The importance of accounting for testing and positivity in surveillance by time and place: an illustration from HIV surveillance in Japan.

The number of tests performed is an important surveillance indicator. We illustrate this point using HIV surveillance data, focusing on Tokyo and Okinawa, two prefectures with high HIV notification rates in Japan. Restricting to data reported from local public health centres and affiliate centres where testing data are accessible, we assessed HIV surveillance data during 2007-2014, based on the annual HIV notification rate (per 100 000 population), HIV testing rate (per 100 000 population) and proportion testing HIV-positive (positivity). Nationally, testing activity and positivity showed an inverse relationship; in 2008, the testing rate peaked, but positivity was lowest. While notification rates were higher for Tokyo (median = 0.98, range = 0.89-1.33) than Okinawa (median = 0.61, range = 0.42-1.09), Okinawa had slightly higher testing rates (median = 187, range = 158-274) relative to Tokyo (median = 172, range = 163-210). Positivity was substantially lower in Okinawa (median = 0.34%, range = 0.24-0.45%) compared with Tokyo (median = 0.57%, range = 0.46-0.67%). Relative to the national testing rate (median = 85, range = 80-115) and positivity (median = 0.34%, range = 0.28-0.36%), Tokyo had higher positivity, despite more testing. In 2014 in Okinawa, all three indicators increased, providing a strong reason to be concerned as positivity increased despite more testing. Together with other information, accounting for testing and positivity improve interpretation of surveillance data to guide public health assessments.