Increasing impact of HIV infection on hospitalizations in the United States, 1983-1988.

During 1983-1988, hospitalizations of patients with a diagnosis of human immunodeficiency virus (HIV) infection increased from 1.3 to 33.7 per 100,000 persons. We used the National Hospital Discharge Survey, which is based on a representative sample of discharges from nonfederal short-stay hospitals, to describe illnesses among hospitalized patients with HIV infection. Of 222,200 such hospitalizations during 1983-1988, most occurred among persons who were 25-44 years of age (79%), white (66%), and male (90%). Among men 25-44 years of age, HIV admissions increased from 8.5 to 148.6 per 100,000 persons during 1983-1988; among black men 25-44 years of age, HIV hospitalizations increased from 43.1 to 387.4 per 100,000 persons. Among women, hospitalizations increased 3.4-fold. Frequently listed illnesses in the Centers for Disease Control (CDC) AIDS case definition were Pneumocystis carinii pneumonia (30%), candidiasis (20%), and Kaposi's sarcoma (13%). Other frequently listed illnesses included infections (39%) such as pneumonia, sepsis, and urinary tract infections; blood dyscrasias (30%) such as anemia, thrombocytopenia, and agranulocytosis; metabolic (17%), gastrointestinal (16%), and respiratory disorders (12%); and drug abuse (9%). These data provide a minimum estimate of HIV hospitalizations because for some patients HIV infection may not be specified on the discharge record. HIV hospitalizations are increasing markedly and are associated with a broad spectrum of severe morbidity.

The completeness of AIDS surveillance.

AIDS surveillance data are widely used in setting HIV intervention policies, and the effectiveness of these data depend on their completeness. We reviewed studies conducted by state and local health departments on the completeness of AIDS reporting. These studies identified AIDS cases through alternate data sources, such as death certificates, hospital discharge records, disease registries, or medication records. In most instances greater than 80% of AIDS cases detected through these studies had been reported, although lower levels of reporting were found in some outpatient settings. A comparison of vital records and AIDS surveillance confirmed that AIDS surveillance is identifying 70-90% of all HIV-related deaths in men 25-44 years of age. Historically, AIDS surveillance has emphasized reporting from hospitals. Efforts to maintain current levels of reporting, or to improve reporting, are challenged by the growth of the epidemic and by the increasing role of outpatient diagnosis of AIDS.

Impact of the 1987 revision of the case definition of acquired immune deficiency syndrome in the United States.

To measure the impact of the 1987 expansion of the definition of acquired immune deficiency syndrome on the number and characteristics of cases in the United States, we reviewed the 28,920 cases diagnosed since the revision and reported through 1988. The proportion meeting only new criteria was 28% overall, with a range of 0 to 82% in different states and territories. This proportion was 26% in the last quarter of 1987 and increased to 31% in the last quarter of 1988. It was higher in heterosexual intravenous drug abusers (IVDAs) (43%) and lower in homosexual male non-IVDAs (21%) than in other groups. The new criteria, but not earlier (pre-1985) criteria, generally require a positive test for human immunodeficiency virus (HIV) infection; however, use of HIV testing varies among states, as demonstrated by differences in the percentage of pre-1985-criteria cases with a reported HIV test (39 to greater than 95%). The revision has changed the distribution of characteristics of cases (e.g., heterosexual IVDAs composed 18% of cases meeting old criteria, 35% of cases meeting only new criteria, and 23% of all cases). Interpretation of trends in both the number and characteristics of cases should take into account the variable impact of the revision on reporting.

HIV-related deaths from selected infectious diseases among persons without AIDS in New Jersey.

This study sought to quantify HIV-related deaths among persons not classified as having AIDS, in an area where AIDS incidence among injecting drug users (IDUs) is high. Death certificates of persons who were aged 25-44 years at death, died in 1987 in two New Jersey counties, and had certain infectious conditions were compared with names in the AIDS Registry. Hospital and/or Medical Examiner records were reviewed for nonmatching cases. Cases were considered as confirmed HIV infection if there was laboratory evidence of such infection and as suggestive HIV infection if the decedent had oral thrush or a combination of certain other clinical findings were present. Of 412 deaths meeting the above criteria, 165 (40.0%) were in the AIDS Registry. We investigated 205 of the remainder; of these, 7.3% were found to have AIDS, 21.5% had confirmed HIV infection without AIDS, and 15.1% had suggestive HIV infection. This increased the HIV-related mortality in excess of deaths due to AIDS in this age group by 9.2% for confirmed HIV infections and 15.6% for both confirmed and suggestive HIV infections, with deaths among IDUs increasing 12.3% for confirmed HIV infections and 18.9% for both confirmed and suggestive HIV infections. Thus, in addition to AIDS indicator diseases, a variety of other infectious conditions can lead to death in HIV-infected persons, particularly in IDUs; however, the extent of such deaths may be less than previously described.

Epidemiology of progressive multifocal leukoencephalopathy in the United States: analysis of national mortality and AIDS surveillance data.

We analyzed progressive multifocal leukoencephalopathy (PML) mortality data from 1979 to 1987 and data on persons with acquired immunodeficiency syndrome (AIDS) reported to the Centers for Disease Control (CDC). Based on analyses of multiple-cause-of-death vital statistics, deaths related to PML have increased fourfold from 1.5/10,000,000 persons in 1979 to 6.1/10,000,000 persons in 1987. The increase in the PML annual death rate began in 1984, occurred primarily in men 20 to 49 years of age, and was greatest in states known to have a high incidence of AIDS. In 1987, 56% of death certificates that listed PML as a cause of death also listed human immunodeficiency virus (HIV) infection. Analysis of AIDS case reports to the CDC from 1981 through June 1990 demonstrated that 0.72% of persons with AIDS were reported as having PML. Although most persons with AIDS who had PML were 20 to 49 years of age (84.6%), PML was reported more frequently among persons with AIDS greater than or equal to 50 years old than less than 50 years old. In addition, PML was reported more frequently among persons with AIDS who were exposed to HIV by blood transfusion than those in all other exposure categories. These data demonstrate that the increase in PML mortality from 1979 to 1987 was associated with the large increase in immunosuppressed persons with AIDS.

Impact of the human immunodeficiency virus epidemic on mortality in children, United States.

To assess the effect of the human immunodeficiency virus (HIV) epidemic on mortality in US children younger than 15 years of age and to identify associated causes of death, the authors examined final national mortality statistics for 1988, the most recent year for which such data are available. In 1988, there were 249 deaths attributed to HIV/acquired immunodeficiency syndrome (AIDS) in children younger than 15 years of age. Associated causes of death listed most frequently on 270 death certificates with any mention of HIV/AIDS included conditions within the AIDS surveillance case definition (30%), pneumonia (excluding Pneumocystis carinii pneumonia) (17%), septicemia (10%), and noninfectious respiratory diseases (8%). The impact of HIV/AIDS as a cause of death was most striking in the 1-through 4-year-old age group and in black and Hispanic children, particularly in the Northeast. By 1988 in New York State, HIV/AIDS was the first and second leading cause of death in Hispanic and black children 1 through 4 years of age, accounting for 15% and 16%, respectively, of all deaths in these age-race groups. With an estimated 1500 to 2000 HIV-infected children born in 1989, the impact of HIV on mortality in children will become more severe.

Public health surveillance of non-infectious chronic diseases: the potential to detect rapid changes in disease burden.

The usefulness of surveillance in relating chronic disease trends to recent changes in risk exposures is often questioned on the grounds that these trends respond slowly, reflecting long periods between aetiological exposures and clinical onset of disease. We challenge this preconception on the basis of a review of several important risk factors and diseases: alcohol and liver cirrhosis; tobacco and stroke, cardiovascular disease, and lung cancer; and oestrogens and endometrial cancer. Data from cohort, cross-sectional, and modelling studies demonstrate that the time between removal of exposures and the onset of decline in morbidity or mortality is not defined by the time between initial exposure and disease occurrence. Rather, the pattern of lifetime exposures (with recent exposures often having a dominant effect), the dynamics of the disease process, and the segment of the population with reduced exposures determine how soon the decline begins.

Teenagers at risk of human immunodeficiency virus type 1 infection. Results from seroprevalence surveys in the United States.

OBJECTIVE: To describe the seroprevalence of human immunodeficiency virus type 1 (HIV-1) and risk factors for HIV-1 infection among teenagers attending selected clinics. DESIGN: Anonymous, cross-sectional serosurveys conducted in 130 clinics in 24 cities. SETTINGS: Adolescent medicine clinics, sexually transmitted disease clinics, clinics in juvenile detention and correctional facilities, and homeless and runaway youth centers. PATIENTS: Teenagers in whom serum samples were drawn as part of routine medical services. MAIN OUTCOME MEASURES: Prevalence of HIV-1 infection and reported HIV risk behaviors. RESULTS: From January 1, 1990 through December 31, 1992, serum specimens were collected from 79,802 teenagers; 591 of these specimens were positive for HIV-1 antibody. Seropositive test results were found in all 24 cities surveyed, and in 95 (73%) of the 130 clinics surveyed. The median clinic-specific prevalence was 0.2% (range, 0% to 1.4%) in 22 adolescent medicine clinics, 0.3% (range, 0% to 6.8%) in 33 correctional facilities, 0.5% (range, 0% to 3.5%) in 70 sexually transmitted disease clinics, and 1.1% (range, 0% to 4.1%) in five homeless youth centers. Rates exceeded 1% in 37 sites (28%). Excluding sites with many men reporting sex with men, rates in women were similar or somewhat higher than rates in men. Rates were highest among young men reporting sex with men, with clinic rates ranging from 16% to 17% in two homeless youth sites and 13% to 17% in two sexually transmitted disease clinics. Most teenagers with risk information reported heterosexual activity as their only potential risk exposure to HIV-1. CONCLUSIONS: Seroprevalence of HIV was generally low but varied by type of clinic and geographic area. The highest rates were observed among young women and gay men in some settings, suggesting that targeted prevention messages are needed.

Older maternal age and infant mortality in the United States.

We used data from the National Infant Mortality Surveillance project to examine the effect of older maternal age on infant mortality for the 1980 United States birth cohort. The 1,579,854 births and 14,591 deaths of singletons who were black or white and whose mothers were 25-49 years of age were included. Direct standardization was used to calculate birth-weight-adjusted relative risks of neonatal and postneonatal mortality, using the birth weights of infants with maternal age 25-29 as the standard. We found that the risk of infant mortality was nearly equal for infants born to mothers 25-29 and 30-34 years of age; infants born to mothers 35-39 years of age were at a slightly elevated (18% higher) risk, and those born to mothers 40-49 years of age were at a much more elevated (69% higher) risk. Among whites, the higher neonatal mortality associated with a maternal age of 35-39 was mostly due to an increased prevalence of low birth weight; among blacks, it was due to higher birth-weight-specific risks. Neither white nor black postneonatal mortality risks were much elevated until a maternal age of 40-49, and this last elevation was mostly due to higher birth-weight-specific risks. These findings suggest that infertility and fetal mortality aside, and considering only the effect on infant mortality, it is relatively safe for women to postpone childbearing into their middle, and perhaps late, thirties.

The role of linked birth and infant death certificates in maternal and child health epidemiology in the United States.

Linked birth and infant death certificates allow measurement of birthweight-specific infant mortality. Jack Smith, MS, to whom this issue of the American Journal of Preventive Medicine is dedicated, played a key role in the National Infant Mortality Surveillance (NIMS) project. NIMS provided national data on birthweight-specific infant mortality for the 1980 birth cohort, updated data previously collected by the National Center for Health Statistics (NCHS) for the 1960 birth cohort, and supported NCHS's implementation of an annual linked file in 1983. NIMS illustrated themes in infant mortality that remain important: the role of low birthweight (LBW) as a contributor to infant mortality, the contribution of disparities in LBW and birthweight-specific mortality to black-white gaps in infant mortality, and the nation's greater success in reducing mortality among LBW infants than in preventing LBW. Linked birth and infant death records are used nationally and by states to study an array of maternal and infant health topics, from the quality of vital records to the impact of therapeutic and public health interventions. By supplementing birth and infant death records with linkages to program and hospital discharge data, epidemiologists and health service researchers are extending the utility of vital statistics data to monitor maternal and infant health.

Socioeconomic differences among people with AIDS: results from a Multistate Surveillance Project.

To characterize the socioeconomic status of persons with acquired immunodeficiency syndrome (AIDS), 11 U.S. state and city health departments interviewed 2,898 persons > or = 18 years of age reported with AIDS between June 1, 1990, and January 31, 1993. Among men who have sex with men, white men reported the lowest percentage (9%), and Central/South American (50%) and Mexican men (40%) reported the highest percentages not completing 12 years of school. Among intravenous drug users (IDUs), 35% of white men, 64% of black men, 67% of Puerto Rican men, 29% of white women, and 63% of black women had not completed 12 years of school. Overall, 77% of the men and 90% of the women were unemployed; we also found racial/ethnic differences by employment but to a lesser degree than differences in education. Among women, but not among men, differences in household income by race and ethnicity were marked; 76% of white and 91% of black female IDUs reported a household income of $10,000. Human immunodeficiency virus (HIV) prevention programs must be targeted toward the educational level of the populations served, and HIV services must adapt to the financial circumstances of their clientele.

The supplement to HIV-AIDS Surveillance project: an approach for monitoring HIV risk behaviors.

A variety of surveillance methods are used to characterize the epidemic of HIV infection and AIDS. Such surveillance includes AIDS case reporting, reporting of diagnosed HIV infections, and HIV seroprevalence surveys among targeted sentinel populations. The need for additional surveillance systems to monitor HIV-related risk behaviors has been increasingly evident. One approach to behavioral surveillance, the CDC's Supplement to HIV-AIDS Surveillance project, uses the infrastructure of HIV infection and AIDS case reporting to collect additional information on risk behaviors among HIV-infected persons, who by definition represent those at highest risk.

The reporting of race and ethnicity in the National Notifiable Diseases Surveillance System.

The authors used 1987 data from the Epidemiologic Surveillance Project (ESP) of the Centers for Disease Control to examine the completeness of race-ethnicity reporting in the National Notifiable Diseases Surveillance System. And, to the extent possible, they used ESP to assess racial and ethnic disparities in the occurrence of selected notifiable infectious diseases. For the 30 reporting areas (29 States and the District of Columbia) that provided data to ESP for all of calendar year 1987, approximately 60 percent of case reports were accompanied by specified race-ethnicity for affected persons. This percentage varied widely by disease and State. In general, non-Hispanic whites had morbidity rates (cases per 100,000 population per year) that were among the lowest compared with rates for other groups, and Native Americans commonly had rates that were among the highest. The ranking of morbidity rates among blacks, Hispanics, and Asians and Pacific Islanders varied by disease, although the last group had strikingly higher rates for malaria and tuberculosis. The age distribution of persons with cases was often lower among minority groups than among non-Hispanic whites, but the authors were unable to calculate age-specific or age-adjusted rates. Potential biases that limit interpretation of the findings are reviewed. Efforts to eliminate racial-ethnic disparities in the occurrence of infectious diseases would be aided by effective surveillance data. For the ESP to meet its potential in this regard, however, substantial improvements in the reporting of race-ethnicity for notifiable diseases are needed.

Overview of the National Infant Mortality Surveillance (NIMS) project--design, methods, results.

The recent slowdown in the decline of infant mortality in the United States and the continued high risk of death among black infants (twice that of white infants) prompted a consortium of Public Health Service agencies to collaborate with all States in the development of a national data base from linked birth and infant death certificates. This National Infant Mortality Surveillance (NIMS) project for the 1980 U.S. birth cohort provides neonatal, postneonatal, and infant mortality risks for blacks, whites, and all races in 12 categories of birth weights. (Note: Neonatal mortality risk = number of deaths to infants less than 28 days of life per 1,000 live births; postneonatal mortality risk = number of deaths to infants 28 days to less than 1 year of life per 1,000 neonatal survivors; and infant mortality risk = number of deaths to infants less than 1 year of life per 1,000 live births.) Separate tabulations were requested for infants born in single and multiple deliveries. For single-delivery births, tabulations included birth weight, age at death, race of infant, and each of these characteristics: infant's live-birth order, sex, gestation, type of delivery, and cause of death; and mother's age, education, prenatal care history, and number of prior fetal losses at 20 weeks' or more gestation. An estimated 95 percent of eligible deaths were included in the NIMS tabulations. The analyses focus on three components of infant mortality: birth weight distribution of live births, neonatal mortality, and postneonatal mortality. The most important predictor for infant survival was birth weight, with an exponential improvement in survival by increasing birth weight to its optimum level. The nearly twofold higher risk of infant mortality among blacks was related to a higher prevalence of low birth weights and to higher mortality risks in the neonatal period for infants weighing 3,000 grams or more, and in the postneonatal period for all infants, regardless of birth weight. Regardless of other infant or maternal risk factors, the black-white gap persisted for infants weighing 2,500 grams or more.

Regional differences in birth weight-specific infant mortality, United States, 1980.

To describe regional differences in birth weight-specific infant mortality in the United States, we used data from the National Infant Mortality Surveillance project. The infant mortality risk (IMR) for the nation was 11.0 deaths per 1,000 live births. The risk (with 95 percent confidence intervals [CI]) for the four U.S. Census regions were West 9.9 (9.7 to 10.1), Northeast 10.4 (10.1 to 10.6), North Central 10.8 (10.6 to 11.0), and South 12.1 (11.9 to 12.3). In all regions, the IMR for blacks was approximately twice that of whites. Seventy-two percent of the higher IMR in the South was due to a higher proportion of black births compared with the remainder of the nation, reflecting the higher mortality rates suffered by black infants, and 28 percent to higher mortality among southern whites. The IMR for whites in the South was significantly higher than in the remainder of the nation: 9.8 versus 9.1 (relative risk = 1.09, CI = 1.06 to 1.11). Thirty-six percent of this excess in IMR was due to a higher frequency of low birth weight (less than 2,500 grams), 18 percent was due to higher IMR in infants with birth weight less than 2,500 grams, and 46 percent due to higher IMR in infants with birth weights of 2,500 g or more. Black infants born in the West had a lower risk of death than black infants in the other regions. When compared with the Northeast and South, 36 percent of the lower risk in the West among black infants was due to a lower frequency of low birth weight, 38 percent due to lower IMR in infants w'ith birth weight less than 2,500 g, and 26 percent to lower IMR in infants with birth weight of 2,500 g or more.

Birth weight-specific infant mortality, United States, 1960 and 1980.

National statistics on the risk of infant mortality by birth weight were collected most recently in 1980 and 1960. (Infant mortality risk is the number of deaths of infants under 1 year of age per 1,000 live births.) In this 20-year period, the infant mortality risk (IMR) for single-delivery infants declined 53 percent, from 23.3 deaths per 1,000 live births to 11.0; 91 percent of this decline was due to lower IMRs within birth weight categories, and 9 percent was due to reduced frequency of low birth weight. The greatest reduction in neonatal mortality (under 28 days)--73 percent--occurred among infants of 1,500-1,999 grams (g) birth weight, whereas the greatest reductions in postneonatal mortality (28 days to under 1 year)--51 percent to 54 percent--occurred among infants of 3,500 g or more birth weight. Trends in IMR for black and white infants were similar, and the twofold gap between the races in IMR persisted from 1960 to 1980. For whites, reductions in the frequency of low birth weights contributed to the decline in the IMR. For blacks, the percentage of infants with birth weights of less than 1,500 g increased, and the total reduction in the IMR was attributable to lower birth weight-specific mortality risks. In some regions of the United States, failure to observe an increase in birth weight for blacks may be a reporting artifact, reflecting improved reporting of births of very small black infants in 1980. Examination of changes in perinatal mortality risks (from 20 weeks gestation to less than 28 days of life) did not suggest that infant mortality trends were substantially affected by changes in the distinction between fetal and neonatal deaths over the 20-year period. Reducing the number of low birth weight infants remains the greatest potential for future reductions in infant mortality.

Birth weight-specific causes of infant mortality, United States, 1980.

To describe underlying causes of infant death by birth weight, we used data from the 1980 National Infant Mortality Surveillance project and aggregated International Classification of Diseases codes into seven categories: perinatal conditions, infections, congenital anomalies, injuries, sudden infant death syndrome (SIDS), other known causes, and nonspecific or unknown causes. Compared with heavier infants, infants with birth weights of 500-2,499 grams (g) are at increased risk of both neonatal and postneonatal death for virtually all causes. Sixty-two percent of neonatal deaths (under 28 days of life) were attributed to "conditions arising in the perinatal period," as defined using codes from the International Classification of Diseases. Prematurity-low birth weight and respiratory distress syndrome (RDS) were the leading causes of such deaths among infants with birth weights of 500-2,499 g, while birth trauma-hypoxia-asphyxia and other perinatal respiratory conditions were the leading causes among heavier infants. For all birth weight groups, congenital anomalies were the second leading cause, representing 27 percent of neonatal deaths. Although perinatal conditions caused nearly one-third of postneonatal deaths (28 days to under 1 year of life) among infants with birth weights of 500-1,499 g, for the other birth weight groups these conditions were much less important; predominant causes of postneonatal death were sudden infant death syndrome (SIDS), congenital anomalies, infections, and injuries. Black infants had a roughly twofold higher risk of neonatal and postneonatal death than did white infants for all causes except congenital anomalies, which occurred with almost equal frequency in blacks and whites. However, for infants with birth weights of 500-2,499 g, blacks had lower risks of neonatal death from RDS and congenital anomalies. Between 1960 (the latest year for which national birth weight-specific mortality statistics had been available) and 1980, SIDS emerged as a major diagnostic rubric. Otherwise, except for infections and congenital anomalies among infants with birth weights of 500-1,499 g, all causes of death declined in frequency among all birth weight groups.

Birth weight-specific infant mortality due to congenital anomalies, 1960 and 1980.

The impact of mortality due to congenital anomalies in single-delivery births was compared in 1960 and 1980 birth cohorts; data were used from the 1960 National Center for Health Statistics national linkage of birth and death certificates and the 1980 National Infant Mortality Surveillance project. In 1960 there were 14,714 deaths due to congenital anomalies, compared with 8,674 in 1980, a 41 percent reduction. The infant mortality risk (IMR) due to congenital anomalies fell 31 percent. This is in contrast with the observed 54 percent decline in IMR due to all causes. This reduction in mortality due to congenital anomalies occurred for both whites and blacks in the postneonatal period and for whites only in the neonatal period. Changes ranged from a 1.8 percent increase for the black neonatal mortality risk to a 46.6 percent decrease for the white postneonatal mortality risk. In spite of these relative reductions, the absolute percentage of all infant deaths due to congenital anomalies had increased from 15.8 percent in 1960 to 24.1 percent in 1980. Two categories, cardiovascular and central nervous system anomalies, accounted for 72 percent of infant deaths due to congenital anomalies in 1960 and for 59 percent in 1980; cardiovascular anomalies accounted for 48 percent of all deaths due to congenital anomalies in 1960 and 40 percent in 1980. Infant mortality risks in the United States showed a 2:1 black to white ratio in both 1960 and 1980. However, for infant mortality due to congenital anomalies, the black and white mortality risks were approximately equal in both 1960 and 1980. For infants with birth weights of 500-2,499 g, the risk of neonatal mortality for blacks was less than half the risk for whites.

Variation in state-specific infant mortality risks.

Data from the National Infant Mortality Surveillance project were used to examine the State-specific variations in infant, neonatal, and postneonatal mortality and to examine some of the factors affecting the risks of death. The infant mortality risk, defined as the risk of death before 1 year of age to an infant born in the 1980 birth cohort, in the highest risk State was nearly three times that in the lowest risk State. Mortality risk ratios of two or greater were found when comparing high and low States for overall black infant mortality risks, overall neonatal mortality, neonatal mortality risks for black and white infants examined separately, neonatal mortality risks for low birth weight infants regardless of race, and overall postneonatal mortality and postneonatal mortality for white infants. The lowest State-specific black mortality risks were higher than the highest white risks for overall infant mortality and neonatal mortality. The differences between State extremes in mortality risks are greater than the differences between the United States and the Scandinavian countries with the lowest infant mortality.

Differences in neonatal and postneonatal mortality by race, birth weight, and gestational age.

In recent decades, neonatal and postneonatal mortality rates have declined overall in the United States. Yet, the mortality rates for black infants continue to be approximately twice those for white infants. With the use of data from 45 of the 53 vital statistics reporting areas that participated in the 1980 National Infant Mortality Surveillance project, we extended previous State analyses to describe differences, nationally, in neonatal and postneonatal mortality risks for black and white infants according to gestational age and birth weight. After restricting our analysis to single-delivery infants with known and plausible combinations of gestational age of 26 or more weeks and birth weights of 500 grams (g) or more, the neonatal mortality risk (NMR)--that is, the number of deaths to infants less than 28 days of life per 1,000 live births--for black infants was 1.6 times higher than the NMR for whites. This difference was largely explained by two findings: First, although the NMR was lower for black than for white infants with gestational ages of less than 38 weeks and birth weights less than 3,000 g, that advantage was heavily outweighed by the higher percentage of such births among blacks, accounting for roughly two-thirds of the overall difference in NMR between blacks and whites. Second, most of the remaining difference in NMR was accounted for by higher NMRs among black infants with gestational ages of 38 or more weeks and birth weights of 3,000 g or more. A comparison of the lowest mortality risk for any combination of birth weight and gestational age showed that the black NMR was 1.89 times higher than the white NMR. The postneonatal mortality risk (PNMR)--PNMR equals the number of deaths to infants 28 days to less than 1 year of life per 1,000 neonatal survivors--for black infants was 2.09 times the PNMR for white infants. Black infants had higher PNMRs than white infants for nearly all combinations of birth weight and gestational age. Higher PNMRs among infants with gestational ages of 38 or more weeks and birth weights of 2,500 g or more accounted for 43 percent of the difference in PNMR between black infants and white infants. Eliminating the U.S. black-white infant mortality disparity will require not only reducing the higher frequency of prematurity and low birth weight among black infants, but also improving the survival during both the neonatal and postneonatal periods of term black infants with normal birth weights.