Characterization of the Association of Time to ART Initiation from HIV Diagnosis and Survival to Fatal and Non-Fatal Severe Events: An Analysis of the HIV Atlanta Veterans Affairs Cohort Study Open Access

Rabold, Elizabeth (Fall 2019)

Permanent URL:


Studies in low-resource settings have shown benefits in viral load suppression and clinical outcomes in people living with HIV who initiate antiretroviral treatment (ART) immediately after diagnosis, but few studies report on early treatment initiation in high-resource settings. This analysis aimed to compare time to death and other severe non-fatal events between veterans who initiated ART within 60 days of HIV diagnosis (early) and intermediate (61-365 days) or late (>365 days) initiation. Using data from the HIV Atlanta Veterans Affairs Cohort Study (HAVACS) and Clinical Case Registry, we identified all treatment-naive veterans who initiated ART at the Atlanta Veterans Affairs Medical Center between January 2005 and May 2018; veterans missing key dates such as HIV diagnosis or ART initiation date were excluded from the analysis. Using Cox proportional hazards models, we calculated the unadjusted and Veterans Aging Cohort Study (VACS) index-adjusted hazard ratios (HR) for the primary composite outcome (death, AIDS-defining diagnosis, malignancy, severe renal or liver disease, atherosclerotic cardiovascular disease, or invasive infection) and all-cause mortality. In total, 520 veterans (2991.8 person-years) were included in this analysis with 159 total events and 65 deaths. Unadjusted models for the composite outcome and all-cause mortality were not statistically significant. Compared to veterans with early initiation, only late initiation was associated with an increased adjusted hazard of the composite outcome (adjusted HR 1.27; 95% confidence interval [CI], 0.76-2.12 and 1.81; 95% CI, 1.15-2.87 for intermediate and late initiation, respectively). Both intermediate and late initiation increased the adjusted hazards of all-cause mortality (adjusted HR 2.49; 95% CI, 1.05-5.94 and 2.81; 95% CI, 1.24-6.33, respectively). Shorter delay in ART initiation is associated with improved event-free survival, independent of baseline characteristics such as CD4 count, viral load, age, renal function, liver function, anemia, and hepatitis C status. These findings highlight the importance of rapid diagnosis and linkage to care, with a 2-month delay in treatment initiation associated with an increased hazard of death. Adopting strategies to streamline treatment initiation soon after diagnosis may lead to improved morbidity and mortality, potentially serving to close the gap in survival in people with and without HIV.

Table of Contents


Overview.. 4

HIV Epidemiology. 5

Approach to HIV Treatment 6

Rapid Start 7

HIV Pathogenesis and Relationship to Other Conditions. 8

Atlanta Veterans Affairs Medical Center. 10

Existing Systems for Referral 11

Research question. 11


Study Design and Setting. 12

Primary and secondary analyses. 12

Data Source. 12

Inclusion Criteria. 13

Exclusion Criteria. 13

Primary Outcome. 13

Secondary outcome. 14

The secondary outcome was all-cause mortality. 14

Additional analyses. 14

Independent variable. 14

Demographic variables. 14

Diagnoses. 15

Social/behavioral variables. 15

Treatment variables. 15

Loss to follow up, transfer out of care, and censoring. 16

Statistical Analysis. 16

Ethical Considerations. 18


Descriptive data. 19

Primary and secondary outcomes. 20

Survival analysis. 21

Discussion. 23

Limitations. 29

Conclusions. 33

Figures and Tables. 34

Figure 1: Eligible veterans included in analysis. 34

Table 1: Baseline characteristics of veterans at the time of initiating ART, by time since HIV diagnosis. 35

Table 2: Primary outcomes by time from HIV diagnosis to ART initiation. 38

Table 3: Transfer and loss to follow up by time from HIV diagnosis to ART initiation. 38

Figure 2: Composite outcomes, stratified by time from HIV diagnosis to ART initiation. 39

Table 4: Person-years contributed by group, stratified by time from HIV diagnosis to ART initiation. 40

Figure 3: Log(-Log) of Estimate Survivor Function for Composite Outcome. 41

Figure 4: Log(-Log) of Estimate Survivor Function for All-Cause Mortality. 42

Table 5: Pearson Correlation Coefficients. 43

Table 6: Unadjusted and adjusted hazard ratios of primary outcomes by time between HIV diagnosis and ART initiation*. 43

Figure 5: Unadjusted Cox Proportional Hazards Model for Composite Outcome. 44

Figure 6: Unadjusted Cox Proportional Hazards Model for All-Cause Mortality. 45

Table 7: Subdistribution hazards model for competing events by time between HIV diagnosis and ART initiation. 46

Figure 7: Cumulative Incidence Function of Subdistribution Model of Composite Outcome. 47

Appendix A.. 48

RCTs included in Cochrane Review* [30]. 48

Summary of findings in Cochrane Review for Rapid ART compared to standard care for people living with HIV* [30]. 49


Diagnoses and Associated ICD-9/ICD-10 Codes. 50


Veterans Aging Cohort Study Index. 51

Works Cited. 52

About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
Subfield / Discipline
  • English
Research Field
Committee Chair / Thesis Advisor
Committee Members
Last modified

Primary PDF

Supplemental Files