De-implementation of inappropriate thyroid ultrasound

PROJECT SUMMARY/ABSTRACT
Inappropriate use of thyroid ultrasound (iTUS) is an important driver of thyroid cancer overdiagnosis and
overtreatment, which involves high-risk procedures and long-term therapeutics that cause medical,
psychosocial, and financial hardships for patients. Cumulative annual cost of well-differentiated thyroid cancer
care in the U.S. has been estimated to exceed $1.5 billion and is projected to reach $3.5 billion by 2030, and
the potential cost after 5 years of thyroid cancer diagnosis is $50,000 per patient. Thyroid cancer is one of the
fastest-growing cancers in the U.S, but mortality remains very low. Approximately 25% of new cases are
attributable to the identification of small thyroid cancers that are unlikely to cause harm if they were left
undiagnosed and untreated. The biggest driver of small thyroid cancer diagnosis is iTUS use in asymptomatic
people, a practice discouraged by clinical guidelines. The pervasiveness of iTUS despite recommendations
against it suggests the need for active strategies to eliminate it. The process of eliminating practices that are
not evidence-based is known as de-implementation. To date, no studies have provided a replicable and useful
way for health systems to identify their iTUS practices, and there has been no systematic evaluation of
multilevel factors driving it, such that we lack key information about targeted, acceptable, and feasible de-
implementation strategies. Without them, overuse will persist. To fill this gap, we will leverage a
multidisciplinary team with vast experience in computer phenotyping expertise, machine learning, and mixed
method research. We will also use two unique databases: the Rochester Epidemiology Project, a medical
record-linkage system that captures health care information from the entire population of 27 counties in
Minnesota and Wisconsin, and the Patient-Centered Clinical Research Network (PCORnet) that shares a
common data model to organize data into a standard structure. There are three aims. Aim 1: Using the REP
and two PCORnet sites, to develop a replicable computer phenotype to identify patients receiving iTUS. Aim 2:
Using 4 PCORnet sities, to identify patient, clinician, and practice factors associated with iTUS in a
representative sample of healthcare practices. Aim 3: Using mixed methods, to understand factors and identify
potential strategies for iTUS de-implementation acceptable to the patient, clinician, and health system
stakeholders. This proposal is responsive to the objectives of NOT-CA-20-021 to explore de-implementation of
ineffective or low-value clinical practices along the cancer care continuum. At the end of this study, we will
have developed and validated a computer phenotype to identify iTUS across diverse settings, as well as a list
of acceptable strategies likely to decrease iTUS. These findings will be broadly disseminable and will pave the
way for studies—deployed in diverse health systems and targeting patients, clincians, and organizations—that
test the effectiveness of the de-implementation strategies identified here.

Grant Number: 5R37CA272473-04
NIH Institute/Center: NIH
Principal Investigator: Juan Brito Campana