USING TOKENIZATION TO OVERCOME LIMITATIONS OF CLAIMS DATA TO IDENTIFY USE OF REMOTE PATIENT MONITORING AMONG PATIENTS WITH PERITONEAL DIALYSIS
Author(s)
Firas M. Dabbous, MS, PhD1, Divya Shridharmurthy2, Rhonda Shaw, BEc3, Samuel Huse, MS4, Kathryn Evans, MS, MPH5, Vanessa Danielson, MSc6, Ariel Berger, MPH7.
1Thermo Fisher Scientific, Libertyville, IL, USA, 2San Mateo, CA, USA, 3LivaNova, Amstelveen, Netherlands, 4Thermo Fisher Scientific, Cambridge, MA, USA, 5Thermo Fisher Scientific, Bethesda, MD, USA, 6Vantive, Liverpool, United Kingdom, 7Thermo Fisher Scientific, Waltham, MA, USA.
1Thermo Fisher Scientific, Libertyville, IL, USA, 2San Mateo, CA, USA, 3LivaNova, Amstelveen, Netherlands, 4Thermo Fisher Scientific, Cambridge, MA, USA, 5Thermo Fisher Scientific, Bethesda, MD, USA, 6Vantive, Liverpool, United Kingdom, 7Thermo Fisher Scientific, Waltham, MA, USA.
OBJECTIVES: Claims databases are often used for real-world evidence (RWE) generation, but their ability to accurately identify clinical events is constrained by availability and specificity of relevant Current Procedural Terminology (CPT) codes. By linking data from the SHARESOURCE Platform (SSP), a two-way remote patient monitoring (RPM) platform for peritoneal dialysis (PD), this study assessed whether available procedure codes accurately captured SSP use.
METHODS: Adult PD patients with confirmed SSP use in the SSP database (1 January 2020-31 December 2023) were tokenized and linked separately to Komodo and Norstella claims databases. The date of SSP initiation was designated the index date, and patients without ≥6 months of continuous enrollment before this date were excluded. Patient characteristics in the 6-month pre-index period were summarized. RPM-related procedure codes were identified in claims occurring within ±30 days of index date.
RESULTS: A total of 3,486 tokenized patients with confirmed SSP use were identified in Komodo; for Norstella it was 2,066. Mean age was 63.1 in Komodo and 56.8 years in Norstella; 57.0% and 53.9%, respectively were male. The most common payer type was Medicare Advantage (69.6%) in Komodo; and commercial insurance (33.6%) and Medicaid (32.3%) in Norstella. RPM-related CPT codes ±30 days of index date were identified in 38 patients (1.1%) in Komodo and 20 patients (1.0%) in Norstella databases respectively.
CONCLUSIONS: The findings demonstrate substantial underutilization of RPM-related codes in clinical practice, resulting in significant under-reporting of RPM use in administrative claims data. Without deterministic linkage through tokenization to external data sources such as SSP, identifying RPM users in secondary databases is not currently feasible. Improved reimbursement for RPM services and consistent use of RPM-specific billing codes may encourage more complete coding of RPM use, enabling more accurate identification of RPM users in claims data and enhancing the utility of these data for RWE generation.
METHODS: Adult PD patients with confirmed SSP use in the SSP database (1 January 2020-31 December 2023) were tokenized and linked separately to Komodo and Norstella claims databases. The date of SSP initiation was designated the index date, and patients without ≥6 months of continuous enrollment before this date were excluded. Patient characteristics in the 6-month pre-index period were summarized. RPM-related procedure codes were identified in claims occurring within ±30 days of index date.
RESULTS: A total of 3,486 tokenized patients with confirmed SSP use were identified in Komodo; for Norstella it was 2,066. Mean age was 63.1 in Komodo and 56.8 years in Norstella; 57.0% and 53.9%, respectively were male. The most common payer type was Medicare Advantage (69.6%) in Komodo; and commercial insurance (33.6%) and Medicaid (32.3%) in Norstella. RPM-related CPT codes ±30 days of index date were identified in 38 patients (1.1%) in Komodo and 20 patients (1.0%) in Norstella databases respectively.
CONCLUSIONS: The findings demonstrate substantial underutilization of RPM-related codes in clinical practice, resulting in significant under-reporting of RPM use in administrative claims data. Without deterministic linkage through tokenization to external data sources such as SSP, identifying RPM users in secondary databases is not currently feasible. Improved reimbursement for RPM services and consistent use of RPM-specific billing codes may encourage more complete coding of RPM use, enabling more accurate identification of RPM users in claims data and enhancing the utility of these data for RWE generation.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
MT34
Topic
Medical Technologies
Topic Subcategory
Digital Health
Disease
Urinary/Kidney Disorders