AIDS treatment costs during the last months of life: evidence from the ACSUS

AIDS treatment costs during the last months of life: evidence from the ACSUS – AIDS Costs and Service Utilization Survey

Fred J. Hellinger

The relatively long period from the time of diagnosis to the time of death for most diseases makes it difficult to estimate lifetime medical care costs. In an article that estimates the medical care cost of treating persons with AIDS, Scitovsky and colleagues state that “unfortunately, while there are data on annual medical care costs for some other diseases, there are no data on their lifetime costs” (Scitovsky, Cline, and Lee 1986, 3,106).

Nonetheless, there is evidence that, as the time of death approaches, the cost of treatment increases (Riley et al. 1987). Baker and colleagues have shown that the medical care cost of lung and breast cancer increases tenfold during the time period immediately preceding death (Baker et al. 1991), and Lubitz and Riley (1993) have shown that between 27 and 30 percent of Medicare payments are attributable to persons in their last year of life.

Existing studies report that AIDS patients receive substantial amounts of health care services during the period immediately preceding death (Scitovsky et al. 1986; Scitovsky and Rice 1987; Solomon et al. 1989; Lee, Durbin, and Kahn 1991; Merzel, Sambamoorthi, and Crystal 1991; Solomon and Hogan 1992). There also is evidence that the cost of care immediately following diagnosis is high (Scitovsky, Cline, and Lee 1986; Lee, Durbin, and Kahn 1991). The care received during the period of time between diagnosis and death is less intensive than the care received immediately following diagnosis or the care immediately preceding death because patients spend much of this time outside of the hospital. This results in a U-shaped cost curve.

Information from existing studies on the cost of treating persons with AIDS (PWAs) reflects patterns of care that existed several years ago; all existing studies of resource utilization for PWAs immediately preceding death reflect patterns of care before 1990 and are based on data obtained from a single geographic area or from a single insurer. This study examines the utilization and cost of medical services by PWAs who were respondents in the AIDS Costs and Service Utilization Survey (ACSUS), which was conducted in 1991-1992.

This study compares the cost of services consumed by PWAs who at the end of data collection were still alive (survivors) to the cost of services consumed by PWAs who died during the last three months of the study (decedents). This study also examines characteristics of decedents that help explain variations in their use of services.

BACKGROUND

In one of the first studies of the cost of treating PWAs, Scitovsky and colleagues analyzed data on 201 AIDS patients who received all their inpatient and outpatient care at San Francisco General Hospital in 1984 (Scitovsky, Cline, and Lee 1986). They found that the lifetime cost curve for AIDS care was U-shaped. For those PWAs who were diagnosed in 1984 (n = 83), the mean monthly charges were $2,617 ($12,040/4.6); for those who were diagnosed prior to 1984 and who lived all 12 months (n = 30) in 1984, mean monthly charges were $586 ($7,026/12); and for those PWAs who died in 1984 (n = 88) mean monthly charges were $3,660 ($23,425/6.4).

In a subsequent study, Scitovsky and colleagues analyzed the expenditures of seven PWAs who did, and seven PWAs who did not, receive AZT during a 12-month period from 1986 to 1987 at San Francisco General Hospital (Scitovsky, Cline, and Abrams 1990). Although the focus of this study was on the effect of AZT use on medical care costs, the study revealed information about the level of spending during the months before death. All of the patients on AZT survived the 12-month study period and all but one of the patients who did not receive AZT died during this period. Thus, the cost of treating persons who did not receive AZT was essentially equivalent to the cost of treating decedents, and the cost of treating persons on AZT was equivalent to the cost of treating survivors. The mean monthly charges for those receiving AZT were $1,873 and the mean monthly charges for those not receiving AZT were $4,793. The differences in these costs reflect the effect of AZT on the cost of care as well as the impact of death on the cost of care, and it is not possible to attribute the differences in costs between these two sources.

Solomon and colleagues analyzed data from Michigan Medicaid, the Michigan General Assistance Medical Program, and the Michigan Resident County Hospitalization Program for services provided on or after January 1, 1984 and before November 30, 1987 to persons with HIV (Solomon et al. 1989). Information from payment records was received on 71 persons who died during this period. These persons were identified from a review of death certificates with AIDS listed as a cause of death. The authors found that HIV-related monthly payments grew from about $1,500 per month three months before the patient’s death to more than $8,000 in the last month of life (Solomon et al. 1989). In a subsequent study of 149 decedents with AIDS who received services from Michigan Medicaid between January 1, 1985 and December 31, 1990, and for whom death records were available, Solomon and Hogan confirmed the finding that the cost of treating persons with AIDS climbs swiftly just prior to death (Solomon and Hogan 1992).

In a longitudinal study of 322 enrollees in New Jersey’s Medicaid home and community-based services waiver program for persons with symptomatic HIV (the AIDS Community Care Alternatives Program-ACCAP), who died between March 1, 1987 and February 28, 1989, Merzel and colleagues found that 50 percent of the expenditures were incurred during the last two months of life (Merzel, Sambamoorthi, and Crystal 1991). ACCAP afforded reimbursement for services not normally covered by Medicaid, including private duty nursing and personal care assistance services provided to enrollees in their homes (Merzel, Crystal, Sambamoorthi, et al. 1992). Expenditures during the last two months of life did not vary by exposure category (i.e., gay/bisexual men, injection drug users, women) but did vary by race and gender. Black enrollees and women enrollees incurred higher expenses during the last two months of life because they were hospitalized more often and had longer lengths per hospital stay.

In a subsequent study that employed, data on all 801 enrollees in the ACCAP between March 1, 1987 and February 28, 1989, Merzel and colleagues found that 9 percent of enrollees incurred costs in excess of $50,000 (these cases were termed “cost outliers”) (Merzel, Crystal, Sambamoorthi, et al. 1992, 38). Of the group of decedents who were cost outliers, 92 percent were hospitalized during the two-year study period; of the group of decedents who were not cost outliers, only 60 percent were hospitalized during the two-year study period. This study also found that 92 percent of the cost outlier decedents received services from home health agencies and 60 percent of the decedents who were not cost outliers received services from home health agencies. The reverse was true for services received from spouses. It was found that 4 percent of the cost outlier decedents received care from a spouse and 13 percent of the decedents who were not cost outliers received services from a spouse.

Based on an analysis of the computerized coverage and claims files from Blue Shield of California, Lee and colleagues found that medical care charges for PWAs were U-shaped over the course of the disease (Lee, Durbin, and Kahn 1991). For patients who received most of their care in Los Angeles and were diagnosed with AIDS in 1989, the mean monthly charges for the three-month time period immediately following diagnosis were $5,204 ($15,612/3); the mean charges for the 12-month time period immediately following diagnosis for those diagnosed in 1989 were $2,304 ($27,651/12); and the mean charges for the three-month time period prior to death for patients who died in 1989 were $9,922 ($29,767/3). For patients who received most of their care in San Francisco these figures were $2,185, $1,570, and $6,423, respectively.

The study by Lee and colleagues used billed charges as a measure of costs. Yet the amount received by providers of care may be substantially less than billed charges. The amount of the billed charges deemed “allowed” is often a better approximation of the amount actually received by providers. Yet often the level of allowed charges is viewed as proprietary by health insurers and is unavailable to researchers. The level of allowed charges was not available to the researchers in the study of Blue Shield of California’s computerized claims files.

In our study, the level of billed charges also is used as a measure of costs. Thus, our study also overestimates the amount actually received by providers. Cost estimates derived using billing data overstate the cost of medical care to insurance companies, public insurers, and individuals. In addition, cost estimates based on billing data do not measure the cost to society (i.e., the “real” cost, or what economists refer to as the “opportunity” cost) because charges (whether billed or allowed) are not accurate measures of the real cost of resources.

METHODS

The contract to develop and conduct the ACSUS was awarded to Westat, Inc., Rockville, Maryland. The ACSUS sample was obtained using a three-stage sampling design (selection of cities, providers, and study participants). The ACSUS includes data on persons with HIV from ten U.S. cities (Baltimore, Chicago, Houston, Los Angeles, Miami, Newark, New York City, Philadelphia, San Francisco, and Tampa).

A list of institutions (hospitals and freestanding clinics) as well as physicians who treat persons with HIV was obtained from the Centers for Disease Control and Prevention (CDC) National AIDS Information Clearinghouse and from local health departments, and estimates were made of the number of infected persons receiving care from each provider. In general, the providers in each area who treated the most persons with HIV were selected. Thirty-two providers were selected for the ACSUS sample, and 26 agreed to participate in the ACSUS. Of these 26, five were from New York City, three each from Newark, Tampa, and Miami, and two from each of the remaining cities.

The third stage involved selecting patients with HIV infection from each site. A systematic probability sample was drawn at each site that involved obtaining basic information (e.g., HIV status, gender, insurance status, likely route of infection, and symptoms) from all patients who received care at a site over a two- to four-month time period. Screening forms were distributed to 5,811 persons with HIV infection. The screener forms were distributed by study coordinators at each site. Patients who were not infected with HIV were automatically excluded from the sample.

The screener form did not have any patient-identifying information other than a number that could be used to call back the patient in case he or she was selected for the sample. Using information in the screening form, the site coordinator placed each patient in a sampling stratum, and an algorithm developed by statisticians at Westat, Inc., was used to select the sample. The sampling algorithm ensured an adequate representation of persons by stage of illness (asymptomatic, symptomatic, and AIDS), exposure group (gay/bisexual, intravenous drug user [IDU], and other), and source of payment. Of those asked to participate in the ACSUS, 88 percent agreed. The lowest recruitment rate was 80 percent for asymptomatic homosexual men.

The ACSUS collected information on all services consumed by persons with HIV. Participants were interviewed every three months during an 18-month time period from spring 1991 to fall 1992. The first interview included questions about the use of services during the time period between March 1, 1991 and the date of the first interview, and the last interview included questions about the use of services from the time of the fifth interview through August 31, 1992. The short time period between interviews was selected to reduce recall bias. It is important to survey HIV-infected persons as frequently as possible because persons with HIV may experience problems that affect memory.

At the first interview participants were asked to identify all providers of their health care, and at each subsequent interview participants were asked to identify new providers. Participants were asked to sign release forms for each health care provider (e.g., physicians, clinics, hospitals, pharmacies, and nursing homes) so that billing and medical record data could be obtained.

ACSUS has been successful in following patients from interview to interview. During each of the six interview periods, only about 7 percent of the persons interviewed during one time period were not interviewed the next time, and almost all of these either were determined to have died or to have moved out of the area. Only about 1 percent of the participants were not reinterviewed in each period because their whereabouts were unknown or they refused to be reinterviewed.

This study uses data from the fifth wave of interviews, which were conducted during the spring and early summer of 1992. Respondents in the fifth wave of interviews were divided into those who did, and those who did not, survive the sixth time period. (Note that all of the data used in this study were obtained from respondents who survived through the fifth time period). Of the 688 PWAs interviewed in the fifth wave, 609 survived the sixth time period and 79 died during the sixth time period. If it is assumed that the dates of death of these 79 PWAs were uniformly distributed across the sixth time period, then the average amount of time between the end of the fifth time period and the date of death was one-and-a-half months.

Billing records from hospitals, physicians’ offices, clinics, home health providers, and pharmacies for all services received by respondents during the period between March 1, 1991 and December 31, 1991 were abstracted by a team of field workers between March 1, 1992 and July 1, 1992, during the first round of the provider survey. The second round of the provider survey collected billing data from providers for all services received by respondents between January 1, 1992 and August 31, 1992. This effort began in April 1993 and ended in July 1993. Data from the second round of the provider survey were unavailable at the time of this study. Data from the first round of the provider surveyor were used in this study to derive estimates of the average charge for a hospital day, an outpatient visit, and a home health visit as well as an estimate of the average amount spent on prescription drugs per respondent.

RESULTS

Table 1 compares characteristics of PWAs who died during the sixth time period with those who survived the sixth time period. Table 1 shows that the proportion of decedents identified as gay/bisexual men, IDUs, or other is similar to the proportion of survivors identified as gay/bisexual men, IDUs, or other. The information on exposure category was obtained from the screener form and a large proportion of both decedents and survivors indicated that they were neither gay/bisexual men nor IDUs. It is likely, however, that some of those who indicated that they were neither gay/bisexual nor IDUs were not accurately reporting this information.

Table 1: Information on Study Population

Decedents (N = 79) Survivors (N = 609)

Exposure Category

Gay/Bisexual 57.1 (*) 53.6

Intravenous drug user 19.5 18.3

Other 23.8 27.8

Gender

Female 10.7 17.2

Male 89.3 82.9

Insurance

Private 27.7 29.8

Medicaid 57.8 51.7

Other public 8.4 14.6

None 6.0 4.0

Race

White 46.4 46.1

Black 31.0 27.6

Hispanic 22.6 24.4

Other 0.0 2.0

* Figures are percentages of decedents (survivors) in each category. The

percentages do not add up to 100 due to rounding error.

Table 1 also reveals that a lower proportion of decedents (11 percent) than survivors (17 percent) are women, and that the proportion of decedents and survivors insured by private companies, Medicaid, or other public payers, and those without insurance, is similar to the proportion of survivors in these categories. Fifty-two percent of the survivors were covered by Medicaid while 58 percent of decedents were covered by Medicaid. Four percent of survivors and 6 percent of the decedents were uninsured.

The racial/ethnic composition of decedents and survivors also is presented in Table 1. White persons comprise 46 percent of both decedents and survivors, while Black persons comprise 31 percent of decedents and 28 percent of survivors. Hispanic individuals comprise about 23 percent of decedents and 24 percent of survivors. Two percent of survivors were neither White, Black, nor Hispanic, and there were no persons in the “other” category among decedents.

Seventy-one percent of decedents were hospitalized during the fifth time period and 28 percent of survivors were hospitalized during this time period. The average length of stay was 19.3 days for decedents and 10.3 days for survivors. The average standardized number of hospital admissions was .70 for decedents and .28 for survivors. The average standardized number of hospital days for decedents during the fifth time period was 13.6, and for survivors it was 2.9. Numbers of hospital days and admissions are “standardized” to a three-month time period because interview instruments collect information about service use between the date of the prior interview and the date of the current interview, and this time period is not always exactly three months.

Table 2 compares hospital use, ambulatory visits, emergency room visits, home health visits, and prescription drug use for decedents and survivors. The average standardized number of emergency room visits during the fifth time period was .4 for decedents and .5 for survivors. The average standardized number of ambulatory visits (i.e., community clinic visits, hospital clinic visits, and physician office visits) was 4.5 for decedents and 5.2 for survivors.

The average standardized number of formal home health visits received by decedents during the fifth time period was 11.2; it was 2.7 for survivors. Formal home health visits include visits by physicians, nurses, social workers, case managers, therapists, and other paid helpers that provide medical care services. Formal home health visits do not include meals delivered nor visits by persons who provide housekeeping services. They also do not include visits by friends and relatives who provide medical care services. Thirty percent of decedents and 11 percent of survivors received formal home health visits during the fifth time period.

Table 2: Utilization of Services for Persons with AIDS during the Fifth Time

Period

Decedents (N = 79) Survivors (N = 609)

Ambulatory medical visits

(excluding emergency room) 4.5 5.2

Emergency room visits 0.4 0.5

Home health visits 11.2 2.7

Hospital days 13.6 2.9

Drugs 5.2 5.3

The average number of prescription drugs taken by decedents during the fifth time period was 5.2, and for survivors it was 5.3. This number includes all drugs that were purchased by the respondent, all that were given to the respondent free of charge by a physician or at a clinic, and all that were refilled during this time period. This number measures the number of distinct medications and not the number of prescriptions.

Data on the estimated cost of care received by decedents and survivors appear in Table 3. Estimates of the cost of each service were obtained from the first round of the provider survey. The estimated cost is $216 for an ambulatory visit, $336 for an emergency room visit, $1085 for a hospital day, $125 for a home health visit, and $50 for each prescription drug. The utilization rates in Table 2 were multiplied by the per unit costs to derive the estimated cost of services received during the fifth time period. The estimated cost figures in Table 3 represent the cost of care for three months.

Table 3 reveals that the cost of treatment in the fifth time period for PWAs who died in the sixth time period was more than three times as large as the cost of treatment in the fifth time period for PWAs who were alive at the end of the sixth time period. The higher costs are attributable to a greater likelihood of being hospitalized, longer lengths of hospital stay, and greater use of home health services. The utilization rates of ambulatory care services, emergency room care, and drugs are similar for decedents and survivors. The average monthly cost of treating decedents during the fifth time period is $5,840 ($17,522/3), and the average monthly cost of treating survivors during the fifth time period is $1,680 ($5,041/3).

Table 3: Cost of Care for Persons with AIDS during the Fifth Time Period

Decedents (N = 79) Survivors (N = 609)

Ambulatory medical visits

(excluding emergency room) $972 (6)(*) $1,123 (22)

Emergency room visits 134 (1) 168 (3)

Home health visits 1,400 (8) 338 (7)

Hospital days 14,756 (84) 3,147 (62)

Drugs 260 (1) 265 (5)

Total 17,522 5,041

* Figures in the parentheses represent the percentage of cost attributable to

the service. The numerals in the parentheses do not add up to 100 due to

rounding errors.

Thirty-one percent of the decedents utilized informal home health services during the fifth time period as did 20 percent of survivors. Informal home health care services include all health care-related services (e.g., changing dressings, ensuring that appropriate medications are taken at proper time intervals, cleaning and operating infusion systems, and taking the patient’s temperature and monitoring other vital signs) provided by friends and relatives in the patient’s place of residence.

Using data from the sample of decedents only, it was discovered that decedents who had received informal home health services spent fewer days in the hospital. Decedents who received informal home health services spent an average of 7.5 days in the hospital whereas those who did not receive informal home health services spent an average of 16.2 days in the hospital. Conversely, it was found that decedents who received formal home health care services spent more days in the hospital (11.9) than decedents who did not receive formal home health care services (3.9).

Again, using data only from the sample of decedents it was found that whether or not a decedent died in the hospital was closely related to the number of hospital days. Death certificates and medical and billing records were scrutinized in order to determine the site of death for a random sample of 31 decedents. If the date of death for a decedent was the same as the date of discharge from the hospital then it was assumed that the person died in the hospital. It was found that 20 decedents died outside of the hospital and 11 died in the hospital. Decedents who died in the hospital spent an average of 27.3 days in the hospital during the fifth time period while those who died outside of the hospital averaged 12.5 days in the hospital during the fifth time period.

DISCUSSION

The cost of treating PWAs who survived a subsequent three-month period was compared to the cost of treating PWAs who did not survive a subsequent three-month period in this study. We were unable to obtain information about the use of services during the weeks immediately preceding death because interview data were not available from decedents for the sixth time period. If the dates of death for the decedents in our study are randomly distributed through the sixth time period, then information about the utilization of services for an average of one-and-one-half months is missing. It is likely that, if information about the utilization of services up to the moment of death had been available, the differences in utilization between decedents and survivors would have been greater.

Nonetheless, this study suggests that the cost of treating PWAs rises rapidly as the time of death approaches. The cost of treating decedents was found to be about three times as great as the cost of treating survivors during the fifth time period. This occurred because decedents used disproportionately large amounts of home health and hospital services compared to survivors. No difference appeared in their utilization of emergency room services, ambulatory medical visits, or drugs.

There is considerable interest in the cost of treating persons during the months immediately preceding death. This interest stems from the belief that much of the cost of treating terminally ill patients is attributable to the use of expensive technologies that do not enhance the quality of life (Lubitz and Riley 1993). To a large extent, passage of the Medicare hospice benefit was predicated on this belief and was designed to provide emotionally beneficial treatment outside of the hospital to persons with terminal illnesses (Lubitz and Prihoda 1984).

In many instances it is impossible to ascertain in advance which services will not extend a patient’s life. Nonetheless, it may be useful to determine which factors explain variations in the cost of treating decedents. For example, it is shown here that decedents who received health care services from friends and relatives used fewer hospital services than decedents who did not receive health care services from friends and relatives. The reduced use of hospital services by decedents with this type of strong social support system may reflect both a greater resolve to remain at home and a greater capacity to remain at home. Further, PWAs with strong social support systems who prefer to remain at home as long as possible may be less likely to demand an aggressive approach to treatment because it frequently entails long periods of hospitalization. This finding suggests that furnishing “buddy” services to seriously ill persons with HIV may not only provide needed emotional support but may also reduce costs.

Data also reveal that decedents who use formal home health care services consume more hospital services than decedents who do not use formal home health care services. This was unexpected because home health care services (whether formal or informal) often are viewed as a cost-efficient substitute for hospital care (Abramovitz 1991). To some extent, this finding may reflect the greater use of all types of health services by PWAs with poor social support systems. And to some extent, the greater use of both formal home health care services and hospital services by some decedents may reflect an aggressive approach to treatment.

It was also found that persons who died in the hospital were hospitalized more than twice as many days as persons who died at home. For some patients, the site of death may reflect an explicit decision about preferences concerning where and how to die. For patients without adequate support services at home, there may be no choice.

CONCLUSION

Using data from a sample of 609 survivors and 79 decedents from the fifth wave of the ACSUS, this study shows that the cost of treating decedents is more than three times the cost of treating survivors. Decedents experienced more than four times the number of home health visits and were hospitalized more than four times as many days as survivors. Both the average length of stay (19.3 days for decedents and 10.3 for survivors) and the frequency of hospitalization (.70 for decedents and .28 for survivors) were higher for decedents than for survivors. The levels of outpatient care (including emergency room care) and of prescription drug use were similar for decedents and survivors.

Previous studies have demonstrated that the cost of treating persons with AIDS increases as the time of death approaches. Yet this study is the first to include persons with AIDS covered by different insurers and the first to include PWAs from a variety of geographic areas. It also is the only study that examines the cost of care since 1990 (this is important because treatment regimens for patients with HIV infection change rapidly [Hellinger 1993]), and the first to examine the use of emergency room care, outpatient medical office visits, prescription drugs, and home health care services by decedents and survivors.

This study compares the use of services by respondents with AIDS during the fifth time period who did, and did not, survive the sixth time period. This group of respondents with AIDS is not subdivided by the length of time since an AIDS diagnosis, by T-cell count, or by the existence of given opportunistic infections. It would be interesting to compare the use of resources by survivors and decedents who had been diagnosed with AIDS for a similar time period, who had similar T-cell counts, and who had the same opportunistic infections. However, this would require a larger sample of decedents.

This study includes a relatively small number of decedents (N = 79). The small number also makes it impossible to estimate the impact of geographic variations on the use of services by decedents. In addition, a larger sample is needed to explore the influence of gender, exposure category, insurance status, and race on the utilization of services by decedents.

REFERENCES

Abramowitz, M. “Paying the Price for Home Care Costs Is Lower than at Hospitals, but Lack of Regulation Raises Concerns.” Washington Post, 6 October 1991, Business Section, H1, H4, and H5.

Baker, M. S., L. G. Kessler, N. Urban, and R. C. Smucker. “Estimating the Treatment Costs of Breast and Lung Cancer.” Medical Care 29, no. 1 (January 1991): 40-49.

Hellinger, F. J. “The Lifetime Cost of Treating a Person with HIV.” Journal of the American Medical Association 270, no. 4 (28 July 1993): 474-78.

Lee, P. R., M. Durbin, and J. G. Kahn. Cost of HIV and AIDS for Privately Insured Persons in California, 1985-1990, Institute for Health Policy Studies, University of California, San Francisco. Final Report submitted to the Health Resources and Services Administration, U.S. Public Health Service, Contract no. 282-88-0018, September 30, 1991.

Lubitz, J. D., and R. Prihoda. “The Use and Costs of Medicare Services in the Last Two Years of Life.” Health Care Financing Review 5, no. 3 (Spring 1984): 117-31.

Lubitz, J. D., and G. F. Riley. “Trends in Medicare Payments in the Last Year of Life.” The New England Journal of Medicine 328, no. 15 (15 April 1993): 1092-96.

Merzel, C., S. Crystal, U. Sambamoorthi, D. Karus, and C. Kurland. “New Jersey’s Medicaid Waiver for Acquired Immunodeficiency Syndrome.” Health Care Financing Review 13, no. 3 (Spring 1992): 27-44.

Merzel, C., U. Sambamoorthi, and S. Crystal. “Trends in the Cost of Services in the Terminal Stages of HIV.” Abstract from the Annual Meeting of the American Public Health Association, Atlanta, GA, November 10-14, 1991.

Riley, G., J. Lubitz, R. Prihoda, and E. Rabey. “The Use and Costs of Medicare Services by Cause of Death.” Inquiry 24 (Fall 1987): 233-44.

Scitovsky, A. A., M. W. Cline, and D. I. Abrams. “The Effects of the Use of AZT on the Medical Care Costs of Persons with AIDS in the First 12 Months.” Journal of Acquired Immune Deficiency Syndromes 3, no. 9 (1990): 904-12.

Scitovsky, A. A., M. W. Cline, and P. R. Lee. “Medical Care Costs of Patients with AIDS in San Francisco.” Journal of the American Medical Association 256, no. 22 (12 December 1986): 3103-3106.

Scitovsky, A. A., and D. P. Rice. “Estimates of the Direct and Indirect Costs of Acquired Immunodeficiency Syndrome in the United States, 1985, 1986, and 1991.” Public Health Reports 102, no. 1 (January-February 1987): 5-17.

Solomon, D. J., and A. J. Hogan. “HIV Infection Treatment Costs under Medicaid Michigan.” Public Health Reports 107, no. 4 (July-August 1992): 461-68.

Solomon, D. J., A. J. Hogan, R. R. Bouknight, and C. T. Solomon. “Analysis of Michigan Medicaid Costs to Treat HIV Infection.” Public Health Reports 104, no. 5 (September-October 1989): 416-24.

COPYRIGHT 1994 American College of Healthcare Executives

COPYRIGHT 2004 Gale Group