An Economic Impact Model for Estimating the Value to Health Systems of a Digital Intervention for Diabetes Primary Care: Development and Usefulness Study

Abstract

Background Diabetes is associated with significant long-term costs for both patients and health systems. Regular primary care visits aligned with American Diabetes Association guidelines could help mitigate those costs while generating near-term revenue for health systems. Digital interventions prompting primary care visits among unengaged patients could provide significant economic value back to the health system as well as individual patients, but only few economic models have been put forth to understand this value. Objective Our objective is to establish a data-based method to estimate the economic impact to a health system of interventions promoting primary care visits for people with diabetes who have been historically unengaged with their care. The model was built with a focus on a specific digital health intervention, Precision Nudging, but can be used to quantify the value of other interventions driving primary care usage among patients with diabetes. Methods We developed an economic model to estimate the financial value of a primary care visit of a patient with diabetes to the health system. This model requires segmenting patients with diabetes according to their level of blood sugar control as measured by their most recent hemoglobin A1c value to understand how frequently they should be visiting a primary care provider. The model also accounts for the payer mix among the population with diabetes, documenting the percentage of insurance coverage through a commercial plan, Medicare, or Medicaid, as these influence the reimbursement rates for the services. Then, the model takes into consideration the population base rates of comorbid conditions for patients with diabetes and the associated current procedural terminology codes to understand what a provider can bill as well as the expected inpatient revenue from a subset of patients likely to require hospitalization based on the national hospitalization rates for people with diabetes. Physician reimbursement is subtracted from the total. Finally, the model also accounts for the level of patient engagement with the intervention to ensure a realistic estimate of the impact. Results We present a model to prospectively estimate the economic impact of a digital health intervention to encourage patients with documented diabetes diagnoses to attend primary care visits. The model leverages both publicly available and health system data to calculate the per appointment value (revenue) to the health system. The model offers a method to understand and test the financial impact of Precision Nudging or other primary care–focused diabetes interventions inclusive of costs driven by comorbid conditions. Conclusions The proposed economic model can help health systems understand and evaluate the estimated economic benefits of interventions focused on primary care and prevention for patients with diabetes as well as help intervention developers determine pricing for their product.

Mapping Factors That Affect the Uptake of Digital Therapeutics Within Health Systems: Scoping Review

“…An important aspect of these incentives is the development of reimbursement codes for digital therapeutics [ 2 , 53 , 99 , 148 , 153 , 169 , 170 , 212 ]. The United States created reimbursement codes for Medicare and Medicaid to be able to better reimburse digital health, but these payment codes are used for digital health as a whole and not specifically for digital therapeutics [ 53 , 160 ]. After certification, the uptake of digital therapeutics is facilitated by the ability of health care professionals or health insurers to identify accredited digital therapeutics through an openly accessible register or inventory [ 4 , 59 , 224 , 226 , 229 , 231 , 232 ].…”

“…At the health system level, digital therapeutics can influence health care costs and resource use [ 41 , 47 , 48 , 53 , 69 , 71 , 72 , 80 , 81 , 94 , 95 , 105 , 107 , 111 , 113 , 121 , 124 , 128 , 131 , 133 , 147 , 153 , 154 , 160 , 164 , 171 , 172 , 180 , 189 , 207 , 211 , 212 , 228 , 231 , 232 , 237 , 238 , 255 - 258 , 261 , 262 , 265 , 269 , 273 ] and affect the workload and workflow of the health workforce, potentially bringing about significant improvements [ 19 , 47 , 60 , 62 , 78 , 80 , 90 , 105 , 116 , 124 , 126 , 128 , 134 , 146 , 147 , 176 , 188 , 194 , 196 , 221 , <...…”

Leveraging behavioral science and artificial intelligence to support mental health in the workplace: a pilot study

“…Each employee only saw messages likely to be meaningful to them based on their past interaction patterns. This presentation style reduces user burden by delivering pointed, relevant content, and may support health equity by facilitating the inclusion of barriers experienced by under-served groups (32).…”

“…At the same time, tailored, personalized interventions are difficult to scale. Recent advancements in behavioral science and artificial intelligence have shown promise in using a behavioral reinforcement learning (BRL) agent to assemble personalized health communications and deliver them at scale across a variety of health arenas ( 31 – 33 ). Specifically, a BRL agent was used to deliver behavioral science-based messaging to patients overdue for their mammogram, resulting in increasing the number of scheduled and attended screenings in a population overdue for this prevention visit ( 31 ).…”

“…Specifically, a BRL agent was used to deliver behavioral science-based messaging to patients overdue for their mammogram, resulting in increasing the number of scheduled and attended screenings in a population overdue for this prevention visit (31). Artificial intelligence has also been used to deliver personalized messages prompting individuals to complete their COVID vaccine series (32). Reinforcement learning specifically offers the ability to increase the frequency of a desired behavior based on data feedback (34).…”