AIDS surveillance: a direct assessment of under-reporting.

Objective: To assess directly the extent of under-reporting of AIDS cases to the National AIDS surveillance system.

Design: All AIDS cases diagnosed from 1 January 1982 to 1 August 1989 in Riverside Health Authority were identified from a local register of HIV infection and cross-checked against records of AIDS cases reported to the national AIDS surveillance system.

Setting: Riverside Health Authority, London, UK.

Main Outcome Measures: An estimate of under-reporting was made by identifying both the number of AIDS cases that had not been reported and the number of AIDS cases reported after August 1989 (allowing for reporting delay). Changes in the timeliness, 1982-1989, of AIDS reporting by Riverside physicians were measured by comparing reporting delay (time between diagnosis and report) and the proportion of AIDS cases reported more than a year after diagnosis (non-reports).

Results: A total of 807 AIDS cases were identified. Under-reporting of AIDS cases was found to be 10%; less than 3% (20 cases) of the AIDS identified had not been reported by September 1991. The timeliness of AIDS reporting from Riverside improved significantly from 1987, when the median reporting delay fell from 7 to 4 months, and the proportion of non-reports fell from 36 to 17% (Mann-Whitney U test P < 0.001, chi 2 test P < 0.001, respectively).

Conclusions: Our estimate of 10% AIDS under-reporting is half that used to adjust the previous forecasts of the AIDS epidemic in the UK, confirms current thinking that under-reporting lies between 5 and 15%, and supports the view that AIDS reporting is more complete than the reporting of most other infectious diseases.