AUA 2017: Best of Health Services Research
There are certainly some strengths to using large, established datasets. The data is already collected, so there is minimal economic burden to the investigator to collect and categorize large amounts of data. The data may be of higher quality than what can be collected by a small research team studying a specific question. Indeed, most datasets are collected by large governmental or private entities that dedicate immense resources to the collection and categorization of the data. And finally, these datasets have considerable breadth, often with thousands of variables that can be analyzed.
Conversely, already-collected data can also be a pitfall. Because the study design and data collection are already complete, you are limited in the kind of investigative questions you can ask. Going back to add more data to these datasets is often cumbersome and prohibitive. Furthermore, the desired outcome (or variable you are studying) may not have been collected, and as such the datasets may lack the appropriate depth of information needed to study a particular question. Lastly, the use of large datasets requires advanced knowledge of methodology and statistics, which is not possible without certain kinds of professional support.
Several important papers were highlighted that evaluated patterns of care with regard to prostate cancer (PC). Tyson et al. showed huge variation in the use of active surveillance among national providers. Quantifying this variation helps us understand the challenges that need to be overcome to standardize and improve patient care. Gray et al. used NCDB data to show that the rate of observation in PC is increasing, but so is the use of radical prostatectomy and subsequent decline in use of radiation. Krishna et al. found that the rate of active surveillance in the US is lower among African American men than white men. Sadly, African American men were also less likely to receive appropriate surveillance follow-up after being placed on a “surveillance” regimen.
Important comparative studies of the year included Barocas et al. who published in JAMA data from SEER that was supplemented with data on patient-reported outcomes. They found significant differences in patient-reported outcomes between those who underwent surgery, radiation, and active surveillance. Studies such as this help inform clinicians and provide data for counseling patients. Dinh et al. published important data that patients on androgen deprivation (ADT) were more likely to have depression than those not on ADT.
In the health policy realm, Herrel et al. used medicare claims to show that patients treated within fully integrated insurance markets had better overall care along a multitude of quality metrics. Fedesa et al. found a decrease in PSA screening following USPSTF recommendations against routine PSA screening. However, since that time, PSA screening appears to have remained stable. Hu et al. demonstrated that the incidence of metastatic prostate cancer appears to be increasing. Unfortunately, there is lack of granularity in the dataset, but one wonders if this is also related to the USPSTF recommendations. Dr. Trinh highlighted work by his group at Harvard that demonstrated the trends in adoption of robotic prostatectomy compared to open prostatectomy. CMS actually cited this article in its decision to increase the RVUs generated by robotic prostatectomy after an initial period of devaluation.
HSR research can answer large questions and provide guidance for important policy decisions. It is an exciting time in the field to be involved in this space.
Presented By: Quoc-Dien Trinh, MD, Brigham and Women’s Hospital
Written By: Shreyas Joshi, MD, Fox Chase Cancer Center, Philadelphia, PA
Twitter: @ssjoshimd
at the 2017 AUA Annual Meeting - May 12 - 16, 2017 – Boston, Massachusetts, USA