Dissertations and Theses

Date of Degree


Document Type


Degree Name

Doctor of Philosophy (Ph.D.)


Epidemiology and Biostatistics


Denis Nash

Committee Members

Sarah Braunstein

Don Hoover

Sheng Li

Subject Categories

Epidemiology | Public Health


HIV; seroconversion; viral suppression; antiretroviral treatment (ART); universal test and treat (UTT);


BACKGROUND: Voluntary HIV testing followed by immediate antiretroviral treatment (ART) initiation (universal testing and treatment) has become an integral part of strategies to eliminate HIV and control the HIV/AIDS epidemic. Minimizing the time from HIV infection to ART initiation is essential for universal test and treatment to be optimally effective. In New York City, an epicenter of the HIV epidemic, the ‘treat all’ recommendation, immediate treatment for all people diagnosed with HIV, was made in late 2011, and efforts to ‘treat all’ were contemporaneous with large scale HIV testing initiatives in NYC. The overarching goal of this dissertation was to examine trends in the timeliness of diagnosis and ART initiation using data from the population-based New York City HIV surveillance registry before, during and after the “treat all” recommendation.

METHODS: I utilized data from the New York City population-based HIV surveillance registry to assess the timing of diagnosis and ART initiation in New York City. For aim 1, to describe and quantify trends in early diagnosis (e.g., examine median CD4 count at diagnosis and proportion of acute HIV cases among all new diagnoses) and early ART initiation (e.g., proportion with CD4 count >500 at ART initiation), I used data on NYC residents diagnosed from 2012-2015. For aim 2, to estimate the time from seroconversion to diagnosis, we applied published estimates of CD4 decline after infection from seroconverter cohorts to our population, NYC residents newly diagnosed with HIV. To verify the assumption that the square root of the CD4 cell count decreases linearly over time prior to antiretroviral treatment (ART) initiation, we compared estimates of diagnosis delay based on first versus second pre-ART CD4 counts, using data on NYC residents diagnosed from 2006-2015 (sub-aim 2a). Finally, using methods developed in Aim 2, we estimated the time from HIV seroconversion to diagnosis and to ART initiation among NYC residents diagnosed from 2006-2015 for aim 3.

RESULTS: In the first aim, I examined the timeliness of diagnosis and treatment initiation in the universal test and treatment era. Among 9987 NYC residents with HIV diagnosed from 2012 to 2015, diagnosis was early (a CD4 cell count ≥500/μL or diagnosed with acute HIV infection) in 35%, and 87% started ART by June 2017. The annual proportion of persons with early diagnosis did not increase appreciably (35% in 2012 vs 37% in 2015; P = .08, Cochran-Armitage test for trend). Overall, 69% of persons had started ART at 6 months after diagnosis. The time from diagnosis to ART initiation decreased from year to year. Within 6 months of diagnosis, 62%, 67%, 72% and 77% of persons with HIV diagnosed in 2012, 2013, 2014, or 2015, respectively, had started ART, with median (interquartile range) times to ART initiation of 3.34 (1.34–12.75), 2.62 (1.28–10.13), 2.16 (1.15–7.11), and 2.03 (1.11–5.61) months, respectively.

In the second aim, I adapted a CD4 decline model to estimate diagnosis delay (time from seroconversions to diagnosis). Among 12,849 NYC residents who were diagnosed with HIV from 2006 to 2015 with at least 2 pre-ART CD4 count measurements around time of diagnosis, the average diagnosis delays based on the first or second pre-ART CD4 count were similar (4.93 years (95% Confidence intervals (CI):4.84-5.03) and 4.85 years (95% CI:4.76-4.95), respectively, p-value=0.09, Wilcoxon signed-rank).

In the third aim, I used methods developed in Aim 2 to estimate the timing of seroconversion and estimated the timeliness of diagnosis and treatment initiation. Among 28,162 people diagnosed with HIV during 2006-2015, 89% initiated ART by June 2017. The median CD4 count at diagnosis increased from 326 (Interquartile range (IQR):132-504) to 390 (IQR:216-571) cells/µL from 2006-2015. The average time from estimated seroconversion to ART initiation decreased by 33% from 8.0 years (95% confidence interval [CI]:7.8-8.2) in 2006 to 5.4 years (95%CI: 5.1-5.6) in 2015. Contributing to the reduction in time to ART initiation, the average time from estimated seroconversion to diagnosis decreased by 22%, from 6.5 years (95% CI:6.3-6.7) to 5.1 years (95% CI:4.9-5.4) from 2006-2015, and the average time from diagnosis to ART initiation reduced by 87%, from 1.5 years (95% CI:1.4-1.5) to 0.2 years (95% CI:0.2-0.3) from 2006-2015.

DISCUSSION: The time to ART initiation was reduced in tandem with expanded HIV testing and treatment efforts in New York City. We found considerable progress in rapid ART initiation: a) the proportion of persons initiating ART within 6 months of diagnosis increased from 2012 to 2015, b) the time from seroconversion to ART initiation decreased by 33% over a 10 year period, and c) the time from diagnosis to ART initiation decreased by 87%, and is now on average very short. Despite these improvements, disparities persist in the ART initiation delay so efforts should focus on subgroups for whom progress still needs to be made. Finally, substantive efforts are needed to reduce delays in diagnosis (i.e., the time from seroconversion to diagnosis). Targeted HIV testing strategies are needed to more rapidly identify people with undiagnosed HIV soon after HIV seroconversion in order to achieve further reductions in HIV incidence and mortality in key subgroups who continue to be negatively impacted by the HIV epidemic.

Included in

Epidemiology Commons



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.