Digital Pill Bottles a Weak Player in the HeartStrong Trial

Volpp KG, Troxel AB, Mehta SJ, et al. Effect of Electronic Reminders, Financial Incentives, and Social Support on Outcomes After Myocardial Infarction: The HeartStrong Randomized Clinical Trial. JAMA Intern Med. 2017; EPub Ahead of Print.

The Bottom Line

The use of electronic pill reminders via smart pill bottles did not significantly reduce time to rehospitalization for a vascular event, medication adherence, or cost in a fairly large trial involving 1500 participants. This study tempers the digital health hype surrounding medication adherence via novel devices being developed.

Background on Adherence Research as of Late

It goes without saying that medication adherence has become the golden goose of medical research as of late. Finding a tool or service to help patients adhere to their medication therapy with the goal of reducing negative outcomes due to primary and secondary prevention through drug therapy poses a multi-billion dollar idea that many researchers and companies are looking to solve. However, recent studies created more questions than answers, with many results demonstrating little to no benefit based on posed interventions.

One recent example published in JAMA Internal Medicine, the Remind Trial, demonstrated no real benefits to sending simple devices (e.g. pill boxes, medication tabs, pill bottle alarms) to patients to improve adherence based on monthly possession ratio collected from insurance data. My take on that study was not surprising, as I really see no benefit to simply handing out such tools to random patients hoping to change their habits made no sense. For some time I have been advocating the use of other tools, such as gamification and smart pill bottles to help patients improve their adherence, and it was for that reason I had eagerly looked forward to results of the HeartStrong Trial using Vitality's GlowCap system.

The HeartStrong Study

The premise of the study was to determine if the advent of new technologies (e.g. smart pill bottles) to track adherence and automate communication with patients, combined with a reward system, would help to reduce cardiovascular events amongst patients post-AMI by being more adherent to key therapy regimen (Statin, BB, ASA, Anti-platelet agents [e.g. Plavix]). Their outcomes were:

  1. Time until first vascular readmission (defined as an inpatient hospitalization with a diagnosis of AMI, unstable angina, stroke, CHF, or death).
  2. Time until first admission, total number of readmissions, medication adherence, and total cost.

Methods and Interventions

The HeartStrong Study was a 2-arm RCT. Patients were randomized 2:1 (intervention:control), with the control group receiving standard of care and the intervention group receiving:

  1. Up to 4 Vitality Glow Cap Pill Bottles (for the aforementioned key therapy regimen). Patients were awarded $25 for activating them.
  2. Daily lottery incentives (1/5 chance of winning $5 and 1/100 chance of winning $50)
  3. Option of enlisting a friend/family member/caregiver to support medication adherence (based on alerts)
  4. Social Work resources
  5. Staff engagement advisor to provide close monitoring, feedback, and reinforcement of adherence

Inclusion criteria included patients with 5 pre-identified insurance coverages, 18-80 years old, currently prescribed 2 of 4 medications (aforementioned), were hospital inpatients for 1-180 days, discharged with a ICD code of AMI. Exclusion criteria included diagnosis of metastatic cancer, ESRD w/Dialysis, dementia, or enrollment in a similar trial. Patients were allowed to enroll up to 60 days post discharge. Participants were enrolled from March 2013 until January 2015 and followed for one year.

Data was collected from claims issued by the insurer partners of the study. Medication adherence (aside from ASA due to OTC status) was estimated using proportion of days covered (PDC), which is basically a way to calculate adherence based on days of medication coverage (for a good write up comparing PDC versus MPR [used in the REMIND Trial) see this Link). Interestingly, the investigators used multiple ways to measure adherence making several assumptions, such as if a patient filled one of the cardio-study medications, odds were they were using them for an annual period. 


A total of 1509 patients were randomized, with 1003 being placed into the intervention arm. This was only slightly below their calculate power analysis (1014 Intervention to 506 Control for 1520 total participants).  Demographic data at baseline was similar with no significant differences.  The population was predominately men, around 60 years of age, mostly on the East Coast, and had an average enrollment time into the study after 40-days. 

There was no statistically difference to detect a 6% difference in rate between the intervention and control arm with regards to primary outcome of first time hospitalization for a vascular event or death. By subgroups, men and those with a lower PHQ-2 score seemed to potentially perform better. There was no difference in medication adherence and medical spending as well.

The Authors Thoughts

The authors posited several thoughts regarding their results. In honesty, it was interesting reading the language of their discussion and conclusion, as it really seemed they themselves were surprised by the outcome of this study. 

  1. The biggest issue is that these interventions just do not work. 
  2. Intervention may have worked if applied earlier. As noted, the members had up to 60 days post discharge to enroll, with most patients enrolling at 40-days. This was due to relying on insurance claims data, but a difference may have occurred if this intervention happened right at discharge in stead of a delay.
  3. Maybe a significant difference would have been detected with another outcome outside of AMI. This is based on the changes in observation rates vs admissions that have changed dramatically in the pst few years that could have impacted collected data.
  4. The current (circa 2013) generation of smart pill bottles just aren't good enough.
  5. If everyone had followed the plan it would have done something. Indeed, the per-protocol analysis did detect some difference, however, this doesn't necessarily emulate real-life experiences. So, in a perfect world this may work, but then adherence wouldn't be an issue either.
  6. The Inclusion/Exclusion process left behind only the most determined AMI survivors who may have better outcomes to begin with, and thus biassed the study.
  7. A bigger sample is needed, and that the current sample size is not enough to detect perhaps the small difference that really occurs (which begs the question how much that would be to fund).

The authors concluded that:

The fact that this intervention had negative results is important in highlighting that some of the approaches we might expect to significantly improve adherence do not in the context of a health plan-based intervention for patients after AMI... further investigation in this area remains critical because the population value of thereapetuic advances depends fundamentally on identifying ways to improve adherence to them. - HeartStrong Investigators

My Thoughts

The Remind Trial and the HeartStrong Trial are increasingly supporting my personal views that technology alone or standalone tools themselves will not solve the adherence problem. Other studies that have used Vitality's GlowCaps have also seen not to positive results, such as "Automated Reminders and Physician Notification to Promote Immunosuppression Adherence Among Kidney Transplant Recipients: A Randomized Trial" which was published last year and covered in depth by #NephJC. That study looked at 120 patients post kidney transplant on tacrolimus to determine impact on adherence. Interestingly, there was an increase in the intervention arm of adherence versus the control group, but tacrolimus groups levels were the same between both groups. Interestingly, self-reported adherence rates were also rated high between both groups. 

Looking at the HeartStrong Trial, one of the interventions that captured my attention was #5 - Staff Engagement Advisors. Pulling the supplement, these individuals were responsible to hep the patient getting started with the GlowCap set-up, to coordinate with providers when patients were non-adherent for greater than 14-days. In between they tried to reach feedback partners, mailing a letter to remind them to use the intervention. Interestingly, it seemed they would have reached out to the provider earlier at 7 days. 

Most of these trials are using the tool as the sole intervention it feels. While scalability to reach a larger population without needing clinician input is a definite benefit, I really agree with the authors thoughts that the current technology isn't there yet. These new devices are more or less just tools to help healthcare practitioners identify at risk patients and serve as a communication tool I feel. Reevaluation of digital health interventions using technology should look beyond a sole device reminder and look to use it as a tool as an overall engagement strategy platform is created for the patient. The authors allude to this with:

This intervention was designed to mimic what health plans could do following an AMI hospitalization without and direct involvement of clinicians; perhaps that involvement is need to better engage the difficult-to-engage patient. - HeartStrong Investigators 

Other considerations I have is the sample size, I agree with one of the authors thoughts that a larger trial will be needed. The biggest issue is cost really, which I feel will be a definite problem for small start-ups looking to jump into this space. Which comes back to one of my largest concerns with digital health and study design in this era - Time and development. The GlowCap is arguably different compared to some of the newer digital health tools coming to market. More products have entered the market since the study investigators probably set this up in 2012. With almost 5 years of time to get this data out to the public, I am curios what issue this poses for others. Looking at ClinicalTrails.Gov, there are many companies with studies currently in the pipeline which I am looking forward to seeing demonstrate similar findings or large differences in outcomes. 

Takeaway Points

The HeartStrong Trial tempers the large-scale hype surrounding digital health to fix medication adherence. The trial demonstrates that more work will undoubtedly need to be done, and we will have to keep an eye on studies coming from other companies in this space.