Texting Tool Failed to Up Pill Adherence in ACS Patients: TEXTMEDS

Patients who received “motivational and supportive” text messages in the year following ACS were no more likely to take their medications than those who get usual care, show results from the randomized TEXTMEDS study. Yet there were some positive shifts in lifestyle measures.

The trial’s researchers say there are lessons to be gleaned for future interventions, citing an unmet clinical need.

While secondary prevention is crucial, it’s “poorly implemented globally,” they write in their paper, published online yesterday in Circulation. “Simple and low-cost methods that are scalable may help address the implementation gap. More than 4.7 billion people own mobile phones and the use of text messages is ubiquitous.” Earlier studies have shown promise with text-based interventions, but these tended to be small and short, they note.

Lead author Clara K. Chow, MBBS, PhD (The University of Sydney, Westmead Hospital, Australia), told TCTMD that their study is novel by virtue of being multicenter and involving a 12-month intervention. “We were disappointed to not be able to impact on medical adherence,” she said in an email, but the good news is that adherence rates were quite high overall. What they learned is “that structural barriers to adherence such as costs . . . need to be addressed to improve the use of prevention medications.”

Feedback from this and similar studies has shown there are some less-tangible benefits derived from the text interventions, said Chow. “Patients value the ongoing connection with the hospital and their healthcare providers, the provision of additional information, the nudges which remind them of other aspects of their care, and the feeling that they are being supported.”

Erica Schorr, PhD, BSBA, RN (University of Minnesota, Minneapolis), chaired the writing group behind a recent American Heart Association scientific statement promoting the use of mobile technologies for secondary prevention in older patients. To her, the negative results also were unexpected. On the positive side, though, “it was a really large trial, and the researchers reported high engagement and high emotional support by participants,” Schorr commented to TCTMD. “And so it was really interesting that although participants felt very supported with the intervention, it actually didn’t have an effect on medication behavior.”

There’s no obvious explanation for why this attempt didn’t work, she said, but what leads people to take their medicine—or not—involves many factors. While texts can serve as reminders or inspiration, their influence can be surpassed by things like education, income level, underlying attitudes about health and illness, and side effects. It’s not uncommon for patients to not take their drugs as prescribed, Schorr pointed out. “In the United States, this is a challenging problem to say the least. . . . One in five prescriptions that people get are never even filled, and of those that are filled, only half of patients are taking the medications correctly.”

Chow stressed that the lack of success in the current study, which took place in Australia, doesn’t detract from the potential for digital-health interventions. “Virtual engagement methods via text message are implementable, scalable, and useful to our patients post-ACS,” she added. The question is exactly how best to make that connection.

TEXTMEDS

Chow and colleagues enrolled 1,424 patients (mean age 58 years; 79% male) with ACS at 18 Australian public teaching hospitals, randomizing them to receive either usual care (secondary prevention as determined by the treating clinician) or the text-based intervention. In the texting arm of the trial, patients received various messages—directed at general secondary prevention (eg, cholesterol targets, mental health, and healthcare access), lifestyle (diet, exercise, and smoking if relevant), and medications (how the drugs work, side effects, and tips for taking)—and had the opportunity to communicate with their healthcare team by text or telephone. The frequency was four texts per week in the first 6 months, decreasing to three per week in the next 6 months.

Self-reported medication adherence, the study’s primary endpoint, was defined as taking more than 80% of up to five cardioprotective drugs. Use of these medications was high at baseline: aspirin (98.8%), beta-blockers (87.0%), ACE inhibitors/ARBs (78.5%), statins (96.8%), and second antiplatelet agents (86.6%). It remained so over the course of follow-up.

At 12 months, there was no difference in overall adherence between the texting and control groups. Nor were there any differences in rates for the individual medications. Systolic blood pressure and LDL levels were similar, as were the rates of smoking and regular exercise.

However, the researchers did see some small but significant improvements in a few lifestyle risk factors. Patients who received the text-based intervention were more likely to have a body mass index < 35 kg/m² (21% vs 18%; P = 0.01), eat at least five servings of vegetables per day (9% vs 5%; P = 0.03), and eat at least two servings of fruit per day (44% vs 39%; P = 0.01).

Healthcare costs and use of medical services were mostly similar in the two groups, though patients in the texting group were slightly more likely to undergo HDL cholesterol testing and to visit their general practitioner.

‘We Should Not Discount These Small Effects’

The researchers propose several possible reasons why medication adherence didn’t improve, such as drug costs, side effects, and pill burden/treatment complexity, as well as the fact that there was little room for improvement to begin with, since baseline drug therapy was so high and patients were enrolled soon after discharge. Importantly, major pharmacies in Australia already have texting and app-based programs in place to remind patients about their prescriptions. “There also is a possibility that an intervention on healthy lifestyles may give patients a feeling that if they have a healthy lifestyle they can take fewer medications,” the investigators suggest.

To TCTMD, Chow said that it’s possible the intervention would have been more effective had it focused only on medical therapy. “However,” she added, “it was important to us to be able to deliver a simple and comprehensive program that supported the multiple aspects of their preventive care needs.”

Their group is examining strategies for digital communication that go beyond texts to include apps, email, and interactive voice response technology, she said. There’s also the possibility that “further personalization and customization of digital content can maintain engagement in the longer term.”

Patients value the ongoing connection with the hospital and their healthcare providers, the provision of additional information, the nudges which remind them of other aspects of their care, and the feeling that they are being supported. Clara K. Chow

Regarding the shifts in BMI and diet, Schorr said, “we should not discount these small effects. These are all lifestyle risk factors for CVD. When you think about the large proportion of Americans who have a body weight that puts them into overweight and obese BMI categories, any positive effects we can have on those would be significant.” She cautioned, however, that this information is self-reported so may be an overly rosy assessment of the TEXTMEDS participants’ true behavior.

Schorr suggested that one explanation for why the texts didn’t sway medication adherence is that behavior-change techniques sometimes help initially and then start to lose steam over time. “You might need to switch things up after a few months,” she said. “When I read this study it made me think about ‘alert fatigue’ that’s similar to incessant notifications that healthcare workers receive in electronic medical record systems. At first you pay attention to these and they serve as flags and make you think twice about your choices, your behavior, the medications you’re ordering, the things you’re doing, and then over time you become desensitized to them.”

People respond differently to different interventions, she noted. Text reminders may work for some, but bidirectional messaging, which was a part of TEXTMEDS, can serve as another layer of motivation, said Schorr.  “To be able to communicate back or ask questions makes it more like a conversation you would have and makes it feel more real life and personalized.”

There’s a need for affordable, feasible, scalable interventions to reduce CVD risk, she stressed. Beyond medication adherence, there could be other beneficial but less easily measured effects such as quality of life and overall health outcomes.

Schorr and colleagues are working on a mobile app, set to launch soon, called CVD Study. Adults who participate in the research will receive texts and app notifications for medication adherence and exercise based on the activity recorded by their smartwatch or phone, plus general CVD educational information every other day for 6 months. The idea, she told TCTMD, is to take a multipronged approach.

Abhinav Sharma, MD, PhD (McGill University Health Centre, Montreal, Canada), and Robert Avram, MD (Montreal Heart Institute, Canada), in an accompanying editorial, agree TEXTMEDS offers useful information for next steps. “Although there are certainly hype and enthusiasm around mobile devices and healthcare apps, the medical community and policy makers must require high-quality evidence before wider adoption and implementation within healthcare or public settings,” they write.

Future trials might consider focusing on one domain of behavior change to raise the odds of success, Sharma and Avram suggest. Also, “drawing on examples of commercial ventures that leverage rapid A/B randomization testing, academic studies evaluating mobile health intervention should explore adaptive randomization across multiple features simultaneously, which has been described in pharmaceutical trials,” they say. This would enable a closer look at how the timing, frequency, and phrasing of messages affect outcomes.

Another tactic is algorithms that adapt messaging to each patient’s characteristics and behavior, while machine learning could be used to identify populations with shared characteristics in order to better target interventions. And finally, they note, it’s important to take steps to ensure equitable access to these digital strategies.