The influence of maternal risk factors on birth outcomes for HIV-infected pregnant women has grown in importance with the spread of human immunodeficiency virus type 1 (HIV) among women of childbearing age. While the primary concern is to reduce the risk of maternal-child transmission, the effect of maternal HIV infection on infant birth weight has substantial significance because low birth weight (LBW) has a strong influence on infant mortality and morbidity (1,2). Prior research has produced conflicting results regarding the role of HIV infection on obstetrical outcomes. Several studies from the United States and Europe have reported a limited impact of maternal HIV on birth weight, after adjusting for illicit drug use and other maternal characteristics that affect neonatal outcomes adversely (3-5). In contrast, several African studies suggest that HIV-infected women have a greater risk of LBW than do seronegative women (6-8). Studies of sub-Saharan women are less likely to be confounded by the coexistence of HIV and illicit drug use but may be confounded by maternal malnutrition or access to prenatal care.
We compared the risk of having a LBW infant in a cohort of 772 HIV-infected women to that of a general sample of women without evidence of HIV infection. The HIV-infected and general cohorts delivered in the same time frame and were enrolled in the New York State (NYS) Medicaid program. NYS Medicaid is an important source of information on HIV-infected pregnant women because it is the primary payer of care for indigent pregnant women as well as persons infected with HIV (9-11). In addition, NYS has more women and children reported with the acquired immunodeficiency syndrome (AIDS) than any other state (12). Our large study population enabled us to examine the risk of LBW associated with HIV infection after adjusting for diverse maternal clinical, demographic, and health care delivery factors.
MATERIALS AND METHODS
The study population was drawn from longitudinally linked NYS Medicaid claims and eligibility files for women delivering a live-born singleton in federal fiscal years (FFY) 1989 and 1990. Diagnostic information on claims data are coded according to the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM). To identify HIV-infected women, we applied a tested case-finding approach that searches for HIV-related medical care or a diagnosis of HIV infection, AIDS, or an AIDS-related condition (13) on ≥2 outpatient or ≥1 inpatient claim(s) (11). Our study eligibility criteria excluded women covered by a NYS Medicaid global fee program that offers a single payment for all prenatal care (6% of HIV-infected and 19% of general Medicaid women). Our eligibility criteria also required that vital statistics data had been obtained for the delivery (accomplished for >90% of both HIV and general samples). Finally, we analyzed only singleton births. Our case-finding and eligibility criteria identified 772 HIV-infected women delivering in the study years.
Regarding the development of data for the general cohort, NYS routinely assembles claims and eligibility files for all women who deliver in the last quarter of each FFY. From women delivering in the last quarter (that is, July through September) of FFY 1989 and 1990, we excluded all women with evidence of HIV infection on review of up to 10 years of Medicaid files. We performed a random selection to identify a general sample of women that was ≈3 times the size of the HIV population. We stratified our selection of the general population to have a similar geographic distribution as the HIV-infected women, with ≈80% residing in New York City (14). The random selection resulted in a general sample of 2,377 NYS Medicaid-enrolled women.
From the Medicaid longitudinal claims and enrollment data, we identified use of illicit drugs during pregnancy, duration of Medicaid enrollment during pregnancy (that is, <3 or ≥3 months), date of delivery, delivery by cesarean section, and residence (that is, in or out of New York City). History of illicit drug use was determined from an algorithm applied to Medicaid claims that searches for the receipt of methadone maintenance services or for ICD-9-CM codes indicating illicit drug dependence or abuse. Validation studies using data from charts has shown that nearly 90% of individuals whose HIV risk factor was illicit drug use were identified by using this algorithm (15). Vital statistics data offered information on birth weight, mother's age at delivery, education, self-reported racial/ethnic group, parity, cigarette smoking during pregnancy, medical risk factors, and self-reported prenatal care (that is, month of initiating prenatal care and number of prenatal visits). Duration of pregnancy was measured from the physician's estimate of infant gestational age or, when unavailable, from the mother's estimate of her last menstrual period.
Lacking a “gold standard,” we examined two measures of the adequacy of the number of prenatal visits: (a) the Kessner index (16) and (b) the Adequacy of Prenatal Care Utilization (APNCU) index developed by Kotelchuck (17). These measures differ primarily in their approach to judging the adequacy of the number of visits for women initiating prenatal care later in pregnancy. The Kessner index requires “catch-up” in the number of visits (for example, nine visits for a delivery at 36 weeks or more regardless of the month that care was initiated), whereas the APNCU does not (for a delivery at 40 weeks, for example, 14 visits needed if care began in the first month, but only nine visits if care began in the fifth month). We report the proportions with first-trimester care separate from the adequacy of the number of visits because many indigent women initiate prenatal care later due to the need to enroll first in Medicaid (14).
Patients were identified as having medical or obstetrical risk factors other than HIV infection if their vital statistics records showed the diagnosis of any of the following conditions during their pregnancy: eclampsia, anemia, diabetes, Rh sensitization, uterine bleeding, cardiac disease, lung disease, kidney disease, chronic hypertension, pregnancy-related hypertension, genital herpes, incompetent cervix, polyhydramnios, or hemoglobinopathy. Because all these conditions occurred at low rates, we created an indicator of having any of one of these medical or obstetrical risk factors for our analyses.
Our study outcome was a birth weight of <2,500 g. The chisquare statistic was used to compare clinical and demographic characteristics of the HIV-infected and uninfected women, as well as to examine associations between these characteristics and LBW. Logistic regression was used to determine the association of maternal HIV infection with the odds of LBW. We report two models for the entire study population and a third examining only the subset with a term delivery because preterm birth (<37 weeks) is strongly related to LBW. We control for illicit drug use, maternal age, educational level, racial/ethnic group, FFY of delivery, Medicaid enrollment during pregnancy, cigarette smoking during pregnancy, previous live births, New York City delivery, adequate number of prenatal visits, and other medical risk factors. We created a “missing data” indicator for three variables—educational attainment, parity, and cigarette smoking during pregnancy—because information was not available for a substantial proportion of the population. Analysis of interactions showed only one of statistical significance (p < 0.05), between HIV infection and medical/obstetrical risk factors. Therefore, we created four variables to show the effect of this interaction: HIV+ with medical/obstetrical risk+; HIV- with medical/obstetrical risk+; HIV+ with medical/obstetrical risk-; and HIV- with medical/obstetrical risk-. Interestingly, we did not find an interaction between HIV infection and drug abuse (p = 0.20). All reported logistic models had a satisfactory fit (p > 0.50) by the Hosmer-Lemeshow statistic.
As shown on Table 1, a LBW infant was delivered by 29% of the HIV-infected women compared to only 9.3% of the general sample (p < 0.001). A higher proportion of the HIV-infected population were non-Hispanic African-American than the general sample (46% vs 37%, p < 0.001). As expected, the HIV-infected women were more likely to be identified as illicit drug users (p < 0.001). In addition, the HIV-infected women were less likely than the general sample to be teenagers (4% vs 13%, p < 0.001) or to have completed high school (49% vs 57%, p = 0.001 excluding missing values). On the other hand, HIV-infected women were more likely to have had a previous live-born delivery (58% vs 70%, p < 0.001). Among women with self-reported cigarette smoking data, >44% of the HIV-infected group reported smoking during pregnancy compared to only 15% of the general sample. Over twice as many HIV-infected women had medical/obstetrical risk factors reported on vital statistics (excluding HIV) than the general sample (37% vs 18%, p < 0.001).
Only 37% of the HIV-infected women had an adequate number of prenatal visits according to the Kessner index, compared to 52% of the general sample. Using the APNCU measure, the adequacy rate for prenatal visits increased to 55% of the HIV-infected group and 65% of the general sample, but many women could not be evaluated because of the absence of information on date of their first prenatal visit. Only one-third of both groups had first-trimester care. Delivery by cesarean section was similar in both groups, occurring in 17.6% of the HIV-infected group and in 18% of the general sample (p = 0.78). Six percent of the general sample compared to 16% of the HIV-infected women had <3 months of Medicaid enrollment during pregnancy (p < 0.001).
In both the HIV-infected group and the general sample, African-American women had approximately twice the LBW rate of non-Hispanic whites (Table 1). Also in both groups, illicit drug users and cigarette smokers during pregnancy had higher rates of LBW. Having any medical or obstetrical risk factor (other than HIV) was associated with a higher rate of LBW only in the general sample (p < 0.001). LBW was more strongly associated with an adequate number of prenatal visits by the Kessner index than with receipt of first-trimester care or adequate care by the APNCU index. Consequently, the Kessner index was used in our multivariate models.
The logistic model displayed in Table 2 shows four groups of women categorized by HIV status and presence of other medical/obstetrical risk factors (for example, eclampsia or pregnancy-related hypertension). We created these four categories because of a significant (p < 0.05) interaction between these variables. All reported odds ratios are adjusted for the other factors reported in Table 2. Compared to women in the general population without any medical risk factors, HIV-infected women with or without other medical/obstetrical risk factors had 2.2 times higher adjusted odds of a LBW baby. Among women in the general population, the adjusted odds of LBW were also increased for women with medical risk factors compared to those with none. However, the adjusted odds for women in the general population with medical risk factors were not as high as for HIV-infected women (1.9 vs 2.2). We did not observe an additive effect of HIV and medical risk factors; thus, for women with both HIV and other medical risk factors, the combined effect on LBW was about the same as for HIV alone.
Women with a history of illicit drug use had twofold higher adjusted odds of LBW than non-drugusers and no interaction appeared between drug use and HIV infection. Even after adjustment for many maternal factors, African-Americans still had nearly twofold higher adjusted odds of LBW than non-Hispanic whites. Relatively minor effects are observed for other maternal characteristics with the notable exception of cigarette smoking. Compared to nonsmokers, cigarette smokers and women without information on smoking had an 80% increase in the adjusted odds of LBW. Brief enrollment on Medicaid during pregnancy was associated with a 40% increase in the adjusted odds of LBW, whereas having an adequate number of prenatal visits by the Kessner index was associated with a 46% reduction in the adjusted odds of LBW.
To examine the independent effect of HIV, we also estimated a model including all of the variables reported in Table 2, except we deleted the medical risk factor variable because of the interaction. In this model (not shown), women with HIV still had two fold higher risk of LBW [odds ratio (OR) = 2.04 and 95% confidence interval (CI) = 1.54, 2.69]. The relationships of other variables to the adjusted odds of LBW also did not differ substantially from the model reported on Table 2.
Finally, because preterm delivery is an important risk factor for LBW, we estimated a third model for only full-term deliveries (n = 2,700) to determine whether the association between maternal HIV infection and LBW persisted in this subset. Premature deliveries occurred for 13.6% of all women in the study, but were more common among HIV-infected women than the general sample (25.9% vs 9.6%, p < 0.001). Among full-term deliveries, the LBW rate was 15.2% and 3.9% for the HIV-infected and uninfected women, respectively. The adjusted odds (not shown) of LBW were even higher among HIV-infected women with term deliveries (OR = 2.7; 95% CI = 1.6, 4.3). Increased adjusted odds of LBW persisted for African-Americans compared to non-Hispanic whites (OR = 2.6; 95% CI = 1.4, 4.7), and smokers (OR = 1.5; 95% CI = 0.99, 2.4). Other variables showed similar associations to those reported on Table 2.
Nearly one-third of our large cohort of Medicaid-enrolled, HIV-infected women delivered a LBW infant. This rate was >3 times greater than that observed for our general sample of Medicaid-enrolled women. By comparison, only 7% of all newborns in the United States have been reported as LBW (18). After controlling for illicit drug use, racial/ethnic group, cigarette smoking, adequacy of prenatal care and other maternal factors, HIV-infected women had over twice the odds of delivering a LBW infant than a general sample of socioeconomically similar women who did not have other medical or obstetrical risk factors. Minkoff et al. (3) have reported that the pregnancy outcomes of HIV-infected women appear to be influenced more by maternal drug abuse and smoking behavior than by any direct antepartum viral influence. In our analysis, a substantially higher risk of delivering a LBW baby for HIV-infected women persisted despite adjustment for many other maternal factors associated with LBW. Although our analysis cannot prove causal relationships, it lends further support to African studies suggesting that maternal HIV infection itself may be a contributory factor to this important adverse birth outcome.
Previous analyses addressing this topic in the United States have been conducted on relatively small study populations in selected health care settings. Thus, concerns about the power to detect significant differences and the representativeness of these other study populations are raised. We observed a 15% LBW rate among the 572 HIV-infected women in our study population who delivered full-term infants, compared to only 8% reported by Selwyn et al. (4) in a study of only 39 HIV-infected women. Similarly, Minkoff et al. (3) compared outcomes of only 91 HIV-infected and 126 uninfected women. African studies have included larger samples. For example, Ryder et al. (8) examined 475 HIV-infected and 616 uninfected women from Zaire and found LBW rates of 32% for women with AIDS, 16.9% for HIV-infected women without AIDS, and 9.8% for uninfected women. In a study of 177 HIV-infected and 326 uninfected women from Kenya, Braddick et al. (6) found that the odds of LBW for infected women were 3.0 (95% CI = 1.3, 6.7). Lepage et al. (19) reported a 40% higher rate of LBW among Rwandan infants of HIV-infected women (n = 215), but this rate did not quite achieve statistical significance.
We lacked information on CD4 T-lymphocyte counts to evaluate the impact of maternal immunosuppression on infant LBW and, in this analysis, we also did not have information on infant HIV status. Infants who are HIV-infected may be more likely to be LBW (20). Another limitation relates to our approach to identifying drug users. Although our method of finding drug users classified nearly 90% of cases correctly in a chart review project (15), our approach may be particularly susceptible to missing women who use noninjection drugs. In our HIV-infected cohort, 51% were found to use illicit drugs, similar to the proportion of injection drug users (54%) in a large cohort of pregnant HIV-infected women enrolled in an observational study in New York City (21). But, in that study, an additional 6% of women used crack. Cocaine and methamphetamine are both associated with an increased risk of LBW (22-24). We cannot rule out differential rates of using these drugs by HIV-infected and uninfected women as factors contributing to observed differences in the risk of LBW. Additional potential confounders that we could not adjust for include maternal weight gain during pregnancy, use of specific HIV-related services, and less severe HIV-associated symptoms such as oral candidiasis that are unreliably recorded in claims files.
The demonstrated increased risk of LBW for infants of HIV-infected women may be partially attributable to in utero HIV infection. HIV may result in intrauterine growth delay comparable to that seen with other intrauterine viral infections such as cytomegalovirus (25,26). The time period for the deliveries evaluated in this study preceded both investigational and routine use of zidovudine in HIV-infected pregnant women. The significant results from AIDS Clinical Trial Group Protocol 076 offer the potential to reduce maternal-child transmission of HIV through the use of zidovudine in the antepartum and peripartum periods, with a brief course for the infant (27). Hopefully, reductions in maternal viral load through the use of zidovudine will lead both to a lower risk of transmission and to a lower rate of adverse birth outcomes such as LBW.
Our findings emphasize the need to address other risk factors for LBW in HIV-infected pregnant women. It is well established that illicit drug use during pregnancy leads to higher rates of adverse birth events, with ≈14% of the LBW infants born in this country attributable to maternal drug addiction (28). We did not observe a significant interaction between maternal drug abuse and HIV infection with regard to LBW, indicating that the effects of these two factors are additive. Thus, our data suggest that the combined effect of maternal HIV infection and illicit drug use can produce as much as fivefold increase in the odds of having a LBW baby compared to women without these factors.
Smoking during pregnancy is also a major risk factor for LBW (29). While smoking cigarettes may seem to be a relatively trivial issue for women who are HIV-infected and perhaps even using illicit drugs, our data suggest that smoking in pregnancy presents an important additional threat to the birth weight status of the child. Cigarette smoking by HIV-infected women may also have important implications for HIV transmission. Burns et al. (30) found a threefold increased risk of vertical transmission of HIV among women with low CD4 T-lymphocyte counts (<200/ml) who smoked after the first trimester. Our group has also observed a 45% increase in the adjusted risk of having an infected child associated with cigarette smoking in a cohort of >900 HIV-infected women (31).
Birth weight is highly influenced by socioeconomic and ethnic factors (32-34). We observed an increased risk of LBW for African-American women, consistent with other research (35). Although we were able to adjust for many risk factors for LBW, several other potentially influential factors were not measured, including nutritional status of the mother during pregnancy, alcohol consumption, and prepregnancy weight. Moreover, a recent report by Rawlings et al. (36) concluded that short intervals between pregnancies may also be a risk factor for LBW.
The adequacy of the total number of prenatal visits, as measured by the Kessner index, had an important impact on reducing the risk of LBW for both the HIV-infected and the general cohorts. We did not observe a significant association between LBW and the adequacy of prenatal visits as measured by the APNCU index. This disparity indicates that an index that requires “catch-up” in the number of visits for women who initiate prenatal care later in pregnancy, such as that of Kessner, may be a better predictor of LBW in this population.
Our large population-based study relied on vital statistics records and coded diagnoses available in Medicaid histories. The quality of diagnostic data from secondary databases have been widely evaluated. Several groups have reported that patient coding for serious conditions such as HIV infection, cancer, and myocardial infarction have relatively high rates of accuracy (37-39).
Possibly a small number of women in the general population could have been HIV-infected but lacked any evidence of HIV in their claims histories. Many HIV-infected women are unaware of their serostatus during pregnancy (40). A validation study of our maternal HIV case-finding methodology indicated that the approach was highly specific (100%) but only moderately sensitive (50-75%) (14). Newborn seroprevalence in NYS was estimated in 1989 to be 5.8 per 1,000 deliveries (41). Even if, among Medicaid enrollees, this rate were 1 per 100, we would expect only 24 HIV-infected women in our general sample of 2,377 pregnant women. With a 50% sensitivity for our HIV screen, only ≈12 women in our general sample would have been HIV-infected. This level of misclassification would have had negligible effects on our results.
In summary, our study highlights the challenge of treating HIV-infected pregnant women who often have multiple risk factors associated with LBW and other adverse birth outcomes. We observed a significant independent effect of maternal HIV on the risk of LBW. If this is attributable to a primary viral effect, then the use of zidovudine during pregnancy (27,42) may have a dual benefit of reducing the risk of LBW and HIV infection in the child. However, an increased risk of LBW was also associated with maternal use of illicit substances and cigarette smoking by both HIV-infected and uninfected women. Prenatal care services are central to any effort to address these behaviors and other risks for LBW. Indeed, improvements in birth weight have been associated with receipt of prenatal care among at-risk population groups (33,43,44). Unfortunately, we found serious deficiencies in the adequacy of the number of prenatal visits, especially for the HIV-infected women in our study population. These deficiencies must be addressed before the high rate of LBW observed in HIV-infected women can be reduced.
Acknowledgment: This work was supported by the National Institute on Drug Abuse (ROI DA07904).
1. Healthy people 2000: National health promotion and disease prevention objectives. Washington, DC: US Department of Health and Human Services, 1990; DHHS publication no. (PHS) 91-50212.
2. Shiono PH, Behrman RE. Low birth weight: analysis and recommendations. Future Child
3. Minkoff HL, Henderson C, Mendez H, et al. Pregnancy outcomes among mothers infected with human immunodeficiency virus and uninfected control subjects. Am J Obstet Gynecol
4. Selwyn PA, Schoenbaum EE, Davenny K, et al. Prospective study of human immunodeficiency virus infection and pregnancy outcomes in intravenous drug users. JAMA
5. Anonymous. Perinatal findings in children born to HIV-infected mothers. Br J Obstet Gynaecol
6. Braddick MR, Kreiss JK, Embree JE, et al. Impact of maternal HIV infection on obstetrical and early neonatal outcome. AIDS
7. Temmerman M, Plummer FA, Mirza NB, et al. Infection with HIV as a risk factor for adverse obstetrical outcome. AIDS
8. Ryder RW, Nsa W, Hassig SE, et al. Perinatal transmission of the human immunodeficiency virus type 1 to infants of seropositive women in Zaire. N Engl J Med
9. Green J, Arno PS. The `Medicaidization' of AIDS: trends in financing of HIV-related medical care. JAMA
10. Hellinger FJ. The lifetime cost of treating a person with HIV. JAMA
11. Fanning TR, Cosler LE, Gallagher P, et al. The epidemiology of AIDS in the New York and California Medicaid programs. J Acquir Immune Defic Syndr
12. AIDS surveillance quarterly update. Albany, NY: Bureau of HIV/AIDS Epidemiology, New York State Department of Health, December 1994.
13. Centers for Disease Control and Prevention. Revision of the CDC surveillance case definition for acquired immunodeficiency syndrome. MMWR
14. Turner BJ, Markson L, Hauck W, Cocroft J, Fanning T. Prenatal care of HIV-infected women: analysis of a large New York State cohort. J Acquir Immune Defic Syndr
15. Fanning TR, Turner BJ, Cosler LE, et al. Quality of Medicaid data for HIV/AIDS research: examination of a statewide database. AIDS Public Policy J
16. Kessner DM. Infant death: an analysis by maternal risk and health care
. Washington, DC: National Academy, 1973.
17. Kotelchuck M. An evaluation of the Kessner adequacy of prenatal care index and a proposed adequacy of care utilization index. Am J Public Health
18. United States Bureau of the Census. Statistical abstracts of the United States
. 110th ed. Washington, DC: Government Printing Office, 1990.
19. Lepage P, Dabis F, Hitimana D, et al. Perinatal transmission of HIV-1: lack of impact of maternal HIV infection on characteristics of live births and on neonatal mortality in Kigali, Rwanda. AIDS
20. Minkoff H, Mofenson LM. The role of obstetric interventions in the prevention of pediatric human immunodeficiency virus infection. Am J Obstet Gynecol
21. Thomas PA, Weedon J, Krasinski K, et al. Maternal predictors of perinatal human immunodeficiency virus transmission. Pediatr Infect Dis
22. Neerhof MG, MacGregor SN, Retsky SS, Sullivan TP. Cocaine abuse during pregnancy: peripartum prevalence and perinatal outcome. Am J Obstet Gynecol
23. Gillogley KM, Evans AT, Hansen RL, Samuels SJ, Batra KK. The perinatal impact of cocaine, amphetamine, and opiate use detected by universal intrapartum screening. Am J Obstet Gynecol
24. Spence MR, Williams R, DiGregorio GJ, Dirby-McDonnell A, Polansky M. The relationship between recent cocaine use and pregnancy outcome. Obstet Gynecol
25. Galask RP, Larsen B, Ohm MJ. Infection in maternal-fetal medicine. In: Sciarra JJ ed. Gynecology and obstetrics
. Philadelphia: JB Lippincott, 1993:1-18.
26. Alford CA Jr, Stagno S, Reynolds DW. Perinatal infections caused by viruses, toxoplasma, and treponema pallidun. In: Aladiem S, Brown AK, eds. Clinical perinatology
. St Louis: CV Mosby, 1974:183-204.
27. Conner EM, Sperling RS, Gelber R, et al. Reduction of maternal-infant transmission of human immunodeficiency virus type 1 with zidovudine treatment. N Engl J Med
28. Mittendorf R, Herschel MA, Williams MA, Hibbard JU, Moawad AH, Lee K. Reducing the frequency of low birth weight in the United States. Obstet Gynecol
29. Lieberman E, Gremy I, Lang JM, Cohen AP. Low birth-weight at term and the timing of fetal exposure to maternal smoking. Am J Public Health
30. Burns DN, Landesman S, Muenz LR, et al. Cigarette smoking, premature rupture of membranes, and vertical transmission of HIV-1 among women with low CD4+ levels. J Acquir Immune Defic Syndr
31. Turner BJ, Markson LE, Hauck WW, Fanning TR. Dangers of smoking: association with maternal-child HIV transmission. Presented at 123rd Annual Meeting of the American Public Health Association, San Diego, CA, October 1995.
32. Mustard CA, Roos NP. The relationship of prenatal care and pregnancy complications to birthweight in Winnipeg, Canada. Am J Public Health
33. Singh GK, Yu SM. Birthweight differentials among Asian-Americans. Am J Public Health
34. Poland ML, Ager JW, Olson KL, Sokol RJ. Quality of prenatal care: selected social, behavioral, and biomedical factors, and birth weight. Obstet Gynecol
35. McCormick MC. The contribution of low birth weight to infant mortality and childhood morbidity. N Engl J Med
36. Rawlings JS, Rawlings VB, Read JA. Prevalence of low birth weight and preterm delivery in relation to the interval between pregnancies among white and black women. N Engl J Med
37. Rosenblum L, Buehler JW, Morgan MW, et al. HIV infection in hospitalized patients and Medicaid enrollees: the accuracy of medical record coding. Am J Public Health
38. Fisher ES, Whaley FS, Krushat WM, et al. The accuracy of Medicare's hospital claims data: progress has been made but problems remain. Am J Public Health
39. Meehan TP, Hennen J, Radford MJ, Petrillo MK, Elstein P, Ballard DJ. Process and outcome of care for acute myocardial infarction among Medicare beneficiaries in Connecticut: a quality improvement demonstration project. Ann Intern Med
40. Centers for Disease Control and Prevention. U.S. Public Health Service recommendations for human immunodeficiency virus counseling and voluntary testing for pregnant women. MMWR
41. Ellerbrock TV, Buch TJ, Chamberland ME, Oxtoby MJ. Epidemiology of women with AIDS in the United States, 1981 through 1990. JAMA
42. Boyer PJ, Dillon M, Navaie M, et al. Factors predictive of maternal-fetal transmission of HIV-1. JAMA
43. Racine A, Joyce T, Anderson R. The association between prenatal care and birth weight among women exposed to cocaine in New York City. JAMA
44. Hulsey TC, Patrick CH, Alexander GR, Ebeling M. Prenatal care and prematurity: is there an association in uncomplicated pregnancies? Birth