Incidence of HIV-related deaths in the United States: Seasonality and trend
Bacchetti, P.
Statistics in Medicine 16(6): 645-652
1997
ISSN/ISBN: 0277-6715 PMID: 9131753 DOI: 10.1002/(sici)1097-0258(19970330)16:6<645::aid-sim440>3.0.co;2-yDocument Number: 268135
This paper examines possible short-term patterns and distortions in the incidence of deaths with AIDS in the United States, using methods previously applied to incidence of AIDS diagnoses. The variation in death counts by calendar month models fairly well as a seasonal pattern consistent with seasonal variation in deaths from all causes. In addition, three apparently non-biological short-term effects in AIDS incidence - a workday effect, a jump between December and January, and a spike in June - are not apparent in death incidence, and analysis of death counts in subgroups does not show any strong evidence for non-biological influences on time of death. Deseasonalized death incidence shows a steady increase over time. Because death incidence is not subject to definition change and is apparently less susceptible to other non-biological influences than AIDS incidence, it may have value for monitoring the HIV epidemic.