The introduction of pre-exposure prophylaxis (PrEP) has been a groundbreaking advancement in the drive to eliminate HIV [1]. Despite its efficacy, the overall impact of PrEP is heavily influenced by adherence levels among users [2]. High adherence to PrEP is crucial for its effectiveness in preventing HIV infection, yet PrEP adherence over time is poor, particularly among youth [3,4]. To reap the potential benefits, interventions for young men who have sex with men and young transgender women who are disproportionately impacted by HIV in the United States that can enhance and maintain PrEP adherence over time are needed [5,6].

In recent years, digital health interventions (DHIs) have emerged as promising tools in promoting health behaviors, including medication adherence [7]. These interventions offer innovative ways to engage users, provide information, and support adherence through interactive and personalized digital platforms. Engagement in DHIs has been shown to play a critical role in their efficacy in supporting behavior change across a variety of health domains [8-11]. Research has consistently demonstrated that increased user interaction with digital platforms is associated with improved outcomes, particularly in the context of HIV prevention and care [12-15].

The mechanisms through which engagement enhances efficacy are multifaceted. Active participation in DHIs can lead to increased knowledge, stronger motivation, and better self-management skills among users [16-18]. These interventions often use reminders, motivational messages, and educational content, which help sustain users’ commitment to their health regimens [16,17]. Promotion of engagement through incentives and gamification has also shown promise, though studies documenting their ability to maintain engagement over time are limited [19-22].

The Adolescent Medicine Trials Network for HIV Interventions (ATN) conducted a study using the Prepared, Protected, emPowered (P3) app, a comprehensive PrEP adherence DHI, and its enhanced version, P3+, which incorporates in-app adherence coaching [23]. The P3 app included multiple features aimed at increasing user engagement, including a social wall, daily quests, and medication tracking, as well as both in-app “virtual rewards” in the form of unlocking video series produced specifically for the study and real-world rewards in the form of monetary incentives awarded based on principles of behavioral economics [24-26]. This study provided a rich dataset to explore how digital engagement correlates with PrEP adherence among a young demographic at high risk for HIV. This paper seeks to analyze data from a subset of the P3 study participants—those who were part of the intervention arms using P3 and P3+. The focus is on understanding how different levels of user engagement with the app’s features were associated with adherence to PrEP as well as the costs. This analysis is crucial, as it contributes to optimizing digital health strategies to support sustained PrEP use.

The findings from this analysis underscore the critical role that user engagement plays in the effectiveness of DHIs aimed at enhancing adherence to PrEP among young individuals at high risk for HIV. A key finding from this study is the significant association between higher levels of engagement with the P3 intervention and increased adherence to PrEP. This aligns with prior research, which has consistently demonstrated that user engagement is a pivotal factor in the success of DHIs across various health domains, including HIV care and prevention [13,14]. Despite the high levels of engagement observed in this study, it is challenging to draw direct comparisons with other DHIs due to the lack of standardized metrics for measuring engagement. A recent scoping review focused on engagement in DHIs for sexual and gender minorities noted that only a small fraction of trials reported engagement metrics, and even fewer studies defined an “optimal” level of engagement necessary to achieve significant health outcomes [35].

The P3 app employed several strategies to boost user engagement, including financial incentives and gamification [19,23]. These methods are supported by a growing body of evidence that suggests such approaches can effectively maintain user interaction over extended periods. For example, Looyestyn et al [20] conducted a systematic review that concluded gamification often increases engagement with online health programs, although the long-term effects on health outcomes remain uncertain. However, enhancing engagement strategies through more robust and personalized approaches has the potential to further improve adherence rates [36]. Moreover, conducting longer-term studies could offer valuable insights into the lasting impact of these strategies, thereby enabling the refinement and adaptation of interventions for sustained long-term outcomes [37].

Interestingly, this study observed that while the P3+ intervention, which included in-app adherence coaching, led to increased engagement (eg, more time spent in the app and higher incentive earnings), this did not translate into significantly better adherence compared to the standard P3 arm. This, coupled with the increased cost and complexity of P3+ delivery, indicates that further studies are necessary to determine whether this intensified intervention is the appropriate fit. This finding may indicate a potential ceiling effect where additional engagement beyond a certain threshold does not necessarily result in proportional improvements in adherence. Using sequential multiple assignment randomized trials could be particularly useful in this context, as they would allow for the systematic evaluation and optimization of the intervention’s components (and human resources) to better tailor the approach to individual needs [35]. Our results here show that increasing frequency of use was of relatively low cost, while increasing intensity of use cost US $45 an hour. As these are interim outcomes, we cannot declare whether they are cost-effective as we have no standard against which to measure the results as the literature in this area is extremely limited. Although the P3+ compared ot P3 did not increase adherence, it may have produced other health benefits that were not measured as part of the trial.

Given the challenges of sustaining engagement in standalone digital health interventions, future research may explore whether adherence support features could be effectively delivered through widely used social and dating platforms. These platforms attract substantial daily engagement from populations at high risk for HIV, suggesting potential reach. However, our study did not collect data on participants’ platform preferences, usage patterns, or privacy concerns, and thus any discussion of integrating interventions into these environments remains hypothetical. Importantly, any such approach would require additional research to assess user acceptability, data security considerations, and potential risks, as well as partnerships with platform developers. If found feasible and acceptable, embedding adherence support within high-traffic commercial apps could reduce the costs of sustaining standalone interventions and may leverage existing infrastructure for personalization and scalability.

This study has several limitations that must be acknowledged. Collection of biologic outcome data was impacted by the COVID-19 pandemic, potentially affecting the accuracy of adherence measurements and the robustness of the findings. The cost analysis presented is based on data collected over a limited period, with potential disruptions also caused by the COVID-19 pandemic. Additionally, due to the small number of participants who were initiating PrEP for the first time, the study team was unable to conduct meaningful comparisons between novice and experienced PrEP users. This limited the ability to explore whether patterns of adherence or engagement differed by prior PrEP experience. Future studies with a more balanced distribution of new and experienced users are needed to examine these potential differences. Moreover, this study was unable to adjust for behavioral confounders such as sexual activity frequency, PrEP experience, and partner status, as well as demographic confounders that are known to influence the relationship between engagement and adherence. This may result in an estimated association that reflects the influence of confounders rather than the true effect of the exposure. Due to the limited sample size, we were unable to explicitly examine how these confounders affect engagement and adherence, or how they interact with one another. Future research with larger samples is warranted to enable comprehensive adjustment for these factors and to better understand their combined impact. Furthermore, variability in site-specific factors, such as participant recruitment and staff composition, likely contributed to the wide range of cost estimates. While staff were asked to report time for non–study-related tasks, separating the two may have been challenging, particularly in relation to staff meetings. In addition, some tasks conducted in a study setting, such as recruitment, may not be a good proxy for real-world costs and are likely an overestimate. A notable limitation of this cost analysis is the exclusion of initial app development expenses, which are often among the most substantial costs for digital health interventions. This decision was made to provide a clearer estimate of the ongoing operational and maintenance costs associated with scaling the P3 and P3+ interventions in community settings, as these are the costs most likely to be absorbed by public health agencies or clinical partners. Including development costs would significantly increase the apparent total cost, potentially skewing the analysis away from the more immediate, recurring expenses that directly impact long-term sustainability. Moreover, app development costs can vary widely based on factors such as platform selection (eg, proprietary vs open-source), feature complexity, integration with existing health care systems, and ongoing user feedback. For instance, the P3 app incorporated multiple interactive elements, including social walls, daily quests, and personalized adherence coaching, each of which adds to the initial investment. Given this variability, including these costs would complicate direct cost comparisons across different digital health interventions. Nonetheless, future studies should consider a more comprehensive economic evaluation that includes both fixed and variable costs, as this will be critical for assessing the true scalability and long-term viability of such digital health solutions in real-world settings. Finally, this study provided financial incentives for app engagement which may not be sustainable outside of research contexts.

In conclusion, this study highlights the critical role of user engagement in enhancing PrEP adherence among young individuals at high risk for HIV. While increased engagement through DHIs like the P3 app correlates with higher adherence rates, the findings suggest that simply intensifying interventions, such as with the P3+ model, may not necessarily lead to proportionate improvements. The additional costs and complexities associated with more intensive approaches require careful consideration and further investigation. Future research should prioritize the development of more personalized and adaptive engagement strategies, leveraging methodologies such as sequential multiple assignment randomized trials designs to optimize intervention components in real-time. By focusing on long-term adherence and the sustainability of engagement, these efforts can better inform the design of scalable, cost-effective digital health solutions that are both effective and responsive to the evolving needs of at-risk populations.