Hypertension (HTN) is a key risk factor for heart disease and stroke [1], with an estimated 78 million Americans aged 20 years and older suffering from this condition [2]. It is associated with significant health disparities [3], as HTN is more prevalent among non-Hispanic blacks than non-Hispanic whites (42.0% vs 28.8%, respectively [3]), and African Americans suffer from greater disease severity, with earlier onset and more complications than age-matched whites [4].

Adherence to medication regimens is an important component of HTN management [5]; however, only half of all hypertensive patients are considered adherent [6,7]. Forgetting to take medications is one of the most commonly cited reasons for nonadherence [8]. Mobile health (mHealth) strategies, such as text message reminders, could be a low-cost and effective way to improve medication adherence that has broad reach. Cell phone use is widespread, with text messaging even more common. Among American adults, 90% own a cell phone [9] and 81% send text messages [10]. Mobile interventions could be particularly effective among African Americans as studies suggest African-American adults are more likely to own a mobile phone (70% vs 61%) [11] and use it as their primary source of Internet access [11].

The goal of this project was to determine the feasibility and acceptability of BPMED, an automated text messaging system designed to improve medication adherence among African Americans with uncontrolled HTN. We also sought to determine the preliminary effectiveness of this approach compared to usual care controls at one-month follow-up. Our primary outcome measure was medication adherence, with secondary outcome measures of blood pressure (BP) and medication adherence self-efficacy. To account for different ways that African-American patients might interact with the health care system, we conducted two parallel pilot randomized controlled trials (RCTs) with participants recruited from primary care and emergency department (ED) settings.

We developed BPMED, an automated text message medication reminder system, to assist African Americans with uncontrolled HTN in remembering to take their HTN medications. BPMED’s development and study protocol are described in detail elsewhere [12]; however, key elements are summarized subsequently. The Wayne State University Institutional Review Board (#0410810B3E) approved this study.

We sought to document the feasibility, acceptability, and preliminary efficacy of text messages for antihypertensive medication reminders. The BPMED participants were found to be very satisfied with and enthusiastic about the program. This finding is consistent with previous text message medication adherence studies for chronic disease that found moderate to high levels of participant satisfaction [19,20] and general comfort with the technology and message content [21]. Demand for such technology is growing with the existence of a large number of apps (n=193) and websites to send medication reminders and log adherence [22,23]. In 2014, 55% of these medication support apps were tailored toward HTN self-monitoring [22,23].

With expanding interest in mHealth interventions, and increasing emphasis on prevention, there have been calls for large-scale RCTs aimed specifically at chronic conditions such as HTN [24]. Our data support this, showing greater numerical improvements in medication adherence at one-month follow-up among individuals randomized to BPMED. Although this was not statistically significant, our sample was underpowered to detect modest differences, and intervention effects may be less pronounced over a short-term follow-up period. Existing literature related to text messaging for medication adherence in chronic disease is mixed [24], but a recent meta-analysis of 16 RCTs found a positive overall effect, suggesting a 17.8% increase in medication adherence rates (from an assumed 50% baseline adherence rate to 67.8%, OR 2.11, 95% CI 1.52-2.93, P<.001) over a mean follow-up period of 12 weeks (range 4-48 weeks) [20]. Thus, we are encouraged to continue research into the potential benefits of text messaging on medication adherence in this population, and are currently conducting a well-powered RCT of mHealth support to improve BP in a cohort similar to our ED trial, with one-year follow-up (NCT02955537). Our finding of near universal support for continued use of BPMED among ED participants lends further credence to this approach.

Regarding medication adherence self-efficacy (a secondary outcome of interest), only minor, nonsignificant improvements were seen among control and BPMED participants. This finding was unexpected as there is an established connection between medication adherence self-efficacy and medication adherence [25,26]. Our small sample size may have been insufficient to reveal differences in medication adherence self-efficacy. Individuals who received BPMED had numerically greater, yet nonsignificant, improvements in SBP and DBP at follow-up. Although there are few studies of text messaging to improve medication adherence targeting patients with uncontrolled HTN, or evaluating direct effects on BP, this trend is consistent with previous work that has established greater reductions in cholesterol and SBP at six months for patients with heart disease who received a mHealth program compared to controls [27].

These pilots were intended to demonstrate feasibility and acceptability of our approach. As such, we had relatively small sample sizes that contributed to our lack of statistically significant effects of BPMED on our primary and secondary outcomes. As noted, these limitations are consistent with other work focused on text message reminders for medication adherence in chronic disease. This highlights the need for larger RCTs of text message medication reminders with longer duration follow-up, a need that has been noted in the literature [20]. The measurement of medication adherence utilizing the MMAS was also a limitation. Although its use is well documented [20], the MMAS is a self-reported measure, not an objective assessment, which can suffer from overestimation of medication adherence [28-30]. Future work should incorporate better measures, including instrumented pill bottles/caps [31] and/or biomarkers that may provide better approximations of adherence [32,33]. Because control participants knew they were in a study about medication adherence, a Hawthorne effect may have contributed to the lack of statistically significant between group differences. This may equal out over longer follow-up periods, particularly with less overt measures of medication adherence. Ultimately, BPMED uses a single component approach to improve medication adherence, perhaps the most important limiting factor. Previous work suggests that multicomponent interventions are more effective at improving medication adherence [23]. Interventions that include bidirectional texting, allowing participants to respond with adherence information, have been found to be significantly more effective than those employing unidirectional text message reminders (relative risk 1.0 vs 1.2, respectively) [34,35]. Moreover, technology-augmented mechanisms, such as instrumented pill bottles, have shown to increase adherence more than text messaging alone [36]. Other similar interventions that utilize pill top monitors and triggered text message reminders when doses are late or skipped found that 87.3% of intervention participants reached 95% or more on-time adherence compared to 61.8% among controls [31]. Future work should take a more robust approach to improving medication adherence, as single component interventions may not be as effective.

Our results demonstrate that text message reminders to improve medication adherence among African Americans with uncontrolled HTN are feasible and acceptable. Our data support the need for more robust trials of mHealth in patients with uncontrolled chronic HTN that are fully powered with longer follow-up periods, more rigorous measures of medication adherence, and multimodal interventions.