IIR 14-093
Personalized Life Expectancy to Encourage High Value Prostate Cancer Care
John Thomas Leppert, MD MS VA Palo Alto Health Care System, Palo Alto, CA Palo Alto, CA Funding Period: March 2017 - February 2021 Portfolio Assignment: Research Methods Development |
BACKGROUND/RATIONALE:
As an integrated system with access to granular and longitudinal data, the Veterans Health Administration (VHA) is ideally positioned to advance the understanding of the life expectancy, to improve prostate cancer screening strategies, and to generate models of personalized risk-adjusted life expectancy estimates to provide critical information to inform prostate cancer treatment decisions. Veterans receiving care in the Veterans Health Administration may have higher prostate cancer risk due to a family history, race, or exposure to toxins such as Agent Orange and burn pits. Prostate cancer presents a clinical, health policy, and population health challenge. Prostate cancer is the most common male cancer, presents in older men that may have additional medical conditions, and often follows an indolent course. It is estimated that 60% of all prostate cancer cases represent an "overdiagnosis" of clinically insignificant tumors. For prostate cancer patients, "overdiagnosis" refers to the diagnosis of a disease process that would otherwise not go on to cause symptoms or death. Similarly, "overtreatment" refers to the treatment of prostate cancers that would not otherwise go on to cause symptoms or death. OBJECTIVE(S): Our objective is to leverage the power of the standardized electronic health record in the VHA to generate personalized risk-adjusted life expectancy estimates. We will use these estimates to provide critical information to inform prostate cancer screening and treatment medical decision-making. These efforts have the potential to deliver higher quality prostate cancer care, by treating patients most likely to benefit, and while avoiding futile treatment and minimizing treatment-related side effects. The study include three specific aims. Aim 1: To develop general and prostate cancer specific life expectancy estimates for maleVeterans receiving care in the VHA. Aim 2: To develop highly personalized life expectancy estimates as a function of patient-level features including age, race/ethnicity, prior medical claims, disease severity, exposure, health habits, pharmacy, and laboratory data for Veterans with prostate cancer in the VHA. Aim 3: To evaluate the appropriateness of prostate cancer care in the VHA among men with limited estimated life expectancy. METHODS: We will work to develop life expectancy estimates for all male veterans receiving care in the VHA and evaluate how a diagnosis of prostate cancer may modify these estimates. (Aim 1) Next, we use several approaches to generate personalized life expectancy estimates for patients with a diagnosis of prostate cancer. These estimates will use data from the electronic health record including age, race/ethnicity, prior medical claims, disease severity, exposure, health habits, pharmacy, and laboratory data for military beneficiaries receiving care in the VHA. (Aim 2) These estimates will include efforts to use machine learning approaches to generate the best-fitting model of overall survival. Finally, we will estimate the overdiagnosis and overtreatment of prostate cancer in the VHA using the general and personalized life expectancy estimates. (Aim 3) FINDINGS/RESULTS: In the first year, we have made steady progress in several areas. First, we have examined data quality and compared data sources for some important prostate cancer specific variables (e.g., Prostate Specific Antigen). Our group has now shown that there can be errors in the PSA reported by the cancer registry, and that obtaining the PSA directly from the electronic health record is preferable. We have published a study identifying the rate and type of PSA mismatches (Mittikanti et al, JCO, 2016), and recently published a study how cohorts constructed using cancer registry data differ from those using electronic health record data (Guo et al, J Urol, 2018). Second, we have begun exploring the ethical implications of life expectancy estimation. This work grew directly from comments received during review of the grant. We have now published an editorial exploring the ethical pitfalls of life expectancy estimation in this population (Leppert et al, J Urol, 2018). Last, we have begun to construct cohorts of prostate cancer patients receiving care in the VHA in order to develop population-wide life expectancy estimates. These simple life tables are nearing completion. Moreover, we are working to develop comorbidity-adjusted life expectancy estimates. We have applied established general measures of comorbidity (e.g., Charlson), as well as the Prostate Cancer Comorbidity Index (manuscript in preparation). IMPACT: These efforts have the potential to deliver higher quality prostate cancer care, by treating patients most likely to benefit, and while avoiding futile treatment and minimizing treatment-related side effects. We also hope to shape the conversation about the ethical considerations when applying life expectancy estimation to prostate cancer screening and treatment. External Links for this ProjectNIH ReporterGrant Number: I01HX002126-01A1Link: https://reporter.nih.gov/project-details/9192674 Dimensions for VA![]() Learn more about Dimensions for VA. VA staff not currently on the VA network can access Dimensions by registering for an account using their VA email address. Search Dimensions for this project PUBLICATIONS:Journal Articles
DRA:
Cancer
DRE: Treatment - Observational, Prognosis, TRL - Applied/Translational Keywords: Cancer, Guideline Development and Implementation, Predictive Modeling, Research Measure Development, Statistical Methods, Surveillance MeSH Terms: none |