The Quitting Matters Pilot demonstrated that it is feasible to reach a demographically and geographically diverse sample of PWH for the purpose of remotely delivering smoking cessation interventions for this population using a mobile digital device (e.g., DTx). Furthermore, it demonstrates that DTx can lead to high levels of treatment engagement and utilization with these technologies in this population. Our remote trial was able to reach PWH from 10 US states, as well as HIV clinics across the Eastern US and engage with a sizeable portion of contacted clinics. Relative to QuitGuide, a standard of care DTx for smoking cessation developed by the National Cancer Institute that has shown efficacy in prior trials, (Bricker et al., 2020) our tailored DTx for PWH (LTQ-H) achieved comparable levels of device usability, but significantly higher rates of days of use, total amount of device engagement, and targeted interactions with active smoking cessation content. These levels of engagement were assessed during a 90-day period using state-of-the-art statistical approaches for intensive longitudinal data. This finding is important, since in a previous DTx study, we demonstrated that device engagement with LTQ, and not QuitGuide, causally mediated smoking outcomes at trial endpoint (Browne et al., 2021). Further, more than 75 % of PWH recruited in this study had high school or fewer years of education which demonstrates that the LTQ-H device engaged a population traditionally underrepresented in technology-based interventions (Mackert et al., 2016; Kontos et al., 2014). Individuals with lower educational attainment often face barriers to accessing and benefiting from digital health tools (Berkman et al., 2011). Therefore, successfully engaging this group suggests that the LTQ-H device is both usable and engaging across diverse educational backgrounds, ensuring that digital innovations do not exacerbate existing health disparities (Latulippe et al., 2017).

While the Quitting Matters Pilot trial was not powered to detect statistical differences between arms in smoking outcomes, LTQ-H had descriptively higher rates of biochemically confirmed smoking abstinence, and larger smoking reductions from baseline to trial endpoint. It is important to note that our randomization procedure failed to equally distribute participants across baseline features known from previous research to predict low quit rates, such as cannabis use, mental health symptoms, and smoking intensity (Ni et al., 2020; Streck et al., 2020; Rogers et al., 2020). More specifically, participants randomized to LTQ-H used cannabis on twice the number of days in the prior thirty days compared to participants randomized to QuitGuide, had three times greater levels of psychological stress than QuitGuide participants, and smoked more CPD at baseline. Moreover, biochemical verification of one participant assigned to the LTQ-H arm failed likely due to self-reported combustible cannabis use prior to the verification test, which is consistent with the fact that 18 % of LTQ-H participants (vs 5 % for QuitGuide) reported at baseline mixing cannabis and combustible tobacco products.

The results of this pilot trial suggested that conducting a remote clinical trial of digital interventions for smoking cessation in this population was acceptable but may require a more robust recruitment strategy (e.g., higher incentives and/or clinic involvement). Further, the use of biochemical verification improved the accuracy of our outcomes. However, combustible cannabis use limits this approach to biochemical verification, which may have reduced our overall quit rate for the LTQ-H arm. The Quitting Matters pilot suggests that future studies should use a combination of CO and cotinine samples to minimize false positives, which has been tested in remote studies (Raiff et al., 2022; Kong et al., 2017) and is consistent with recent guidelines (Benowitz et al., 2020).

The largest study examining the use of a digital intervention for smoking cessation in this patient population found quit rates of 14.9 % (Shuter et al., 2022). Other pilot studies like the one reported here found quit rates ranging from 15 % to 40 %, (Schnall et al., 2022; Poudel et al., 2023; Bui et al., 2022) yet those studies used different endpoints (e.g., 2-month follow-up), (Bui et al., 2022) international populations (e.g., Cambodia) (Poudel et al., 2023; Bui et al., 2022) or reported different primary smoking outcomes (e.g., smoking reductions) (McClure et al., 2024). Together with the fact that LTQ-H demonstrated high levels of device engagement during the three months trial period, these smoking outcomes suggest that LTQ-H holds great promise for the remote delivery of smoking cessation interventions among PWH.

From an implementation science standpoint, the levels of HIV clinic outreach of the Quitting Matters Pilot are particularly significant in the context of the COVID-19 pandemic. One of the clinic directors stated that, “My professional life has been completely overtaken by COVID.” We believe this statement represented the experience of many infectious disease clinics in the U.S. at the time we recruited for this pilot study. In that context, the proportion of clinics that expressed interest in our study provides preliminary support for the implementation of LTQ-H in HIV clinics throughout the country.

Our pilot trial had several limitations. First, our recruitment rates were slightly low relative to the total number of candidates that were initially contacted, which is typical of trials recruiting patients for cancer treatment and low prevalence diseases or conditions such as HIV (Lara et al., 2001; Haider, 2022). However, these low recruitment rates call for a more robust national recruitment strategy in this patient population. Our two-step procedure of participant enrollment and verification (i.e., one screening visit followed by a randomization visit) may have reduced our final enrollment rates, but this procedure also mitigated participant deception, a common occurrence in online clinical trials (Heffner et al., 2021). Low recruitment rates can also be expected from pilot studies with a moderate budget for personnel and advertisement such as this pilot. Overall, one year of recruitment generated a sample size that was fully adequate for the goals of a pilot trial (Eldridge et al., 2016a). Second, our biochemical verification relied solely on expired breath CO, which has inherent limitations (e.g., it cannot confirm abstinence beyond 3–5 days) and may have been inappropriate for a population with high levels of combustible cannabis use. Rogers et al., (2020); Rice et al., (2022). Data from this pilot study also suggests a more adaptable biomarker collection procedure (both CO and cotinine) to avoid sample contamination from e-cigarettes, or NRT use, as conducted in prior remote trials (Vilardaga et al., 2019; Brooke, 1996). The use of anatabine/anabasine to capture combustible tobacco products could also be considered. A more adaptable biomarker collection approach is consistent with most recent SRNT taskforce guidelines (Benowitz et al., 2020). Third, notable imbalances between treatment arms in patient characteristics known to predict low quit smoking rates may have biased our study in favor of the QuitGuide app. Larger trials may consider stratified randomization to account for the strong influence of these patient characteristics. Fourth, both LTQ-H and QuitGuide directly encouraged the use of NRT. However, adherence to the NRT patch was low (<32 %), and did not differ between arms. These low adherence rates were consistent with prior literature, (Pacek et al., 2018) and in this study could also be explained by the lack of in-person contact with a provider (low-touch medical oversight), a clinical tradeoff between more intensive treatment modalities and remote digital trials that ultimately favors scalability and high reach. Finally, while the study was able to collect a significant number of device interactions with LTQ-H and QuitGuide, an unexpected software upgrade led to a loss of 32 % of device interactions in the LTQ-H arm. However, the statistical approach we chose to analyze this data provided robust parameter estimates to account for individual and group variability in intensive longitudinal datasets and ample statistical power to evaluate our device engagement hypotheses.

In conclusion, the Quitting Matters Pilot trial demonstrated that LTQ-H, a novel digital intervention, has the potential to engage a racially and demographically diverse population of PWH with low literacy, may achieve higher user engagement and acceptability than a DTx designed for the general population, and shows promising smoking cessation outcomes. This pilot also established the feasibility of conducting a fully remote trial with biochemical verification in this population and fostered community partnerships with HIV clinics. Building on these findings, a fully powered, nationwide remote hybrid Type 1 effectiveness-implementation trial is currently underway to evaluate the effectiveness and implementation potential of LTQ-H among a larger sample of PWH across the U.S. (NCT06883097).