The completeness, validity, and timeliness of AIDS surveillance data

PURPOSE

To assess the completeness, validity, and timeliness of the AIDS surveillance system after the 1993 change in the surveillance case definition.

METHODS

To assess completeness of AIDS case reporting, three study sites conducted a comparison of their AIDS surveillance registries with an independent source of information. To evaluate validity, the same sites conducted record reviews on a sample of reported AIDS cases, we then compared agreement between the original report and the record review for sex, race, and mode of transmission. To evaluate timeliness, we calculated the median delay from time of diagnosis to case report, before and after the change in case definition, in each of the three study sites.

RESULTS

After expansion of the case definition, completeness of AIDS case reporting in hospitals (> or = 93%) and outpatient settings (> or = 90%) was high. Agreement between the information provided on the original case report and the medical record was > 98% for sex, > 83% for each race/ethnicity group; and > 67% for each risk group. The median reporting delay after the change was four months, but varied by site from three to six months.

CONCLUSIONS

The completeness, validity, and timeliness of the AIDS surveillance system remains high after the 1993 change in the surveillance case definition. These findings might be useful for programs implementing integrated HIV and AIDS surveillance systems.

HIV in the United States at the turn of the century: an epidemic in transition

OBJECTIVES

The current status of and changes in the HIV epidemic in the United States are described.

METHODS

Surveillance data were used to evaluate time trends in AIDS diagnoses and deaths. Estimates of HIV incidence were derived from studies done during the 1990s; time trends in recent HIV incidence were inferred from HIV diagnoses and seroprevalence rates among young persons.

RESULTS

Numbers of deaths and AIDS diagnoses decreased dramatically during 1996 and 1997 but stabilized or declined only slightly during 1998 and 1999. Proportional decreases were smallest among African American women, women in the South, and persons infected through heterosexual contact, HIV incidence has been roughly constant since 1992 in most populations with time trend data, remains highest among men who have sex with men and injection drug users, and typically is higher among African Americans than other racial/ethnic groups.

CONCLUSIONS

The epidemic increasingly affects women minorities, persons infected through heterosexual contact, and the poor. Renewed interest and investment in HIV and AIDS surveillance and surveillance of behaviors associated with HIV transmission are essential to direct resources for prevention to populations with greatest need and to evaluate intervention programs.

A method for classification of HIV exposure category for women without HIV risk information

An increasing number of cases of human immunodeficiency virus (HIV) and acquired immunodeficiency syndrome (AIDS) among women is reported to state and territorial health departments without exposure risk information (i.e., no documented exposure to HIV through any of the recognized routes of HIV transmission). Because surveillance data are used to plan prevention and other services for HIV-infected persons, developing methods to accurately estimate exposure risk for HIV and AIDS cases initially reported without risk information and assisting states to analyze and interpret trends in the HIV epidemic by exposure risk category is important. In this report, a classification model using discriminant function analysis is described. The purpose of the classification model is to develop a proportionate distribution of exposure risk category for cases among women reported without risk information. The distribution was estimated based on behavioral and demographic data obtained from interviews with HIV-infected women; the interviews were conducted in 12 states during 1993-1996. Variables used in the analysis were alcohol abuse, noninjection-drug use, and crack use; year of HIV/AIDS diagnosis; age; employment; and region. As a result of the classification procedure, nearly all cases among women with no reported risk were classified into an exposure risk category: 81%, heterosexual contact; and 16%, injection-drug use. These proportions are higher than the current redistribution fractions (calculated from risk reclassification patterns and weighted by demographic characteristics) and reflect the increasing proportion of cases among women attributable to heterosexual contact with an infected partner. This report provides one method that could be applied to HIV surveillance data at the national level to estimate the proportion of cases in exposure risk categories. However, because the study in this report is limited in sample size and geographic representativeness, other models are also needed for adjusting risk exposure data at the national, state, and local levels.

Survival after AIDS diagnosis in adolescents and adults during the treatment era, United States, 1984-1997

CONTEXT

Declines in the number of acquired immunodeficiency syndrome (AIDS) deaths were first observed in 1996, attributed to improvements in antiretroviral therapy and an increase in the proportion of persons receiving therapy.

OBJECTIVE

To examine national trends in survival time among persons diagnosed as having AIDS in 1984-1997.

DESIGN, SETTING, AND SUBJECTS

Retrospective cohort study using data from a population-based registry of AIDS cases and deaths reported in the United States.

MAIN OUTCOME MEASURE

Months of survival after AIDS diagnosis through December 31, 1998, compared by year of diagnosis.

RESULTS

Among 394 705 persons with an AIDS-defining opportunistic illness (OI) diagnosed in 1984-1997, median survival time improved from 11 months for 1984 diagnoses to 46 months for 1995 diagnoses. Among persons with an OI diagnosed in 1996 and 1997, 67% were alive at least 36 months after diagnosis and 77% were alive at least 24 months after diagnosis, respectively. Among 296 621 AIDS cases diagnosed during 1993-1997, 65% were based on immunologic criteria and 35% on OI criteria; 80% were among men; and 42% were among non-Hispanic blacks, 40% among non-Hispanic whites, 17% among Hispanics, 1% among Asians/Pacific islanders, and less than 1% among American Indians/Alaska natives. The probability of surviving at least 24 months increased from 67% for those with immunologic diagnoses in 1993 to 90% in 1997 and from 49% for those with OI diagnoses in 1993 to 80% in 1997. Survival time increased with each year of diagnosis from 1984 to 1997 for blacks, whites, and Hispanics. The greatest annual survival gains occurred among persons receiving an AIDS diagnosis in 1995 and 1996.

CONCLUSIONS

Survival time after AIDS diagnosis improved from 1984 to 1997. While AIDS incidence is declining, improved survival times present a growing public health challenge as the number of persons living with chronic human immunodeficiency virus disease/AIDS increases.

HIV prevalence and associated risks in young men who have sex with men. Young Men's Survey Study Group

CONTEXT

Studies conducted in the late 1980s on human immunodeficiency virus (HIV) infection among older men who have sex with men (MSM) suggested the epidemic had peaked; however, more recent studies in younger MSM have suggested continued high HIV incidence.

OBJECTIVE

To investigate the current state of the HIV epidemic among adolescent and young adult MSM in the United States by assessing the prevalence of HIV infection and associated risks in this population in metropolitan areas.

DESIGN

The Young Men's Survey, a cross-sectional, multisite, venue-based survey conducted from 1994 through 1998.

SETTING

One hundred ninety-four public venues frequented by young MSM in Baltimore, Md; Dallas, Tex; Los Angeles, Calif; Miami, Fla; New York, NY; the San Francisco (Calif) Bay Area; and Seattle, Wash.

SUBJECTS

A total of 3492 15- to 22-year-old MSM who consented to an interview and HIV testing.

MAIN OUTCOME MEASURES

Prevalence of HIV infection and associated characteristics and risk behaviors.

RESULTS

Prevalence of HIV infection was high (overall, 7.2%; range for the 7 areas, 2.2%-12. 1%) and increased with age, from 0% among 15-year-olds to 9.7% among 22-year-olds. Multivariate-adjusted HIV infection prevalence was higher among blacks (odds ratio [OR], 6.3; 95% confidence interval [CI], 4.1-9.8), young men of mixed or other race (OR, 4.8; 95% CI, 3. 0-7.6), and Hispanics (OR, 2.3; 95% CI, 1.5-3.4), compared with whites (referent) and Asian Americans and Pacific Islanders (OR, 1. 1; 95% CI, 0.5-2.8). Factors most strongly associated with HIV infection were being black, mixed, or other race; having ever had anal sex with a man (OR, 5.0; 95% CI, 1.8-13.8); or having had sex with 20 or more men (OR, 3.0; 95% CI, 2.0-4.7). Only 46 (18%) of the 249 HIV-positive men knew they were infected before this testing; 37 (15%) were receiving medical care for HIV, and 19 (8%) were receiving medical drug therapy for HIV. Prevalence of unprotected anal sex during the past 6 months was high (overall, 41%; range, 33%-49%).

CONCLUSIONS

Among these young MSM, HIV prevalence was high, underscoring the need to evaluate and intensify prevention efforts for young MSM, particularly blacks, men of mixed race or ethnicity, Hispanics, and adolescents. JAMA. 2000;284:198-204

Tracking the HIV epidemic: current issues, future challenges

The emergence of a new infectious disease, AIDS, in the early 1980s resulted in the development of a national AIDS surveillance system. AIDS surveillance data provided an understanding of transmission risks and characterized communities affected by the epidemic. Later, these data provided the basis for allocating resources for prevention and treatment programs. New treatments have dramatically improved survival. Resulting declines in AIDS incidence and deaths offer hope that HIV disease can be successfully managed. However, to prevent and control HIV/AIDS in the coming decades, the public health community must address new challenges. These include the defining of the role of treatment in reducing infectiousness; the potential for an epidemic of treatment-resistant HIV; side effects of treatment; complacency that leads to relapses to high-risk behaviors; and inadequate surveillance and research capacity at state and local levels to guide the development of health interventions. Meeting these challenges will require reinvesting in the public health capacity of state and local health departments, restructuring HIV/AIDS surveillance programs to collect the data needed to guide the response to the epidemic, and providing timely answers to emerging epidemiologic questions.

Efficacy of trimethoprim-sulphamethoxazole prophylaxis to decrease morbidity and mortality in HIV-1-infected patients with tuberculosis in Abidjan, Côte d'Ivoire: a randomised controlled trial

BACKGROUND

There is a high incidence of opportunistic infection among HIV-1-infected patients with tuberculosis in Africa and, consequently, high mortality. We assessed the safety and efficacy of trimethoprim-sulphamethoxazole 800 mg/160 mg (co-trimoxazole) prophylaxis in prevention of such infections and in decrease of morbidity and mortality.

METHODS

Between October, 1995, and April, 1998, we enrolled 771 HIV-1 seropositive and HIV-1 and HIV-2 dually seroreactive patients who had sputum-smear-positive pulmonary tuberculosis (median age 32 years [range 18-64], median CD4-cell count 317 cells/microL) attending Abidjan's four largest outpatient tuberculosis treatment centres. Patients were randomly assigned one daily tablet of co-trimoxazole (n=386) or placebo (n=385) 1 month after the start of a standard 6-month tuberculosis regimen. We assessed adherence to study drug and tolerance monthly for 5 months and every 3 months thereafter, as well as rates of admission to hospital.

FINDINGS

Rates of laboratory and clinical adverse events were similar in the two groups. 51 patients in the co-trimoxazole group (13.8/100 person-years) and 86 in the placebo group (25.4/100 person-years) died (decrease In risk 46% [95% CI 23-62], p<0.001). 29 patients on co-trimoxazole (8.2/100 person-years) and 47 on placebo (15.0/100 person-years) were admitted to hospital at least once after randomisation (decrease 43% [10-64]), p=0.02). There were significantly fewer admissions for septicaemia and enteritis in the co-trimoxazole group than in the placebo group.

INTERPRETATION

In HIV-1-infected patients with tuberculosis, daily co-trimoxazole prophylaxis was well tolerated and significantly decreased mortality and hospital admission rates. Our findings may have important implications for improvement of clinical care for such patients in Africa.

Short-course oral zidovudine for prevention of mother-to-child transmission of HIV-1 in Abidjan, Côte d'Ivoire: a randomised trial

BACKGROUND

In Africa, the risk of mother-to-child transmission of HIV-1 infection is high. Short-course perinatal oral zidovudine might decrease the rate of transmission. We assessed the safety and efficacy of such a regimen among HIV-1-seropositive breastfeeding women in Abidjan, Côte d'Ivoire.

METHODS

From April, 1996, to February, 1998, all consenting, eligible HIV-1-seropositive pregnant women attending a public antenatal clinic in Abidjan were enrolled at 36 weeks' gestation and randomly assigned placebo or zidovudine (300 mg tablets), one tablet twice daily until the onset of labour, one tablet at onset of labour, and one tablet every 3 h until delivery. We used HIV-1-DNA PCR to test the infection status of babies at birth, 4 weeks, and 3 months. We stopped the study on Feb 18, 1998, when efficacy results were available from a study in Bangkok, Thailand, in which the same regimen was used in a non-breastfeeding population.

FINDINGS

280 women were enrolled (140 in each group). The median duration of the prenatal drug regimen was 27 days (range 1-80) and the median duration of labour was 7.5 h. Treatment was well tolerated with no withdrawals because of adverse events. All babies were breastfed. Among babies with known infection status at age 3 months, 30 (26.1%) of 115 babies in the placebo group and 19 (16.5%) of 115 in the zidovudine group were identified as HIV-1 infected. The estimated risk of HIV-1 transmission in the placebo and zidovudine groups were 21.7% and 12.2% (p=0.05) at 4 weeks, and 24.9% and 15.7% (p=0.07) at 3 months. Efficacy was 44% (95% CI -1 to 69) at age 4 weeks and 37% (-5 to 63) at 3 months.

INTERPRETATION

Short-course oral zidovudine was safe, well tolerated, and decreased mother-to-child transmission of HIV-1 at age 3 months. Substantial efforts will be needed to ensure successful widespread implementation of such a regimen.

Trends in HIV prevalence among childbearing women in the United States, 1989-1994

We used data from a national serosurvey to describe national and regional trends in the prevalence of HIV among women giving birth in the United States from 1989 through 1994, and to estimate the number of women between 15 and 44 years old with HIV infection who had not yet developed opportunistic infections defining AIDS. We compared these estimates with AIDS prevalence and mortality estimates from the national AIDS case surveillance system. HIV seroprevalence among childbearing women remained stable nationwide from 1989 through 1994, ranging from 1.5 to 1.7/1000 women. In the Northeast, seroprevalence declined significantly after 1989. Seroprevalence increased significantly in the South through 1991 and then stabilized, although seroprevalence among black women continued to increase through 1994 in some southern states. Although AIDS prevalence and mortality increased nationwide each year from 1989 through 1994, the number of women infected with HIV who had not yet developed AIDS changed little and was approximately 86,000 in 1994. Our data suggest that new HIV infections among women of reproductive age are occurring at a rate that offsets losses from this population due to aging, disease progression, and death.

HIV infection in disadvantaged out-of-school youth: prevalence for U.S. Job Corps entrants, 1990 through 1996

To describe HIV infection prevalence and prevalence trends for disadvantaged out-of-school youth in the United States, we analyzed the HIV prevalence for and demographic characteristics of youth, aged 16 through 21 years, who entered the U.S. Job Corps from January 1990 through December 1996. Job Corps is a federally funded jobs training program for socially and economically disadvantaged out-of-school youth. All 357,443 entrants residing at Job Corps centers during their training were tested for HIV infection; 822 (2.3 per 1000) were HIV-positive. HIV prevalence was higher for women than for men (2.8 per 1000 versus 2.0 per 1000; relative risk [RR]=1.4; 95% confidence interval [CI]=1.2-1.6). Among racial/ethnic groups, prevalence was highest for African Americans (3.8 per 1000). Prevalence was higher for African American women (4.9 per 1000) than for any other gender and racial/ethnic group. From 1990 through 1996, standardized HIV prevalence-stratified by age, race/ethnicity, home region, population of home metropolitan statistical area, and year of entry--declined for women and for men: for women, from 4.1 per 1000 in 1990 to 2.1 per 1000 in 1996 (p=.001); and for men, from 2.8 per 1000 in 1990 to 1.4 per 1000 in 1996 (p=.001). These data suggest that HIV prevalence for disadvantaged out-of-school youth declined from 1990 through 1996. However, considering their youth, prevalence was still high, particularly for women and African Americans, most notably African American women. These data support the need for ongoing HIV prevention programs targeting such youth.

The current status of methods for estimating the prevalence of human immunodeficiency virus in the United States of America

The prevalence of human immunodeficiency virus (HIV) infection can be estimated by two distinct methods. One method, back-calculation, is a complex statistical procedure that estimates the HIV epidemic curve. The second method is based on data from population-based surveys, which provide estimates of the proportion of persons infected with HIV within subgroups, and on the known or estimated population totals for these subgroups. Estimates from these methods are subject to substantial uncertainty and bias, both of which are difficult to quantify. We review recent use of these procedures to estimate HIV prevalence in the United States of America. We also summarize new data on the uncertainty and the bias in these estimates. Reliable estimates of HIV prevalence can be made only by synthesizing estimates from several procedures and by a comprehensive evaluation of relevant data. Future estimates of HIV prevalence will require modifications of these methods or the development of new methods.

Estimating the number of AIDS-defining opportunistic illness diagnoses from data collected under the 1993 AIDS surveillance definition

Expansion of the surveillance definition for AIDS in the United States in 1993 caused a substantial distortion in the trend in AIDS incidence, mainly because CD4-positive (CD4+) T-lymphocyte criteria were added to the definition. To evaluate trends in the rate at which HIV-infected persons develop the opportunistic illnesses listed in the AIDS surveillance definition (AIDS-OIs), we developed a procedure for estimating the incidence of these diseases. This estimate is based primarily on the probability distributions of the time from a CD4+ count in given ranges to the diagnosis of the first AIDS-OI. Our estimates of AIDS-OI incidence change by <4% during most calendar quarters during 1991 through 1995 if we also include the estimated effects of unreported AIDS-OIs among persons with AIDS reported based on the CD4+ criteria. Our procedure eliminates the transient effect of adding the CD4+ criteria to the AIDS surveillance definition and permits us to evaluate trends in the incidence of AIDS-OIs.

Human immunodeficiency virus infection and stroke in young patients

OBJECTIVE

To determine the association between human immunodeficiency virus (HIV) infection and stroke among young persons.

DESIGN

Retrospective case-control study.

SETTING

Large, inner-city public hospital.

PARTICIPANTS

All patients aged 19 to 44 years with a diagnosis of stroke, whose HIV status was determined, admitted from January 1990 through June 1994. Controls matched for age and sex were selected from patients who were admitted during the same period for status asthmaticus whose HIV status was known.

MAIN OUTCOME MEASURE

The associations of HIV infection with all strokes and with cerebral infarction, after adjustment for other cerebrovascular risk factors, were evaluated by Mantel-Haenszel stratified analyses. The subtypes and causes of stroke in HIV-infected patients were compared with HIV-seronegative patients.

RESULTS

The HIV infection was associated with stroke (odds ratio [OR], 2.3; 95% confidence interval [CI], 1.0-5.3) and cerebral infarction (OR, 3.4; 95% CI, 1.1-8.9), after adjustment for other cerebrovascular risk factors. Among patients with stroke, cerebral infarction was more frequent in HIV-infected patients than in HIV-seronegative patients (20 [80%] of 25 vs 48 [56%] of 88, P = .04). The frequency of cerebral infarctions associated with meningitis (P < .001) and protein S deficiency (P = .06) was higher in HIV-infected patients than in seronegative patients.

CONCLUSIONS

Our study suggests that HIV infection is associated with an increased risk of stroke, particularly cerebral infarction in young patients. This risk is probably mediated by increased susceptibility of HIV-infected patients to meningitis and protein S deficiency.

Effect of the human immunodeficiency virus epidemic on mortality from opportunistic infections in the United States in 1993

To measure the effect of the human immunodeficiency virus (HIV) epidemic on mortality from opportunistic infections (OIs) in 1993, national multiple-cause death certificate data were examined using two approaches. First, for each OI, the percentage of deaths with HIV infection reported as the underlying cause was calculated. Second, the age-adjusted rate of death per million population was compared with the rate predicted from a model of rates in 1970-1980 or 1979-1981, as available. The percentage of deaths with HIV as the underlying cause and the ratio of observed to predicted death rates were as follows: toxoplasmosis, 91% and 86 (5.24/0.06); cryptosporidiosis/isosporiasis, 90% and infinite (1.61/0.00); progressive multifocal leukoencephalopathy, 87% and 19 (2.58/0.13); pneumocystosis, 82% and 18 (15.44/0.87); cytomegalovirus disease, 82% and 17 (12.60/0.74); nontuberculous mycobacteriosis, 79% and 18 (15.51/0.84); cryptococcosis, 76% and 4 (5.80/1.35); and histoplasmosis, 68% and 6 (1.36/0.23). Thus, the HIV epidemic has greatly increased mortality from several OIs.

Late postnatal mother-to-child transmission of HIV-1 in Abidjan, Côte d'Ivoire

BACKGROUND

HIV-1 can be transmitted from an infected mother to her infant through breastfeeding, although the precise risk of transmission by this route is unknown. A long-term follow-up of children born to HIV-infected women in Abidjan, Côte d'Ivoire, has enabled us to estimate this risk.

METHODS

Children born to 138 HIV-1-seropositive women, 132 HIV-2-seropositive women, 69 women seroreactive to both HIV-1 and HIV-2, and 274 HIV-seronegative women were enrolled at birth and followed up for as long as 48 months. All children were breastfed (median duration 20 months). Blood samples for either or both HIV PCR and HIV serology were obtained at 1, 2, and 3 months of age, and every 3 months thereafter. Early HIV infection was defined as a positive HIV-1 PCR result obtained in the first 6 months of life. Late postnatal transmission was diagnosed when a child had a negative PCR at 3 or 6 months of age, followed by either or both a positive HIV-1 PCR at 9 months or older, or persistently positive HIV-1 serology at 15 months or older.

FINDINGS

82 children born to HIV-1-seropositive mothers and 57 children born to mothers seropositive for both HIV-1 and HIV-2 had PCR results for samples taken within the first 6 months. By 6 months of age, 23 (28%; 95% CI 19-39) of the 82 children born to HIV-1-seropositive mothers and ten (18%; 95% CI 9-30) of the 57 children born to dually seropositive mothers were HIV-1 infected. Among children whose PCR results were negative at or before age 6 months, and who were followed up beyond 6 months, an additional four (9%) of the 45 children born to HIV-1-seropositive mothers and two (5%) of the 39 children born to dually seropositive mothers became HIV infected. The estimated rates of late postnatal transmission, with account taken of loss to follow-up and the observed pattern of weaning, were 12% (95% CI 3-23) and 6% (0-14), respectively. One of the five children whose mothers seroconverted from HIV-negative to HIV-1, and one of seven children whose mothers seroconverted from HIV-2 to dual reactivity, became HIV-1 positive. No case of late postnatal transmission occurred in children born to HIV-2-positive or persistently HIV-negative mothers.

INTERPRETATION

Breastfed children born to mothers seropositive for HIV-1 alone or seropositive for HIV-1 and HIV-2 in Abidjan are at substantial risk of late postnatal transmission. Early cessation of breastfeeding at 6 months of age should be assessed as a possible intervention to reduce postnatal transmission of HIV.

Update on the seroepidemiology of human immunodeficiency virus in the United States household population: NHANES III, 1988-1994

To update the estimate of seroprevalence of HIV from the third National Health and Nutrition Examination Survey (NHANES III), data from the second phase of the survey were combined with previously published data to produce a more precise estimate. The testing was performed anonymously on 11,203 individuals 18-59 years of age examined from 1988 to 1994. Fifty-nine individuals were HIV positive, for an overall prevalence of 0.32%. The number of individuals living in households with HIV infection based on this estimate was 461,000, with a 95% confidence interval of 290,000-733,000. Analysis of nonresponse demonstrated that white and black men 40-59 years of age were least likely to participate in the survey. A sensitivity analysis demonstrated that this nonresponse may have biased the NHANES III estimate downward by 190,000 persons. Data from the second phase of the survey were used to analyze the association between drug use and HIV infection. Black women who used cocaine were 12 times more likely to be HIV positive compared with all tested black women (6.5% vs. 0.55%). This survey provides an estimate of HIV prevalence for individuals who reside in households but excludes some persons who are at higher risk for HIV infection, including prisoners and the homeless not residing in shelters.

Prevalence of HIV infection in the United States, 1984 to 1992

OBJECTIVE

To estimate the number of persons infected with the human immunodeficiency virus (HIV) living in the United States and the change in HIV infection prevalence since 1984.

DESIGN

We estimated HIV prevalence from 3 data sources. We estimated past HIV infection rates from a statistical procedure based on national acquired immunodeficiency syndrome (AIDS) case surveillance data and estimates of the time from HIV infection to AIDS diagnosis. We also analyzed HIV prevalence data from 2 national surveys, a survey of childbearing woman and a household survey of current health status. We used other data sources to adjust these survey estimates to include groups not covered in the surveys.

RESULTS

Approximately 0.3% of US residents (650,000-900,000 persons) were infected with HIV in 1992. Approximately 0.6% of men (including adolescent boys > or = 13 years of age) were infected, including approximately 2% of non-Hispanic black men and 1% of Hispanic men. Approximately 0.1% of women (including adolescent girls > or = 13 years of age) were infected, including approximately 0.6% of non-Hispanic black women. Approximately half of all infected persons were men who had sex with men, and one fourth were injecting drug users. The prevalence of HIV infection increased from 1984 to 1992, with a greater relative increase among women than men.

CONCLUSIONS

The 3 different data sources and methods are consistent in estimating that 650,000 to 900,000 persons were infected with HIV in the United States in 1992. Among adolescents and adults of both sexes, the proportion infected was substantially higher among non-Hispanic blacks and Hispanics than among non-Hispanic whites. HIV-related illness will be a major clinical and public health problem in the United States for years to come.

Prevalence and incidence of vertically acquired HIV infection in the United States

OBJECTIVE

To estimate human immunodeficiency virus (HIV) type I prevalence among childbearing women, HIV incidence in infants, and the number of children living with HIV infection and acquired immunodeficiency syndrome as a result of transmission from mother to infant (vertical transmission).

DESIGN

The national HIV serosurvey of childbearing women was used to estimate the incidence of vertically acquired HIV infection in children born between 1988 and 1993. Data from the national acquired immunodeficiency syndrome case surveillance system and a multicenter pediatric HIV surveillance project were modeled to estimate incidence in children born between 1978 and 1987.

SETTING

Surveillance conducted by the Centers for Disease Control and Prevention, Atlanta, Ga, in collaboration with state and local health departments.

RESULTS

Approximately 14,920 HIV-infected infants were born in the United States between 1978 and 1993. Of these, an estimated 12,240 children were living at the beginning of 1994; 26% were younger than 2 years, 35% were aged 2 to 4 years, and 39% were aged 5 years or older. Approximately 6530 HIV-infected women gave birth in the United States in 1993, and, based on a 25% vertical transmission rate, an estimated 1630 of their infants were HIV infected.

CONCLUSIONS

These results provide a basis for estimating medical and other resource needs for HIV-infected women and their children and for measuring the impact of interventions to reduce vertical transmission of HIV.

The seroepidemiology of human immunodeficiency virus in the United States household population: NHANES III, 1988-1991

To provide an estimate of the seroprevalence of human immunodeficiency virus (HIV) in a representative sample of the U.S. household population, serum samples from participants in the third National Health and Nutrition Examination Survey (NHANES III) were tested for HIV antibody. The testing was performed anonymously on 5,430 individuals 18-59 years old from phase 1 of NHANES III conducted from 1988 to 1991. Twenty-nine individuals were HIV positive. The total weighted prevalence was 0.39%. The population estimate of infected individuals was 547,000, with a 95% confidence interval of 299,000-1,020,000 infected persons. Black participants were four times more likely to be HIV positive than white/other individuals and three times more likely than Mexican Americans. Men were three times more likely to be infected than women. Higher nonresponse to the survey and to phlebotomy was observed in young white men; therefore these data provide a conservative estimate of HIV infection in the general household population. This estimate does not include individuals who do not live in households and who may be at higher risk of infection, such as persons in penal institutions, the homeless, or certain hospitalized patients.

High HIV-1 incidence in young women masked by stable overall seroprevalence among childbearing women in Kinshasa, Zaïre: estimating incidence from serial seroprevalence data

OBJECTIVE

To describe the dynamics of the HIV-1 epidemic in childbearing women in Kinshasa, Zaïre, by estimating incidence from serial seroprevalence studies.

METHODS

In 1986 and 1989, 5937 and 4623 pregnant women, respectively, were screened for HIV-1 in Kinshasa. We estimated age-specific incidence from two seroprevalence surveys by using a birth-year cohort analysis and adjusting for differences in mortality and fertility between HIV-1-infected and uninfected women. Mortality and fertility data were measured in a cohort of women recruited from the survey in 1986 and followed until 1989.

RESULTS

While the overall HIV-1 seroprevalence changed little (5.8% in 1986 and 6.5% in 1989; P = 0.17), the prevalence increased in birth-year cohorts of women under 25 years of age in 1989 from 3.2 to 6.2% (P < 0.001), but decreased for women above 25 years of age from 6.9 to 6.7% (P = 0.7). In addition, new HIV infections between 1986 and 1989 were balanced by a higher mortality and lower fertility observed in HIV-infected women. After adjusting for these effects, we estimated an overall 3-year cumulative HIV-1 incidence of 2.8 per 100 uninfected women [95% confidence interval (CI), 1.4-4.2]. The highest incidence, 5.7 per 100 (95% CI, 3.5-8.0), was in women aged 20-24 years in 1989.

CONCLUSION

Despite an overall relatively stable HIV-1 prevalence in childbearing women in Kinshasa between 1986 and 1989, approximately 40% of all HIV-1 infections detected in the 1989 survey occurred between 1986 and 1989, and 60% occurred in women under 25 years of age in 1989.

Trends in HIV prevalence among disadvantaged youth. Survey results from a national job training program, 1988 through 1992

OBJECTIVE

To describe trends in the prevalence of human immunodeficiency virus (HIV) among socially and educationally disadvantaged US youth.

DESIGN

Analysis of demographic and geographic trends of HIV infection among Job Corps students from January 1988 through December 1992.

SETTING

The Job Corps is a national training program for disadvantaged and out-of-school youth. POPULATION SCREENED: Youths aged 16 to 21 years who entered the Job Corps residential training centers during the survey period.

MAIN OUTCOME MEASURE

Trends in prevalence of HIV infection among Job Corps students stratified by sex, age, race, and region of the country.

RESULTS

Of the 269,956 Job Corps students screened, 812 (0.3%) tested positive for the antibody to HIV type 1. Seroprevalence of HIV for young men decreased from 3.6 per 1000 in 1988 to 2.2 per 1000 in 1992 (chi 2 test for trend, P < .001). Seroprevalence for young women increased from 2.1 per 1000 in 1988 to 4.2 per 1000 in 1990 (P = .001), with seroprevalence remaining stable from 1990 through 1992. The decreasing trends in HIV prevalence among men and increasing trends among women were primarily due to changes in seroprevalence in African-American students.

CONCLUSIONS

The overall prevalence of HIV infection of three per 1000 is high, given the youth of Job Corps students. The significant rise in HIV rates among female Job Corps students provides evidence of the increasing risk of infection for socioeconomically disadvantaged young women. Reasons for the declining trend in HIV prevalence among male Job Corps students are not clear. Efforts to prevent the spread of HIV infection among adolescents must focus on the group that is hardest to reach--out-of-school and impoverished youth.

Effect of routine use of therapy in slowing the clinical course of human immunodeficiency virus (HIV) infection in a population-based cohort

Clinical trials have shown that the prophylactic use of zidovudine and aerosolized pentamidine (or other antibiotics used as prophylaxis against Pneumocystis carinii pneumonia) in acquired immunodeficiency syndrome (AIDS)-free human immunodeficiency virus (HIV)-infected persons delays the development of AIDS, but the effectiveness of such therapy in general use in the population still remains largely undocumented. To help answer this question, the authors estimate the effectiveness of this therapy in a population-based cohort of HIV-infected homosexual and bisexual men in San Francisco. The authors use a continuous-time Markov process to model the decline of CD4+ T-lymphocytes (T4-cells) measured in cells/microliter in HIV-infected persons. The model partitions the HIV (type 1) infection period into six progressive T4-cell count intervals (stages), followed by a seventh stage: AIDS diagnosis. The authors use maximum likelihood methods to fit the model to the observed transitions for 428 HIV-infected men during June 1984 to March 1991, from the San Francisco Men's Health Study. Since zidovudine was not widely used before 1988, the model has a component that controls for calendar time-related biases. The fitted model provides statistical estimates and confidence intervals for measuring therapy effectiveness. The authors estimate that prophylactic therapy reduces the progression rate from stage 4 (T4-cell count, 350-499) to stage 5 (T4-cell count, 200-349) by a factor of 0.26 (95% confidence interval (CI) -0.22 to 0.55); from stage 5 to stage 6 (T4-cell count < 200) by a factor of 0.33 (95% CI 0.04-0.54); and from stage 6 to 7 (AIDS) by a factor of 0.62 (95% CI 0.47-0.73). In addition, therapy started by an HIV-infected person in stage 4 is estimated to reduce the risk of developing AIDS by a factor of 0.83 (95% CI 0.46-0.94) at 6 months and 0.68 (95% CI 0.35-0.89) at 24 months after entering stage 4. Therapy started by HIV-infected persons in more advanced stages is estimated to reduce the risk of developing AIDS by factors ranging from 0.70 (95% CI 0.39-0.90), early in stage 5, to 0.28 (95% CI 0.14-0.45), late in stage 6. Thus, the prophylactic use of zidovudine and pentamidine in routine medical care has a strong, consistent, and significant effect in slowing the clinical course of HIV infection in a population-based cohort.

An analysis of AIDS incidence data by clustering trends

AIDS incidence trends vary greatly among geographic areas in the United States. We define clusters of areas within which AIDS incidence trends are similar, as areas within a cluster may have similar human immunodeficiency virus epidemic patterns and thus may lead to similar prevention/intervention strategies. Methods of exploratory data analysis are used to define such clusters from reported quarterly AIDS incidence to December 1990 (adjusted for estimated reporting delays) in homosexual and bisexual men not using intravenous drugs in 39 metropolitan statistical areas (MSAs) in the United States. After smoothing AIDS incidence in each MSA, we define groups from cluster analysis based on a measure of similarity between pairs of MSAs. A log-linear model gives estimates of the scale factors and the common trend for the MSAs in each group. Alternative metrics and simulated data suggest that the clustering is fairly robust to variations in AIDS incidence data. The resulting clusters separate MSAs with different trends, for example, MSAs in which AIDS incidence shows signs of reaching a plateau are separated from MSAs in which incidence continues to increase rapidly.

Projections of the number of persons diagnosed with AIDS and the number of immunosuppressed HIV-infected persons--United States, 1992-1994

This report presents projections of the number of persons who will initially be diagnosed with a condition included in the 1987 surveillance definition for acquired immunodeficiency syndrome (AIDS) in the United States during the period 1992-1994. The report also presents estimates and projections of the prevalence of persons infected with the human immunodeficiency virus (HIV) who have CD4+ T-lymphocyte (T-cell) counts < 200/microL and who have not been diagnosed with a condition listed in the 1987 AIDS surveillance definition. These estimates and projections are used to predict the effect of expanding the AIDS surveillance definition to include all HIV-infected persons with a CD4+ T-cell count < 200/microL. Approximately 58,000 persons were diagnosed with AIDS in the United States during 1991. During the period 1992-1994, the number of persons newly diagnosed with AIDS is expected to increase by at most a few percent annually, with approximately 60,000-70,000 persons diagnosed per year. Although AIDS diagnoses among homosexual and bisexual men and among injecting drug users are projected to reach a plateau during this period, the number of AIDS diagnoses among persons whose HIV infection is attributed to heterosexual transmission of HIV is likely to continue to increase through 1994. The number of living persons who have been diagnosed with AIDS is expected to increase from approximately 90,000 in January 1992 to approximately 120,000 in January 1995. There is, however, considerable uncertainty in these projections. For example, the plausible range for the number of persons initially diagnosed with AIDS in 1994 is 43,000-93,000. CDC estimates that, as of January 1992, 115,000-170,000 U.S. residents had severe immunosuppression (a CD4+ T-cell count < 200 cells/microL without a diagnosis of AIDS in an HIV-infected person). Only about 50,000 of these persons were receiving medical care for HIV-related conditions and were known to have a CD4+ T-cell count < 200 cells/microL. The number of persons with severe immunosuppression is expected to increase to 130,000-205,000 by January 1995, with the actual number more likely to be in the lower half of this range than the upper half. The expanded AIDS surveillance definition, which includes severe immunosuppression, is predicted to result in an increase of approximately 75% in the number of persons reported during 1993, but an increase of < 20% in 1994 compared with the number of persons who would have been reported had the definition not been changed.(ABSTRACT TRUNCATED AT 400 WORDS)

Population-based monitoring of an urban HIV/AIDS epidemic. Magnitude and trends in the District of Columbia

OBJECTIVE

To assess the extent of the human immunodeficiency virus (HIV)/acquired immunodeficiency syndrome (AIDS) epidemic in the District of Columbia and demonstrate an approach to monitoring HIV infection and projecting AIDS incidence at a community level.

DESIGN

Backcalculation methods to reconstruct HIV incidence from AIDS incidence in subgroups. Results were compared with directly measured HIV seroprevalence in selected sentinel populations: childbearing women, civilian applicants for military service, and hospital patients admitted for conditions unrelated to HIV infection.

RESULTS

Between the start of the epidemic in 1980 and January 1, 1991, one in 57 District of Columbia men aged 20 to 64 years was diagnosed with AIDS. Unlike the plateau projected for the nation, AIDS incidence for the District of Columbia was projected to increase by 34% between 1990 and 1994. Models of HIV infection incidence suggested two broad epidemic waves of approximately equal size. The first occurred in men who have sex with men and peaked during the period from 1982 through 1983. The second began in the mid-1980s in injecting drug users and heterosexuals. We estimated that among District of Columbia residents aged 20 to 64 years, 0.3% of white women, 2.9% of white men, 1.6% of black women, and 4.9% of black men were living with HIV infection as of January 1, 1991. These estimates are broadly consistent with survey data: among black childbearing women in their 20s, HIV prevalence doubled to 2% between the fall of 1989 and the spring of 1991; from military applicant data, we estimated that over 5% of black men born from 1951 through 1967 were HIV-positive; in the sentinel hospital, HIV prevalence rates among male patients aged 25 to 34 years were 11.3% in white men and 16.9% in black men.

CONCLUSION

Backcalculation and surveys yielded quantitatively consistent estimates of HIV prevalence. Many injecting drug users and heterosexuals in the District of Columbia were infected after January 1, 1986. Similar monitoring of the epidemic in other localities is needed to focus efforts to reduce the incidence of HIV transmission.

A simulation model of AIDS in San Francisco: II. Simulations, therapy, and sensitivity analysis

The HIV and AIDS incidences each year for homosexual men in San Francisco are estimated from data. A computer simulation model for HIV transmission dynamics and progression to AIDS is used to reconstruct the HIV epidemic. Using some a priori parameter estimates, simulations are found that give good fits to the incidence data. In the stimulations the populations is divided into risk groups whose sexual activities are found to be strongly connected. There is saturation in the high-risk group, but changes in sexual behavior are more important in obtaining adequate fits. The simulation modeling yields useful parameter estimates, but the remaining uncertainty in parameter values implies that the simulation forecasts are also uncertain. Changes in HIV incidence lead to changes in AIDS incidence about 6-10 years later. Simulation models with and without zidovudine treatment both fit the incidence data; thus the effects of therapy on AIDS incidence are unclear. The fits of the simulation model are most sensitive to the yearly migration rate, the number of stages in the progression to AIDS, and the average number of new sexual partners per month; thus better estimates of these parameters would be desirable.

Using provisional diagnoses in monitoring a clinical trial

In clinical trials there is delay between the occurrence of an event and the recording of that event as an end point in the trial data base. Delays are especially likely if events are reviewed carefully to determine whether diagnostic criteria for end points are satisfied. As a result, the value of a statistic used to evaluate efficacy during a trial may differ from the value of that statistic based on the true end point status of all events which have already occurred. Simulating the process causing delays can be useful in evaluating interim efficacy data by providing a quantitative estimate of the uncertainty in a monitoring statistic. These ideas are illustrated using data from the Lipid Research Clinics Coronary Primary Prevention Trial.

The increasing frequency of heterosexually acquired AIDS in the United States, 1983-88

Of 88,510 cases of acquired immunodeficiency syndrome (AIDS) reported in adults in the United States from 1983 through 1988, the percentage attributed to reported heterosexual contact with persons known to be infected or at increased risk of infection with human immunodeficiency virus (HIV) has increased steadily from 0.9 percent in the first quarter of 1983 to 4.0 percent in the fourth quarter of 1988, from 0.1 (in 1983) to 1.4 percent (in 1988) among men, and from 13 (in 1983) to 28 percent (in 1988) among women. Among women, the cumulative incidence of AIDS attributable to heterosexual contact per million population is over 11 times greater for Blacks and Hispanics than for Whites. Among men, this incidence is over 10 times greater for Blacks and four times greater for Hispanics than for Whites. The pattern of distribution of heterosexually acquired AIDS parallels the distribution of other heterosexually transmitted diseases, which are also more frequent in Black and Hispanic inner-city populations. Drug use, exchange of sex for drugs or money, and early onset of sexual activity in adolescents are increasingly associated with heterosexually transmitted infections and are likely to be very important in heterosexual transmission of HIV in inner-city US populations.

Black-white differences in exercise blood pressure. The Lipid Research Clinics Program Prevalence Study

After exclusion of persons on blood pressure medication or with prevalent cardiovascular disease, we studied 83 black and 2,548 white men and 113 black and 1,519 white women 20-69 years old from the Lipid Research Clinics population sample who had performed a standardized treadmill exercise test. Resting systolic and diastolic blood pressures were similar in black and white men, but the diastolic pressure was significantly higher in black than in white women (81.4 vs 77.4 mm Hg). Body weight was higher in black than in white women, and reported physical activity was higher in black than in white men. The proportion of smokers was somewhat higher in blacks than in whites. During the treadmill exercise test with a modified Bruce protocol, mean systolic blood pressure at stage 2 was 174 mm Hg in black men and 166 mm Hg in white men (p less than 0.02), but stage 2 blood pressures did not differ between black and white women (153 and 152 mm Hg, respectively). Even after adjustments were made for levels of baseline characteristics (age, weight, resting systolic blood pressure, smoking, low density lipoprotein cholesterol, physical activity, and alcohol intake), black men responded with a 7-mm Hg higher systolic blood pressure during exercise than white men (p less than 0.01). Another new finding was a highly significant positive association between stage 2 systolic blood pressure and low density lipoprotein cholesterol in men. The findings suggest a higher systemic vascular resistance during exercise in the selected sample of black men, which is consistent with the higher incidence of hypertension in black men.

Detection of aberrations in the occurrence of notifiable diseases surveillance data

The detection of unusual patterns in the occurrence of diseases and other health events presents an important challenge to public health surveillance. This paper discusses three analytic methods for identifying aberrations in underlying distributions. The methods are illustrated on selected infectious diseases included in the National Notifiable Diseases Surveillance System of the Centers for Disease Control. Results suggest the utility of such an analytic approach. Further work will determine the sensitivity of such methods to variations in the occurrence of disease. These methods are useful for evaluating and monitoring public health surveillance data.

Predictive value of the exercise tolerance test for mortality in North American men: the Lipid Research Clinics Mortality Follow-up Study

More than 3600 white men, from 30 to 79 years old and without a history of myocardial infarction, underwent submaximal treadmill exercise tolerance tests as part of their baseline evaluation for the Lipid Research Clinics Mortality Follow-up Study. The exercise test was conducted according to a common protocol and coded centrally; depression of the ST segment by at least 1 mm (visual coding) and/or 10 microV-sec (ST integral, computer coding) signified a positive test. Concurrent measurements of age, blood pressure, history of cigarette smoking, and plasma levels of lipids, lipoproteins, and glucose, as well as other coronary risk factors, were obtained. Cumulative mortality from cardiovascular disease was 11.9% (22/185) over 8.1 years mean follow-up among men with a positive exercise test vs 1.2% (36/2993) over 8.6 years mean follow-up among men with a negative test. Three-quarters (43) of these deaths were due to coronary heart disease. The relative risk for cardiovascular mortality associated with a positive exercise test was 9.3 before and 4.6 after age adjustment. Cardiovascular mortality rates were especially elevated (relative risk 15.6 before and 5.1 after age adjustment) among the 82 men whose exercise tests were adjusted "strongly" positive based on degree and timing of the ischemic electrocardiographic response. A positive exercise test was also moderately associated with noncardiovascular mortality; the relative risk for all-cause mortality was 7.2 before and 3.4 after age adjustment. The relative risk for cardiovascular mortality associated with a positive exercise test was not appreciably altered by covariance adjustment for known coronary risk factors other than age. A positive exercise test was a stronger predictor of cardiovascular death than were high plasma levels of low-density lipoprotein cholesterol, low plasma levels of high-density lipoprotein cholesterol, smoking, hyperglycemia, or hypertension. Its impact on risk of cardiovascular death was equivalent to that of a 17.4 year increment in age.

Accuracy and comparability of long-term measurements of cholesterol

Laboratories of the Lipid Research Clinics Program (LRC) maintained the accuracy of their measurements of total cholesterol by using seven pooled serum calibrators supplied by the Centers for Disease Control (CDC). Over the 11-year life of the LRC, each calibrator was prepared in succession and a target value was assigned by the CDC reference method for cholesterol. The results of a special experiment in which six of the seven calibrators were analyzed simultaneously demonstrated that the target values were accurately assigned. Deviations of the target values from the experimental means ranged from zero to 1.7% of the original target value. The experiment revealed no evidence of drift in the bias of the reference method over the life of LRC and demonstrated the accuracy, consistency, and the comparability of the values assigned to the successive calibrator pools used by the LRC laboratories during more than eight years. It demonstrated the reliability of a reference method and the suitability of frozen serum pools for maintaining an accurate measurement base for serum cholesterol.

Accuracy and comparability of long-term measurements of cholesterol

Laboratories of the Lipid Research Clinics Program (LRC) maintained the accuracy of their measurements of total cholesterol by using seven pooled serum calibrators supplied by the Centers for Disease Control (CDC). Over the 11-year life of the LRC, each calibrator was prepared in succession and a target value was assigned by the CDC reference method for cholesterol. The results of a special experiment in which six of the seven calibrators were analyzed simultaneously demonstrated that the target values were accurately assigned. Deviations of the target values from the experimental means ranged from zero to 1.7% of the original target value. The experiment revealed no evidence of drift in the bias of the reference method over the life of LRC and demonstrated the accuracy, consistency, and the comparability of the values assigned to the successive calibrator pools used by the LRC laboratories during more than eight years. It demonstrated the reliability of a reference method and the suitability of frozen serum pools for maintaining an accurate measurement base for serum cholesterol.

Assessment of 2,3,7,8-tetrachlorodibenzo-p-dioxin exposure using a modified D-glucaric acid assay

An enzyme-inhibition assay was evaluated and modified to quantify D-glucaric acid in a population exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD). The modified assay combined improvements described separately in previous reports, including pH adjustment by standard addition of buffers rather than by titration, an optimum pH of 2.3 for converting D-glucaric acid to 1,4-glucarolactone, and the use of the relation reciprocal of absorbance versus concentration for calculating unknowns. Reference limits for adult males were 0.06-5.90 mmol D-glucaric acid/mol creatinine and for adult females 0.87-6.23 mmol D-glucaric acid/mol creatinine. Children under the age of 15 yr had a reference range of 0-8.34 mmol D-glucaric acid/mol creatinine. Persons on anticonvulsant therapy excreted from 3 to 10 times the upper reference limits of D-glucaric acid. Urinary concentrations of D-glucaric acid in persons identified as being at high risk for exposure to 2,3,7,8-TCDD in Times Beach, Mo., were not significantly different from concentrations in those identified as being at low risk for such exposure.

Coronary risk factors and exercise test performance in asymptomatic hypercholesterolemic men: application of proportional hazards analysis

The association of established coronary risk factors with submaximal graded treadmill exercise test performance was examined in 6,850 asymptomatic, white 34--59-year-old hypercholesterolemic men screened between 1973 and 1976 at 12 North American Lipid Research Clinics for participation in their Coronary Primary Prevention Trial. The prevalence of ischemic electrocardiographic responses (greater than or equal to 1 mm S-T segment depression) was 8.6%. The Cox proportional hazards method was adapted so as to take into account the level of exercise at which ischemic responses occurred and to which subjects without ischemic responses were exposed. The results were compared with those obtained by standard logistic regression. In both models, age, blood pressure, plasma cholesterol, and (inversely) plasma high-density lipoprotein cholesterol and alcohol consumption were significant independent predictors of an ischemic response to exercise. Surprisingly, ischemic responses were less frequent in smokers than in nonsmokers. However, when the proportional hazards method was used, cigarette smoking was weakly but significantly (p less than 0.01) predictive of an ischemic response on the treadmill. Results from this model differed from those of the logistic model because the former takes into account the reduced exercise capacity of smokers, which renders them less likely to reach workloads sufficient to induce myocardial ischemia. The proportional hazards model similarly demonstrated a possible beneficial effect of habitual physical activity which was not apparent in the logistic model. Quetelet index and plasma triglyceride were only weakly associated with the probability of an ischemic response, and did not contribute significantly to either model.

Estrogen use and all-cause mortality. Preliminary results from the Lipid Research Clinics Program Follow-Up Study

The association of exogenous estrogen use and hysterectomy status with all-cause mortality was examined in 2,269 white women, aged 40 to 69 years, who had been followed up for an average of 5.6 years in the Lipid Research Clinics Program Follow-up Study. A total of 72 deaths occurred during this period. The relative risk of death in estrogen users compared with nonusers was 0.54 in gynecologically intact women, 0.34 in hysterectomized women, and 0.12 in bilaterally oophorectomized women. The risk of death in estrogen users, irrespective of hysterectomy status, was 0.37 times that in nonusers (3.4/1,000 v 9.3/1,000). The significant negative association of estrogen use with mortality persisted after multivariate adjustment for confounding factors. Hysterectomy status alone was not a significant predictor of death. Some, but not all, of the lower risk of mortality in estrogen users can be accounted for by increased levels of high-density lipoprotein cholesterol.

The use of epidemiologic data for personal risk assessment in health hazard/health risk appraisal programs

Health Hazard/Health Risk Appraisal (HHA/HRA) programs employ personal risk assessment as an educational and motivational technique to encourage the adoption of healthier lifestyles by health education clients. We have reviewed the scientific basis of of the risk assessments provided in HHA/HRA. There are severe limitations in both the data and the risk estimation procedures. Various undocumented assumptions enter into the projections; several key aspects are arbitrary. Proposals for improvement of the procedure generally require data that are not available. Attention has largely focused on increasing mathematical sophistication (e.g. adjustment for competing risks) while ignoring problems in accuracy of the client data from which projections are calculated and serious questions about the health education messages implicit in the appraisal results. Health Hazard/Health Risk Appraisal programs should be candid about the limitations of the technique. Use of appraisal results for evaluation purposes is questionable. The health education messages produced by such programs deserve attention, since their appropriateness, especially for clients who are not white, middle class and middle-aged, has not been established.

In defense of matching

This paper discusses the simplified situation of an epidemiologic study involving disease, exposure, and a single (possibly confounding) extraneous factor, all of which are dichotomous. The question is: In studying the association between disease and exposure, should the comparison group be selected by random sampling or by matching on the extraneous factor? An example is used to demonstrate the general principle that matching controls confounding in estimating the risk ratio in a follow-up study, but not in estimating the exposure odds ratio in a case-control study. Calculations based on a probability model show that, despite the possible reduction in sample size which may be associated with matching, matching will often lead to a more precise estimate of the effect measure than random sampling and is not likely to result in a significant loss in precision in situations of practical importance. Therefore, selection of the referent group by matching should be given serious consideration for both follow-up and case-control studies.

Matching in epidemiologic studies: validity and efficiency considerations

Validity and efficiency issues are considered with regard to the use of matching and random sampling as alternative methods of subject selection in follow-up and case-control studies. We discuss the simple situation involving dichotomous disease and exposure variables and a single dichotomous matching factor, and we consider the influence on efficiency of a possible loss of subjects due to matching constraints. The decision to match or not should be motivated by efficiency considerations. An efficiency criterion based on a comparison of confidence intervals under matching and random sampling for the effect measure of interest (the risk ratio and risk difference in follow-up studies, and the odds ratio in case-control studies) leads to the following conclusions when the sampling method does not influence the size of the comparison group. In follow-up studies, matching on a confounder is expected to lead to a gain in efficiency over random sampling, while matching on a nonconfounder is not expected to result in a loss of efficiency. In case-control studies, the same conclusions hold, except that matching is not as advantageous as in follow-up studies and can lead to a loss of efficiency in some situations (usually of little practical importance). When matching reduces the size of the comparison group, there is likely to be a meaningful gain in efficiency due to random sampling only when the matched comparison group is at most 40-50% the size of the randomly-sampled comparison group is a follow-up study, and at most 50-65% the size in a case-control study.

Autoradiography of protein turnover in subcrestal versus supracrestal fibre tracts of the developing mouse periodontium

Eighteen-day-old C57Bl mice were injected with 5 muCi/g body weight [3H]-proline and killed at intervals of 4 h to 7 wk later. Grain counts in three fibre tracts revealed that (1) half-lives of labelled protein in the developing periodontium were much shorter than those found previously for the mature periodontium , and (2) the half-life of labelled protein in the dento-gingival region was longer than the half-lives in the trans-septal and dento-alveolar fibre tracts (half-lives: dento-alveolar = 2.5 days, trans-septal = 3.8 days; dento-gingival = 7.9 days). Eight mice were given injections of [3H]-proline on days 11, 13, 15 and 17, which encompasses the formation of the three fibre tracts, and killed 4 h, 2.5, 5 and 8 wk after the last injection. All label incorporated during the formation of the periodontium had been lost by 5 wk post-injection, showing that a stable core fibre (i.e. one which is not metabolized) was not present.