Abstract
Surveys revealed increases in the prevalence of HIV-infected patients in the US end-stage renal disease (ESRD) program in the 1980s and early 1990s, with clustering in young black men 25 to 44 yr old. Since the availability of highly active antiretroviral therapy in 1996, the prognosis of HIV-infected patients has improved, and therapy has been shown to change the course of classic HIV-associated nephropathy. We used the United States Renal Data System database to determine if the incidence and prevalence of HIV-infected patients with renal disease has increased in the ESRD program, by means of principal diagnoses and comorbid AIDS-defining diagnoses.
As the number of US patients living with AIDS increased 57% from 214,711 in 1995 to 337,017 in 2000, and the number of incident ESRD patients increased 29.9% from 72,827 to 94,602, the number of incident HIV-infected patients increased only by 3.5%, from 1133 to 1171. Over this time, the percentage of incident ESRD patients with HIV infection fell from 1.56% to 1.24%. Among black men 25 to 44 yr of age, HIV infection as a proportion of incident ESRD cases fell from 8.5% to 6.2% from 1995 to 2000. The incident rate per million of AIDS or HIV infection in black men aged 25 to 44 fell from 107 in 1995 to 78 per million in 2000. The incidence rate for HIV-infected women in the ESRD program rose 14% while it declined 7% in men. Almost 2000 HIV-infected women, or 28.8% of the population, have initiated therapy for ESRD with hemodialysis. The number of prevalent cases increased in absolute numbers 81.3% from 2687 to 4871 (0.90% to 1.16% of the ESRD program). One-year survival rates for HIV-infected incident ESRD patients increased from 53.1% to 67.1% from 1995 to 2000.
Although these values may be underestimates because of underreporting due to confidentiality concerns and lack of biopsy confirmation, we conclude that although the prevalence of HIV infection is increasing in the US ESRD population, the increase as a proportion of the program is minimal and is due to better survival after development of renal failure. The incidence of HIV infection in the US ESRD program is stable. Highly active antiretroviral therapy may be responsible for the change in epidemiology of HIV infection in the US ESRD program.
Several different renal diseases have been noted to occur in patients with HIV infection (1,2⇓) since the first description of an AIDS-associated nephropathy 20 yr ago (3). Surveys by the Centers for Disease Control and Prevention (CDC) revealed exponential increases in the prevalence of HIV-infected patients from very low levels in the US end-stage renal disease (ESRD) program in the 1980s (4) and early 1990s (5). Assessment of the seroprevalence of HIV infection in US dialysis units in 1986 by the CDC revealed a prevalence of 0.98%, with a high cumulative incidence of infection in geographic areas in which a high prevalence of AIDS had been reported (6). In 1994, a survey by the CDC revealed the prevalence of known HIV infection in all US dialysis units to be 1.5% (7). Clustering of cases of ESRD associated with HIV infection and AIDS was noted in young black men 25 to 44 yr old (8). A review of the cumulative statistics of the ESRD program between 1992 and 1997 showed 0.97% of patients had HIV infection or AIDS, and that 87.8% of those patients were African American (9).
In 1996, an editorial by Winston and Klotman (8) that used data from the Health Care Finance Administration (HCFA) warned of the possibility of an epidemic of HIV infection in the US ESRD program. Highly active antiretroviral therapy (HAART) was initiated shortly thereafter, in 1996, with the introduction of the protease inhibitor indinavir in the United States (10). Since the availability of a variety of protease inhibitors and the widespread use of HAART, the prognosis of HIV-infected patients has improved (11), and case reports have indicated therapy can dramatically change the course of classic HIV-associated nephropathy (HIVAN) (12,13⇓). Several studies from different groups have suggested the prognosis of HIV-infected ESRD patients treated with HAART has improved (14–18⇓⇓⇓⇓).
We used data from the United States Renal Data System (USRDS) to determine what the incidence and prevalence of HIV infection and AIDS-associated renal diseases were in the US ESRD program from 1995 to 2000.
Materials and Methods
The data for this study were taken from the 2002 update of the standard analytic files from the USRDS standard analytic files (USRDS, 2002). The characterization of persons with AIDS and/or HIV infection and ESRD followed a number of steps. It was our intent to maximize inclusion of as many AIDS or HIV-infected patients in the ESRD program as possible. As such, we had no exclusion criteria for patients identified as having HIV infection or AIDS.
First, since 1983, the HCFA 2728 Medical Evidence Form has included AIDS as a cause of renal failure. This form was revised in 1995 to include comorbid conditions, among which are AIDS and HIV infection. For the calculation of incidence rates, persons were designated as having HIV infection or AIDS (HIV/AIDS) if the principal cause of ESRD was AIDS and/or there was an indication of AIDS or HIV as a comorbid condition. The addition of the HIV/AIDS comorbidity indicator increases the number of patients included by about 20%. Incidence as defined by a principal diagnosis of HIV/AIDS and/or HIV/AIDS as a comorbid indicator was defined as method 1.
It is well known that there can be underreporting of HIV. For example, on the Medical Evidence Form for comorbid conditions, “cannot disclose” is reported for about 70% of cases for AIDS and HIV. Thus, it is likely that the 2728 form will underreport HIV-related renal diseases. To help counter this downward bias, we examined the Medicare billing data for the AIDS ICD9 code (042). By means of a method previously validated by Hebert et al. (19), we assigned an AIDS/HIV code to persons with at least one hospitalization with an AIDS or HIV code or at least two outpatient encounters with an AIDS or HIV code. The latter is done to avoid “rule-out” diagnoses. To estimate prevalence rates of HIV infection and/or AIDS, we used both the 2728 definition (method 1) as well as a more expanded definition that used the billing data (method 2). Thus, for method 2, the prevalence estimates are based on the combination of method 1 and billing data. A person could be identified as having HIV infection or AIDS through either mechanism. The intent of analyses that used method 2, examining the billing data, is to identify the largest number of HIV/AIDS patients.
This method will identify the maximum number of patients with HIV infection or AIDS in the ESRD database. However, it is also true that not all ESRD patients in the USRDS database will have billing data. First, for persons who are not Medicare entitled at the time of renal failure, there is a 90 d waiting period until Medicare coverage begins. Second, there is also no bill data for Medicare beneficiaries who are enrolled in health maintenance organizations. Most importantly, however, is the effect of legislative provisions for secondary payers. Secondary payer means that if a person has other insurance, that other insurer pays for care, and Medicare is the secondary payer. Originally, this provision extended for 12 mo after the start of ESRD. It was subsequently extended to 18 mo and then 30 mo. Beginning in 1997, the period was made indefinite. As a result, Medicare is the primary payer for a decreasing percentage of patients, 74% in 1995 and 63% in 2000. Thus, there will be an unknown number of potential patients with HIV/AIDS that cannot be identified either through HCFA 2728 or billing data.
Regional variation in the incidence and prevalence of ESRD associated with HIV/AIDS was examined by identifying high-impact AIDS Metropolitan Statistical Areas (MSA) from the CDC public use data site (http://wonder.cdc.gov/). Fifteen MSA (Atlanta, GA; Baltimore, MD; Boston, MA; Chicago, IL; Dallas, TX; Fort Lauderdale, FL; Houston, TX; Los Angeles, CA; Miami, FL; New York, NY; Newark, NJ; Philadelphia PA; San Diego, CA; San Francisco, CA; and Washington, DC), accounting for half of all AIDS cases since 1980, were selected.
Mortality was calculated with 90 d after the date of renal failure used as the start date, per the USRDS methodology. A standard life table method was used for time until death with censoring at July 1, 2001. All calculations were performed by SAS software, version 8.2.
Results
The population at risk of developing ESRD due to HIV infection or AIDS has increased in a linear fashion in recent years, as the total number of persons living with AIDS increased 57%, from 214,711 in 1995 to 337,017 in 2000 (Figure 1). Despite the increase in the population of people with AIDS, the number of deaths due to AIDS has decreased, from a peak of 51,670 in 1995 to 16,672 in 2000 (20) (Figure 1). This great improvement in mortality is thought to be due to the widespread use of HAART (2). From 1995 to 2000, the prevalent ESRD population increased by 38.4% from 299,664 to 414,765. From 1995 to 2000, the incident ESRD population increased 29.9%, from 72,827 to 94,602.
Figure 1. Estimated persons living with AIDS and number of AIDS deaths, 1995 through 2000, United States. Data from the CDC.
The growth rate of ESRD patients with AIDS or HIV infection was greater than that of people living with AIDS. The prevalence of a diagnosis of HIV/AIDS in the ESRD population on the basis of Medical Evidence Form 2728 data (method 1) is shown in Table 1. From 1995 to 2000, the number of prevalent US ESRD patients identified with HIV/AIDS increased from 1347 to 3482, or 159% (Figure 2). This number of prevalent patients has grown much more rapidly than has the incidence (see below). The growth rates were greatest for persons between the ages of 45 and 64 yr and for women (Table 1). The number of women with ESRD and AIDS increased almost threefold during this time period. The percentage increase in prevalence was greater for black patients than for white patients. The prevalence rate of a diagnosis of HIV/AIDS in the ESRD program increased more than twofold from 5.1 to 12.7 per million during this 5-yr time span. HIV/AIDS accounted for 0.45% of the total ESRD prevalent population in 1995, increasing to 0.83% in 2000 (method 1) (Figure 3).
Table 1. Prevalent cases with HIV/AIDS, 1995–2000
Figure 2. Prevalent cases of ESRD with HIV/AIDS, 1995 to 2000.
Figure 3. Percentage of total prevalent ESRD patients with HIV/AIDS, 1995 to 2000.
Analyses that used the most expansive definition of renal disease associated with AIDS or HIV infection—that is, including persons identified with HIV infection or AIDS by use of a principal diagnosis or using a comorbid diagnosis of AIDS or HIV on the Medical Evidence Form or from billing data (method 2) (19)—suggest a lower rate of growth in the prevalence of AIDS or HIV infection in the ESRD program, from 2687 to 4871, or an 81% increase over the 5-yr period (Table 2, Figure 2), although the absolute number of patients included in this analysis is greater. The increase in prevalence was 100% for black and 14% for white patients. The prevalence rate of a diagnosis of HIV or AIDS in the ESRD program increased 73.5% from 10.2 to 17.7 per million during this 5-yr time span. By means of this definition (method 2), ESRD patients with AIDS or HIV infection accounted for 0.90% of ESRD patients in 1995 and 1.16% of patients in 2000.
Table 2. Prevalent cases with HIV/AIDS, 1995–2000
The number of incident patients identified with HIV/AIDS with it used as a diagnosis of the cause of the renal disease (method 1) decreased 9.6% from 1995 to 2000, from 941 in 1995 to 851 in 2000, as the general incident ESRD population increased (Figure 4). The number of incident patients identified with HIV/AIDS with it used as a principal diagnosis or its presence as a comorbid condition (method 2) remained essentially constant from 1995 to 2000, ranging from a high of 1133 in 1995 to a low of 1075 in 1997 (Table 3, Figure 4). From 1995 to 2000, 71% of incident patients with AIDS or HIV infection were male and 87% were black (method 1) (Figure 5); 61% were black men. However, even among black men, the number of incident cases of ESRD with AIDS or HIV infection is stable or declining, from 729 (46.4 per million) in 1995 to 694 (41.4 per million) in 2000 (method 1).
Figure 4. Total number of incident cases of ESRD with HIV/AIDS, 1995 to 2000.
Table 3. Incident cases with HIV/AIDS, 1995–2000
Figure 5. Percentage of ESRD incident cases with HIV/AIDS by demographic subgroup, 1995 to 2000.
Sixty-seven percent of patients with AIDS or HIV infection entering the ESRD program from 1995 to 2000 were between 25 and 44 yr of age (method 1). African American men 25 to 44 yr old account for only 2% of the United States population, but accounted for 40% of incident ESRD cases due to AIDS or HIV infection from 1995 to 2000 (method 1) (Table 4, Figure 5). HIV-infected black men between the ages of 25 and 44 comprised 8.1% of the cumulative incident ESRD patients in this population between 1995 and 2000 (method 1). The incident rate per million due to HIV/AIDS in black men aged 25 to 44 fell from 107 in 1995 to 78 per million in 2000 (method 1) (data not shown). Incident rates of ESRD per million due to AIDS or HIV infection in black men tended to drop from 1995 to 2000 in younger age categories (method 1) (data not shown). Among black men 25 to 44 yr of age, AIDS or HIV infection as a proportion of all incident ESRD cases fell from 8.5 to 6.2% from 1995 to 2000 (method 1) (Figure 6). In 45 to 64 yr olds, however, the incidence rate per million rose 29% from 5.0 to 6.4 per million (method 1) (data not shown).
Table 4. Number of incident cases of end-stage renal disease with HIV/AIDS by age, sex, and race, 1995–2000
Figure 6. Percentage of ESRD incident cases with HIV/AIDS for African American men ages 25 to 44, 1995 to 2000.
The incidence rate for HIV-infected women in the ESRD program rose 14% while it declined 7% in men (method 1) (Table 3). Almost 2000 HIV-infected women, or 28.8% of the population, have started therapy for ESRD. AIDS and/or HIV infection accounted for 1.56% of all incident ESRD cases in 1995 and decreased to 1.24% in 2000 (method 1). Incident rates for a diagnosis of AIDS or HIV infection remained at about 4 per million during this time period (method 1) (Table 3). Incident rates were approximately 2 1/2 times greater for men than for women, and were over 50 times greater for black compared with white patients (method 1).
When incidence in the ESRD program is categorized as diabetic disease, hypertensive disease, glomerulonephritis, polycystic kidney disease, other, unknown and missing, AIDS- and HIV-associated renal diseases are the fifth leading discrete cause of chronic kidney disease (method 1) (Figure 7). Renal disease with AIDS or HIV infection accounted for a cumulative 1.3% of all incident ESRD from 1995 to 2000 (Figure 7). Renal disease with AIDS or HIV infection accounted for 3.9% of all incident ESRD in African Americans from 1995 to 2000 (method 1). Renal disease with AIDS or HIV infection accounted for 18% of all incident ESRD in African American men aged 25 to 44 in 1995, but only 13% in 2000 (method 1).
Figure 7. Percentage of total ESRD incident cases with HIV/AIDS by cause of renal failure, 1995 to 2000 combined.
There were differences in HIV/AIDS identification of ESRD patients by region of the country, as measured by MSA. For example, the New York, NY, MSA accounted for about 15% of all AIDS cases from the CDC and accounted for 15% of all ESRD due to HIV/AIDS. However, ESRD due to HIV/AIDS appears to be undercounted in California. The MSA of Los Angeles, San Diego, and San Francisco account for over 10% of all AIDS cases reported by the CDC, but less than 2% of ESRD HIV/AIDS in the USRDS database.
One-year survival rates for HIV-infected incident ESRD patients increased from 53.1% to 67.1% from 1995 to 2000 (method 1) (Figure 8).
Figure 8. One-year mortality rate after entry into the ESRD program for incident cases, 1995 to 1999.
Discussion
Prevalent cases of ESRD in patients with HIV infection increased approximately twofold between 1995 and 2000. The incidence still clusters in black men aged 25 to 44 yr. HIV infection or AIDS as a percentage of the incident cases and incident rates per million in the entire population, and in black men, however, decreased from 1995 to 2000. In black men, the decrease was about 20% in the 25-to-44-yr-old age group. The incidence may have risen slightly in black men over the age of 44.
These prevalence values may be underestimates because of concerns regarding confidentiality in different jurisdictions, and because of lack of biopsy confirmation (2,21⇓). However, there is no reason to believe that underreporting would have changed during the time period covered by this study. Therefore, the trends are likely to be accurate.
Although the prevalence of renal disease in HIV-infected patients in the US ESRD program has increased, this is largely due to increased survival after the development of kidney failure. The incidence of HIV infection in the ESRD program remained unchanged and is decreasing as a percentage of all ESRD cases.
Although the incidence of the complication is decreasing, HIV infection remains an important risk factor for the development of ESRD. In 1995, the rate of ESRD among people living with AIDS, 5227 per million, was approximately 20-fold as great as that in the general population, 267 per million. In 2000, although the rate of ESRD among people living with AIDS had decreased 33.7% to 3467 per million, it was still approximately tenfold greater than that in the general population, 337 per million.
Recent work has closely linked the pathogenesis of HIVAN to infection of renal cells (1,2,12,13,21,22⇓⇓⇓⇓⇓). Case reports have emphasized the possible reversibility of established HIVAN in patients treated with HAART (12,13⇓). Recent epidemiologic reports have suggested that HAART is associated with decreased prevalence of renal disease and amelioration of progression in the general population and in HIV-infected patients with chronic renal disease (1,2,20,21,23–27⇓⇓⇓⇓⇓⇓⇓⇓). The increased prevalence and the decreased incidence of HIV infection in the US ESRD program may reflect earlier and more widespread treatment of patients with HIV infection with HAART, as well as improved survival of HIV-infected ESRD patients.
Although AIDS- and HIV-associated disease is the fifth leading discrete cause of ESRD, it is clear that as a proportion of the ESRD program, the association of HIV infection with ESRD is stable and that there is no epidemic increase of new HIV-associated renal diseases. (Missing, other, and unknown categories each comprise a larger number of patients than those listed as HIV/AIDS.) Our data on survival corroborate a recent report from Ahuja et al. (15). The quite dramatic 39.7% increase in survival has had a substantial effect on the epidemiology of the ESRD program. The finding cannot be plausibly attributed to improvement in renal replacement therapeutic techniques. Recent studies, however, have suggested that medications such as HAART and angiotensin-converting enzyme inhibitors, which might be associated with improved renal function and survival in this population are not maximally utilized in the ESRD program, suggesting survival statistics might be amenable to even greater improvement in the future (18,28,29⇓⇓). These findings are even more salient because it is well appreciated that newer antiretroviral agents can also cause nephropathy (1,2,17,27⇓⇓⇓).
The percentage increase in prevalence of HIV infection in the ESRD program was greater for black than white patients, and a disproportionate number of HIV-infected patients in the incident ESRD population are men of African descent. Although the huge, 50-fold disparity in the incidence of HIV-associated ESRD could be associated with socioeconomic factors, including intravenous drug use and access to care and antiretroviral drugs (1,2,16,17,21⇓⇓⇓⇓), it is conceivable that genetic differences between populations in the expression of renal chemokine receptors (30), involved as secondary mediators of HIV infection (30,31⇓), play a pathogenic role in the epidemiologic differences noted (2,21,31–33⇓⇓⇓⇓). This hypothesis is bolstered by the finding of clustering of renal disease in the families of HIV-infected ESRD patients (34). Special care must be taken regarding the evaluation of young African American men (and in light of the present findings older African American patients as well) with chronic kidney disease to make a diagnosis of an HIV-associated renal disease and institute appropriate therapy.
The prevalence rate for HIV-infected women in the ESRD program almost doubled or tripled (Tables 1 and 2⇑) during the study period, depending on the analytic technique used. The incidence of HIV-infected women in the ESRD program rose while it declined in men (Table 3), reflecting the demographics of the epidemic in the general population (35). Almost 2000 women, or more than a quarter of the cumulative population of HIV-infected patients, have initiated therapy for ESRD with hemodialysis, emphasizing that although the disease is more common in men, it certainly occurs in women. The relative risk for developing ESRD is 0.42 for women compared with men. Gender is not a reason to rule out a diagnosis of HIVAN or other HIV-associated renal disease (1,2⇓).
It is almost certain that the true burden of AIDS in the ESRD population is underestimated in the USRDS database. The HCFA 2728 reporting form shows that in many areas, reporting of AIDS and/or HIV infection is prevented by patient confidentiality. We attempted to overcome this limitation by examining billing data for indications of AIDS diagnoses. However, if another payer is responsible for health care costs, then billing data will not be available. Of the 1347 prevalent patients identified in 1995 (Table 1), 968, or 71.9%, had Medicare entitlement at the time of enrollment in the ESRD program. Of these patients, 84% had subsequent AIDS billing, compared with only 6% of the persons without initial Medicare entitlement. We thus conclude that the 2728 AIDS/HIV designation has relatively good sensitivity.
The specificity of the 2728 criteria is much less certain. It is evident from the regional variation that there is underreporting of AIDS/HIV in certain states, notably California. Thus, the use of billing data likely adds to the case ascertainment.
It is not likely that a “gold standard” for the presence of AIDS/HIV in the universe of patients with ESRD will be developed. Patient confidentiality issues in a number of states will prevent the identification of many patients through the 2728 mechanism (36). The identification through billing data is also problematic because many patients have other sources of insurance coverage. Efficiencies in data collection and abstraction, such as the Vision/SIMS system in Medicare, will improve data transmission and accuracy, but they do not override the confidentiality issues raised above. ESRD networks are exploring the use of data collection through the E-lab system, but the system does not mandate HIV testing.
The lack of complete billing data for all persons suggests that our enumeration was less than complete. Still, there is no reason to hypothesize that the level of reporting has changed significantly during these years, so we believe that the trend data are probably accurate.
In addition, the validity of this analysis is bolstered because it is consistent with CDC dialysis facility surveillance data, using different ascertainment strategies (1,2,5⇓⇓). Tokars et al. (5) used a questionnaire mailed to all chronic hemodialysis units licensed by the HCFA in conjunction with the HCFA annual facility survey for calendar year 2000, covering, among other issues related to hemodialysis practices and infectious diseases, the number of patients with HIV infection. They recorded a 0.7% prevalence of AIDS and a total 1.4% prevalence of HIV infection in 1995, and a 0.4% prevalence of AIDS and a total 1.5% prevalence of HIV infection in 2000. By use of the USRDS database, we found AIDS or HIV infection accounted for a cumulative incidence of 1.3% from 1995 to 2000, and a prevalence of 1.16% in 2000. Thus, we believe the trend data reported here to be accurate.
Another inherent limitation of this study is the inability to link information regarding antiretroviral drug prescription and use with outcomes. Such data, however, are only available in much smaller population samples (28). Another limitation is that we cannot ascertain that the patients have HIVAN, compared with other renal diseases (1,2⇓). We also cannot accurately ascertain the number of hepatitis C virus (HCV)-infected patients coinfected with HIV, who may have different prognostic outcomes (37), by use of these databases. The 2728 form, which codes for underlying renal disease, and the comorbid conditions section do not include HCV infection, although they do include HIV/AIDS. Billing data would only delineate HCV infection if it were specifically treated by a nonnephrologist Medicare billing physician. Analysis of diagnostic data in the USRDS regarding HCV infection in the HIV-infected population (data not shown), not surprisingly, led to counterintuitive results, indicating different patient populations designated in the database. Many HIV-infected ESRD patients with HCV as a coexisting infection may not be coded as such on the 2728 forms. More extensive use of biopsies in this population would also improve the ability to establish trends in large databases. However, our definitions have been designed to be inclusive and present a worst-case analysis regarding HIV infection in the ESRD program. Under these constraints and limitations, it is clear with more prevalent use of HAART and better preventive measures, HIV infection is unlikely to be epidemic in the US ESRD program.
- © 2004 American Society of Nephrology
References
Jump to section
More in this TOC Section
Cited By...
- Association of Kidney Function and Early Kidney Injury With Incident Hypertension in HIV-Infected Women
- Association of Urine {alpha}1-Microglobulin with Kidney Function Decline and Mortality in HIV-Infected Women
- Race, Kidney Disease Progression, and Mortality Risk in HIV-Infected Persons
- Association Between Kidney Function and Albuminuria With Cardiovascular Events in HIV-Infected Persons
- NPHS2 Variation in Sporadic Focal Segmental Glomerulosclerosis
- Racial Differences in End-Stage Renal Disease Rates in HIV Infection versus Diabetes
- Chronic Nephropathies of Cocaine and Heroin Abuse: A Critical Review
- Highly Active Antiretroviral Therapy and the Epidemic of HIV+ End-Stage Renal Disease