Early initiation of antiretroviral therapy (ART) for patients infected with the human immunodeficiency virus (HIV) and subsequent lifelong viral suppression have important clinical and prevention benefits. [1, 2] HIV treatment and prevention guidelines stress the importance of HIV testing, linking persons diagnosed with HIV infection to care programs, and initiation and maintenance of ART. [3–5] Despite these guidelines, a substantial number of persons with HIV (PWH) in the United States have not been engaged in HIV care services. Centers for Disease Control and Prevention (CDC) data from 2014 show that of the estimated 1.1 million PWH, 51% were not virally suppressed, [6] including an estimated 57% of black, 52% of Hispanic, and 43% of white PWH. Unsuppressed HIV infection, whether from lack of diagnosis, treatment, or medication adherence, is the most important factor contributing to HIV incidence, currently estimated at >35,000 new infections per year in the United States. [6] Although treatment initiation, medication adherence, and retention in care are key for controlling HIV infection from clinical and prevention perspectives, there are few evidence-based behavioral interventions to support these behaviors, especially among black and Hispanic PWH.

CDC’s Compendium of Evidence-Based Interventions and Best Practices for HIV Prevention has identified 24 interventions with varying levels of efficacy for improving either medication adherence (N = 14) or retention in, linkage to, and re-engagement in HIV care (N = 11) among PWH. [7] Whether delivered to individuals or groups, these interventions all require multi-session attendance and relatively intensive organizational resources to implement. There is a clear need for simple adherence and retention interventions that can be sustained at relatively low cost and brought to scale quickly to achieve maximum patient coverage. [8]

Two widely disseminated video-based interventions, VOICES/VOCES (V/V) and Safe in the City (SITC), were tested in sexually transmitted disease (STD) clinics and found to be cost-beneficial and effective in reducing STD incidence. [9–14] It is therefore reasonable to expect that a waiting room video having a similar format and user-informed development process as V/V and SITC [15] would have a positive effect on HIV-related health outcomes and be acceptable to both patients and clinic staff. The 25-minute V/V [11] and SITC [16] education-entertainment videos are available in English and Spanish, can be viewed in clinic waiting rooms or small group sessions, and comprise a set of intervention videos. Drawing on the successful model developed for this set of videos, we created Taking Care of Me (TCOM), an intervention video with supporting posters for use in HIV clinic waiting rooms. This paper describes its impact on treatment initiation, medication adherence, and retention in care.

We used a commercial institutional review board (IRB) because CDC was not involved in data collection and none of the study sites had an affiliated IRB. On April 6, 2016 the IRB approved the study with a waiver of informed consent since patients and providers would not be actively recruited or surveyed, and only routinely collected data from each clinic’s EMR would be used (S1 Appendix). The intervention was registered as a clinical trial after data collection began because no participants were enrolled in the study. The authors confirm that all ongoing and related trials for this intervention are registered.

IRB information:

Western Institutional Review Board | 1019 39th Avenue SE Suite 120 | Puyallup, WA 98374–2115 Office: (360) 252–2500 | Toll Free: (800) 562–4789 www.wirb.com

Study number: 1163249

Protocol number: 20160574

Online tracking: 11–434725

Trial registration: http://www.ClinicalTrials.gov (NCT03508310)

There were a total of 4003 patients across both study conditions (Atlanta = 2629, Huntsville = 1374) (Fig 1, S2 Appendix, Table 1). Of the patients in Atlanta 1706 were in one study condition, and 923 (54.1%) were in both conditions. Of the patients in Huntsville 794 were in one study condition, and 580 (73.0%) were in both conditions. Patient sex, gender identity, and HIV risks were similar across sites and conditions. Compared to Atlanta, Huntsville had slightly more (4 percentage points) younger patients (18–24 year category) and somewhat more (8 percentage points) patients who received their HIV diagnosis over 10 years ago. The main difference between sites was race and ethnicity. Atlanta’s clinic population was 90% black, while Huntsville’s was 48% black and 44% white. No adverse events were reported during the study.

For HIV treatment initiation we found a significant (p < 0.01) overall increase of 10.4 percentage points between study periods and significant intervention effects at both sites. (Table 2) Significant but differential intervention effects were observed for sex, race/ethnicity, age, HIV risk exposure, and time since HIV diagnosis. Treatment initiation increased among males (10.7 percentage points) and females (9.4 percentage points) between study periods (both p < 0.01). Significant increases were observed among blacks (9.7 percentage points, p < 0.01), Hispanics (18.1 percentage points, p = 0.03), whites (9.5 percentage points, p < 0.01), and persons of ‘other’ race or ethnicity (28.7 percentage points, p = 0.01). There were significant increases (p < 0.01) among persons 18–34 years of age (14.3 percentage points) and persons 35–87 years of age (8.2 percentage points). Treatment initiation increased between study periods among persons reporting heterosexual behavior (7.5 percentage points, p < 0.01), male-to-male sex (11.4 percentage points, p < 0.01), and persons reporting ‘other’ risk behaviors (14.1 percentage points, p = 0.02). Significant increases were found among those diagnosed <1 year (23.1 percentage points, p < 0.01), 1–5 years (8.0 percentage points, p < 0.01), 6–10 years (6.5 percentage points, p = 0.03), and >10 years (7.4 percentage points, p = 0.01) ago who were exposed to the intervention.

Differences in retention in HIV care, although observed in the anticipated direction for both sites combined (1.4 percentage point increase), were not statistically significant and did not differ substantially by site (Table 3). Retention increased 28.0 percentage points between study periods for patients in the ‘other’ racial category. Although the number of patients in this sub-group was small, this difference was statistically significant (p = 0.03). No statistically significant differences were observed when examined across sex/gender, remaining race/ethnicity categories, age, HIV risk, or time since HIV diagnosis.

Viral suppression was the primary measure of medication adherence. At Huntsville and Atlanta combined, viral suppression improved 6.0 percentage points (p < 0.01) between study periods. (Table 4) However, viral suppression in Atlanta increased by 17.5 percentage points (p < 0.01), but decreased in Huntsville by 8.6 percentage points (p < 0.01). In Atlanta, all of the percentage point changes for subgroups were increasing between study periods. Significant increases were observed for patients who were male or female; black, Hispanic, or white; between 35–87 years old; had heterosexual or male-to-male sexual risk; and were diagnosed one or more years ago. (Table 5) Conversely, in Huntsville, all of the percentage point changes for subgroups were decreasing between study periods. Significant decreases were observed for patients who were female or transgender; black or Hispanic; between 35–87 years old; had male-to-male sexual risk; and were diagnosed six or more years ago.

Similar findings were observed with respect to the secondary measure of laboratory-confirmed undetectable viral load. At both sites combined, undetectable viral load improved 3.8 percentage points (p = 0.03) between study periods. (Table 6) The sites again had divergent results. Undetectable viral load in Atlanta improved by 15.2 percentage points (p < 0.01), but dropped in Huntsville by 7.6 percentage points (p = 0.01). Again in Atlanta, all of the percentage point changes for subgroups were increasing between study periods. (Table 7) Significant increases were observed for patients who were male or female; black; between 25–34 or 35–87 years old; for any transmission risk category; and for any time since diagnosis. Conversely, in Huntsville, all of the percentage point changes for subgroups were decreasing between study periods. Significant decreases were observed for patients who were female or transgender; black or Hispanic; between 35–87 years old; had male-to-male sexual risk; and were diagnosed 6–10 years ago.

In our evaluation of the TCOM video, we found a significant overall increase in the percentage of treatment initiators among patients who visited the sites while the video played and posters were displayed (intervention condition), compared to patients who visited without the video and posters (comparison condition). The overall positive change in treatment initiation also was observed in sub-group analyses of demographic (i.e., sex, race/ethnicity, age) and behavioral characteristics (i.e., HIV risk, time since HIV diagnosis). The nature of the study design allows us to draw conclusions about association but not causation. Given that TCOM featured multiple vignettes that modeled PWH of diverse racial/ethnic backgrounds and sexual orientations, it is not surprising that the video showed evidence of changing behavior among a heterogeneous patient population. However, the overall effects for retention in HIV care were statistically non-significant. Medication adherence, measured by viral suppression and undetectable viral load, was significantly greater during the intervention period than the comparison period in Atlanta and both sites combined. Other co-occurring factors may have contributed to observed results, although their potential effect is unclear. For example, the findings from the Strategic Timing of Antiretroviral Therapy (START) study, which showed the benefit of early ART initiation, were published in August 2015. [1] START’s results were widely known prior to publication and likely integrated in clinical practice before our study’s comparison period. Also, HIV treatment guideline recommendation to initiate ART for all PWH was upgraded to the highest level (i.e., A1) in January 2016. [18] Finally, health insurance coverage increased nationwide during the study period; however, most PWH are Medicaid-eligible. Alabama and Georgia are not Medicaid expansion states, so increased national Medicaid coverage likely had no impact on the numbers of PWH attending the study clinics.

Unlike Atlanta, Huntsville’s adherence results were negative. The reason is unclear. We were unable to explore influences of socioeconomic status, educational attainment, or health insurance coverage because the study did not collect these EMR data. Stratifying by race/ethnicity, age, HIV risk, and time since HIV diagnosis did not explain the difference between sites. Sub-group analyses of adherence in Huntsville showed negative results across all groups, suggesting systemic (e.g., clinic hours) or environmental (e.g., local policies) factors at work. We contacted the Huntsville clinic to inquire about any such changes between study conditions. The Huntsville clinic confirmed there had been a 23.5% increase in the number of new patients during the intervention period but reported no environmental, political, policy, or other changes in the clinic or Huntsville that could have contributed to the observed intervention effects (S3 Appendix). It is highly probable that the Huntsville clinic patients who were new to HIV treatment in the intervention condition did not have time to become virally suppressed during the study measurement period. If we had used a less stringent medication adherence measure (e.g., any viral load suppression within study period), the results may have been higher. Ceiling effects also may have contributed to the between-site difference. Huntsville had 77.7% suppression and 73.4% undetectable proportions during the comparison period; these proportions were only 41.1% and 9.7%, respectively, in Atlanta. Thus, Huntsville had less room for improvement.

The results presented in this paper extend the findings of Warner et al. (2008) [13] by showing that a video-based intervention focused on HIV treatment can be effective at improving HIV clinical outcomes. Having an effective tool that can be easily delivered in a waiting room affords a high degree of audience-targeting with relatively low effort. As an effective approach to supporting improved HIV clinical outcomes, this intervention has the potential to be widely disseminated to HIV clinics and other appropriate settings. Additionally, it is reasonable to expect that the acceptability among patients and clinics seen with SITC [19] would be true of TCOM.

Video as an intervention tool continues to be prominent in current public health approaches. [20] TCOM represents the first stand-alone video-based intervention we are aware of that is effective at improving HIV-related health outcomes for PWH. The relative overall change in treatment initiation (i.e., starting a new behavior) for TCOM is comparable to the level of change in incident STDs (i.e., resulting from adopting new behaviors) for SITC (10%). [13] Although video-based intervention effect sizes may be modest when compared to those achieved from more resource-intensive interventions, video-based interventions deliver their content consistently and can be scaled more efficiently, because they are rapid to disseminate, easy to implement, and inexpensive to sustain. For example, 260 agencies initiated V/V in the first 20 months of its availability [21], and over 2100 clinics requested SITC in its first 15 months. [19] TCOM has the potential to be an important addition to the V/V and SITC set of videos. The time and resources required for rigorous evaluation are barriers to developing and evaluating new videos, and changes in societal norms and advances in HIV prevention and treatment may influence a video’s continued relevance. Given the relative scarcity of public health resources, the trade-offs between faster dissemination and lengthy evaluation should be considered. By forgoing intensive evaluation of videos that follow successful models, new prevention videos could be made available sooner. This could likely have a significant positive impact on the health of PWH, reduce future health care costs, and help achieve national HIV prevention, care, and treatment goals.

Our study suggests that there were significant intervention effects among specific sub-groups for treatment initiation and medication adherence outcomes. For treatment initiation, most demographic and behavioral characteristic groups appeared to have benefited from TCOM exposure. It is plausible that the observed intervention effects are due partly to the video’s embedded prevention messages and its appeal to a wide patient audience. However, the intervention may have had an indirect effect if clinicians watched the video and learned how to effectively dialogue with patients about the benefits of early treatment. For both measurement levels of medication adherence, males, blacks, and older patients appeared to have benefited from exposure to TCOM. The significant intervention effects for both adherence measures seen among Atlanta, but not Huntsville, patients are difficult to explain. In addition to a possible ceiling effect, it is probable that Huntsville’s simultaneous increases in patient population and treatment initiation temporarily increased clinic-wide viral load levels. Retention in care is a difficult outcome to measure in a standard way. HRSA’s definition needed to be adapted for this study design, and the study’s 10-month intervention period may not have been enough time to see the full effects of the intervention.