Reply to Marks et al

Abstract

TO THE EDITOR—The letter from Marks and colleagues contributes to the accumulating data on the spectrum of engagement in human immunodeficiency virus (HIV) infection care and its implications for test-and-treat strategies for HIV prevention [1]. In our study, we estimate that 19% of HIV-infected individuals in the United States have an HIV-1 RNA level (viral load) of <50 copies/mL [2]. By utilizing meta-analytic data and previously unpublished data from multiple US HIV care centers, Marks and colleagues estimate that 29% of HIV-infected individuals in the United States have a viral load of <75 copies/mL and that 34% have a viral load of <400 copies/mL [1]. On the basis of their estimate or ours, the majority of HIV-infected individuals in the United States has detectable HIV-1 RNA levels and thus can contribute to the forward transmission of HIV infection. The estimates of the proportion of US HIV-infected individuals with undetectable HIV-1 RNA levels by Marks and colleagues [1] fall within the range of our simulations (Figure 3 in our article [2]). Our simulations were admittedly simplistic because most varied the estimate of successful transition across a single step in the engagement cascade. We used this approach because these were the best available estimates in the published literature and to demonstrate the effect of uncertainty in our estimates of successful transition at each stage of the engagement in care spectrum. Although Marks and colleagues add important information to the available data on the spectrum of HIV care, we believe their analyses have limitations that may cause an overestimation of the proportion of US HIV-infected individuals with an undetectable HIV-1 load. The major limitation is that rates of initial linkage to HIV care were not clearly taken into account. Marks and colleagues use an estimate of 59% for retention in HIV care over time [3]. In the meta-analysis on which this number is based, it was also found that just 69% of individuals with diagnoses of HIV infection in the United States entered HIV medical care over time [3]. It is very challenging for individual care centers to estimate the number of individuals who fail initial linkage because the total number of new HIV infections is difficult to ascertain. In that regard, studies utilizing state databases frequently can account for linkage failure and retention failure at the same time. For this reason, these studies were preferentially used in our review [2]. In the meta-analysis used by Marks and colleagues [1], studies that utilized state databases estimated that 45% of individuals were retained in HIV care [3]. We have other smaller concerns regarding the analyses by Marks and colleagues. In the quoted meta-analysis, more recent engagement in care studies performed since 2003 demonstrated that only 42% of individuals were adequately retained in care [3]. Furthermore, included in the 59% estimate are studies that utilized patient self-report of engagement in HIV care. In these studies, 72% of individuals were engaged in care, which may be an overestimate [3]. Thus, more recent and more objective measures of engagement in care suggest that the 59% figure overestimates retention in care. Finally, we have some concern that the cohort data presented in the letter of Marks and colleagues may have excluded patients who may be less likely to use and adhere to antiretroviral therapy by requiring a clinic visit coupled with a viral load measurement within 90 days of that visit. Although we list these potential limitations of the data presented by Marks and colleagues, we feel that their work adds important information on rates of virologic suppression in real-world clinic settings in the United States. We would argue that their results are not substantially different from ours. On the basis of their estimate or ours, 66%–81% of HIV-infected individuals in the United States have detectable HIV-1 RNA levels, which is inadequate to achieve the goals of the National HIV/AIDS Strategy and to prevent HIV transmission at the population level. Strategies to improve all aspects of the engagement in care spectrum are needed to decrease the ‘national viral load, which has the potential to incrementally decrease HIV infection incidence in the United States in years to come.