Faculty Bibliography

BACKGROUND: HIV incidence estimates are essential for understanding the evolution of the HIV epidemic and the impact of interventions. Tests for recent HIV infection allow incidence estimation based on a single cross-sectional survey. The BED IgG-Capture Enzyme Immunoassay (BED assay) is a commercially available and widely used test for recent HIV infection. METHODS: In a systematic literature search for BED assay studies, we identified 1181 unique studies, 1138 of which were excluded based on titles or abstracts. We conducted reviews of the 43 remaining publications and a further 23 studies identified on conference Web sites or by colleagues. Thirty-nine articles were included in the final review. We investigated the sensitivity of incidence values to various estimation methods and parameter choices. RESULTS: BED assay surveys have been conducted on 5 continents in general populations and high-risk groups, using 1 or more of 10 distinct incidence formulae. Most studies used estimators that do not account for assay imperfection. Those studies that correct for assay imperfection commonly do not use locally valid assay parameters. Incidence estimates were very sensitive to methodological and parameter choices. Most confidence intervals provided good assessment of uncertainty due to counting error, but only a few incorporated parameter uncertainty. CONCLUSIONS: BED assay surveys can produce valid HIV incidence estimates, but many studies have not sufficiently accounted for assay imperfection. Future studies should (1) report all information necessary for incidence point and uncertainty estimation, (2) use an unbiased estimator with locally valid assay calibration parameters, and (3) compute confidence intervals that take into account parameter uncertainty.

Disparities in breast cancer survival have been observed between African American and white women. There are also known differences in mean baseline white blood cell (WBC) count among racial and ethnic groups. If the WBC count falls below conventionally defined treatment thresholds for patients undergoing adjuvant chemotherapy, reduced doses or treatment delays may occur, which could lead to race-based differences in treatment duration. We used the tumor registry at Columbia-Presbyterian Medical Center to identify 1178 women with newly diagnosed stage I and II breast cancer from whom we collected base-line information for 73 African American women and 126 age- and tumor stage-matched white women. Of these women, 43 African American and 93 white women underwent adjuvant chemotherapy. African American women had statistically significantly lower WBC counts than white women at diagnosis (6.2 x 10(9)/L for African American women versus 7.4 x 10(9)/L for white women, difference = 1.2, 95% confidence interval [CI] = 0.2 to 1.2; P =.02) and after treatment (5.3 x 10(9)/L for African American women versus 6.4 x 10(9)/L for white women, difference = 1.1, 95% CI = 0.2 to 2.5; P =.03). Overall, African American women required a statistically significantly longer duration of treatment than white women (19 weeks versus 15 weeks, respectively, difference = 4 weeks, 95% CI = 0.5 to 7.2 weeks; P =.03). The lower baseline WBC counts and longer duration of treatment for early-stage breast cancer in African American women compared with those in white women result in lower dose intensity of treatment for African American women, possibly contributing to observed racial differences in breast cancer survival