Causes of Death in the Highly Active Antiretroviral (HAART) Era: A Retrospective Comparison between a Hybrid HIV and Hematology/Oncology Practice and the Adult & Adolescent Spectrum of HIV-Related Diseases (ASD) Project.

Background: Our practice at Virginia Mason Medical Center (VMMC) includes 600 HIV+ patients, 60 of whom died in the last decade. Our intimate doctor-to-patient care allows for increased precision when determining the underlying causes of patient mortality. Large cohort studies such as the ASD project may not allow for such detail because of dependence on medical records or death certificates to determine causes of death.

Objective: To determine variances in death between a single provider VMMC patient dataset, and a larger public health cohort during the HAART era.

Methods: We contrasted two datasets. The first was the Seattle/King County ASD dataset (n=4721), which recorded 351 patient deaths during 1996–2004. The second was the 1996–2006 VMMC HIV mortality cohort. Abstracted data include patient demographics, causes of death, co-morbidities, treatment adherence, CDC AIDS classification, and relevant laboratory data. We used X² and Fisher Exact test for our statistical analysis.

Results: Of the 60 VMMC patients who died, 57 (95%) were male, 16 (27%) injection drug users (IDU), 50% with significant mental illness, and 44 (73%) with a C2/C3 CDC AIDS classification. Median time between HIV diagnosis to death was 11 years (range, 0–22). There were 33 (55%) patients with poor/moderate adherence.

Of the 351 ASD patients who died, 301 (86%) were male, 43 (12%) IDU, 250 (71%) with significant mental illness, and 285 (81%) with a C2/C3 CDC AIDS classification. Median time between HIV diagnosis and death was 6 years (0–18). Of 92 patients for whom adherence data was collected, 69 (75%) had poor/moderate adherence.

39 (65%) VMMC patients died from non-opportunistic illness (OI), 18 (30%) from OI, and 3 (5%) from both (see table). The most common OIs were wasting, non-Hodgkin’s lymphoma, and progressive multi-focal leukoencephalopathy (PML). The most common non-OIs were malignancy, liver failure, and pneumonia. 11 of 60 patients (18%) died despite a non-detectable HIV viral load (NDVL) and median CD4+ count of 216 cells/μL (range, 16–952).

301 of 351 ASD patients had a known cause of death. 135 (45%) died from non-OI, 105 (35%) from OI, and 61 (20%) from both non-OI and OI (see table). The most common OIs were mycobacteria, dementia, and cytomegalovirus. The most common non-OIs were liver failure, pneumonia, and sepsis. 35 of 351 patients (10%) died despite a NDVL and median CD4⁺ count of 223 cells/μL (5–1616).

Conclusions: Males and those with substance abuse, mental illness, poor/moderate adherence, and a C2/C3 AIDS designation were heavily represented in both datasets. The VMMC patients had a longer interval between HIV diagnosis and death than those in the Seattle/King County ASD project.

Liver failure and pneumonia were the dominant non-OIs in both datasets. Malignancy as a cause of death was over-represented in VMMC due to the concentration of such patients in a Hem/Onc practice. ASD had a greater proportion of patients without a known cause of death, suggesting greater difficulty designating the underlying cause of death when patients are not intimately known.