A Randomized Trial of Incentivization to Maximize Retention For Real-Time Symptom and Activity Monitoring Using Ecological Momentary Assessment in Pediatric Concussion

Abstract

ObjectiveTo determine the incentivization strategy that maximizes patient adherence to report symptoms and activity via ecological momentary assessment (EMA) following pediatric concussion.BackgroundConcussion is a common pediatric injury. Traditionally, outcome assessment has occurred at discrete points-in-time, days or weeks apart, relying on patient's subjective recall of symptoms and activity. EMA is a behavioral measurement approach that allows for reporting of real-time symptoms and behaviors in real-life settings. While feasible in adolescents, the ideal strategy to maximize responsiveness from the emergency department (ED) setting is unknown.Design/MethodsThis was a randomized controlled trial of patients age 13–18 with concussion presenting to an urban, academic pediatric ED within 5 days of injury. Patients were randomized to one of 4 incentive-based arms: 2 dynamic (loss-based and streak) and 2 flat-rate (monetary and electronic device). Through the ReCoUPS app, patients reported symptoms 3 times per day and cognitive activity once each evening for 3 weeks. Physical activity (step count) and sleep were monitored using a FitBit (which was kept by the participant in the electronic device flat-rate arm). The primary outcome was proportion of prompts to which patients responded. Secondary outcomes included daily symptom change and time to symptom resolution.ResultsThirty participants were enrolled, median age 15.5 years, 60% female. Median proportion completed was 81% in the loss-based arm, 59% in the streak accrual arm, 50% in the FitBit-received arm, and 57% monetary flat rate arm. Retention was higher in the dynamic compared to the flat arms (68% v. 54%, p = 0.065). There was no significant difference between morning, afternoon, and evening symptoms. Sixty-four percent of participants had symptom resolution during the 3-week follow-up.ConclusionsDynamic incentivization showed higher rates of response to tri-daily symptom prompts compared with flat-fee incentivization. This data shows tracking concussed youth using EMA from ED is feasible using a dynamic incentivization strategy.

Acceptability and feasibility of mobile phone-based ecological momentary assessment and intervention in Uganda: A pilot randomized controlled trial

“…Our study used training, participant follow-up and financial incentives to support data submission, but saw wide variation in response rate. Our findings are consistent with other EMA research, demonstrating response rates ranging from 31-90%, inclusive of studies incentivizing prompt-based responses [14,[26][27][28]. While limited by the small pilot sample size, our data suggest that greater familiarity with smartphone applications at enrollment may be associated with EMA response success.…”

“…Participants were 48% female with a median age of 31 years (IQR: [25][26][27][28][29][30][31][32][33][34][35][36][37][38]. Everyone owned a cell phone at enrollment, with 77% reporting the ability to keep their phone charged every PLOS ONE day, and just over half having ever used a smartphone app.…”

Uses of mHealth in Injury Prevention and Control: a Critical Review

“…This includes the construction of population-based cohorts that can be used for understanding injury patterns and conducting interventions at scale [17] and daily symptoms tracking to measure concussion recovery [18]. Given the need for frequent assessments of symptoms, there is also ongoing work to test different engagement strategies to promote more complete assessments [19]. Finally, there is emerging evidence on the benefit of mHealth-facilitated interventions for mitigating post-concussive symptoms.…”

Response Rate Patterns in Adolescents With Concussion Using Mobile Health and Remote Patient Monitoring: Observational Study

“…Specific to concussion, several studies examined mHealth technologies to track symptoms or activity after injury. Reporting behaviors (eg, response rate) ranged from 50% to 90% due to differing assessments, tracking frequencies, and duration of follow-up [ 15-17 21 22 undefined undefined ]. Furthermore, these studies tend to reduce response rates to a single number (eg, the percentage of days in which participants completed responses divided by the number of prompts received [ 15 23 ]), with limited exploration of variability in response patterns within a day, across recovery duration, or by certain patient or injury characteristics.…”