve a 3 month moving percentage is used.
Table 5 gives the number and percent of births for black and white mothers across the 4 categories of birth interval. In both of the shorter intervals, 0 to 2 months and 3 to 5 months, the percentage of black women is higher than the percentage of white women. For example, 7.21% of the black births occur in the 3 to 5 month interval, while 4.08% of the white births occur in this interval. Black women are therefore more likely to have shorter birth intervals.
It is evident that some of the differences in LBW percentages between black and white births are due to the shorter average birth intervals for black births. Table 6 is an attempt to quantify this. In table 6 the black births are re-distributed across the 4 birth interval categories based on the percentages in these categories for white births. In other words, the column labeled "Adjusted births" shows what the number of black births in each category would be if they had the same birth interval pattern as the white births. The LBW percent for black births in each category is then applied to the adjusted black births to get the adjusted black LBW births of 2318. This is in contrast to the actual black LBW births of 2367.
If the number of black LBW births were 2318 then the LBW percent would be 9.64% instead of the actual 9.85%. The difference between the actual LBW percent of 9.85% and the adjusted LBW percent of 9.64% represents the difference due to shorter average birth intervals. Even though the differences due to birth interval have been adjusted out of the 9.64%, this is still much higher than the white LBW percent of 4.41%. Because of the adjustment, the difference between the adjusted black LBW percent of 9.64% and the white LBW percent of 4.41% is due to factors other than birth interval. Therefore the shorter average birth interval for the black births accounts for only a small part of the difference in the black and white LBW percentages.
One of the reasons the birth interval is interesting is because, unlike many of the other risk factors for LBW, the birth interval is theoretically subject to influence from public health programs and interventions. If it were possible for the forces of public health to have absolute influence over birth intervals, the decision might be made that there would be no birth intervals shorter than 9 months. Table 7 shows how this would affect the LBW rate.
On the line labeled "Adjusted" it can be seen that the total births of 100618 are assumed to have the same LBW percent as the actual LBW percent for the 9+ category in the table. This results in 5542 LBW births instead of the actual 5777. The difference of 235 are the estimated LBW births that would be avoided if the forces of public health successfully decreed that all birth intervals will be 9 months or more. The 235 potentially avoidable LBW births represent a small proportion of the approximately 14000 LBW births that occur annually.
In summary, two useful inferences may be drawn from this data. One is that only a small part of the difference between black and white LBW percentages can be accounted for by the differences in birth intervals. And another is the number of total LBW births would be reduced somewhat but not greatly if ideal birth intervals could be achieved for all Florida births.
It is also worth noting that the patterns seen in the Florida data are much different than the patterns observed in the Rawlings study. Specifically, in the Rawlings study there were no significant differences between LBW percentages for Black and White births in the shortest and longest birth interval categories. In contrast, there were large and significant differences across all birth interval categories in the Florida data. This should serve as a reminder that extreme caution must be used when attempting to make inferences about one population based on studies of another.
HSI 8/22/95 TABLE 1 LOW BIRTHWEIGHT (<2500 GRAMS) BY BIRTH INTERVAL ALL RACES Florida Resident Births 1994 (excluding first births) Birth to LMP Births LBW LBW Risk Date in Births Percent Ratio Relative (months) to 9+ Category 0-2 1,866 177 9.49 1.72 3-5 4,885 394 8.07 1.46 6-8 5,888 360 6.11 1.11 9+ 87,979 4,846 5.51 1.00 Total 100,618 5,777 5.74 TABLE 2 LOW BIRTHWEIGHT (<2500 GRAMS) BY BIRTH INTERVAL MOTHER'S RACE WHITE Florida Resident Births 1994 (excluding first births) Birth to LMP Births LBW LBW Risk Ratio Date Births Percent Relative (months) to 9+ Category 0-2 1,092 71 6.50 1.50 3-5 3,035 167 5.50 1.27 6-8 3,995 167 4.18 0.96 9+ 66,348 2,882 4.34 1.00 TOTAL 74,470 3,287 4.41 TABLE 3 LOW BIRTHWEIGHT (<2500 GRAMS) BY BIRTH INTERVAL MOTHER'S RACE BLACK Florida Resident Births 1994 (excluding first births) Birth to LMP Births LBW LBW Risk Ratio Date in months Births Percent Relative to 9+ Category 0-2 738 105 14.23 1.52 3-5 1,734 221 12.75 1.36 6-8 1,777 185 10.41 1.11 9+ 19,788 1,856 9.38 1.00 TOTAL 24,037 2,367 9.85 TABLE 4 BIRTHS BY BIRTH INTERVAL Florida Residents 1994 Birth to Births Births to White Black Black LMP Date to White Black LBW LBW to White (Months) Mothers Mothers Percent Percent Risk Ratio 0-2 1,092 738 6.50 14.23 2.19
TABLE 5 BIRTHS BY BIRTH INTERVAL Florida Resident Births 1994 Birth to Births Births Percent Percent Black LMP Date to White to Black Births Births to White Months) Mothers Mothers to White to Black Risk Mothers Mothers Ratio 0-2 1,092 738 1.47 3.07 2.09
TABLE 6 LOW BIRTH WEIGHT RATES BY BIRTH INTERVAL MOTHER'S RACE BLACK Florida Resident Births 1994 Adjusted Actual Birth to Births LBW LBW Adjusted LBW Minus LMP Date Births Percent Births Births Adjusted PT-LBW Births 0-2 738 105 14.23 352 50 55 3-5 1,734 221 12.75 980 125 96 6-8 1,777 185 10.41 1,289 134 51 9+ 19,788 1,856 9.38 21,415 2,009 -153 TOTAL 24,037 2,367 9.85 24,037 2,318 49 ADJUSTED LBW PERCENT =9.64 TABLE 7 LOW BIRTHWEIGHT RATES BY BIRTH INTERVAL ALL RACES Florida Resident Births 1994 Birth to LBW Percent LMP Date Births Births LBW (Months) 0-2 1,866 177 9.49 3-5 4,885 394 8.07 6-8 5,888 360 6.11 9+ 87,979 4,846 5.51 TOTAL 100,618 5,777 5.74 ADJUSTED 100,618 5,7542 5.51 LBW AVOIDED 235