Surprising Causes of Death in Texas Hospital Study: Safety Net Questions

May 30, 2003

This article is part of The Body PRO's archive. Because it contains information that may no longer be accurate, this article should only be considered a historical document.

A study of the changing causes of death of people with HIV at Parkland Memorial Hospital, a major hospital in Dallas, Texas, found that pneumocystis (also called PCP) is still a major cause of death. And more than half of those with HIV who died of all causes in the study period of 1999-2000 were not receiving modern antiretroviral treatment. During this period pneumocystis caused 17% of the deaths, end-stage liver disease 13%, and non-Hodgkin's lymphoma 7%. Bacterial pneumonia not considered HIV associated, sepsis, and other non-AIDS-defining infections caused 18% of the deaths, and a group of conditions considered probably immunodeficiency related caused 9%. In a comparison period in 1995, before modern antiretroviral treatment (HAART) was available, more of the deaths were from AIDS-related conditions. But end-stage liver disease caused 10% of the deaths in the earlier period, showing that it is not a new problem.

There was a large decrease in deaths of HIV-infected persons overall -- from 119 deaths in 1995 to 44 in 1999 and 47 in 2000.


It is often hard to draw conclusions from statistical comparisons of deaths, because the numbers can depend on many factors (like hospital admissions policies) not related to medical care. But the fact that pneumocystis remains the leading cause of death of people with HIV, at one major hospital at least, raises questions about how well the safety net has been working.

There has long been a widespread assumption that almost anyone in the U.S. can get HIV treatment one way or another. We do not know how much this is true. Perhaps the belief persists because those who cannot get treatment also cannot get public attention.

Pneumocystis prophylaxis costs very little, and failure to use it is not due to the expense of the drugs. In this study many patients were not on prophylaxis because their HIV was not diagnosed -- suggesting lack of medical care, due either to lack of access or to the patients' decisions.

Adherence to HAART was a problem, with 39% (18 patients) of those who died in 1999-2000 without HAART listed as not receiving HAART because they were not adherent -- and 26% not receiving HAART because they were diagnosed shortly before death. We know from general experience that many adherence problems result from difficulty in obtaining a continuing supply of medicine -- including inflexible reimbursement rules that may make it difficult to replace lost medicines, or that leave too short a window to refill a prescription when patients have many other balls in the air. Physicians may not know whether non-adherence is due to economic obstacles.

Parkland Memorial Hospital is well regarded and accepts patients on an ability-to-pay basis. But Texas has long been seen as one of the worst states for access to HIV care (though improving now, due to grassroots organizing).

Cause-of-death studies can give us unique information about how well the medical safety net is working or not working. This one suggests that access to care may be less than generally believed, even before the funding crisis that is developing now.


Jain M.K., Skiest D.J., Cloud J.W., Jain C.L., Burns D., and Berggren R.E. Changes in mortality related to human immunodeficiency virus infection: Comparative analysis of inpatient deaths in 1995 and in 1999-2000. Clinical Infectious Diseases. April 15, 2003; number 36, pages 1030-1038.

ISSN # 1052-4207

Copyright 2003 by John S. James. Permission granted for noncommercial reproduction, provided that our address and phone number are included if more than short quotations are used.

This article was provided by AIDS Treatment News. It is a part of the publication AIDS Treatment News.


The content on this page is free of advertiser influence and was produced by our editorial team. See our content and advertising policies.