Click for next page ( 69


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 68
6 Estimation of the Number of Income-Eligible Pregnant and Postpartum Women Because the number of pregnant and postpartum women cannot be directly observed with Current Population Survey (CPS) data, current USDA methods use the number of income-eligible infants to infer the number of women who are pregnant and who are up to 1 year postpar- tum. Various adjustments are made to the core estimates of income-eli- gible infants to obtain estimates of the number of income-eligible women. These include adjustments for multiple births and infant and fetal deaths, adjustments to account for the length of time a woman is pregnant, and adjustments to account for the percentage of postpartum women who breastSeed their infants. The methods also make assumptions about how the number of income-eligible infants translates into the number of in- come-eligible pregnant and postpartum women. For example, the method assumes that the family income for an infant is the same as family income during the time the mother was pregnant with the infant. This chapter reviews several ofthe adjustment procedures and assump- tions currently used to estimate the number of income-eligible pregnant and postpartum women based on the number of income-eligible infants. Assumptions about multiple births and fetal and infant deaths, about in- come during the time before and after a child is born, and about breast- feeding rates are reviewed. 68

OCR for page 68
INCOME-ELIGIBLE PREGNANT AND POSTPARTUM WOMEN PREGNANT WOMEN 69 A pregnant woman is income eligible for WIC if her family income is at or below 185 percent of federal poverty guidelines. Alternatively, income eligibility can be achieved if the woman is adjunctively eligible that is, if she is enrolled in Medicaid, the Food Stamp Program, or Temporary Assis- tance for Needy Families (TANF). The current USDA method to estimate the number of income-eligible pregnant women is based on the number of income-eligible infants: the number of income-eligible pregnant women is estimated by multiplying the number of income-eligible infants by 0.75 to account for a pregnancy of 9 of the 12 months of a year. This method assumes that the number of fetal and infant deaths equals the number of multiple births (i.e., there is no adjustment for fetal or infant deaths or multiple births). In this section, we assess the validity of this assumption. Assumptions Regarding Fetal and Infant Deaths and Multiple Births The current USDA estimation methodology assumes that the number of fetal and infant deaths and the number of multiple births cancel each other out. On one hand, using the number of infants to estimate the num- ber of pregnant women without accounting for fetal and infant deaths would understate the number of pregnant women. On the other hand, using the number of infants to estimate the number of pregnant women without accounting for the presence of multiple births would overstate the number of pregnant women. USDA (1999a) cites evidence from the Second WIC Eligibility Study (WES II) that multiple births are slightly more common than fetal and infant deaths. According to the WES II study, which based its findings on data from the late 1980s and early 1990s, an adjustment factor of 0.74 would be more accurate, instead of the current factor of 0.75. Using more recent but somewhat limited data, the panel made crude estimates of fetal and infant deaths and multiple births. These estimates show that the effects of multiple births and fetal and infant deaths do nearly cancel each other out. Vital statistics data for the year 2000 show 64,000 live births that were second- or higher-order births in multiple deliveries. These data also show 28,000 infant deaths. We do not, however, have good data on the number of fetal deaths since data on fetal deaths are not collected consistently at the national level due to many differing definitions by states. Thus, any esti- mate of fetal deaths will be problematic. To compare the number of fetal

OCR for page 68
70 ESTIMATING ELIGIBILITYANDPARTICIPATIONFOR THE WICPROGRAM deaths with the number of infant deaths, the panel chose to look at fetal and infant death reports in one state, North Carolina.1 In North Carolina, the number of reportable fetal deaths each year approximately equals the number of infant deaths. Assuming that this also is true for the United States as a whole, there would be 56,000 infant deaths and fetal deaths in 2000 (28,000 plus 28,0001.2 The 2000 census measured 3,806,000 infants in the United States (all income levels). Subtracting the 64,000 second- or higher-order multiple births from the 3,806,000 infants and adding the estimated 56,000 fetal and infant deaths results in an adjustment factor of 0.9979, which is essentially 1.0. It could be argued that these results for the entire population are not applicable to the WIC population, which has lower income and therefore is likely to experience a higher infant death rate. Data on multiple births and fetal and infant deaths by family income level are not available. However, we can repeat the analysis above for blacks, who on average are from lower income households and for whom infant mortality is higher than for whites. In 2000 in the United States there were 10,700 second- or higher-order multiple births and 8,500 infant deaths for blacks. We again assume that the number of fetal deaths is equal to the number of infant deaths. Sub- tracting the second- or higher-order multiple births and adding the esti- mated fetal and infant deaths to the 2000 population of 549,000 black infants (with one race listed) results in an adjustment factor for blacks of 1.012. Fetal and infant mortality rates of blacks are higher than those of whites. But blacks constitute approximately one-fourth of all WIC partici- pants (23 percent in 1998; U.S. Department of Agriculture, 2000b). Thus, it is likely that fetal and infant mortality in the WIC-eligible population is not as high as it is for the black population, and thus, the adjustment of 1.012 is probably too high for all WIC infants. The WES II study found that multiple births were slightly more com- mon than fetal deaths (U.S. Department of Agriculture, 1999a). Our crude analysis indicates that multiple births are slightly more common than fetal tin North Carolina and a number of other states, a fetal death is reportable if it occurs after 20 or more weeks of gestation. 2Even if consistent data were available for fetal deaths at all gestational ages, it is not clear that very early spontaneous fetal deaths should be counted for purposes of this method- ology (some occur before the woman knows she is pregnant).

OCR for page 68
INCOME-ELIGIBLE PREGNANT AND POSTPARTUM WOMEN 71 and infant deaths in the general population, but that fetal and infant deaths may be slightly more common than multiple births in low-income popu- lations. Income Variability The current USDA methodology does not directly utilize data on women to estimate the number of income-eligible and adjunctively eligible pregnant women. Instead, constant "multipliers" are applied to the esti- mated number of WIC-eligible infants. The current method used to estimate income-eligible pregnant women assumes that family income during pregnancy is similar to family income after the birth of a child no adjustment is made for changes in family income that occur around the time of childbirth. The method assumes that the number of income-eligible pregnant women is equal to 75 percent of the number of income-eligible infants. Variation in family income around the birth of a child was raised as a possible flaw in the current USDA estimation methodology (USDA, 1999a). For example, a woman may temporarily drop out of the labor force when the child is born, result- ing in diminished earnings when the child is an infant. In this example, the assumption that family income during pregnancy is the same as family income in the first year after the birth of the child will result in an overes- timation of the number of pregnant women eligible for WIC. But in an- other example, couples not previously married or living together but hav- ing a child together may marry or move in together, which may increase family income around the time the child is born. With Survey of Income and Program Participation (SIPP) longitudi- nal data, pregnant women can be identified by observing the birth of a child and counting back 9 months to collect the income and other infor- mation for the mother. The panel has considered two estimates of the ratio of the average number of months of eligibility for pregnant women relative to the average number of months of eligibility of infants. The first estimate is from the 1996 SIPP panel. The other is from a paper prepared for the panel to examine the variability of pregnant women's income. Using data from the 1996 SIPP panel, the ratio of the number of income-eligible and adjunctively eligible pregnant women to the number of income-eligible and adjunctively eligible infants was calculated for the years 1997 and 1998. In 1997, the ratio was 90.7 percent and in 1998 the ratio was 92.7 percent. These ratios indicate that between 7 and 9 percent

OCR for page 68
72 ESTIMATING ELIGIBILII~YANDPARTICIPATIONFOR THE WICPROGRAM of infants had mothers who were not eligible for the full 9 months of preg- nancy. The panel also commissioned a paper on the variability of income around the birth of a child, titled Income Variability and WIC Eligibility: Evidence from the SIPP (Yelowitz, 20021. This paper exploits the monthly longitudinal data from SIPP to track the income of women while they were pregnant through the first year after the birth of their child. SIPP data from the 1990, 1991, 1992, 1993, and 1996 panels were pooled. Yelowitz (2002) calculated the probability that a pregnant woman was eligible for WIC in each month during her pregnancy given that her infant was eligible for at least one month between birth and his or her first birth- day. The proportion of women eligible for WIC at least one month during pregnancy who had infants who were eligible at least one month during the first year ranges from 60 percent in the first month of pregnancy up to 80 percent in the month right before birth. Using these results, the weighted average number of months a pregnant woman was eligible was 6.4.3 This implies that the 0.75 adjustment for pregnant women is too high. SIPP- based estimates indicate that an adjustment factor of 0.53 would be more accurate (6.4 divided by 12 months). These results reiterate the conclusion of Chapter 5 that income vari- ability over the course of a year can be significant. They can demonstrate the importance of measuring income on a monthly basis. For example, if an adjustment factor of 0.53 is applied to the 2001 CPS estimates of in- come-eligible infants, the resulting estimate of the number of income-eli- gible pregnant women would be 804,000 instead of 1,138,000 which is the number of income-eligible pregnant women that result from multiply- ing the number of income-eligible infants by 0.75. Other Considerations Although a pregnant woman is eligible as soon as she becomes preg- nant, there is usually a delay between the time a mother conceives and the time she realizes she is pregnant. It is likely that there is also a lag between the time a woman finds out she is pregnant and the time she applies for 3Yelowitz also found that a small number of pregnant women were eligible during pregnancy but their infants were never eligible. This translates into an additional 0.76 of a month of eligibility over all pregnant women.

OCR for page 68
INCOME-ELIGIBLE PREGNANT AND POSTPARTUM WOMEN 73 WIC. The current method of estimating the number of pregnant women does not take either of these lags into account. These lags affect participation rather than eligibility. The USDA pub- lication WIC Participant and Program Characteristics 1998 indicates that nearly half of women who enroll in WIC enroll during their second or third trimester of pregnancy. The current method of using the 0.75 adjust- ment factor to obtain the number of eligible pregnant women based on the number of eligible infants would therefore substantially overstate the num- ber of pregnant women who participate in WIC, but not the number who are eligible for WIC. The assumptions examined in this section must be made because it is not possible to directly identify pregnant women using the CPS. SIPP data allow direct estimation of the number of pregnant women. Since the SIPP data are longitudinal, one can match infants with their mothers and obtain characteristics of the mother when she is pregnant, which is an advantage SIPP has over the CPS. POSTPARTUM WOMEN A postpartum woman is categorically eligible for WIC if she is less than 6 months postpartum or if she is 6 months to a year postpartum and breastEeeding. A categorically eligible postpartum woman is income eligible for WIC if her family income is at or below 185 percent of the federal poverty level, or if she is enrolled in Medicaid, the Food Stamp Program, or TANF (i.e., is adjunctively eligible). Estimates of breastSeeding and nonbreastSeeding women are made sepa- rately because food package costs differ depending on breastEeeding status. Estimates of breastSeeding duration are necessary because some women stop breastEeeding and return to the WIC service site to obtain formula for their infant. If they change their status from breastEeeding to nonbreastEeeding during the first 6 months postpartum, women receive a smaller and thus less expensive food package, although their infant's food package is now more expensive because it includes infant formula. During the second 6 months, nonbreastEeeding women are not eligible to receive a food package for themselves, but their income-eligible infants can receive the full infant food package. Current USDA methodology estimates the number of income-eligible postpartum women based on the number of income-eligible infants, with two adjustments. First, an adjustment of 0.9844 for multiple births and

OCR for page 68
74 ESTIMATING ELIGIBILI~YANDPARTICIPATIONFOR THE WICPROGRAM infant deaths is made. Second, an adjustment is made to account for the percentage of postpartum women who breastSeed their infants and the du- ration of breastSeeding~see Boxes 2-5 and 2-61. The current method to adjust for breastSeeding rates and duration is based on data from the National Maternal and Infant Health Survey (NMIHS), which was fielded in 1988. The NMIHS data show that when breastSeeding rates were fully adjusted for the duration of breastSeeding among postpartum women at a point in time, 37.4 percent were not breastSeeding and were less than 6 months postpartum, and 17.1 percent were breastSeeding (12.5 percent less than 6 months postpartum, and 4.6 percent more than 6 months postpartum). Thus the number of income- eligible infants is multiplied by 0.374 to estimate total nonbreastSeeding women less than 6 months postpartum. To estimate total breastSeeding women, the number of income-eligible infants is multiplied by 0.171. Note that the 0.374 adjustment factor plus the 0.125 adjustment factor for breastSeeding less than 6 months equals 0.50, which is the factor that would be applied to the number of income-eligible infants to estimate the total number of income-eligible women eligible for WIC until 6 months post- partum. The panel examined two types of adjustment factors used to estimate the number of eligible postpartum women those for estimating breast- feeding status and those used to account for fetal and infant deaths and multiple births. We also briefly discuss the method used to account for adjunctive eligibility of postpartum women. Adjustment Factors to Account for Breastfeeding Status Recent Trends in Breastieedting Rates Because the data used to estimate the percentage of breastEeeding women are more than 14 years old, the panel commissioned a paper to examine more current data on breastEeeding rates and duration, review breastSeeding trends and correlates, and review data sources and consider implications for estimating WIC eligibility. This paper, titled Estimating Eligibility for WIC: The Role of Breastieedting Jacknowitz, 2002), clearly demonstrates that the breastSeeding rates from the 1988 NMIHS are out of date. Several nationally representative surveys provide consistent data dem- onstrating that the rate of breastSeeding among all mothers has increased

OCR for page 68
INCOME-ELIGIBLE PREGNANT AND POSTPARTUM WOMEN 75 substantially since the NMIHS-based estimates were produced. Figure 6-1 (from lacknowitz, 2002) shows estimates of breastEeeding initiation rates from various data sources for all new mothers from 1970 to 2000, and Figure 6-2 shows trends in breastEeeding initiation rates from various data sources for women who participate in WIC. The figures show increases during the 1990s in breastEeeding initiation rates for both new mothers and mothers who participate in WIC. For WIC mothers, initiation rates have increased from 37 percent in 1988 (using the NMIHS data) to 54 percent in 1998 (using data from the National Survey of WIC Partici- pants). Data from the Ross Laboratories Mothers Survey, which were col- lected each year from 1988 to 1998, also show the upward trend in breastEeeding among WIC mothers (Figure 6-21. These figures refer to breastEeeding status shortly after delivery. Among the data sources used to prepare these figures, only the Ross Laboratories 80 70 60 - tic 50 . _ (A m 40 30 20 10 / ~ ~ .' RLMS ~ NSFG O NHAANES NMIHS 0- ~ 1 ,~9~ ,~9~ ,~9~ ,~9~ ,~9~oO ,~9OO~ ,~9OO~ ,~9Oo~ ,~9Oo~ ,~9OoO ~99~ ~99~ ~99> ~99~ 99 case Year FIGURE 6-1 Trends in breastSeeding initiation rates for all mothers (1970-2000~. NOTES: RLMS = Ross Laboratories Mothers Survey (Ryan, 2000~; NSFG = National Survey of Family Growth (NCHS, 1998~; NHANES = National Health and Nutrition Examination Survey (Burstein et al., 2000~; NMIHS = National Maternal and Infant Health Survey (Visness and Kennedy, 1997~. The vertical line marks the year of data collection for data used in the current ENS method. SOURCE: Jacknowitz (2002~.

OCR for page 68
76 ESTIMATING ELIGIBILI~YANDPARTICIPATIONFOR THE WICPROGRAM `~ 40 E ~ 30 (a m 20 60- 50 - I ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,I,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,or,,,,/,........................ ~ ,................................................ l .................................... ' 10 - RLMS O NHANES NMIHS NSWP ~ WIC-IFPS O - 1 1 1 1 1 1 1 1 1 1 1 1 1 Ago ,~9~ ,~9OoOO ,~9Oo9 ,~99~ ~99~ ~99~ 99~ 99> 99~ 99~ 9~ 99 999 C, Year FIGURE 6-2 Trends on breastSeeding initiation rates for WIC mothers (1986-2000~. NOTES: RLMS = Ross Laboratories Mothers Survey (Ryan, 2000~; NHANES = Na- tional Health and Nutrition Examination Survey (Burstein et al., 2000~; NMIHS = National Maternal and Infant Health Survey (Visness and Kennedy, 1997~; NSWP = National Survey of WIC Participants and Their Local Agencies (Cole et al., 2001~; WIC-IFP S= WIC Infant Feeding Practices Survey (Baydar et al., 1997~. The vertical line marks the year of data collection for data used in the current ENS method. SOURCE: Jacknowitz (2002~. Mothers Survey provides annual data on breastSeeding status of mothers at later points in their postpartum period. Figure 6-3 shows upward trends in 6-month postpartum breastSeeding rates for all mothers and for WIC-par- ticipating mothers. (Six months postpartum is the time when mothers must be recertified to remain eligible as breastSeeding postpartum women.) The breastSeeding rate for WIC mothers more than doubled over this time pe- riod (from 8 to 20 percent). lacknowitz (2002) used data from the 2000 Ross Laboratories Moth- ers Survey to calculate updated breastEeeding adjustment factors. These updated adjustment factors are 0.190 for women with infants less than 6 months postpartum (compared with the 0.125 currently used), 0.086 for women with infants between 6 to 12 months postpartum (compared with 0.046), and 0.310 for nonbreastSeeding women less than 6 months post- partum (compared with 0.3741. Using USDA:s 1999 estimates of the num-

OCR for page 68
INCOME-ELIGIBLE PREGNANTAND POSTPARTUM WOMEN 40 30 i0 .rJ) 20 ~ . -O 1 All mothers 77 U) m 10 ~ WIC mothers O - 1 1 1 1 1 1 1 1 1 1 1 .~9~ ~9~ ~99~ ~99` ~99~ ~99~ 99> 99~ 99~ 9~ 99~ 999 Year FIGURE 6-3 Trends in breastSeeding rates 6 moths after birth for all mothers and WIC mothers, Ross Laboratories Mothers Survey (1988-2000~. Data sources: Ryan et al. (1991~; Ryan (2000~; and Smith (2001~. SOURCE: Jacknowitz (2002~. her of income-eligible postpartum women and these updated adjustment factors, Jacknowitz found that an additional 63,000 women were income- eligible as breastEeeding women between 6 and 12 months postpartum (there is a corresponding decrease in the number of income-eligible nonbreastEeeding women). 4 Although none of these estimates represents the exact population of interest that is, WIC-eligible postpartum women we expect rates for WIC-eligible mothers to fall somewhere in between the rates for all moth- ers and WIC-participating mothers because we expect the WIC-eligible mothers will have higher income and educational status than WIC-partici- pating mothers but lower income and educational status than all mothers.5 4Because of data limitations, Jacknowitz was not able to adjust the rates for maternal age or for income; moreover, the analysis used breastfeeding rates for WIC participants, not those who were WIC eligible. fin her summary of the literature on correlates of breastSeeding status, Jacknowitz (2002) reports that lower educated mothers are less likely to initiate breastfeeding than more highly educated mothers, but that the relationship between income and breastfeeding initiation is not clear. Bitler et al. (2002) find a strong negative correlation between the likelihood of WIC participation and educational status.

OCR for page 68
78 ESTIMATING ELIGIBILI~YANDPARTICIPATIONFOR THE WICPROGRAM It is clear, however, that breastSeeding rates have changed substantially and that the adjustment factors used by USDA to estimate the number of breastSeeding and nonbreastSeeding postpartum women are out of date. CONCLUSION: The adjustment factors currently used to estimate the number of postpartum women who breastfeed are out of date. More recent estimates of breastfeeding status indicate that a greater percent- age of postpartum women now breastfeed than in the late 1980s, when the adjustment factors were developed. Updiating Estimates of the Numl7er of Breastieedling Postpartum Women The panel recommends that USDA update breastSeeding adjustment factors using more recent data sources. BreastSeeding rates, fully adjusted for duration, should be converted into breastSeeding adjustment factors for women less than 6 months postpartum and for women 6 months to 1 year postpartum in the same manner in which previous adjustment factors have been constructed. These adjustment factors should then be applied to esti- mates of the number of income-eligible women less than 12 months post- partum. Potential data sources for calculating the factors to adjust for breastSeeding rates and duration include the Early Childhood Longitudinal Study-Birth Cohort (ECLS-B), the National Health and Nutrition Exami- nation Survey (NHANES), and the published aggregate totals from the Ross Laboratories Mothers Survey. Each of these data sources has limita- tions to its use (see lacknowitz, 2002, for a more complete discussion). NHANES is a nationally representative survey of the civilian noninsti- tutionalized population ages 2 months and older. ECLS-B is a nationally representative survey of children born in 2001. Both the NHANES and the ECLS-B data collect information on income and program participation status, so it is possible to estimate the number of WIC income-eligible mothers. The NHANES data have a small sample size but are collected annu- ally, and therefore data from several years could be pooled to produce more reliable estimates. ECLS-B has a larger sample size (15,000 children born in calendar year 20011. However, unlike NHANES, it is not certain that updates will be conducted later. Each of these data sets collects different measures of breastSeeding sta- tus and duration. NHANES asks if the sampled child (who is at least 2

OCR for page 68
INCOME-ELIGIBLE PREGNANT AND POSTPARTUM WOMEN 79 months old) is currently breastEed and, if not, when breastEeeding stopped. ECLS-B also asks these questions but asks them over three waves of longi- tudinal data collection (at age 9 months, 18 months, and 30 months). Both sets would allow construction of a measure of the duration of breastSeeding. ECLS-B and NHANES both release public use data. The ECLS-B survey is scheduled to release its first wave of data in 2003. NHANES data waves from 1999 and 2000 are currently being released and will continue to be released in two-year groupings. Another data source, the Ross Laboratories Mothers Survey (RLMS), is produced on a regular basis and collects data on breastEeeding status at different points in the postpartum period (initiation, 1 month postpartum, 2 months postpartum, etc.~. RLMS is designed as a nationally representa- tive sample of mothers with infants up to 12 months old and has a very large sample size (over 400,000 in 20001. However, response rates for this survey have been quite low in the recent past (e.g., 31 percent in 20001. The estimated rates of breastEeeding do, however, track quite closely to those produced from other nationally representative data sources (see Fig- ures 6-1 and 6-21. RLMS does not, however, collect income information and so it is not possible to use the data to estimate breastEeeding rates among WIC income-eligible women. It does collect information about WIC participation. RLMS is a proprietary data source and not available for public use. Thus, it is not clear whether USDA can regularly obtain these data to continually update its breastEeeding adjustment factors. However, results from this survey have been published periodically, including esti- mated rates of breastEeeding status for each month after a child's birth. Given that survey data for breastEeeding status are not available on an annual basis, the panel considered whether synthetic estimates of breastEeeding rates could be produced in each year that the survey is not done. The synthetic estimates would be used to update the latest breastEeeding adjustment factor to account for changes in the distribution of the relevant population by reweighting the subgroup-specific breast- feeding rates. For example, if Hispanic mothers have higher rates of breastEeeding and the share of Hispanic mothers in the income-eligible population is known to have increased, the most recently available esti- mates of breastEeeding rates could be updated to reflect the increases in the percentage of the population that is Hispanic (and likely to have higher rates of breastEeeding). However, as lacknowitz (2002) shows, group-spe- cific breastEeeding rates have changed considerably more than population shares have. Thus, it is more important for USDA to use a method that

OCR for page 68
80 ESTIMATING ELIGIBILI~YANDPARTICIPATIONFOR THE WICPROGRAM reflects changes in breastSeeding rates than one that reflects only changes in population shares. The panel concludes that currently used adjustment factors for breastSeeding rates do not reflect the number of income-eligible postpar- tum women who breastSeed. The panel recommends that new estimates of the number of breastSeeding postpartum women should be made to reflect more recent trends in breastSeeding rates. RECOMMENDATION: Updated adjustment factors for breast- feeding rates amongWIC-eligible populations should be produced and applied to estimates of the number of income-eligible postpartum women to determine the numbers breastfee~ling anal not breastfee~ling. Data from the 1999 and 2000 NHANES should be available soon but will need to be combined with data from 2001 to create a large enough sample size to produce reliable estimates. Until those data are available, the most recently published breastSeeding rates from RLMS for WIC-participating postpartum women at initiation and 6 months postpartum would provide a better basis for the calculations used to set the adjustment factors. Adjustment Factors for Infant and Fetal Deaths and Multiple Births If the CPS continues to be used as the base data set from which eligi- bility estimates for postpartum women are produced, it is reasonable to continue to base these estimates on the number of income-eligible infants. An adjustment for fetal deaths is not needed, since all fetal deaths would already be excluded from the count of infants. Most of the infant deaths would not be observable by the CPS and hence not part of the count of infants. (One-half of all infant deaths in the United States occur in the first day of life, and two-thirds occur in the first month.) However, an income- eligible woman whose infant died would continue to be eligible for WIC until 6 months postpartum. Therefore, not adjusting for infant deaths would result in a very small undercount of eligible postpartum women. A count of infants that includes multiple births would overestimate the number of postpartum women. Using the figures for 2000 presented in the section on pregnant women, subtracting the 64,000 second- or higher- order multiple births from the census count of 3,806,000 infants results in an adjustment factor of 0.9832, which is very close to the current adjust- ment factor of 0.9844. It is reasonable to continue to use this small adjust-

OCR for page 68
INCOME-ELIGIBLE PREGNANT AND POSTPARTUM WOMEN 81 ment for multiple births in estimating the number of postpartum women. Given that the number of multiple births has increased substantially in the past decade (National Center for Health Statistics, 2001), this adjustment factor should be reevaluated periodically using the latest U.S. vital statistics data. SIPP-BASED ESTIMATES OF PREGNANT AND POSTPARTUM WOMEN This chapter's review of methods used to estimate categorical eligibility has primarily assumed that the CPS is the base data set used to estimate eligibility. However, one important advantage of SIPP is relevant to the discussion of estimating categorical eligibility in this chapter. Specifically, the longitudinal data of SIPP can be used to observe the number of cat- egorically eligible pregnant and postpartum women. Income, program par- ticipation, and family living arrangements can be measured at the time that a woman is pregnant and during her postpartum period, meaning that this information does not need to be inferred from information on infants.6 The presence of an infant in the household and information on the relation of household members is used to infer that a woman is the mother of an infant. Furthermore, several waves of data are needed in order to observe the periods over which a mother is pregnant and a child is born to directly estimate the number of pregnant and postpartum women, so the estimates are not as timely as CPS-based estimates. Nonetheless, the advantage of directly observing pregnant and postpartum women is a key one. CONCLUSION: SIPP data allow the direct observation of the num- ber of pregnant and postpartum women and their income, program participation, and living arrangements during the pregnancy and post- partum periods which is a major advantage over CPS data. SUMMARY This chapter has reviewed the methods for estimating the number of income-eligible and adjunctively eligible pregnant and postpartum women. 6Like the CPS, breastfeeding status is not collected in SIPP. Infant and fetal deaths are not observed in SIPP either; however, multiple births can be observed in SIPP and correctly accounted for when identifying the infants' mother.

OCR for page 68
82 ESTIMATING ELIGIBILITYANDPARTICIPATIONFOR THE WICPROGRAM Current methods used to infer the numbers of income-eligible pregnant women from the number of income-eligible infants tend to overstate the number of women in these categories. The panel concluded that the as- sumption that the numbers of infant and fetal deaths are roughly equal to the number of multiple births is reasonable. However, the 0.75 adjustment used to obtain the number of pregnant women from the number of infants does not consider that women may not be income eligible for the entire period of their pregnancy. Estimates from Yelowitz (2002) imply that women whose infants are eligible for WIC are themselves income eligible or adjunctively eligible for 6.4 months of the 9-month pregnancy period. With regard to the estimation of the number of breastEeeding and nonbreastSeeding postpartum women, the panel concludes that current ad- justments used to account for breastEeeding status among postpartum women are out of date they substantially underestimate the number of women who breastSeed and recommends that USDA update these ad- ~ustments. Finally, the panel concluded that the ability to directly estimate the number of pregnant and postpartum women with SIPP is a key advantage of that data set over the CPS-based estimates of pregnant and postpartum women.