The Federal government spent $12.6 billion in fiscal year 2006 on medical care for persons living with HIV in the United States.1 The Health Resources and Services Administration (HRSA), the federal agency that administers funds allocated under the Ryan White Comprehensive AIDS Resources Emergency (CARE) Act, spent $2.1 billion, and the Centers for Medicare and Medicaid spent $9.5 billion for the care of persons living with HIV ($6.3 billion for Medicaid and $2.1 billion for Medicare). A large proportion of the Ryan White CARE Act funds support services intended to reduce the need for hospital care (eg, case management, clinic visits, outpatient drugs, home health care, hospice care, day care, nutrition services).2
This study examines hospital utilization by patients living with HIV. The goal of this study is to compare inpatient utilization and costs in 2000 with inpatient utilization and costs in 2004 to see if initial trends observed in the first few years after the diffusion of highly active antiretroviral therapy (HAART) have persisted further into the HAART era. In addition, this study derives per capita estimates of use and costs on a population basis by combining information about the number of persons living with HIV in a specific area with information about the utilization of hospital services by persons living with HIV in that area.
Early attempts to estimate HIV costs focused on the lifetime cost of treating a person with AIDS, and many early studies combined lifetime cost estimates with the number of persons living with AIDS to estimate the cumulative cost of AIDS to the nation.3
Several cost studies were conducted before the first drug to treat HIV was approved in March 1987. The first widely cited study of the cost of AIDS was conducted by Hardy and colleagues4 at the Centers for Disease Control and Prevention (CDC), and this study estimated the lifetime cost of AIDS care to be $147,000. In the study by Hardy and colleagues,4 hospital costs accounted for virtually all the costs of treatment. Yet, many years later in the current era of HAART, hospital care remains a significant contributor to the cost of treating persons living with HIV. Hospital costs accounted for 38% of the total cost of HIV care in 1998 in a widely cited study by Bozzette and colleagues.5
Various studies have revealed large differences in the use of hospital services by patients living with HIV across geographic areas.6-10 In particular, studies have revealed that HIV-infected patients who receive care in the Northeast are hospitalized for longer periods than HIV-infected patients who receive care in the West. The variation in the length of a hospital stay across geographic areas is not a special characteristic of HIV, and there is evidence indicating that where people live affects how often people are hospitalized and how long they spend in the hospital.11-14
Several recent studies have analyzed the probability of a patient with HIV being hospitalized. Floris-Moore and colleagues15 examined hospitalization rates among a cohort of 604 HIV-infected drug users in the Bronx, New York, and found that patients on HAART had lower hospitalization rates than those not on HAART. They also found that the rate of hospitalization between 1992 and 1996 was lower than the rate between 1997 and 2000.
In contrast, most recent studies have revealed reductions in the probability of a person living with HIV being hospitalized after the introduction of HAART. A study by Hellinger16 of HIV-related hospitalizations in 8 states found that the number of hospitalizations fell from 114,885 in 1996 to 77,694 in 2000, and a subsequent study by Gebo and colleagues17 examining HIV-related hospitalizations in 12 states found that the number of hospitalizations fell from 128,754 in 1996, to 96,811 in 1998, and to 91,398 in 1996. A study by Fleishman and Hellinger18 of HIV-related hospitalizations in 7 states also revealed a decline in the number of inpatient admissions between 1996 and 2000 and found that the rate of decline diminished over this period. In each of the aforementioned studies, the reductions occurred even though the number of persons living with HIV increased over the period.19 In a study of the utilization of hospital services by patients enrolled at 11 of the 17 sites that comprise the HIV Research Network (HIVRN), Fleishman and colleagues20 found that the mean number of hospitalizations per person per year decreased from 0.40 to 0.35 between 2000 and 2002.
This study compares the utilization of hospital care by all persons living with HIV residing in 6 states (California, Florida, New Jersey, New York, South Carolina, and Washington state) in 2000 with the hospital care utilized by all persons living with HIV residing in the same states in 2004. The 6 states in our sample include high- and low-prevalence states and states in each major geographic region of the nation. Moreover, the 6 states in this study include almost 50% of the persons living with AIDS in the United States.19
Data from the hospital discharge abstracts of 91,343 HIV-related admissions in the 6 states in 2000 and data from all 72,829 hospital discharge abstracts of persons living with HIV in the same 6 states in 2004 are used in this study. Information about utilization at all hospitals in a state by persons living with HIV was obtained from the Healthcare Cost and Utilization Project (HCUP), State Inpatient Database (SID), which is maintained by the Agency for Healthcare Research and Quality (AHRQ).21
The HCUP contains hospital discharge data and is a federal-state-industry partnership to build a multistate health care data system.21 Data are obtained from state governments and state hospital associations. The SID includes data on inpatient stays from all community hospitals. The AHRQ secures the data from each statewide data organization and transforms each data set into a common format. The SID file for each state is returned to that state, and distribution of the SID for each state is supervised by the statewide data organization that provided the initial data set to the AHRQ.
Data from the CDC19,22 and 2 state health departments23,24 were used to estimate the number of persons living with HIV. Data from the CDC were used to estimate the number of persons living with HIV in 4 states (Florida, New Jersey, New York, and South Carolina), whereas estimates of the number of persons living with HIV in California were obtained from data provided by the California Department of Health Services23 and estimates of the number of persons living with HIV in Washington state were obtained from data provided by the Washington State Department of Health.24 Data from each of these sources were combined with data from the AHRQ on the utilization of hospital services by persons living with HIV to derive population-based measures of hospital utilization (ie, to measure the average utilization of hospital care per person living with HIV in each state).
The CDC does not report estimates of the number of persons living with HIV in 2000 or 2004 for California or Washington state in its HIV/AIDS surveillance reports because neither state used confidential name-based systems to identify persons living with HIV in 2000.25 Both states used a name-to-code-based system in 2000, and Washington state adopted a confidential name-based reporting system in 2001, whereas California adopted a confidential name-based reporting system in 2006.
In the most thorough study of the accuracy of diagnostic coding for persons living with HIV, it was found that 97% of persons with an HIV diagnosis on their hospital discharge abstract were infected with HIV.26 In that study, more than 7000 hospital records of persons in 6 states with diagnostic codes indicative of HIV were examined; it was determined whether or not an individual was infected using AIDS surveillance data from state health departments and a review of the medical charts. The authors concluded that “Undercoding of HIV did not appear to be a major problem in hospital records… Our findings suggest that, on hospital computerized records, HIV codes were predictive and sensitive for HIV.”26 In contrast, the predictive accuracy of using codes for AIDS-related illnesses was poor. For example, it was determined that only 38% of patients with the diagnostic code for Pneumocystis jiroveci pneumonia (PCP; 136.3) and without an HIV code were infected with HIV, and it was determined that only 23% of patients with a diagnosis of cryptococcosis (117.5) and without an HIV code were infected with HIV.
After October 1994, the International Classification of Diseases, Ninth Revision, Clinical Modifications (ICD-9-CM) system included only 1 code (042) for HIV and AIDS.27 This study uses a conservative procedure for detecting hospitalized patients living with HIV and identifies only those patients with a primary or secondary code diagnosis (most states include up to 6 secondary diagnoses) that includes 042 as patients living with HIV. Thus, patients who are infected with HIV but are not identified as such are not included in this analysis.
The average length of a hospital stay was calculated by the dividing the total number of hospital days for hospital admissions with a 042 diagnostic code by the associated number of hospital admissions. The number of hospital admissions per person living with HIV in each of the 6 states was derived by dividing the number of hospital admissions in each state that included a diagnosis of 042 by the number of persons living with HIV in each state.
Multivariate statistical techniques are used to estimate equations explaining the length of a hospital stay and the cost of a hospital stay.27 A negative binomial regression model is used to estimate the determinants of the length of stay.28 Negative binomial models are used when there is a discrete variable (eg, length of stay) and the mean is considerably less than the variance (ie, when there is overdispersion).29 The cost of a hospital stay was estimated using the usual multivariate linear regression analysis.
The average cost of a hospital admission was obtained by converting hospital charges to cost using a methodology developed by Friedman and colleagues30,31 at the AHRQ. This methodology derives a hospital-specific all-payer inpatient cost-to-charge ratio by matching American Hospital Association (AHA) survey data and Centers for Medicare and Medicaid Services (CMS) accounting database records.
Our estimate of the cost of hospital care is presented in 2004 dollars. Hospital costs in 2000 were translated into hospital costs in 2004 by multiplying by the ratio of the value of the Bureau of Labor Statistics' Hospital Component of the Medical Care Component in the Consumer Price Index for the year 2004 to the corresponding value for the year 2000. The cost equations were first estimated using the cost per hospital admission as the dependent variable and then using the natural logarithm of the cost per hospital admission.
A variety of independent variables were used to estimate the multivariate regression equation explaining the length of a hospital stay and to estimate the multivariate regression equation explaining the cost per hospital admission. One of the independent variables is the insurance status of a patient, and this variable was categorized as Medicare, Medicaid, private, self-pay, or other. Other independent variables are race, which is categorized as black, white, Hispanic, or other, and teaching status, which is a dichotomous variable. The teaching status of each hospital was determined using data from the American Medical Association on the existence of a residency program. If a hospital operated a residency program it was designated as a teaching hospital. To control for disease severity, we calculated the total number of diagnoses listed on the discharge abstract. We also created 2 dummy variables indicating 2 AIDS-related opportunistic conditions that typically require high-intensity resource use: cytomegalovirus retinitis (CMV, indicated by ICD-9 code 078.5, and PCP, indicated by ICD-9 code 136.3).32
Both estimates were derived using traditional linear models,33 and estimates derived using generalized estimating equations (GEE) models33 are derived under the usual assumptions that the relation between the dependent and independent variables is linear and that the error term is normally distributed with mean 0. One of the central suppositions of traditional linear models is the serial independence of the error terms. If a longitudinal data set is used, however, the assumption of serially independent error terms is not credible. Consequently, a GEE model was used because it does not assume that error terms are distributed independently.34 Instead, the GEE model is estimated under the assumption of serial correlation among the error terms associated with repeated measurement of cross-sectional units over time.
Table 1 presents descriptive statistics for the study variables in 2000 and 2004. Over this period, the average age of a hospitalized patient living with HIV rose from 41 to 44 years of age and the average number of diagnoses for each patient rose from 6.0 to 7.4. The average length of an HIV-related hospital stay, the percentage of female patients, and the percentage of black patients remained the same from 2000 through 2004, however.
Table 2 presents data on the number of hospital admissions and the number of persons living with HIV in each of the 6 states for the years 2000 and 2004. Over the 4 years from 2000 through 2004, there was a 20% reduction in the number of hospital admissions by persons living with HIV (91,343 in 2000 and 72,829 in 2004) and there was a 28% increase in the number of persons living with HIV (256,386 in 2000 and 334,721 in 2004) (see Table 2). California, Florida, New Jersey, and New York experienced a reduction in the number of HIV-related hospitalizations and an increase in the number of persons living with HIV. South Carolina and Washington state experienced a slight increase in the number of HIV-related hospitalizations and a greater proportional increase in the number of persons living with HIV.
Table 3 presents data on the average annual number of hospital admissions per person living with HIV in each of the 6 states for the years 2000 and 2004. The reduction in the number of admissions accompanied by the increase in the number of persons living with HIV (see Table 2) resulted in a 39% decrease in the average number of admissions per person living with HIV across the 6 sample states between 2000 and 2004 (see Table 3). The average number of admissions per person living with HIV ranged from a high of 0.51 in New York to a low of 0.17 in Washington state in 2000. These averages fell, and 4 years later, they were 0.26 in New York and 0.16 in Washington state. Again, these averages were the highest and lowest among the 6 states. Nevertheless, the magnitude of these changes varied significantly across the 6 states in our study sample. The largest percentage reduction in the average number of admissions per person living with HIV occurred in New York (49%), whereas the smallest reduction occurred in Washington state (6%).
Table 4 presents data on the average monthly hospital cost per person living with HIV for each of the 6 states in the years 2000 and 2004. The sizable decrease (39%) in the number of admissions per person living with HIV in our sample (see Table 3) led to a commensurately large decrease (44%) in the average monthly hospital cost per person living with HIV (see Table 4). New York experienced the largest reduction in the average monthly hospital cost per person living with HIV (52%), whereas Washington state experienced the smallest reduction (15%). New York still had the highest level of hospital costs per person living with HIV, however, and Washington state had one of the lowest levels (the average monthly hospital cost per person living with HIV in Washington state was $199 in 2004, and it was $330 in New York).
Table 5 presents data on the average length of a hospital stay for each of the 6 states for years 2000 and 2004. The longest average length of stay in 2000 was in New York (9.3 days), and the shortest was in Washington state (6.4 days). In 2004, Washington state still had the shortest average length of a hospital stay for a patient with HIV (6.1 days), but the longest was in South Carolina (9.1 days). Indeed, the average length of stay increased in 4 of the 6 states from 2000 to 2004, and the overall average length of stay increased slightly from 2000 (8.3 days) to 2004 (8.4 days) (see Table 1).
Results from the multivariate regression analyses of the determinants of the length of a hospital stay for HIV-infected patients are presented in Table 6. After taking into account the effect of gender, race, insurance status, teaching status, age, CMV retinitis, PCP, and number of diagnoses, the length of stay in California was found to be 2.5 days shorter than the length of stay in New York (the reference state). Table 6 also indicates that the length of stay in Washington state was 2.75 days shorter than in New York, whereas the length of stay in Florida was approximately 1 day shorter. The length of stay in New Jersey was not found to be significantly different than the length of stay in New York, and the length of stay in South Carolina was approximately 0.25 day longer than in New York. Black patients were hospitalized 0.5 day longer than white patients, and Medicare patients were hospitalized 1 day less than privately insured patients.
Results from the multivariate regression analysis of the determinants of the average cost per hospital admission are presented in Table 7. The dependent variable is equal to the natural logarithm of the average cost per stay, and this model indicates that the cost of a hospital stay in California is 19% lower than in New York. New York is the reference category for the geographic variable, and New Jersey has the highest average cost per hospital stay. Medicare patients experienced lower costs than private-pay patients, and female patients experienced lower costs than male patients. In addition, patients with more diagnoses, a diagnosis of CMV retinitis, and a diagnosis of PCP experienced higher costs.
This study found that the average age of a hospital patient living with HIV increased from 41 years in 2000 to 44 years in 2004. This increase represents the continuation of an important trend. In a study of hospital patients living with HIV in 9 states, Hellinger16 found that the average age of an HIV-infected patient was 38 years in 1996 and 41 years in 2000.
The increasing age of hospitalized HIV-infected patients reflects similar trends in the age of persons diagnosed with AIDS and in the age of persons living with AIDS. In 2000, 15% of persons diagnosed with AIDS were 50 years of age or older, whereas 19% of persons diagnosed with AIDS in 2004 were 50 years of age or older.19,22 Moreover, in 2000, the share of persons living with AIDS who were 50 years or older was 19%, and in 2004, this share had grown to 23%.
The aging of the HIV population has important consequences for those responsible for the treatment and care of persons living with HIV. For instance, there is evidence that older patients with HIV do not achieve the same increase in CD4+ cell count response as younger patients35,36 after initiation of HAART, and there is evidence that persons diagnosed at an older age are more likely to die within 3 years after a diagnosis of AIDS.37 Indeed, the simultaneous increase in the number and age of persons living with HIV is having an important major impact on the scope and cost of HIV care.
The increase in the average age of hospitalized patients with HIV suggests that a growing proportion of these patients are likely to be treated for conditions that affect the general population of hospital patients, such as gastrointestinal disorders, cardiovascular problems, and cancer. A recent study of the patterns of diagnoses in a large cohort of HIV-infected patients (the HIVRN) revealed that most hospitalizations occurred for reasons other than an AIDS-defining condition and that the most common comorbidity for patients with HIV was gastrointestinal disorders.38 The most common gastrointestinal diagnoses were pancreatitis and liver diseases. In a study of hospital admissions caused by HAART-related toxicities in Madrid, Spain, Nunez and colleagues38 found that liver toxicities were the main reason for hospitalization. They also found that the proportion of patients admitted with an opportunistic infection had decreased and that the proportion of patients with drug-related toxicities had increased.
Another important change in the characteristics of hospital patients living with HIV is the increase in the proportion of patients who are covered under the Medicare program (22% in 2000 and 25% in 2004). The rise in the percentage of patients covered by Medicare was accompanied by a drop in the percentage of patients covered by Medicaid (52% in 2000 and 49% in 2004) and a drop in the proportion of patients covered by private insurance (15% in 2000 and 12% in 2004). This finding reflects the fact that many persons living with HIV are surviving long enough to become eligible for Medicare under the disability criteria. Even though these patients often are still eligible for Medicaid, the Medicare program is identified as the primary payer, because Medicare pays the bulk of the hospital bills for dual eligibles. In the early years of the HIV epidemic, few patients survived long enough to qualify for Medicare (in general, a person must be disabled for 29 months to be eligible for Medicare). For example, a 1985 hospital survey found that only 2% of 5393 patients with HIV were covered under Medicare.8
The fact that this study found that the average length of a hospital stay for an HIV-infected patient rose from 8.3 days in 2000 to 8.4 days in 2004 is notable, because the average length of a hospital stay for an HIV-infected patient had been decreasing for at least a decade. A survey of hospitals conducted in 1985 by the National Association of Public Hospitals and the Association of American Medical Colleges found that the average length of a hospital stay for an HIV-infected patient was 19 days,8 whereas interviews with 1164 patients living with HIV in 1992 who were being treated at 26 sites in 10 cities (the AIDS Cost and Service Utilization Survey) found that the average length of a hospital stay for an HIV-infected patient was 13 days.40 A more recent study of hospital abstracts of HIV-infected patients from 9 states in 1996 obtained from the HCUP found that the average length of a hospital stay for an HIV-infected patient was 10 days.16
Moreover, the findings that the proportion of hospital patients living with HIV who were female remained the same (34%) from 2000 to 2004 and that the proportion of patients who were black remained the same (51%) are also notable, because studies have shown that the proportion of hospital patients living with HIV who were female and the proportion of hospital patients living with HIV who were black increased steadily through the 1990s.3,16 For example, a study of HIV-related hospitalizations in 8 states revealed that 28% of hospital patients living with HIV were female and 46% were black in 1996 and that these percentages rose to 32% and 51%, respectively, for the same 8 states in 2000.16
This study also found that hospital care for persons living with HIV varied significantly across states. In particular, the average number of admissions per person living with HIV varied significantly across states, as did the average length of a hospital stay. Nevertheless, it was found that the average number of admissions per person living with HIV decreased between 2000 and 2004 (see Table 3).
There are many possible reasons for variations in the utilization of hospital care by persons living with HIV across states. For example, states have different economic conditions (eg, proportion of population without insurance, who are unemployed, who earn more than the poverty level) and population profiles (eg, states have different racial compositions, income levels, religious beliefs, attitudes towards health care). Moreover, states have different health care regulations (eg, some states have any-willing-provider laws, certificate of need legislation, strong bans on the corporate practice of medicine), health care institutions (eg, some state have many for-profit hospitals and few, if any, large academic health centers), and health care policies and programs.
It is important to understand the limitations of this study. First, the findings in this study are influenced by the stream of patients living with HIV who receive hospital care in a state different from the one in which they reside. This study merges data on the number of persons with AIDS living in each state with data about the consumption of hospital care in that state by persons living with HIV. Nonetheless, we are not capable of detecting particular individuals, and it is apparent that some persons receive care in states other than the state in which they reside. Plainly, these issues are of greater concern for certain states than for others. For example, the problems of patients receiving care in a state other than the state in which they reside may be less of a problem for Florida than for other states, because there is relatively little border crossing in the high-population centers of Florida.
Second, this study provides information about geographic variations in the access of persons living with HIV to only a single health care resource, the hospital, and it is unclear whether or not access to hospital services is related to access to other health care services. It may be argued that geographic variations in access to hospital care should not be as great as geographic variations in access to a variety of outpatient services, because there is generally less discretion regarding the decision to seek hospital care (ie, it is unlikely that a hospital would refuse to admit an HIV-infected individual in serious condition) than there is in the decision to seek outpatient prophylactic pharmaceutic therapies. Also, persons living with HIV who are insured may not have coverage for outpatient drugs or may have limited coverage (eg, persons with Medicare insurance) and may be unable to purchase expensive outpatient drugs.
Third, this study examines only 6 states. It is possible that the relation between race and ethnicity and hospital utilization and outcomes may differ in states not examined in this study. It also is possible that the relations discovered in the 6 states in this study are reflective of other states. In any event, approximately 50% of the persons living with AIDS in the United States were diagnosed in the 6 states examined in this study.
Fourth, data on the population under study (ie, number of persons living with HIV in a given state) were obtained from several different sources (the CDC and 2 state health departments), and the different methods used to estimate the number of persons living with HIV by each source is a limitation of this study. In particular, the estimates of the number of persons living with HIV derived by the 2 state departments of health used name-to-code-based systems, and this may overstate the number of persons living with HIV.
Fifth, the findings of this study reflect the changing nature of the epidemic within each state. Consequently, states with active surveillance and a greater proportion of cases diagnosed at an early stage may have a lower proportion of patients with HIV seeking hospital care. Also, states with high proportions of injecting drug users may have a greater proportion of residents with hepatitis C and may have a greater proportion of patients with HIV seeking hospital care. As a result, it is important to examine trends across states to help mitigate these effects.
Finally, this study is based on administrative data that do not include information about HIV risk factors, CD4 cell counts, HAART use, or other comorbidities (ie, hepatitis C virus [HCV]/hepatitis B virus [HBV]). Indeed, a study of the relation between cost, length of a hospital stay, and characteristics of the patient and hospital based on data from medical records could provide valuable information about the impact of clinical variables, but such a study would likely involve a relatively small number of hospital patients because of the high cost of reviewing and abstracting data from medical records.
Most persons living with HIV depend on public sources to pay for their care, and reliable information about the utilization and outcomes of care for persons living with HIV is important to decision makers who are responsible for allocating resources to treat persons living with HIV.3,41 This study found that the state in which a person living with HIV resides is a major determinant of the average number of hospital admissions per person living with HIV, of the average length of a hospital stay of a patient with HIV, and of the average hospital cost per person living with HIV.
The increasing age of persons living with HIV and the rising number of persons living with HIV in the United States make it difficult for policy makers and planners to allocate resources properly. Using population-based estimates of hospital use in 6 states, this study found a dramatic reduction in the utilization and cost of hospital services by person living with HIV between 2000 and 2004. This study also found that the proportion of patients covered by Medicare rose from 22% to 25%, that the average patient age rose from 41 to 44 years, and that the average number of diagnoses per hospitalized HIV-infected patient rose from 6.0 to 7.4 from 2000 through 2004.
Consequently, as patients with HIV survive longer, experience multiple changes in their drug regimens, and require treatment for many of the same diseases that afflict the general population, the task of providing high-quality care to patients living with HIV is likely to become even more challenging. Nevertheless, access to good information about how patients living with HIV use health care resources, including hospital services, remains a prerequisite for physicians and planners to ensure that persons living with HIV receive an appropriate mix of health care services.
1. Kates J. US Federal Funding for HIV/AIDS: The FY 2007 Request, HIV/AIDS Policy Fact Sheet. Menlo Park, CA: The Henry J. Kaiser Family Foundation; 2006.
2. AIDS Drug Assistance Programs (ADAPs). HIV/AIDS Policy Fact Sheet. Menlo Park, CA: The Henry J. Kaiser Family Foundation; 2002.
3. Hellinger FJ. Cost and financing of care for persons with HIV disease. An overview. Health Care Financ Rev. 1998;19:5-18.
4. Hardy AM, Rauch K, Echenberg D, et al. The economic impact of the first 10,000 cases of acquired immunodeficiency syndrome in the United States. JAMA. 1986;255:209-211.
5. Bozzette SA, Geoffrey J, McCaffrey DF, et al. Expenditures for the care of HIV-infected patients in the era of highly active antiretroviral therapy. N Engl J Med. 2001;344:817-823.
6. Hellinger FJ. Forecasting the medical care costs of the HIV epidemic: 1991-1994. Inquiry. 1991;28:213-225.
7. Andrulis DP, Beers JD, Bentley JD, et al. The provision and financing of medical care for AIDS patients in US public and private teaching hospitals. JAMA. 1987;258:1343-1346.
8. Andrulis DP, Weslowski VB, Gage LS. The 1987 US Hospital AIDS Survey. JAMA. 1989;262:784-794.
9. Andrulis DP, Weslowski VB, Hintz E, et al. Comparisons of hospital care for patients with AIDS and other HIV-related conditions. JAMA. 1992;267:2482-2486.
10. Goldman DP, Leibowitz AA, Shapiro MF, et al. The impact of state policy on the costs of HIV infection. Med Care Res Rev. 2001;58:31-53.
11. Manning WG, Leibowitz A, Goldberg GA, et al. A controlled trial of the effect of a prepaid group practice on use of services. N Engl J Med. 1984;310:1505-1510.
12. Greenfield S, Nelson EC, Zubkoff M, et al. Variations in resource utilization among medical specialties and systems of care. Results from the medical outcomes study. JAMA. 1992;267:1624-1630.
13. Wennberg JE. Future directions for small area variations. Med Care. 1993;5(Suppl):YS75-YS80.
14. Wennberg JE. Practice variations and the challenge to leadership. Spine. 1996;21:1472-1478.
15. Floris-Moore M, Lo Y, Klein RS, et al. Gender and hospitalization patterns among HIV-infected drug users before and after the availability of highly active antiretroviral therapy. J Acquir Immune Defic Syndr. 2003;34:331-337.
16. Hellinger FJ. HIV patients in the HCUP database: a study of hospital utilization and costs. Inquiry. 2004;41:95-105.
17. Gebo KA, Fleishman JA, Moore RD. Hospitalizations for metabolic conditions, opportunistic infections, and injection drug use among HIV patients: trends between 1996 and 2000 in 12 states. J Acquir Immune Defic Syndr. 2005;40:609-616.
18. Fleishman JA, Hellinger FJ. Recent trends in HIV-related inpatient admissions 1996-2000: a 7-state study. J Acquir Immune Defic Syndr. 2003;34:102-110.
19. US Centers for Disease Control and Prevention. HIV/AIDS Surveillance Report: Cases of HIV Infection and AIDS in the United States, 2004, vol. 16. Atlanta, GA: US Department of Health and Human Services; 2005.
20. Fleishman JA, Gebo KA, Reilly ED, et al. Hospital and outpatient health services utilization among HIV-infected adults in care 2000-2002. Med Care. 2005;43(Suppl 9):40-52.
22. US Centers for Disease Control and Prevention. HIV/AIDS Surveillance Report: US HIV and AIDS Cases Reported through December 2000, vol. 12, no. 2. Atlanta, GA: US Department of Health and Human Services; 2001.
23. California Department of Health Services. The Office of AIDS, HIV/AIDS Case Registry Section, HIV Cumulative Surveillance Report. Sacramento, CA: California Department of Health Services; 2005.
24. Washington State Department of Health. Washington State HIV/AIDS Surveillance Report-July 2006. Olympia, WA: Washington State Department of Health, IDRH Assessment Unit; 2006.
25. US Centers for Disease Control and Prevention. HIV/AIDS Surveillance Technical Report: HIV Infection in Areas Conducting HIV Reporting Using Patient Identifiers, 2000, vol. 1, no. 1, Atlanta, GA: US Department of Health and Human Services; 2001.
26. Rosenblum L, Buehler JW, Morgan MW, et al. HIV infection in hospitalized patients and Medicaid enrollees: the accuracy of medical record coding. Am J Public Health. 1993;83:1457-1459.
27. Fasciano NJ, Cherlow AL, Turner BJ, et al. Profile of Medicare beneficiaries with AIDS: application of AIDS casefinding algorithm. Health Care Financ Rev. 1998;19:19-38.
28. SAS Institute. SAS/STAT User's Guide, Version 8, vol. 2. Cary, NC: SAS Institute; 1999.
29. Cameron AC, Trivedi PK. Regression Analysis of Count Data. Cambridge University Press, Cambridge, UK; 1998.
30. Friedman B, De La Mare J, Andrews R, et al. Practical options for estimating cost of hospital inpatient stays. J Health Care Finance. 2002;29:1-13.
32. Moore RD, Chaisson RE. Costs to Medicaid of advancing immunosuppression in an urban HIV-infected patient population in Maryland. J Acquir Immune Defic Syndr Hum Retrovirol. 1997;14:223-231.
33. Greene WH. Econometric Analysis. 2nd ed. New York: Macmillan Publishing Company; 1993.
34. Liang KL, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73:13-22.
35. Moore RD, Keruly JC. CD4+ cell count 6 years after commencement of highly active antiretroviral therapy in persons with sustained virologic suppression. Clin Infect Dis. 2007;44:441-446.
36. Nogueras M, Navarro G, Anton E, et al. Epidemiological and clinical features, response to HAART, and survival in HIV-infected patients diagnosed at the age of 50 or more. BMC Infect Dis. 2006;6:159-167.
37. Hall HI, McDavid K, Ling Q, et al. Determinants of progression to AIDS or death after HIV diagnosis, United States, 1996 to 2001. Ann Epidemiol. 2006;16:824-833.
38. Betz ME, Gebo KA, Barber E, et al. Patterns of diagnoses in hospital admissions in a multistate cohort of HIV-positive adults in 2001. Med Care. 2005;43(Suppl):III-3-III-14.
39. Nunez MJ, Martin-Carbonero L, Moreno V, et al. Impact of antiretroviral treatment-related toxicities on hospital admissions in HIV-infected patients. AIDS Res Hum Retroviruses. 2006;22:825-829.
40. Hellinger F. The lifetime cost of treating a person with HIV. JAMA. 1993;270:474-478.
41. Aschman DJ. HIV Capitation Risk Adjustment Conference Report. Conference sponsored by the Health Resources and Services Administration, US Department of Health and Human Services, in conjunction with the Henry J. Kaiser Family Foundation, Washington, DC, May 28-29, 1997.
© 2007 Lippincott Williams & Wilkins, Inc.