Background: The Coding Causes of Death in HIV (CoDe) Project aims to deliver a standardized method for coding the underlying cause of death in HIV-positive persons, suitable for clinical trials and epidemiologic studies.
Methods: The project incorporates detailed data collection, a classification system, and a centralized adjudication process performed by 2 independent reviewers. The methodology was tested in the Data Collection on Adverse events of Anti-HIV Drugs Study , and independent reviews of causes of death were compared. Logistic regression models identified factors associated with initial agreement by reviewers on underlying cause of death.
Results: A total of 491 reported fatal cases were adjudicated; in only 5% of cases the cause of death remained undetermined after adjudication. Reviewers initially agreed on the underlying cause for 339 (69%) deaths. As compared with deaths due to AIDS-related causes, the odds of agreement were more than 80% lower when deaths were ultimately deemed to be due to non-AIDS-related causes (odds ratio = 0.17 [95% confidence interval = 0.08-0.37]) or undetermined causes (0.11 [0.04-0.36]). The odds of initial agreement were also lower for deaths occurring in subjects with hypertension (0.43 [0.22-0.85]) and depression (0.43 [0.23-0.80]).
Conclusions: The extent and format of data collected in the CoDe Project appear to be sufficient for an informed review, and the proposed coding scheme is adequate for obtaining an underlying cause of death.