BACKGROUND/RATIONALE:
In the initial evaluation of prostate cancer (the most-commonly diagnosed non-cutaneous malignancy among US men), a patient's chances of receiving guideline-concordant imaging may be more dependent on where he lives than on his disease characteristics. Clinicians use imaging variably across regions where the setting on a thermostat may be dialed up high in some areas, where most patients receive scans, and dialed down low in other areas where few are scanned.

This suggests policies focused solely on reducing inappropriate imaging might simply dial the thermostat down, potentially risking a reduction in appropriate imaging. In order to design and implement policies to optimize prostate cancer imaging, it is important to understand regional variation in prostate cancer imaging and the extent to which the thermostat model contributes to it.

This Career Development Award focused on understanding these principles and informing an implementation effort to decrease inappropriate prostate cancer imaging across VHA.

OBJECTIVE(S):
The goals of this research project were to understand the scope, the causes and the effects of regional variation in prostate cancer imaging.

The goals of this CDA were to provide training to the awardee as he became an independently funded leader in urologic health services research and implementation science.

METHODS:
In Specific Aim #1, the candidate tested the thermostat model in data from the VA Central Cancer Registry and SEER-Medicare, systems with different payment models, system structures, and eligibility criteria. Aim #1 also explored imaging patterns in the country of Sweden, another healthcare system without direct physician financial incentives.

To understand further the potential causes of imaging variation the project explored regional prostate cancer imaging rates across delivery systems (Specific Aim #2) and contributed further depth and detail to these data through semi-structured interviews (Specific Aim #3) to determine physicians' and patients' attitudes regarding imaging.

To determine the effects of regional imaging variation, the candidate is constructing a decision analytic model (Specific Aim #4) assessing the tradeoffs in cost and disease detection of various imaging strategies. These model-based strategies will be compared to current care in order to determine the magnitude of possible benefit which a new strategy might deliver.

FINDINGS/RESULTS:
Specific Aim #1: A project assessing the outcomes of a Swedish effort to provide data on prostate cancer imaging use and imaging guidelines to clinicians was associated with a reduction in inappropriate imaging over a 10-year period, as well as slightly decreased appropriate imaging in high-risk patients. This finding confirms the presence of the thermostat model in a model without direct financial incentives for medical providers and has important implications for implementation of a program to reduce inappropriate prostate cancer imaging in the VA.

We also assessed imaging utilization among Veterans using VACCR data. We found high rates of inappropriate imaging among men with low risk prostate cancer (41%) as well as lower than optimal rates of appropriate imaging among men with high risk prostate cancer (70%). Veterans who utilized Medicare-reimbursed care outside the VA had higher rates of inappropriate imaging but not necessarily higher rates of appropriate imaging. Despite the VA providing a centralized organizational structure, facilitating the efficient flow of information and policy, and providing a compensation structure which dissociates payment from the volume of services delivered, the imaging rates seen in the VA were not dissimilar from those observed among patients treated in a fee-for-service Medicare setting, though inappropriate imaging was somewhat reduced.

Specific Aim #2: We sought to determine the effect of health care system on the appropriateness of care. To do this, we had to create a novel dataset combining SEER-Medicare and VA data. During this past year, we completed the clean-up and merging of the VACCR and SEER-Medicare data - no small feat! We created a novel facility volume variable to try and adjust for potential differences in quality. Regardless of whether this variable is included, it seems that, for appropriate care, there is no difference between patients treated in the VA, patients treated in Medicare, or VA patients who sought care outside the VA. However, inappropriate care is more like to happen among patients treated in Medicare or VA patients who sought care outside the VA, as compared to patients cared for only in the VA. The project was presented as a podium presentation at the 2015 HSR&D meeting and the manuscript has been submitted for review.

Specific Aim #3: This qualitative aim sought to determine the reasons patients and physicians order prostate cancer imaging. We used a qualitative approach explore patient and provider knowledge and behaviors relating to the use of imaging. We conducted 39 semi-structured interviews total including 22 interviews with patients newly diagnosed with prostate cancer and 17 interviews with physicians caring for them at three VAMCs representing a broad spectrum of inappropriate imaging rates. After core theoretical concepts were identified, the Theoretical Domains Framework (TDF) was selected to explore linkages between themes within the dataset and existing domains within the framework. Themes from patient interviews were categorized within four TDF domains. Patients reported little interest in staging as compared to disease treatment (goals) and many could not remember if they had imaging at all (knowledge). Participants tended to "trust their doctor" to make decisions about appropriate tests (beliefs about capabilities). Some patients expressed a minor concern for radiation exposure, but concern about cancer outcomes outweighed these fears (emotion). Themes from physician interviews were categorized within five TDF domains. Most physicians self-reported that they know and trust imaging guidelines (knowledge) yet some are still likely to follow their own intuition, whether due to clinical suspicion and/or years of experience (beliefs about capabilities). Additionally, physicians reported that medico-legal concerns, fear of missing associated diagnoses (beliefs about consequences), influence from colleagues who image frequently (social influences) and the facility where they practice influences rates of imaging (environmental context). Interviews with patients and physicians suggest that physicians are the primary (and in some cases only) decision-makers regarding staging imaging for prostate cancer. This finding suggests a physician-targeted intervention may be the most effective strategy to improve guideline-concordant prostate cancer imaging. We have completed the project and published our findings in Implementation Science.

For Specific Aim #4 the decision analytic aim, we have amassed background data to determine the consequences associated with false positives and incidental findings from diagnostic imaging procedures performed in low-risk patients. We retrospectively reviewed positive prostate cancer biopsies from the Manhattan VA Health care system. Imaging (CT, MRI, bone scan) use was assessed among patients with low-risk disease, for whom the 2015 NCCN Prostate Cancer imaging guidelines recommend against advanced imaging and among high risk patient for whom imaging is recommended. The rate of inappropriate imaging was calculated based on the cohort of low risk men; 75% were classified as low risk prostate cancer and 25% as high risk prostate cancer. Of the men with low risk Pca, 17% underwent inappropriate staging imaging. Of these, 63% had follow up imaging for incidental findings: this is the false positive rate, which is alarming high. Highlighting the negative effects of inappropriate imaging will likely aid to decrease overuse of health care resources and improve quality of care. We are currently finalizing this manuscript. We have also completed the literature review and are building the decision tree for a decision analytic model.

IMPACT:
Understanding the scope, causes and effects of unwanted regional imaging variation has begun to suggest important strategies to control this problem. The findings from this project, especially the qualitative data on barriers to utilization of imaging guidelines, have informed a VA Investigator Initiated Research grant 1 I01 HX002038-01A2, "A multi-modal, physician-centered intervention to improve guideline-concordant prostate cancer imaging," which has been recommended for funding by HSRD. This is a 10 site implementation trial of a physician focused behavioral intervention to improve guideline-concordant imaging to stage men with incident prostate cancer. The trial should start in February 2018.

---

External Links for this Project

---

PUBLICATIONS:

None at this time.

DRA: Health Systems Science, Cancer
DRE: Diagnosis, Technology Development and Assessment
Keywords: none
MeSH Terms: none