Skip to main content

Key Findings

As before, CHESS use was initially very heavy as users explored the system, declined to about once a day after several weeks, and declined slowly from there to 2-3 times per week by the end of the study. Total amount of use and its distribution across CHESS services was similar to the first study, with Discussion Group (sending and reading bulletin board letters from fellow patients) again accounting for the majority of use. However, the two cohorts of the present study differed considerably. The first (and more ill) cohort used CHESS less than half as much as the second, but made more use of information services, while the second cohort were exceptionally heavy users of Discussion Group. The Discussion Groups in both cohorts did not seem as successful as in the prior study.

Beneficial effects of CHESS were similar and less consistently apparent than in the previous study. While the CHESS group generally scored higher (covariate-adjusted) than controls on self-reported quality of life measures, only a few measures reached or approached statistical significance, and only then after eight or twelve months had elapsed. It appeared that those who self-selected to use CHESS more frequently had better Quality of Life outcomes than lighter users.

Health care utilization and cost data showed no consistent differences between groups, possibly because of number of indicators suggested that the experimental group was actually sicker at pretest. There is some suggestion of increasing use of ambulatory care services by the control group, particularly later in the study. Hospitalization data also indicated fewer days and lower costs for the experimental group in the final 4-month study period, but this was offset by an increase at the previous period. It appeared that, in minority subjects, those with access to CHESS may have had substantially lower inpatient and higher outpatient costs then did the controls. This finding is consistent with previous research regarding the effects of CHESS in the underserved.

These weak or mixed outcomes may be traceable to some or all of a number of factors present in this study and not in the previous one. First, treatments for HIV infection and management of the disease changed radically during the study, producing enormous variation in the nature, costs, and expectations of treatment. Second, a variety of problems resulted in much lower-quality medical records and billing data than we had reason to expect. Third, this study drew from essentially the same population as the first, but those who had participated in the first study were naturally ineligible. Mortality had not produced large turnover from the first study. The population here may have been depleted of those most motivated to use a system like CHESS and therefore the current study population may have tended to be those less motivated to participate and manage their disease than those in the first study. Fourth, the samples obtained here, especially in Cohort 1, were much more ill than in the previous study. It appeared that this reduced both use and usefulness of CHESS. It may well be that such systems are more useful to patients as they initially adjust to their diagnosis.

Finally, it appeared as though random assignment did NOT produce equivalent experimental and control groups in several ways. Almost all newly-diagnosed subjects (13 of 14 with less than three months between diagnosis as HIV+ and recruitment into the study) were randomly assigned to the control condition. These are people who might be most in need of the information and support provided by CHESS. Also, the CHESS group was generally sicker at pretest than controls. Thus, since AIDS is a generally deteriorating disease, once the downward spiral begins (which may be days or years after diagnosis), it often cannot be reversed, and therefore time since diagnosis cannot be covariate-adjusted away.

All the above problems acted either to reduce use of the CHESS system or to obscure any differences that may have occurred between experimental groups. Thus, while the present study does not contradict the positive results obtained in the first study, its additions to them are more modest than we had anticipated. Possibly future analysis of the data will add to the outcomes.


A previous AHCPR-funded study demonstrated benefits of short-term access to CHESS (the Comprehensive Health Enhancement Support System) for HIV-infected individuals. Those randomly assigned to have the computer-based system in their homes reported better quality of life, less use of health services, and improved feelings of control over their own health care. A previous study of CHESS effects in younger women with breast cancer had similar quality of life findings.

This study aimed to extend the results of these CHESS studies in a number of ways: 1) by extending access to CHESS to 12 months; 2) by focusing on sicker patients who would need more health care; 3) and by assessing health care utilization (and cost) from clinic and hospital records rather than patient self-report.


Men and women with advanced HIV-infection (N=279) treated at HIV clinics in Madison and Milwaukee were randomly assigned to experimental and control groups in two cohorts, the first of which required a clinical AIDS diagnosis, and the second of which required a CD4 count less than 500 at any time (since some recruited in the first were too ill to use CHESS). The data collected included: 1) surveys on quality of life and other issues at pretest and then 4-, 8-, and 12-month post-tests; 2) interviews on medications, income, insurance and health care outside of participating providers at pretest and then at 4-, 8- and 12-months post-test; 3) CHESS use data; 4) billing records from participating clinics and hospitals; and 5) relevant data extracted from medical records. Due to budgetary constraints on the part of AHCPR, the grant funding awarded was 17.4% lower than originally allocated. As a result, we collected all of the data proposed in the grant application, but were not able to fully analyze all of it.

Funding Period:
January 1994 - June 1997
Principal Investigators:
David Gustafson Sr., Ph.D