Wilson, Ellen K. PhD, MPH*; Gavin, Norma I. PhD*; Adams, E Kathleen PhD†; Tao, Guoyu PhD‡; Chireau, Monique MD, MPH§
CONGENITAL SYPHILIS, A SERIOUS AND potentially fatal disease,1 can be prevented if pregnant women infected with syphilis are appropriately treated. For this reason, the American Academy of Pediatrics, the American College of Obstetricians and Gynecologists, and the Centers for Disease Control and Prevention recommend that women be screened for syphilis as early as possible in their pregnancies.1–3 They also recommend that women who are at high risk for syphilis or live in high-prevalence areas be screened again early in the third trimester. In Florida, legislation passed in 1986 requires that all women receive 2 tests during prenatal care (Florida Administrative Code 64D-3.019 and Florida Statute 384.31) (Kidder B, Florida Bureau of Medicaid Services, personal communication, October 19, 2005).
The extent to which the recommendations for syphilis screening are actually being implemented, however, is uncertain. When surveyed, 85% to 98% of physicians report that they routinely provide syphilis screening for pregnant women.4–6 Estimates of the proportion of women actually being screened are few and vary widely from 63% to 98%.7–9 Furthermore, no study to date has investigated whether screening rates have changed since the initiation of the National Plan to Eliminate Syphilis (NPES) in 1998, and none have focused on rates among low-income women—a population at a disproportionately high risk for syphilis.10
In this study, we use Medicaid administrative data to assess rates of prenatal syphilis screening in a low-income population—Florida women who had Medicaid-covered deliveries in fiscal years (FYs) 1995 and 2000—and estimate the change in rates over time. FY 1995 follows major expansions of the Medicaid program to cover more low-income pregnant women but predates other legislation that may affect their coverage, including welfare reform and the State Child Health Insurance Program (S-CHIP, a health insurance program jointly financed by the federal and state governments for low-income families and children). FY 2000 follows implementation of these program changes as well as the NPES.
Medicaid covered nearly half of all deliveries in Florida in the 2 analysis years (45% in 1995 and 44% in 2000).11,12 The income eligibility level for pregnant women in Florida during the analysis period was 185% of the federal poverty level (FPL). Thus, the data capture most deliveries among low-income women in the state.
Florida is a target state for syphilis elimination efforts. The state has implemented a number of strategies, including the creation of a team to review and analyze syphilis data and a mobile team to respond to outbreaks statewide; collaboration between government agencies and communities to develop and deliver syphilis elimination strategies; and development of a protocol for sexually transmitted disease laboratory and provider visitation to ensure quality syphilis prevention and care services.13 Trends in the incidence of congenital syphilis in Florida reflect those nationally declining from 60.5 cases per 100,000 live births in 1995 to 12.6 cases per 100,000 live births in 2003.14 Despite this decline, however, the rate of congenital syphilis in Florida remains above the national average, and some areas in the state experience particularly high rates. For example, among American cities in 2003, Miami had the third highest rate of congenital syphilis with 69.6 cases per 100,000 live births.15
This study addresses the following questions:
1. What proportions of Medicaid-covered pregnant women in Florida received a) any prenatal syphilis screening, b) syphilis screening early in their pregnancies, and c) repeat screenings later in their pregnancies?
2. Did these proportions change after implementation of the NPES?
3. How does the timing of prenatal care initiation, Medicaid enrollment duration, and Medicaid eligibility category affect these rates?
Materials and Methods
We obtained Medicaid enrollment and claims data for Florida for 4 years: 1994 and 1995, which came from the State Medicaid Research Files (SMRF), and 1999 and 2000, which came from the Medicaid Analytic eXtract files (MAX). The MAX files are similar to the SMRF but include recent federal and state changes in Medicaid enrollment categories, including the integration of S-CHIP enrollees and enhanced reporting of the race/ethnicity, diagnosis, procedure, and place of service fields.
Identification of Study Women
Because pregnant women are not readily identifiable in claims data, our analysis focuses on women with claims for Medicaid-covered deliveries. Codes used to identify these women are shown in Table 1. To ensure 9 months of data before delivery for all study women, we included only those women with a delivery from October 1, 1994, through September 30, 1995 (FY 1995) or from October 1, 1999, to September 30, 2000 (FY 2000). Because claims data often do not provide a unique date of delivery, we established a hierarchy of claims dates to choose the most probable date for the delivery.
We found delivery dates for a total of 72,871 women in the FY 1995 claims and 68,714 women in the FY 2000 data. We excluded women who might have incomplete claims data because they had other health insurance coverage concurrent with their Medicaid enrollment either from Medicare (175 women in FY 1995 and 62 in FY 2000) or private health insurance (2,482 in FY 1995 and 2,050 in FY 2000); women with capitated Medicaid managed care enrollment during either pregnancy or delivery (9,734 women in FY 1995 and 4,699 in FY 2000); and undocumented immigrants who are eligible for emergency and delivery-related care only (4,175 in FY 1995 and 7,697 in FY 2000). We also excluded 217 women in FY 1995 and 133 women in FY 2000 with missing enrollment data. If a woman had a second delivery during the 12-month analysis period, we analyzed her first delivery only. The final sample size was 56,088 in FY 1995 and 54,073 in FY 2000.
Syphilis Screening Tests
Because the claims records do not contain the date of conception or gestational age of the child at birth, and because not all pregnancies are full term, we do not know the true length of pregnancy and therefore cannot determine the trimester the woman enrolled in Medicaid or initiated prenatal care. Consequently, we used 9 months before delivery to proxy the prenatal period and searched this timeframe for evidence of syphilis screening tests using the Current Procedural Terminology, Fourth Edition (CPT-4) codes shown in Table 1. We gave the woman credit for a screening test if we found a claim with a CPT-4 code for a syphilis screening test, a confirmatory diagnostic test, or a dark field examination coupled with an International Classification of Diseases, Ninth Edition (ICD-9) code for syphilis.
Because Florida Medicaid reimburses hospitals for both inpatient and outpatient care on a per diem basis, we would miss many, if not all, screening tests that occurred in hospital settings. Analyses are therefore limited to screenings that occurred in clinics and doctors’ offices.
We kept the date of service for each occurrence of one of these codes and created dichotomous variables indicating whether she had: 1) any such tests; 2) a test early in pregnancy, defined as 5 or more months before delivery; and 3) a repeat test, defined as both a test early in pregnancy and a test within 3 months of delivery. We used these variables as our dependent measures.
Prenatal Care and Medicaid Enrollment
Receiving timely prenatal syphilis screening is contingent on receiving timely prenatal care. Prenatal visits were identified using selected CPT-4 and ICD-9 codes in certain combinations as shown in Table 1. The earliest date of service on records meeting these criteria was identified as the date Medicaid-covered prenatal care began.
The likelihood that women are screened may also be affected by the length of time they are covered by Medicaid. We calculated the duration of Medicaid enrollment during pregnancy based on the number of months between the date Medicaid enrollment began and the date of delivery. The duration of Medicaid enrollment among pregnant women is closely associated with the timing of prenatal care initiation.16 Causation occurs in both directions. Early enrollment in Medicaid facilitates access to care and thereby receipt of early prenatal care. On the other hand, prenatal care providers often facilitate enrollment in Medicaid among eligible women during the first prenatal visit; thus, early prenatal care may lead to early enrollment in Medicaid.
Medicaid enrollment duration is likely to differ across Medicaid eligibility categories. Most women with Medicaid-covered deliveries become eligible for Medicaid because of their pregnancies. These women are not enrolled in the program for one to several months after conception. Women with very low income levels who are either disabled or single parents meeting Welfare-related eligibility standards (including family incomes under 23% of the 2005 FPL) and teens eligible under Medicaid S-CHIP are eligible for Medicaid regardless of pregnancy status and therefore may have longer Medicaid enrollment during pregnancy. However, many “Welfare-eligible” women and pregnant teens are not enrolled in Medicaid prepregnancy but are motivated by their pregnancies to enroll.
We computed rates of any, early, and repeat screening based on all study women in each of the study years and compared the rates over time. For the FY 2000 data, we also performed bivariate analyses using χ2 statistics to test the association between the likelihood that women had a claim for prenatal syphilis screening and prenatal care use, Medicaid enrollment duration, and eligibility category. To isolate the effects of these 3 variables, we also computed and tested the statistical significance of differences in the rates of 1) any prenatal care, 2) early prenatal care (5–9 months before delivery), 3) any prenatal syphilis screening among those who had any prenatal care, and 4) early prenatal syphilis screening among those who had early prenatal care by Medicaid enrollment duration and eligibility category. All analyses were performed using Sudaan 9.0.1.
The use of claims data to assess healthcare delivery has numerous advantages: the population size is large, the data provide information across multiple providers, and data collection is both low-cost and unobtrusive. However, an accurate picture of the use of specific health services is hard to get from claims data because of lack of uniformity in coding across providers17,18 and the failure to capture all services received.17,19,20 The data capture only those services that were billed to and paid by the Florida Medicaid program on a fee-for-service basis. For example, the claims data do not include services provided in hospital inpatient or outpatient departments that were reimbursed on a per diem basis. Therefore, the actual proportion of women receiving screening is likely to have been higher than what we found. Because the proportion of women for whom hospital inpatient or outpatient departments were their usual source of care for prenatal care is relatively small (8% in FY 2000), however, it is unlikely that this limitation has a large effect on our estimates. When the analysis is limited to women with clinic- or office-based prenatal care, the proportion of women who had a claim for syphilis screening increases by only 2 percentage points. Because we measure early prenatal care as care received 5 or more months before delivery, another outcome that will be somewhat underestimated among women who delivered prematurely is the extent of early prenatal syphilis screening.
Characteristics of the Study Population
The study population was mainly between the ages of 18 and 34 (85% to 86%); racially diverse, with approximately 30% non-Hispanic black and 20% Hispanic; and predominantly urban, with roughly 87% of the study women living in central cities of large metropolitan areas (Table 2). Only small changes were seen over time. In FY 2000 compared with FY 1995, mean age was slightly younger (24.1 year vs. 24.8 years) (not shown); a slightly larger proportion were Hispanic (21% vs. 19%); and the proportion who lived in Dade County, the largest metropolitan central city in the state, was somewhat smaller (15% vs. 18%).
In FY 1995, 62% of women had their first claims for prenatal care early in their pregnancies (5–9 months before delivery), and an additional 31% had claims for prenatal care beginning later in their pregnancies. Only 8% had no claims for prenatal care. Prenatal care use improved slightly in FY 2000: 64% received prenatal care early in their pregnancies and 6% received no prenatal care.
The proportion of women who enrolled in Medicaid late in their pregnancy (less than 5 months before delivery) was the same in both years (26%), but the proportion who were already enrolled at the start of their pregnancy (9 or more months before delivery) was significantly higher in FY 2000 (20%) than in FY 1995 (15%). The proportion of women who met Welfare-related eligibility requirements for Medicaid decreased substantially, from 41% to 30%, and the proportion enrolled under expansion criteria increased from 55% to 66%.
In contrast to the small improvements we found in prenatal care use, we found large increases in prenatal syphilis screening from FY 1995 to FY 2000. In FY 1995, only 26% of women had at least one claim for a syphilis screening test (Fig. 1). By FY 2000, this proportion had more than doubled, to 57%. The proportion of women who had a claim for syphilis screening early in their pregnancies also increased sharply, from just 11% in FY 1995 to 33% in FY 2000, as did the proportion with repeat screening, from 4% in FY 1995 to 9% in FY 2000.
Receipt of syphilis screening is highly correlated with both the timing of prenatal care and Medicaid enrollment duration. For example, among women with late prenatal care (less than 3 months before delivery), only 36% had any syphilis screening compared with 69% of women with early prenatal care (not shown). Similarly, among women enrolled in Medicaid under 3 months, the proportion with any screening was just 24% compared with 68% of those enrolled 7 to under 9 months (not shown).
Although timely prenatal care is necessary for timely syphilis screening, it is not sufficient. As shown in Table 3, among women who had at least one claim for prenatal care, only 60% had a claim for any syphilis screening. Similarly, among women receiving early prenatal care, only 52% received early syphilis screening.
The receipt of both prenatal care and syphilis screening is correlated with Medicaid enrollment duration. Women enrolled less than 3 months before delivery were significantly less likely to have at least one claim for prenatal care (67%) than those enrolled 3 or more months before delivery (more than 95%). The proportion of women who had a claim for prenatal care early in pregnancy increased with enrollment duration between 5 to 7 months and 7 to 9 months, and then declined for women enrolled prepregnancy (9 or more months of enrollment). Among women with a claim for prenatal care, the percentage with a claim for any syphilis screening increased with enrollment duration up to 7 to 9 months. Receipt of early syphilis screening also increases with enrollment duration: from 50% of women enrolled 5 to 7 months to approximately 60% of women enrolled 7 or more months.
The receipt of prenatal care and syphilis screening is also correlated with Medicaid eligibility category. Welfare-related enrollees were somewhat less likely than expansion-related enrollees to receive any prenatal care (90% vs. 97%), primarily because they were more likely to be enrolled for under 3 months. The difference between Welfare-related enrollees and expansion-related enrollees in the receipt of early prenatal care is more substantial: For every category of Medicaid enrollment duration, expansion-related enrollees are more likely to receive early prenatal care. Overall, 86% of expansion-related enrollees received early prenatal care compared with just 74% of Welfare-related enrollees. Controlling for the receipt of prenatal care, differences between Welfare-related enrollees and expansion-related enrollees in the proportions of women receiving any syphilis screening and early syphilis screening were relatively small (59% vs. 61% and 49% vs. 53%, respectively).
This study found that the proportion of Medicaid-enrolled women in Florida who had a claim for prenatal syphilis screening more than doubled between FY 1995 and FY 2000, from 25% to 57%. Despite the increase, the proportion in FY 2000 remained far below the recommended universal screening. The proportions of women who were screened early in pregnancy (34%) or received a repeat test late in pregnancy (10%) were even lower.
The prenatal syphilis screening rates we found in the Florida claims data are slightly lower than the 63% found in a recent claims-based study of commercially insured women delivering in 1999.8 However, the difference in the estimates is largely attributable to Tao and colleagues’ restriction of their study sample to women who were enrolled in insurance continually for 2 years. Most of the women in this study enrolled in Medicaid only after they became pregnant. Our screening rates for Medicaid women enrolled 7 or more months before delivery are comparable or slightly better than those estimated by Tao and colleagues for commercially insured women.
The proportions of women with documented screenings in both this study and the study by Tao and colleagues were substantially lower than the 83% found by Trepka and colleagues9 in their Dade County study or the 98% found by Schrag and colleagues in their multisite study.7 Our lower rate may reflect true differences in screening or may be related to differences in data collection techniques. Whereas both Trepka and colleagues and Schrag and colleagues used medical records to assess syphilis screening, both this study and Tao and colleagues used claims data.
The proportion of women receiving repeat screening was 9% overall, similar to the 11% found by Trepka and colleagues. Nine percent overall equates to 17% of those women who received a first syphilis test. This proportion is somewhat lower than either the 19% found by Schrag and colleagues or the 23% found by Tao and colleagues. Thus, despite the fact that the Medicaid population is presumably at higher risk for syphilis than the populations of these 2 studies, they are less likely to receive the repeat screenings recommended for high-risk women.
As has been noted elsewhere, timely receipt of prenatal care is essential for women to receive the recommended screenings.1,21,22 Increasing the proportion of women receiving prenatal care early is therefore an important means of increasing screening rates. Our findings indicate, however, that it is not sufficient.
In general, length of Medicaid enrollment duration during pregnancy has a strong positive association with the receipt of both prenatal care and syphilis screening. One explanation for this increase may be that because women enrolled for a longer period had more prenatal care visits, they therefore had more opportunities to be screened. Another possible explanation is that because syphilis screening is recommended at the first prenatal visit, women who were enrolled for shorter periods of time may have been screened at a first prenatal care visit not covered by Medicaid.
Women enrolled for 9 or more months, however, are somewhat less likely to receive any prenatal care than those enrolled for 7 to under 9 months and much less likely to receive early prenatal care. One explanation for this may be that women who were enrolled in Medicaid for 7 to under 9 months may include many women who were not eligible for Medicaid before they became pregnant but were aware that they were pregnant and motivated and able to seek prenatal care early in their pregnancies. Women who were eligible and enrolled in Medicaid before their pregnancies may be comparatively less motivated to receive care. The lower rates of prenatal care among women enrolled 9 or more months could also be explained if these women were more likely to use safety net providers in hospital settings in which case our claims analysis would not have fully captured their receipt of services. This explanation is not supported by the data, however, which indicate that women enrolled 9 or more months are no more likely to have hospitals as their usual source of prenatal care than women enrolled in Medicaid for shorter periods of time.
As noted, Welfare-related enrollees are somewhat less likely to receive any prenatal care and substantially less likely to receive early prenatal care than expansion-related enrollees regardless of enrollment duration. This fact largely explains differences between these 2 groups in the likelihood that they receive syphilis screening. Because Welfare-related enrollees are typically poorer and less educated than expansion-related enrollees, these women may be particularly challenging to reach.
Overall, these data suggest that significant advances have been made in meeting the objectives for universal prenatal syphilis screening in Florida since implementation of the NPES, but considerably more needs to be done. To increase the proportion of pregnant women who receive syphilis screening, we must first increase the proportion of women who receive prenatal care particularly early in their pregnancies. Medicaid enrollment appears to be inadequate to ensuring this need is met. For women who do receive prenatal care, additional efforts are also needed to ensure compliance with recommended screening guidelines. Increasing preconceptional counseling and care, and ensuring that doctors are reimbursed for providing these services, may also contribute to increased screening.
1. Centers for Disease Control and Prevention, 2004. Congenital syphilis—United States, 2002. MMWR Morb Mortal Wkly Rep 2004; 53: 716–719. Available at: www.cdc.gov/mmwr/PDF/wk/mm5331.pdf
. Accessed December 14, 2005.
2. American Academy of Pediatrics, The American College of Obstetricians and Gynecologists. Guidelines for Perinatal Care, 5th ed. Elk Grove Village, IL: American Academy of Pediatrics, 2002.
3. American Academy of Pediatrics and American College of Obstetricians and Gynecologists. Guidelines for Perinatal Care, 3rd ed. Washington, DC: AAP and ACOG, 1992:133–455.
4. Mills WA, Martin DL, Bertrand JR, Belongia EA. Physicians’ practices and opinions regarding prenatal screening for human immunodeficiency virus and other sexually transmitted diseases. Sex Transm Dis 1998; 25:169–175.
5. Hogben M, St. Lawrence JS, Kasprzyk D, et al. Sexually transmitted disease screening by United States obstetricians and gynecologists. Obstet Gynecol 2002; 100:801–807.
6. Weisbord JS, Koumans EH, Toomey KE, Grayson C, Markowitz LE. Sexually transmitted diseases during pregnancy: Screening, diagnostic and treatment practices among prenatal care providers in Georgia. South Med J 2001; 94:47–53.
7. Schrag SJ, Arnold KE, et al. Prenatal screening for infectious diseases and opportunities for prevention. Obstet Gynecol 2003; 102:753–760.
8. Tao G, Patterson E, Lee LM, Sansom S, Teran S, Irwin KL. Estimating prenatal syphilis and HIV screening rates for commercially insured women. Am J Prev Med 2005; 28:175–181.
9. Trepka MJ, Bloom SA, Zhang G, Kim S, Nobles R. Inadequate syphilis screening among women with prenatal care in a community with high syphilis incidence. Sex Transm Dis 2006; 33:670–674.
10. Aral SO. The social context of syphilis persistence in the southeastern United States. Sex Transm Dis 1996; 23:9–15.
11. National Governors’ Association. Table 2. Medicaid births as a percentage of total births, 1994 and 1995. State Medicaid coverage of pregnant women and children. NGA reports online, September 30, 1997. Washington, DC: National Governors’ Association. Available at: www.nga.org/pubs/IssueBriefs/1997/97093MCHUpdate.asp
. Accessed December 14, 2005.
12. NGA Center for Best Practices. MCH update: States protect health care coverage during recent fiscal downturn. 2005. Available at: www.nga.org/Files/pdf/0508MCHUPDATE.PDF
. Accessed December 14, 2005.
16. Egerter S, Braveman P, Kristen M. Timing of insurance coverage and use of prenatal care among low-income women. Am J Public Health 1999; 92:423–427.
17. Cotter JJ, Smith WR, Rossiter LF, Pugh CB, Bramble JD. Combining state administrative databases and provider records to assess the quality of care for children enrolled in Medicaid. Am J Med Qual 1999; 14:98–104.
18. Iezzoni L. Assessing quality using administrative data. Ann Intern Med 1997; 127:666–674.
19. Downey L, Lafferty WE, Tao G, Irwin KL. Evaluating the quality of sexual health care provided to adolescents in Medicaid managed care: A comparison of two data sources. Am J Med Qual 2004; 19:2–11.
20. Bronstein JM, Santer L, Johnson V. The use of Medicaid claims as a supplementary source of information on quality of asthma care. J Healthc Qual 2000; 22:13–18.
21. Southwick KL, Guidry HM, Weldon MM, Mertz KJ, Berman SM, Levine WC. An epidemic of congenital syphilis in Jefferson County, Texas, 1994–1995: Inadequate prenatal syphilis testing after an outbreak in adults. Am J Public Health 1999; 89:557–560.
22. Warner L, Rochat RW, Fichtner RR, Stoll BJ, Nathan L, Toomey KE. Missed opportunities for congenital syphilis prevention in an urban southeastern hospital. Sex Transm Dis 2001; 28:92–8.