Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Research Letters
AIDS 2009, 23:533–540
Estimation of HIV incidence in San Francisco
Susan Scheer, Chi-Sheng Chin, Anthony Buckman and
Willi McFarland
The Centers for Disease Control and Prevention
recently released the first direct national estimate of
HIV incidence. Local jurisdictions have begun to
apply this methodology. The national and local
estimates have been higher than assumed. When
applied to San Francisco, there were 935 new HIV
infections [95% confidence interval (CI) 658 – 1212]
during 2006. We compared this incidence estimate
to an estimate produced in San Francisco in 2006 by
a panel of HIV researchers using an iterative Delphi
method. Results were similar. Further corrobora-
tion of the new method in local areas would
strengthen interpretation and identify HIV risk
variations.
The Centers for Disease Control and Prevention (CDC)
recently released the first direct estimate of HIV incidence
in the United States [1]. Drawing data from 22 states that
participated in a national surveillance system during 2006,
HIV incidence was calculated using the BED HIV-1
capture enzyme immunoassay (BED) that distinguishes
recent from long-standing infection among new diag-
noses and a statistical adjustment for frequency of HIV
testing and then extrapolated to the United States [1,2].
The calculation estimated that 56 300 (95% CI 48 200–
64 500) persons were newly infected in the United States
in 2006. The figure was similar to that produced by a
method of back calculation from AIDS cases, estimating
55 400 (95% CI 50 000 – 60 800). The same methodology
was applied to New York City and the resulting estimate
was 4762 – three times higher than the national rate [3].
Both the national and the New York City estimates were
higher than many public health officials assumed.
San Francisco was participating in the CDC-funded HIV
incidence surveillance system during 2006. However,
the State of California did not implement name-based
HIV case reporting until April 2006. As such, data from
California was not included in the 22 states from which
the national incidence estimate was derived. However, of
the 593 newly diagnosed and reported HIV cases in
San Francisco, California in 2006, only 17 were originally
reported before name-based HIV was implemented in
April 2006 and 14 of these were re-reported later in 2006
by name. As a result, 590 cases (99.5%) are included in San
Francisco’s 2006 HIV case registry, and all were subject to
the same scrutiny regarding checking for duplication and
accuracy of date of diagnosis, for example, that cases
reported in other name-based reporting systems.
After release of the national HIV estimate, CDC released
the methodology [2] and accompanying SAS computer
programs [4], so that individual incidence surveillance
sites could calculate their local estimate. We, therefore,
report here the 2006 HIV incidence estimate for San
Francisco.
Table 1 provides the available data used for the HIV
incidence estimation in San Francisco. Ninety cases were
classified as recently infected with HIV out of a total of
590 new HIV diagnoses reported in 2006. On the basis of
the CDC adjustment and extrapolation methods, we
estimate that there were 935 new HIV infections (95% CI
658–1212) in San Francisco during 2006. We used the
same 20-fold multiple imputation process that CDC used
to calculate the national estimate because HIV testing
history and BED results were not available for all cases in
2006. For the population of men who have sex with men
(MSM) [including MSM who inject drugs (MSM-IDU)],
the estimate was 716 (95% CI 489–944). No other risk
population met the criterion of having a minimum of
200 new HIV diagnoses needed for reasonable precision.
For comparison, a HIV incidence number was produced
for San Francisco in 2006 by the HIV Epidemiology
Section of the San Francisco Department of Public. The
HIV Epidemiology Section periodically examines all
known local data and consults with experts in the HIV/
AIDS field to arrive at a ‘consensus’ estimate of HIV
incidence [5]. Data include studies using different
methods to estimate HIV incidence in diverse risk popu-
lations, including longitudinal cohorts and clinical trials,
the less sensitive assay, trends in new HIV diagnoses,
trends in sexually transmitted infections and risk behavior
and repeat testing histories [6]. An iterative Delphi
method is used by an expert panel to arrive at the
most plausible estimate and upper and lower bounds.
The final estimate is adopted into the Epidemiological
Profile of the city’s HIV Prevention Plan and used for
programme planning. Using this approach, the estimated
number of people acquiring a new HIV infection in San
Francisco in 2006 was 975 adults (upper and lower
plausible bounds, 801 and 1082, respectively). For MSM
(including MSM-IDU), the corresponding figure is 851
(732–1023).
Both methods of estimating HIV incidence have their
limitations. The CDC methodology extrapolates from
a relatively small number of new HIV diagnoses (590),
ISSN 0269-9370 Q 2009 Wolters Kluwer Health | Lippincott Williams & Wilkins
533