Background: Chronic conditions such as diabetes and hypertension pose major public health and economic challenges. Innovative tools, including remote patient monitoring (RPM), aim to address barriers to care by digitally transmitting health data to providers.
Objective: This study assesses the impact of RPM on health care utilization in Medicare beneficiaries with diabetes or hypertension.
Methods: This retrospective cohort study used a longitudinal panel from 2016 to 2020 based on Louisiana Medicare Fee-for-Service claims and Beneficiary Files. Patients were included if they were diagnosed with diabetes or hypertension before 2016. RPM exposure was defined by common procedure terminology/healthcare common procedure coding system codes (eg, 99453, 99454, 99457, 99458, 95250, 95251, 99091, E2100-E2103, E0607), with the corresponding month considered the treatment month. Control beneficiaries were randomly assigned treatment months matching the treatment group. Outcomes were monthly counts per 1000 beneficiaries of outpatient, emergency department, and inpatient visits (all-cause and disease-specific), constructed via revenue center/place-of-service/primary ICD-10 (International Classification of Diseases-10) codes. We estimated inverse-probability-treatment-weighted 2-way fixed-effects models with standard errors clustered at the physician level. Data represent secondary analysis of deidentified claims; no participant recruitment occurred. All confidence intervals are reported at the 95% confidence level.
Results: The study included 5488 beneficiaries in the treatment group and 341,226 in the control group. RPM was associated with increased level of outpatient visits in the treatment group by 2284.01 (95% CI 2059.75-2508.27) visits per month per 1000 people (P<.001) after initiation and decreased the trend of outpatient visits by 37.61 (95% CI -37.98 to -37.24) visits per month per 1000 people compared with the control group (P<.001). Similarly, the level of emergency room visits increased in the treatment compared with the control by 84.60 (95% CI 72.21 to 96.99) visits per month per 1000 people initially after RPM initiation (P<.001) and decreased the trend by 1.40 (95% CI -1.60 to -1.20) visits per month per 1000 people (P<.001). The level of inpatient visits increased by 32.64 (95% CI 24.84-40.44) visits per month per 1000 beneficiaries in the treatment group compared with the control immediately after RPM prescription (P=.002), but the trend decreased by 0.54 (95% CI -0.66 to -0.42) visits per 1000 people each month (P<.001).
Conclusions: This study is among the first large, claims-based evaluations of RPM in older adults with diabetes or hypertension using a beneficiary- and time-fixed-effects design. We identify a pattern of increased health care utilization following RPM initiation, followed by a gradual decline that does not fully offset the initial increase. By characterizing utilization trajectories and not clinical endpoints, this study helps reconcile inconsistent findings in prior RPM research and highlights the importance of implementation factors in Medicare populations. Strategies to maximize the value of RPM may include appropriate population targeting, patient education on digital health technologies, and clinician workflow support to support effective integration into routine care.
Keywords: diabetes; health utilization; hypertension; medicare; remote patient monitoring.
© Debra Winberg, Charles Stoecker, Jian Li, Lizheng Shi. Originally published in the Journal of Medical Internet Research (https://www.jmir.org).