Over the past few years, treatment options for people with HIV have expanded dramatically. As new antiretroviral treatment strategies improve the health of thousands of patients across the nation, the progress with AIDS research and new drug development is undeniable. However, progress in research by itself means little unless it transforms medical practice for the benefit of real patients. To what degree has health care improved for people with HIV?

Little information has been available until recently on the care of HIV infected people in the United States. How many HIV infected people receive medical care, when do they enter medical care, how much care do they receive, how good is it, where and from whom do they get it, what does it cost, and who pays for it?

These questions loom even larger because advances in HIV research have occurred at the same time the American health care system has been in its greatest period of change for a generation. Due to spiraling health care costs in the early 1990s, managed care was widely implemented over the decade in the private insurance market as a means to contain costs. By 1997, 85% of people enrolled in employer-sponsored health plans received their care from a managed care company.⁽¹⁾

This article is the first of a two-part series on health care for HIV infected people in the United States. This issue describes the results of a large, ground-breaking study of HIV health care from the RAND Corporation as well as information from several smaller studies. This article also describes published and unpublished information on the diagnosis of HIV infection in New York State. As more information becomes available about patterns of health care for HIV-infected people, it is increasingly clear the majority of patients under care have advanced disease. Since diagnosis is the first crucial step in medical care, more attention must be paid to why people delay seeking HIV testing, how many people have been tested, and how many HIV-infected people may not have been tested yet.

The second article in this series, which appears next month, will examine changes in the most important health care program for people with HIV, the government-funded Medicaid program.

The RAND Study

The largest study of health care trends in HIV treatment was completed by a study team organized by the RAND Corporation. Published results from the study come in two parts: an analysis of health care for people with HIV during early 1996 when new therapies were just emerging on the market and a follow-up study that tracked the health care of HIV patients from 1996 to 1998.^(2,3)

The original study used statistical sampling techniques to create a snapshot of HIV care in the United States during the first two months of 1996. The study examined multiple factors: demographics of the patients and providers, health care patterns and drug utilization, regional variations, and disease stage.

The study used a three-stage sampling technique to examine people with HIV over eighteen who made at least one medical visit during the study period to private or public health care institutions, except prisons and the military. The first sample was geographic. Twenty-eight metropolitan statistical areas and twenty-four clusters of rural counties were randomly selected. In aggregate, the selected areas accounted for more than seventy percent of reported AIDS cases in the country.

The second stage of the analysis sampled health care providers in the selected areas. Fifty-eight institutional or individual "known HIV providers" in the urban areas and 28 in the rural areas were surveyed. "Known HIV providers," who were identified by local public health officials, were physicians or clinics with a high volume HIV patient census and a primary focus on HIV care. Known HIV providers were compared to 87 "other providers" in urban areas and 23 in rural areas, who had collectively treated more than 4,000 HIV patients but who were not primarily involved in HIV care.

The final stage of the analysis was surveys of HIV patients cared for by the providers in the selected areas. More than 4,000 patients were randomly selected, 2,864 of whom consented to a full interview. More than ninety percent of the interviews were conducted in person, and the remainder were done on the telephone. For the follow-up study, these same patients were followed through 1998.

Basic Findings of the RAND Study

The study team estimates about 231,400 people with HIV received medical care during the first two months of 1996. Seventy-seven percent were male, and almost 90% were under fifty years old. Just under half were white, one-third were black, and about one-sixth were Hispanic. Other ethnic groups accounted for less than 1% of the patient population. One-half were gay men, one-quarter injection drug abusers, 18% heterosexual, and 9% reported no known risk factor for HIV infection.

Economically, the patient population was quite poor, although many had received at least some post-secondary education. The vast majority reported incomes under $25,000 per annum, and almost half had income under $10,000. Just under half of the population had some post-secondary education, and about forty percent were employed at the time of the survey.

The study found a surprisingly high level of private health coverage among the population. Thirty percent had some form of private insurance, mirroring to some extent the percentage of patients who reported full-time employment. Twenty-nine percent relied on Medicaid, the state-federal health insurance program for the poor, and another 20% on Medicare, the federal health program for the elderly and long-term disabled. Medicare recipients in this population usually also received Medicaid, meaning they were both long-term disabled and poor. (Medicare unlike Medicaid does not offer a prescription drug benefit, so dual enrollment is usually sought for all who qualify.) Twenty percent of the patients were uninsured.

The South had the most people with HIV in absolute terms with approximately 83,000 HIV patients in care at the time of the study. The Northeast and West had fewer patients in absolute terms (57,100 and 65,700 respectively) but shared the highest HIV case rate in the country with 112 HIV patients per 100,000 people. The South had an HIV case rate of 90 patients per 100,000 people. The Midwest had the lowest absolute number of patients (25,700) and the lowest case rate at 42 patients per 100,000 people.

The study also demonstrated HIV treatment is not centered in primary care, but rather is concentrated in a few centers and physician practices. HIV treatment has become a specialist discipline. Most HIV patients are seen by clinicians and institutions with a particular focus on HIV treatment. Thirty percent of HIV patients received care from a specialist clinic at a teaching hospital. The vast majority of patients received care from providers who had more than fifty HIV patients. Only 3% of patients were treated by physicians or in clinics that cared for fewer than five HIV patients.

The vast majority of HIV patients in medical care at the time of the survey had advanced disease. One of most disturbing findings is that no more than 10% of the HIV patients could be classified as recently infected or asymptomatic. Despite all the focus on early intervention and the aggressive use of antiretroviral treatment during the initial stages of the disease, 60% of the patients in care at the time of the study met the CDC case definition for AIDS. One-quarter of the patients had fewer than 50 CD4 cells.

The RAND study found some important differences in the study population. Women were less likely to have an AIDS diagnosis and were more likely to be younger and black. The vast majority of women had a high school education or less. Perhaps as a result, women were much less likely to be employed and were more likely to be poor and uninsured. Gay men were far more likely to have received at least some post-secondary education, to be employed, to rely on private health insurance, and to reside in the West.

Important differences also emerged in health insurance coverage. Just slightly more than half of HIV patients in the West had private health insurance. At 41%, patients in the Midwest had a lower but still substantial rate of private health insurance coverage. This stands in stark contrast to the Northeast and South where private health insurance was uncommon at nineteen and twenty-three percent respectively. However, patients in the Northeast without private coverage were most likely on Medicaid. In the South, patients without private health insurance were more likely uninsured. For example, the study team found for every ten patients on Medicaid in the Northeast, only two were uninsured. In the South, for every ten patients on Medicaid, eleven were uninsured.

The first published analysis of the RAND study provides an intriguing snap shot of health care trends for the HIV infected population. The study estimates the 231,400 HIV infected people in the country made 1.95 million outpatient medical visits over a six-month period, approximately 1.4 visits per patient per month. During this same period, twenty percent of the patients were hospitalized for an average of 10.4 days each. One-third visited a hospital emergency department within the six months of the study period for an estimated total of 154,000 visits.

Drug utilization was also measured in the study, which found an increasing but uneven level of treatment over a six-month period. At the start of the study, 85% of patients had used at least one antiretroviral drug. By the end of January 1996, 16% had used one of the newer drugs (i.e. lamivudine or a protease inhibitor) and by the end of 1996 this figure rose to 55%.

The average annual treatment cost for the study population was also estimated. Approximately $5.1 billion was spent in 1996 on medical care for HIV-infected people in the United States, according to the study team. This translates into approximately $22,200 per patient per year. At the time of the study, most of the expenditure, $2.2 billion (43%), was spent on hospitalization costs. Another $2 billion (40%) was spent on pharmaceuticals, $100 million (2%) on emergency department visits, and $800 million (15%) on outpatient services and laboratory tests.

HIV care accounted for less than 1% of the $700 billion total 1996 health care expenditure in the United States. The study authors argue this proportion is not excessive given HIV infection accounts for 7% of potential years of life lost in the United States, more than pneumonia, influenza, chronic obstructive pulmonary disease, diabetes, and chronic liver disease combined. The study states "the crisis in expenditures for patients with HIV disease appears to be one of financing not of cost."

The RAND team conducted follow-up interviews with the surveyed patients at several points through the end of 1998. This follow-up allowed the team to examine trends in the utilization of health care and antiretrovirals during a period of rapid change in HIV clinical management. The study team analyzed changes over time and among all subgroups of patients on six measures of the adequacy of HIV care.

The six measures of adequacy were attending (1) fewer than two outpatient visits in the previous six months, (2) more than one emergency department visit without associated hospitalization within six months, (3) more than one hospitalization in the previous six months, (4) not receiving recommended antiretroviral therapy by the end of 1996 despite having a CD4 count below 500, (5) never receiving antiretroviral treatment, (6) not receiving pneumocystis pneumonia prophylaxis when indicated.

Looking at subgroup differences, the study team found a majority of patients received adequate care on all six measures. However, there were important disparities. Fifteen percent of the patients had fewer than two outpatient visits in the previous six months, 23% had visited an emergency department without an associated hospitalization, and 19% were hospitalized.

The use of care differed by CD4 cell count, age, sex, race, HIV risk category, type of insurance, economic status, and region. Individuals with lower CD4 counts tended to receive more care on all six measures, while those with higher CD4 counts were less likely to receive adequate care. Patients over fifty years old were less likely to visit an emergency department and were less likely to have received indicated PCP prophylaxis. Women were hospitalized more and used emergency department more often, and were less likely to receive indicated PCP prophylaxis or antiretroviral therapy. Blacks and Latinos received less optimal care on four of the six measures. Those with more education received better care, and the pattern was similar for people with higher incomes.

Injection drug abusers and heterosexuals received less adequate care than gay men. Uninsured people and those on Medicaid both received less adequate care than people with private insurance. People on Medicare, who usually also received Medicaid, had high rates of emergency department use and hospitalization, but they were also more likely to receive indicated PCP prophylaxis and antiretroviral treatment than those on Medicaid alone. People in HMOs received the highest level of care overall, and had the highest rate of PCP prophylaxis. Regionally, patients in the West and Midwest had the best level of care. Patients in the Northeast used emergency departments more, were more likely to be hospitalized, and had inadequate use of PCP prophylaxis. Patients in the South were the least likely to receive indicated antiretroviral treatment. Most of the subgroup differences, however, were due in large part to differences in the type of health insurance possessed by individuals in each group.

The follow-up study examined changes in care between 1996 and 1998, with average follow-up of 15.1 months. In general, care improved for all subgroups over time. The proportion of patients who had adequate care on all six measures increased during the study period from 29% to 47%. The percentage of patients who received inadequate care fell at least two measures from 34% to 17% over the study period.

An interesting finding is on the measure of two or more outpatient visits within six months. Adequately treated patients reported fewer outpatient visits over time, even though they still met the baseline definition of two visits within six months. The study team suggests the finding reflects the improving health of the patient population. The use of indicated PCP prophylaxis improved the least of any measure with 26% still not receiving it by 1998 despite having CD4 cells under 200.

The use of antiretroviral treatment also rose rapidly during the follow-up period. Only 17% of patients in the study received indicated antiretroviral treatment in early 1996, and this figure rose to 85% in 1998. Antiretroviral use was inversely related to CD4 count, with those at the lowest CD4 level receiving the most treatment.

While antiretroviral treatment use rose in all subgroups, differences persisted. Certain subgroups were offered antiretroviral treatment at a later point in time than other subgroups, and rates of use were not equalized by the time the study ended. Women and blacks were still less likely to receive antiretroviral treatment, even when differences in baseline CD4 counts were controlled. Privately insured patients were more likely to receive treatment compared to uninsured or Medicaid patients. The rate of antiretroviral use between Medicaid and uninsured patients was almost indistinguishable in 1998.

The study team interprets these differences by quantifying the number of months individuals in each subgroup waited for treatment. An average man with HIV in the study waited 11.2 months from the beginning of 1996 to start antiretroviral treatment, while women waited on average 13.5 months. Whites waited, on average, 10.6 months, while blacks waited 13.5 months. Uninsured patients waited the longest, an average of 13.9 months, Medicaid recipients waited 12.4 months, and the privately insured received treatment the fastest 9.4 months. And Southerners waited longer than patients in any region, an average of 13.4 months.

In summary, the RAND study provides the most complete picture yet of HIV health care trends. Antiretroviral treatment was uncommon in early 1996, but its use rose rapidly over the next two years. By that time, a majority of patients received indicated treatment. Differences among demographic groups in access to and utilization of health care were large at the beginning of the study and declined but were not eliminated over the two year period. Individuals with private health insurance fared much better than those on Medicaid or the uninsured, and people in managed care plans did better than others with private insurance. Most importantly, the vast majority of people with HIV who receive medical care in the United States have advanced disease. Very few people with early stage HIV infection in any region or demographic group received regular medical care from 1996 to 1998.

Diagnosing HIV Infection

Why are so few early stage patients receiving medical care? The RAND team concludes by saying their result "imply that the HIV-infected adults who are not receiving regular care are primarily those with early and quite possibly unrecognized HIV infection." With the growing emphasis on early treatment, this result seems at odds with much of the discussion in the community and among academics. One possible answer is that people find out they are infected with HIV but then delay seeking medical care for a variety of economic, social or psychological reasons. Perhaps the news is so traumatic, many people shelve the information until their health declines.

A 1998 study from Boston examined a group of people with HIV in medical care who had been infected with HIV heterosexually or through intravenous drug abuse to ascertain if and how long they had delayed seeking medical care once they had learned that they were HIV-infected.⁽⁴⁾ The study found 32% of the participants delayed seeking medical care for over a year from the time they were diagnosed with HIV-infection, and 18% delayed treatment for over five years. The reasons associated with delayed entry into medical care include male gender, activate substance abuse, not having a spouse or living mother, having a history of incarceration, and receiving an HIV test result by phone or mail rather than in person.

Another possible explanation for the preponderance of advanced patients in medical care is that the diagnosis of HIV infection occurs late in disease. The AIDS Patient Survey, a federal study of 2,801 people with AIDS across the country, lends support to this argument.⁽⁵⁾ The study examined the time from HIV testing to AIDS diagnosis in this population. Since HIV infection typically takes many years to cause AIDS, the time between HIV testing and AIDS diagnosis is an important indicator of HIV testing delay. The study found most AIDS patients had been infected for more than five years before they had been tested for HIV. The time from HIV test to AIDS diagnosis ranged from two years to less than one, with 20% of the population receiving their HIV test and AIDS diagnosis simultaneously.

Another study demonstrated that a substantial number of at-risk people delayed seeking HIV testing. Beginning in 1996, a research team surveyed people in nine states to understand what factors caused HIV-testing delays.⁽⁶⁾ The study included people who had some sexual or drug-using risk for HIV-infection but who had not yet been tested for HIV. Participants in the study could list multiple reasons why they delayed HIV testing. Just under one-half of the sample (48%) said they were afraid to find out their result, thought they were HIV-negative, or they didn't want to think about HIV. Forty-three percent said they felt they were unlikely to have been exposed, and thirty-two percent said there was little they could do if they were infected. Nineteen percent said they were worried their name would be reported to the government, and seventeen percent feared others might think they had AIDS if they sought testing.

HIV Testing Trends in New York

To help understand how HIV testing is integrated into HIV health care, information from the 1993 and 1996 Behavioral Risk Factor Surveillance Surveys, annual surveys conducted in each state, were reviewed for New York State.⁽⁷⁾ Funded by the Centers for Disease Control, the BRFSS is a random-digit-dial telephone survey of health behaviors in the civilian, non-institutionalized adult population. The survey is a unique source of information on HIV testing trends, particularly in the private sector. The large size of the survey allows a scientifically valid comparison of differences in health behavior among the states. Although this survey looks only at New York State, the same trends in HIV testing have been seen in every state.

In 1993 and 1996, the New York State BRFSS included questions on HIV testing history. The 1993 BRFSS surveyed 1,984 New Yorkers, and the 1996 surveyed 3,594. The size of these surveys is large enough to detect differences in behavior between residents of New York City and residents of the rest of New York State. The surveys can also detect scientifically valid differences by sex and race. The BRFSS, however, only examines English-speaking, domiciled adults who have telephones and may not provide a valid indicator of testing trends in any other particular sub-population, such as age or high-risk category.

How Many New Yorkers Have Been Tested For HIV?

Between 1993 and 1996, the percentage of New Yorkers who say they had been tested for HIV increased substantially from 26 to 40%. Despite this progress, more than fifty percent of New Yorkers had still not been tested by 1996.

Looking more closely at data from 1996, several other important trends emerge. A much higher proportion of African Americans and Hispanics report being tested compared to whites. Men in New York report much higher rates of testing than women. In fact, New York is one of three states with a statistically significant disparity in HIV testing between men and women. New York City residents are also much more likely to test than residents of other parts of the state.

Why Do New Yorkers Get Tested For HIV?

In 1996, the BRFSS asked respondents who reported having been tested for HIV the reason why. The Centers for Disease Control groups these responses into two main categories: "voluntary testing" for personal or health-related concerns and "mandatory testing" for other reasons, such as immigration and military service.

Just over half of New Yorkers (54.7%) reported their last HIV test was voluntary. However, the rate of voluntary testing differed substantially among demographic groups. African Americans and Hispanics had the highest rates of voluntary testing with 65.7% and 70.5% respectively. Less than half (47.8%) of white New Yorkers who have been tested reported the test was voluntary. Females were much more likely to be voluntarily tested than males. Sixty-three percent of tested females in the survey report voluntary HIV testing compared to only 47.8% of males. However, if testing because of pregnancy is excluded from the voluntary category, only 47.6% of females in New York received a voluntary HIV test. New York City residents were more likely to be voluntarily tested (62.4%) than other residents of the state, of whom only 48% had received a voluntary test. (See following tables.)

---

Information on voluntary testing is crucial because research proves individuals who voluntarily receive HIV testing are more likely to undergo counseling and modify their risk behaviors. Mandatory HIV testing represents a significant proportion (45.3%) of HIV testing in New York State. Mandatory testing in this analysis refers to tests that are required by law, such as those for blood donation, immigration and military service, or are conducted for purposes other than for an individual's personal or health-related concerns. The largest single reason for a mandatory HIV test is purchasing a life insurance policy. Mandatory testing is generally not accompanied by risk reduction counseling and is not as likely to lead to behavior change. Like voluntary testing, the rate of mandatory testing differed substantially by demographic group. Whites, males, and residents of New York State outside New York City were more likely to receive a mandatory than a voluntary HIV test (52.2%, 52%, and 51.9% respectively). African Americans, Hispanics, New York City residents, and females were the least likely to be tested mandatorily for HIV.

---

Where do New Yorkers Get Tested For HIV? The 1996 BRFSS asked people who had been tested for HIV where they received their last HIV test. Overall New Yorkers receive HIV testing in many difference venues, with private physician offices the single most frequent site. Private hospitals and out-patient clinics closely follow. About 5% of New Yorkers sought HIV testing at dedicated counseling sites, and these sites were more heavily utilized by African Americans and Hispanics. Overall, 68.9% of New Yorkers have been tested in a private setting, either at their physician's office, a private hospital, a employer or insurance clinic, or at home.

---

---

How Many New Yorkers Have Not Been Tested For HIV? Another way to address the question of HIV testing delay is to examine the absolute number of HIV-infected people who may not yet have received an HIV test. To accomplish this, GMHC constructed a model to make some preliminary tests on the size of this population. The GMHC model extrapolates from existing data on government-funded medical care for HIV-infected people in New York to provide high and low estimates on the untested HIV-infected population. The model relies on the number of New Yorkers with HIV who receive medical care through the state Medicaid and AIDS Drug Assistance Program (ADAP). Assumptions were made about the proportion of HIV-infected New Yorkers receiving care through these two programs, and the percentage of untested HIV-infected people relative to HIV infected people in government-funded medical care. The New York State Department of Health supplied fiscal year 1996 county level Medicaid and ADAP data on the number of HIV-infected New Yorkers receiving care through these government health programs.(8) In the study period, 119,182 New Yorkers with HIV received medical care through the two programs. Although this figure conflicts with the number of patients reported in the RAND survey, the state figures were used since no other estimate was available. Research has shown Medicaid and ADAP provide health care for at least 65% percent and perhaps 85% of HIV-infected New Yorkers. High and low estimates of the total number of HIV-infected New Yorkers receiving either public or private medical care can be extrapolated when the number of people in publicly-funded care is known, and assumptions are made of the relative size of public and private care. Several studies estimate the size of the untested HIV infected population relative to the HIV-infected population who receive medical care. The GMHC model assumes New Yorkers with HIV who receive medical care represent between 67 and 85% of the total HIV-infected population in the state. The CDC estimates that one-third of the nation's HIV infected population have not been tested which was used in this study as the upper bound on the possible number of untested people. The lower bound was assumed to be 15% based a review of published literature. Studies of emergency department patients and hospitalized patients with acute pneumocystis carinii pneumonia have found previously undiagnosed HIV infection in ten to twenty-four percent.(9, 10,11,12,13) The limitations of this analysis are many. First and foremost, in simple arithmetic calculations, such as those used in this model, small changes in any one variable can exaggerate or minimize the final result. Data from the New York State Medicaid and ADAP programs are designed for billing purposes. Using these data for other purposes is highly speculative, opening the model to many possible errors. In fact, Medicaid data from Chautauqua county could not be used in this analysis because of coding errors in the original data set. Coding errors in data from other counties may have been unrecognized. The model does not offer any demographic information on untested New Yorkers, except geographic location. Information on race, sex, age, and risk factor for HIV infection cannot be determined in this model. In addition, the model does not take into consideration HIV-infected New Yorkers who may know their status but who have not entered medical care. The model has a particular problem with calculations for New York City. Unlike data for the rest of the state, Medicaid data for city residents was described by zip code of residence rather than county. An attempt was made to assign zip codes to each borough, but the geographic area for some zip codes could not be determined. In these cases, the zip code was assigned to the borough with the most similar zip code. Although the total estimate for New York City is as valid as any other in the model, the estimates at the borough level have an additional level of uncertainty. Despite these limitations, Dr. Perry Smith, chief of epidemiology at the New York State Department of Health said, "It sounds reasonable. We don't really have estimates of our own." A high and low estimate for HIV-infected New Yorkers who have not been tested in each county can be found on the attached chart. In total, the model estimates between 27,000 and 69,000 New Yorkers with HIV have not yet been tested. The median estimate is 51,000. The county-level estimates range from one to three untested people in Hamilton county to 6,300 to 16,000 in New York county (Manhattan). New York City residents represent almost eighty percent of the untested population in this model. However, Erie, Monroe, Nassau, Onondaga, Suffolk and Westchester counties each have high estimates above one thousand people.

---

Summary Over the past two years, health care for people with HIV has rapidly improved. New therapies have been widely adopted by patients who currently receive medical care. Traditionally disadvantaged populations, such as women, racial minorities and the poor, were initially slower to receive new therapies, but by 1998 antiretroviral treatment had been widely distributed. A large problem remains with many people with HIV unaware of their status, and many more who do not attend medical visits on a regular basis. Footnotes William M. Mercer, Inc. National Survey of Employer-Sponsored Health Plans: A Stratified Random Sample of All US Employers With 10 or More Employees. New York, 1997. Bozzette SA, et al. The care of HIV infected adults in the United States. New England Journal of Medicine. 1998;339:1897-1904. Shapiro MF, et al. Variations in the care of HIV infected adults in the United States. Journal of the American Medical Association. 1999;281:2305-15. Samet JH et al. Trillion virion delay: time from testing positive for HIV to presentation for primary care. Archives of Internal Medicine, 1998 April 13; 158:734-40. Hecht, F. AIDS Patient Survey. Unpublished. Bindman AB, et al. Multistate evaluation of anonymous HIV testing and access to medical care. Multistate Evaluation of Surveillance of HIV (MESH) Study Group. Journal of the American Medical Association. 1998;280:(16):1416-20. New York State Department of Health. Behavioral Risk Factor Surveillance Survey. Unpublished. New York State Department of Health. Unpublished. Trepka MJ, et al. Extent of undiagnosed HIV infection in hospitalized patients. American Journal of Preventive Medicine 1996;12:195-202. Hays RB, et al. Actual versus perceived HIV status, sexual behaviors and predictors of unprotected sex among young gay and bisexual men who identify as HIV-negative, HIV-positive and untested. AIDS 1997;11:1495-1502. Schwarcz SK, et al. Prevention of Pneumocystis carinii pneumonia: who are we missing? AIDS 1997;11:1263-8. Shuter J, et al. Rates of and factors associated with self-reported prior HIV testing among adult medical patients in an inner city emergency department in the Bronx, New York City. Journal of AIDS 1997;14:61-6. Duchin JS, et al. Risk factors for Pneumocystis carinii pneumonia: delayed diagnosis of HIV infection and failure to receive prophylactic therapy. XI International Conference on AIDS 1996; Abstract Tu.B.114.

Back to the GMHC Treatment Issues May/June 1999 contents page.