Physicians’ Decision Choice of Conservative Treatment Impacted by Evidence, Peers, and Financial Incentives Open Access

Liu, Yu (Summer 2021)

Permanent URL:


Active Surveillance had become an increasingly popular disease management strategy for localized prostate cancer between 2001 and 2015. Using active surveillance, rather than other active treatments, for localized prostate cancer patients presented opportunities for health care cost saving. Urologists were gradually adopting active surveillance. In my dissertation, I studied the adoption of active surveillance from three perspectives: urologists’ referral network position, peer influence, and financial influence. I also compared the use of active surveillance with common active treatments, e.g., prostatectomy and intensity modulated radiation therapy. I found that urologists who were at the center of the referral network were more likely to use active surveillance than urologists who were at the periphery of the referral network. I also found that the patients’ selection criteria of peers had different impacts for active surveillance, prostatectomy, and intensity modulated radiation therapy. Lastly, the reimbursement reduction of active treatments reduced urologists’ use of conservative treatment, and the impacts were different for urologists who used different kinds of treatments as their major treatment methods. My research results had three key health policy implications. In the era of precision medicine, patients were more likely to undertake diversified treatment strategies, including conservative treatment methods for low risk cancer. Primary care doctors and specialists who were at the center of the referral network may disseminate information about the efficacy of conservative treatments. Therefore, policy makers may leverage their influence to promote cost effective conservative treatments. Second, policy makers may consider introducing patients’ selection criteria measurements as a physician performance evaluation method. Third, the Centers for Medicare and Medicaid Services shall consider the impacts of reimbursement cut of active treatments on the promotion of conservative treatments, and consider financial incentives to promote conservative treatment. 

Table of Contents

1 Introduction

1.1 Overview of Conservative Treatment for Early Stage Prostate Cancer

1.1.1 Localized Prostate Cancer and Treatment

1.1.2 Trends of Localized Prostate Cancer Diagnosis and Treatments

1.1.3 Active Surveillance Treatment for Localized Prostate Cancer

1.2 Physician Treatment Decision Choice

1.3 Physician’ De-adoption of Low-value Care and Adoption of Conservative Treatment

1.4 Motivations and Implications

1.4.1 The Adoption Pattern of Conservative Treatment: Active Surveillance

1.4.2 Implications for Healthcare System Cost

1.5 Dissertation Outlines

1.6 Innovations

2 Physician Referral Network Position and Decision Choice

2.1 Background

2.1.1 Physician Network Position and Adoption of Treatment

2.1.2 Current Challenges Evaluating Networks’ Impact on Treatment Adoption

2.1.3 Physician’s Network Established by Primary Care Doctors

2.2 Data and Methods

2.2.1 Data and Patient Cohort for Treatment Prediction

2.2.2 Patients’ Probability of Undertaking Different Treatments

2.2.3 Urologist Cohort and Analysis Sample

2.2.4 Urologists’ Practice Style

2.2.5 Urologists’ Referral Network

2.3 Model

2.3.1 A Simple Theoretical Model

2.3.2 Econometric Model

2.4 Preliminary Results

2.4.1 Urologists’ Network Group and Patients’ Characteristics

2.4.2 Urologists’ Network Group’s Impact on Adoption of Active Surveillance

2.5 Subgroup and Sensitivity Analysis

2.5.1 Teaching status, solo-practice, IMRT self-referral subgroups

2.6 Limitations and Future Work

2.6.1 Urologists Types and Adoption of Active Surveillance

2.6.2 Accuracy of Referral Network

2.7 Summary

3 Physician Peer Influence on Active Surveillance and Other Treatment Choices

3.1 Background

3.2 Urologist Cohort and Methods

3.2.1 Urologists’ Co-workers’ Practice Styles

3.2.2 Urologist’s Cohort

3.2.3 Hypotheses and Models

3.3 Preliminary Results

3.3.1 Urologists’ Characteristics Before and After Move

3.3.2 Urologists and Their Co-workers’ Practice Styles

3.3.3 Peers’ Impacts on Urologists’ Selection of Treatment after Moving to a New Practice

3.3.4 Urologist Move to a Practice Because of Practice Style and Division of Labor

3.4 Limitations and Future Work

3.5 Conclusions and Discussion

3.5.1 Comparison of Percentage Usage and KL Score as Practice Style

3.5.2 Impacts of Co-workers for Different Treatments

3.5.3 Targeted location to move

3.5.4 Patients Selection

3.6 Summary

4 Patients’ Volume and Financial Impact on Physicians’ Active Surveillance Usage Choice

4.1 Introduction

4.2 Treatment Costs and Patients’ Volume

4.2.1 Average Cost for Different Treatments

4.2.2 Patients’ Volume

4.2.3 Treatment Distribution and Expected Reimbursement Difference

4.2.4 Control Variables

4.2.5 Type of Urologists for Subgroup Analysis

4.2.6 Econometrics model

4.3 Preliminary Results

4.3.1 Treatment Costs

4.3.2 Changes of Average Reimbursement per Patient and Changes of Patient Volume

4.3.3 Impact of Average Patients’ Reimbursement and Patients’ Volume on Active

4.3.4 Subgroup Analysis

4.4 Limitations

4.5 Conclusions and Discussion


List of Figures

Chapter 1

1.1 Figure 1: Prostate Cancer Screening, Treatment Guidelines and Clinical Trials

1.2 Figure 2: Trends for Newly Diagnosed Localized Prostate Cancer between 2004 and 2015

Chapter 2

2.1 Figure 1: Patients’ Average Entropy Score by Year Diagnosis

2.2 Figure 2: Patients’ Average Entropy Scores by Treatment Options

2.3 Figure 3: Urologists’ Average KL Distance Scores by Year Diagnosis

2.4 Figure 4: Randomly selected 200 urologists’ Network Graph

2.5 Figure 5: Trends of Closeness Centrality Score by Periods

2.6 Figure 6: Probabilities of Undertaking Prostatectomy, IMRT, Active Surveillance by Network Group

2.7 Figure 7: Differences in Probabilities of Treatments between Subgroups

2.8 Figure 8: Active Surveillance Usage Probability Differences between High Network and Low Network Group by Teaching Status

Chapter 3

3.1 Figure 1: Distribution of Year that Urologists Changed Practice

3.2 Figure 2: Number of Years Practiced Before and After Move

3.3 Figure 3: Patient Volume by Periods

3.4 Figure 4: Coworkers’ Percentage of Treatment Impact on Treatment Choice

3.5 Figure 5: Coworkers’ KL Distance Impact on Treatment Choice

3.6 Figure 6: Impact of Percentage of Coworker before Move

3.7 Figure 7: Impact of KL Distance of Coworker before Move

Chapter 4

4.1 Figure 1: Active Surveillance Usage Probabilities Difference

List of Tables

Chapter 1

1.1 Table 1: Localized Prostate Cancer Treatment Cost

1.2 Table 2: National Comprehensive Cancer Network (NCCN) 2010 Guidelines Treatment Options for Localized Prostate Cancer

Chapter 2

2.1 Table 1: Summary Statistics of Patients’ Characteristics by Treatment Options

2.2 Table 2: Prediction Accuracy Rate by Treatment and Year Diagnosis

2.3 Table 3: Analysis Sample Selection Process

2.4 Table 4: Correlations between KL Distance and Treatment Usage Percentage, Aggregated Patients' Entropy Score, and Patients' Volume

2.5 Table 5: Correlations between Different KL Distance Scores

2.6 Table 6: Correlations between Different Usage Percentages

2.7 Table 7: Summary Statistics of Analysis Sample by Network Group

2.8 Table 8: Patients' Probabilities of Undertaking Active Surveillance by Different Network Groups and periods

2.9 Table 9: T-Test Network Scores for Different Groups of Urologists

2.10 Table 10: Patients' Probabilities of Treatments by Network Groups

Chapter 3

3.1 Table 1: Patients’ Characteristics Before and After Urologists’ Move

3.2 Table 2: Urologists’ and Co-workers’ Practice Styles Before and After the Move

3.3 Table 3: Odds Ratio of Practice Style Impact by Being Coworker

3.4 Table 4: Paired Patients' Probabilities T-test P-Value between Own and Coworkers

Chapter 4

4.1 Table 1: Inpatient Input price Index and Medicare Economic Index

4.2 Table 2: Average Cost for Different Treatment

4.3 Table 3: Average Reimbursement Change and Patients’ Volume Change by Year

4.4 Table 4: Active Surveillance Treatment Number and Percentage by Year

4.5 Table 5: Number of urologists and Number of Patients treated by Urologist Type

List of Supplementary Files

Chapter 2

Supplementary Table 1: Coefficients and Standard Errors of Formula 4

Chapter 4

Supplementary File 1: Variables for Cost Calculation

Supplementary Table 1: Coefficients and Standard Errors of Formula 5

Supplementary Table 2: Coefficients and Standard Errors of Formula 6

Supplementary Table 3: Coefficients and Standard Errors of Formula 7

About this Dissertation

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
  • English
Research Field
Committee Chair / Thesis Advisor
Committee Members
Last modified

Primary PDF

Supplemental Files