Top 5 Articles in 2020
2020 was a year like none other! At the Penn Medicine Nudge Unit, many projects were put on hold while the health system quickly adapted to meet the needs of the COVID-19 pandemic. However, it also created some opportunities for our group to focus on new areas of work and rethink the design of our work. Here is a quick look at the Top 5 articles published in 2020 from the Penn Medicine Nudge Unit, based on Altmetric scores and citations.
Top 5 Articles Published in 2020
In this study led by Sujatha Changolkar, latent class analysis using electronic health record (EHR) data identified two phenotypes of clinician practice patterns – one that had a lower clinical workload and one that had a higher clinical workload. We then looked at how these two groups differed in their response to a nudge in the EHR for influenza vaccination. We had previously found that this nudge led to a significant 9.5-percentage point increase in influenza vaccination. However, in this study we found that clinicians with a lower clinical workload had no change in their vaccination rates. Instead, the bulk of the impact was on clinicians with a higher clinical workload. This indicates the nudge may have helped streamline processes that benefited busy clinicians.
4. PLOS One: Association Between Behavioral Phenotypes and Response to a Physical Activity Intervention Using Gamification and Social Incentives: Secondary Analysis of the STEP UP Randomized Clinical Trial
In this study led by Shirley Chen, we used participant data on personality traits, risk preferences, and social support to develop three behavioral phenotypes and evaluated their response to a behaviorally-designed game to improve physical activity which was run among participants in 40 US states. The group with a phenotype of “extroverted and motivated” had increases in activity during the intervention but they went away during follow-up. Those who were “less active and less social” had increases in physical activity during the intervention and it was sustained during follow-up. Those with the phenotype of “at-risk and less motivated” had no response to any intervention. For more insights from this work, see our article in Harvard Business Review.
3. JAMA Oncology: Effect of Integrating Machine Learning Mortality Estimates with Behavioral Nudges to Clinicians to Increase Serious Illness Conversations Among Patients with Cancer: A Stepped-Wedge Cluster Randomized Clinical Trial
In this randomized trial co-led by Chris Manz and Ravi Parikh, we conducted one of the first clinical trials testing the combination of machine learning methods with behavioral nudges within clinical care. Each week, clinicians in the intervention were sent an email with a list of up to 6 patients who they were scheduled to see in the upcoming week that had the highest risk of 6-month mortality. They were asked to precommitment to having a “serious illness conversation” (SIC) to discuss treatment planning and goals of care at the end-of-life. The email had peer comparison feedback on their performance compared with other clinicians. If not appropriate, they could opt a patient out with just one click of a mouse. Otherwise, they would get a text message on the day of the patient’s appointment with them reminding them of the commitment. Among all patients, the intervention tripled the rate of SICs from 1.3% in control to 4.6% in the intervention. For the highest risk group of patients, it increased 5-fold from 3.6% in control to 15.2% in the intervention group. For more insights from this work, see our article in STAT.
In this randomized trial led by Sri Adusumalli, we tested two nudges in the EHR to encourage cardiologists to either prescribe or increase dosing of statins for patients at high risk of major cardiovascular events such as heart attack or stroke. This is one of our first prospective evaluations of randomized nudges in the EHR. The trial found that the interventions had no impact on prescribing overall but had a small increase in the optimal dosing of statins among patients at highest risk who already had atherosclerotic cardiovascular disease.
In this analysis of an ongoing randomized trial, we compared the sustainability of using smartphones or wearable devices to remotely-monitor activity levels among patients discharged from the hospital. At 6 months, 61.5% of patients using a smartphone were still transmitting data while only 46.5% of those using a wearable were doing so. Since most people already have smartphones, these findings suggest we can use the devices already in our pockets to help improve engagement within remote-monitoring programs.
That’s a quick look at the Top 5 articles from the Penn Medicine Nudge Unit in 2020. Stay tuned for an upcoming post with a sneak peek on what’s in store for 2021.
Have a question or want to reach out? Click here to send me a message.