Prostate cancer is a pervasive disease among elderly men but infrequently fatal. Clinically insignificant (indolent) cancer is common and over-diagnosis and treatment are concerns, particularly since the advent of prostate specific antigen (PSA) screening. About one-fifth of men are diagnosed with prostate cancer during their lifetime and there is a 3% lifetime risk of death from the disease. Autopsy studies have revealed that 60-70% of older men have detectable cancer within the prostate. Given the complications and side effects of treatment (including impotence and incontinence), a principal clinical challenge is deciding whether treatment will be beneficial and, if so, which treatment modality is to be preferred. Unfortunately there is a paucity of comparative effectiveness research to guide patients and providers. Evidence-based recommendations for the precise timing of intervention in patients with long life expectancy do not exist. Life-altering decisions about prostate cancer treatment are made in a relatively low-information environment.
Our aim in this one-year pilot study was to gain familiarity with two key aspects of data required for the full study. The two objectives of the pilot study are: Objective P1: Extract and validate Gleason scores from 2000-2004 VA Central Cancer Registry (VACCR) and other VA medical record data. Objective P2: For years 2000-2004, demonstrate geographic practice pattern variation in probability of treatment type as a function of medical center effects.
The study was a retrospective analysis of secondary data. Methods were quantitative. VACCR Gleason score information is available in 2004 as a numerical variable and in prior years in a pathology text field. We extracted Gleason score from the pre-2004 text data and compared distributions available from the literature. We estimated models of prostate cancer treatment across the main treatment modalities: radical prostatectomy (RP), external beam radiation therapy (EBRT), brachytherapy (BT). Independent variables included age, co-morbidities, and VAMC fixed effects. We documented geographic variation in type of treatment.
We were able to extract Gleason scores from pre-2004 VACCR text data. We have found additional Gleason score information in the VA's Text Integration Utilities (TIU). Gleason scores appear consistent with expectations. We found considerable practice prostate cancer treatment pattern variation. For example, three VISNs provide one of the more invasive treatments (RP, EBRT, BT) for at least 80% of patients. Three do so for fewer than 40% of patients.
Prostate cancer is pervasive but infrequently deadly. Clinically insignificant disease is common and over-diagnosis/over-treatment is a concern. Thus, a principal clinical challenge is presented by screening and diagnostic tests that are not conclusive about whether treatment will be beneficial. Because of the lifestyle implications of side effects, choice of treatment is difficult for many patients. Unfortunately there is a paucity of comparative effectiveness research to guide patients and providers. Consequently, life-altering decisions about prostate cancer treatment are made in a relatively low-information environment.
Comparative effectiveness research is a high priority for the Department of Veterans Affairs. The full study this pilot will support will help implement VA's commitment by investigating the comparative effectiveness of major modes of treatment for localized prostate cancer. Though prostate cancer is common, it is infrequently deadly. All treatment options are associated with significant side effects that relate to quality of life. The results of the full study will be of interest to VA clinicians who need to assess the risks and benefits of prostate cancer treatment for each patient. It will provide valuable context in general and useful guidance for targeting and timing treatment. There will be benefits for VA researchers as well because this project would be among the first VA comparative effectiveness study to use local practice patterns as instruments in an instrumental variables analysis of the effects of medication choice on outcomes.
External Links for this Project
Grant Number: I01HX000512-01A1
- Frakt AB. The Media’s Role in Health Policy. Paper presented at: AcademyHealth National Health Policy Conference; 2012 Feb 14; Washington, DC. [view]