Methods

Top Methods Results Discussion Author & Article Info References

We obtained data from multiple-cause mortality tapes prepared by the National Center for Health Statistics from death certificates of U.S. residents from 1987 through 1992, which were filed in all 50 U.S. states and the District of Columbia [1]. We identified diseases by their codes in the International Classification of Diseases, Ninth Revision (ICD-9) [2] and identified HIV infection by supplemental codes introduced in 1987 [3]. Among deaths for which HIV infection was recorded as the underlying cause, infectious diseases and cancers that could be secondary to HIV infection were found as nonunderlying (immediate, intermediate, or contributing) causes of death; this classification allowed more than one such disease per death.

To determine the percentage of HIV-related deaths caused by a given disease, we excluded from the denominator deaths for which information on secondary diseases was missing: deaths for which no disease but HIV infection was recorded and those for which the only other causes recorded were nonspecific (such as cardiac arrest), not associated with HIV infection (such as trauma), or likely to have preceded HIV infection (such as drug abuse). We examined trends in the annual percentage of deaths associated with each infectious disease or cancer that was reported in at least 1.0% of the denominator in the 6-year period. We used Poisson regression analysis [4] to test the statistical significance (P < 0.05) of the trend for each disease.

Results

Top Methods Results Discussion Author & Article Info References

From 1987 to 1992, HIV infection was the underlying cause of 140 461 deaths, of which 104 831 had possible secondary causes of death specified on the death certificates. The proportion represented by the latter (75%) remained stable as the number increased annually from 10 001 deaths in 1987 to 24 230 in 1992. These 104 831 HIV-related deaths provided the denominators for calculating the annual percentage of deaths associated with each disease. Most of these HIV-related deaths (60%) were reported with 1 secondary infectious disease or cancer; 19%, with 2; 6%, with 3 to 8; and 15%, with none (but with other secondary conditions). Twelve infectious diseases and 2 cancers were each reported in at least 1.0% of the HIV-related deaths.

The annual number of deaths associated with each disease generally increased as the total number of HIV-related deaths increased, but the relative rates of increase differed among diseases. These varying rates caused distinctly different trends in their percentages (Table 1; Figure 1). The percentages of HIV-related deaths associated with the following three diseases decreased: pneumocystosis, from 32.5% to 13.8%; cryptococcosis, from 7.7% to 5.0%; and candidiasis, from 2.3% to 1.7%. The percentages of deaths associated with eight diseases significantly increased: nontuberculous mycobacteriosis, from 6.7% to 12.2%; cytomegalovirus disease, from 5.2% to 9.9%; bacterial septicemia, from 9.0% to 11.5%; diffuse non-Hodgkin lymphoma, from 3.9% to 5.7%; tuberculosis, from 2.9% to 4.1%; progressive multifocal leukoencephalopathy, from 0.8% to 1.9%; bacterial pneumonia, from 1.2% to 2.1%; and cryptosporidiosis or isosporiasis, from 0.7% to 1.2%. The percentages of HIV-related deaths associated with unspecified pneumonia, Kaposi sarcoma, and toxoplasmosis (ranges from 17.6% to 18.6%, 10.4% to 12.1%, and 4.9% to 5.5%, respectively) had no significant linear trends during the 6-year period.

View larger version (25K): [in this window] [in a new window]   Figure 1. Trends in the prevalence of infectious diseases and cancers reported among persons dying of human immunodeficiency virus infection in the United States from 1987 to 1992.

As a result of these different trends, the ranking of the diseases by the percentage of HIV-related deaths in which they were reported has changed. Pneumocystosis was the most common of these diseases until 1991, when its frequency decreased below that of unspecified pneumonia (Figure 1). The rank of nontuberculous mycobacteriosis increased from sixth place to third during 1987 to 1992, cryptococcosis dropped from sixth to ninth place, and candidiasis dropped from eleventh to thirteenth place.

Disease trends among black persons (including Hispanic blacks) were similar to trends among white persons (including Hispanic whites); trends among females were generally similar to those among males.

Discussion

Top Methods Results Discussion Author & Article Info References

From 1987 to 1992, the distribution of secondary diseases among persons dying of HIV infection changed markedly. Pneumocystosis, initially the most common of these diseases, accounted for one third of HIV-related deaths in 1987; by 1992, however, it accounted for less than half this proportion. The percentages of cryptococcosis and candidiasis decreased to a lesser extent, those of eight other diseases increased, and those of three others did not change significantly.

The dramatic decrease in the percentage of HIV-related deaths associated with pneumocystosis is consistent with results of studies of persons with the acquired immunodeficiency syndrome (AIDS) and of cohorts of HIV-infected persons [5-7]. This decreasing frequency of death associated with pneumocystosis is probably due to two factors: 1) enhanced prevention, attributable to the increasing use of chemoprophylaxis, and 2) increased survival of persons with Pneumocystis carinii pneumonia, attributable to improved methods of diagnosis and treatment [8, 9]. Despite these advances, pneumocystosis continues to cause a relatively large percentage of HIV-related deaths, probably because many HIV-infected persons do not obtain medical care for HIV infection until the infection is in a late stage, when pneumocystosis may have already developed [10]. In addition, P. carinii pneumonia sometimes develops despite prophylaxis, especially as immunodeficiency becomes more severe [11].

The large decrease in the percentage of HIV-related deaths associated with pneumocystosis should be expected to increase the percentages of HIV-related deaths associated with other diseases, in the absence of other influences. However, the percentages of deaths from cryptococcosis and candidiasis also decreased, perhaps because of successful prophylaxis or treatment with new antifungal agents such as fluconazole. This drug was licensed by the U.S. Food and Drug Administration in 1990 and was first used in clinical trials a few years earlier [12, 13].

The percentages of HIV-related deaths associated with toxoplasmosis, Kaposi sarcoma, and pneumonia caused by unspecified organisms did not change despite the decreasing percentages of deaths associated with pneumocystosis, cryptococcosis, and candidiasis. The increasing use of chemoprophylaxis against pneumocystosis with drugs effective against toxoplasmosis could have prevented increases in the percentage of deaths associated with toxoplasmosis and may also have held down the increases in the percentages with bacterial pneumonia and septicemia [14]. The percentage of HIV-related deaths associated with Kaposi sarcoma could have been suppressed by the decreasing percentage of homosexual or bisexual men (who account for most cases of Kaposi sarcoma) among persons with AIDS and the decreasing percentage of homosexual men with Kaposi sarcoma among all homosexual men with AIDS [15, 16].

Unspecified pneumonias are probably a mixture of cases caused by various unidentified pathogens, including P. carinii. Prophylaxis against pneumocystosis would be expected to decrease this component of unspecified pneumonia, whereas other components might be increasing. The net result is an apparently stable trend overall for unspecified pneumonia. Any changes in clinical practice affecting the specificity of the diagnosis of pneumonia could also have influenced these trends. A limitation of our study is the fact that the relative proportions of the components of unspecified pneumonia are unknown. Nonetheless, because of the decreased percentage of HIV-related deaths associated with P. carinii pneumonia, unspecified pneumonia became the leading secondary cause of death among persons dying of HIV infection, accounting for almost one fifth of HIV-related deaths. Some cases of unspecified pneumonia may represent terminal events in patients whose inevitable deaths were caused primarily by other HIV-related diseases. For other cases, determining the causative organisms and the most effective methods for preventing and treating this vaguely described entity may substantially increase the survival of persons with HIV infection.

As in our study, studies of persons with AIDS found increasing trends in the percentage of patients with a diagnosis of nontuberculous mycobacteriosis (Mycobacterium avium complex infection) [7]. It is too early to see the effect of newly licensed drugs such as rifabutin in preventing and treating M. avium complex infection [17].

For most of the diseases analyzed in our study, the similarity of trends among whites and blacks and among males and females suggests that advances in treatment and prophylaxis have affected all of these groups. However, racial or sexual inequities in access to such care may still exist [18].

The quality of our data depends on how accurately and thoroughly the causes of death were reported on death certificates. Previous studies suggest that deaths for which the underlying cause was reported as HIV infection represent only 66% to 86% of all deaths attributable to HIV infection among men aged 25 to 44 years and 55% to 80% of such deaths among women aged 15 to 44 years [19, 20]. Underestimation of HIV-related deaths would not necessarily affect the prevalence of secondary diseases among HIV-related deaths, but such nonunderlying causes of death could also be under-reported. Disease prevalence based on death certificate data includes only causes of death and thus underestimates total prevalence, particularly for diseases unlikely to cause death, such as candidiasis.

Our findings corroborate the success of prophylaxis and therapy for pneumocystosis shown in other studies and suggest that similar benefits have resulted from the treatment or prevention of cryptococcosis, candidiasis, and perhaps toxoplasmosis. Because drugs for prophylaxis against such opportunistic infections are available, HIV-infected persons should learn of their HIV infection and obtain appropriate medical care as early as possible. As prophylaxis and treatment have made some diseases less common among persons dying of HIV infection, other diseases have become more common; this finding emphasizes the importance of further research to develop methods for preventing and treating HIV-related diseases.