br The call for more practice based research is
The call for more practice-based research is dependent on being able to measure care.16 Analysis of administra-tive datasets and large registries has provided much insight, but access is often limited and delays reduce actionability. Data validity of large administrative datasets
has, also, been questioned.26 We hoped the use of EHRs, which were promulgated in part to increase access to data would facilitate the evaluation of contemporary commu-nity care patterns. Ironically, while prostate cancer is one of the most common cancers among men, the diagnosis and billing coding systems (the World Health Organiza-tion’s International Statistical Classification of Diseases and Related Health Problems-10, and the American Medical Association’s Current Procedural Terminology codes) were not able to capture either the clinically criti-cal components of cancer staging, nor the selection of noninterventional therapies like AS. Accordingly, we found that manual chart abstraction was the only way to reliably capture clinical care and that the use of EHRs by all participating groups, served only to facilitate manual abstraction. We hope that continued improvement in data capture and exchange as well as efforts like the American Urological Association’s Quality Registry will continue to improve access to actionable data for front-line providers.27
There are several limitations to this Oxidopamine study. Neither patients nor sites were randomly selected. Practices self-selected out of an interest to better understand their practice patterns, and all newly diagnosed men from each practice were included, except as noted. Data col-lection was performed by members of each practice who were carefully trained but were not professional
abstractors. Hence, an independent and professional chart abstraction consultancy team was further employed. We were only able to externally validate data from 6 of the original 9 participants, however we have no reason to believe abstraction accuracy differed in these 3 practices and the concordance among the validated groups was very good. Agreement between the primary cohort and validation cohort was 99.2% for age, 77.5% for race, 89.2% for primary therapy choice, and 89.2% for risk group classification. The disagree-ment over primary therapy was primarily due to different interpretations of when a patient was lost to follow-up vs receiving outside therapy. The discordance in risk group classification was most commonly (62%) due to reclassification of disease risk group from very low risk to low risk (eg T1c vs T2a). We furthermore conducted the primary analyses on both the study cohort (pre-sented) and the validation cohort (not presented) with similar results found. While some missing data could have biased the results, only 203 (9.1%) patients were excluded for lack of information on primary therapy. In Figure 1, both outlying practices were among the 6 externally validated practices, as were 2 of the 3 outlier practices in Figure 2. Our analysis of surgery vs radio-therapy is limited by only 1 of the 9 sites in our cohort having no ownership stake in a radiotherapy center. We did not have access to other variable such as
Figure 1. Practice-specific estimated probabilities of active surveillance vs other therapy (surgery, radiotherapy, or other) for men with NCCN very low, low, and intermediate risk prostate cancer. (* = sites with significant variation in the baseline prob-ability of active surveillance compared to other therapy). (Color version available online).
Figure 2. Practice-specific estimated probabilities for utilization of surgery vs radiotherapy as the primary therapy for men with NCCN lower and intermediate risk prostate cancer. Site B was the only site with no ownership stake in radiotherapy. (* = sites with significant variation in the baseline probability of surgery compared to radiotherapy). (Color version available online).
comorbidity, which can influence treatment choice. Biopsy type was not measured, but given the timing of this study, we don’t believe the use of fusion biopsy would have been common or influenced the results.
This study demonstrates that AS was prominently utilized by urologists practicing in a large cohort of geographically diverse independent community urology group practices. Patterns of care delivery appear to be primarily consistent with clinical characteristics. Variation by practice is an opportunity for additional investigation and quality improvement efforts.
Acknowledgment. With gratitude for additional statistical advice from Heidi Reichert and Xiaohui Jiang (EpidStat Insti-tute).
12. Mitchell JM. Urologists' self-referral for pathology of biopsy speci-mens linked to increased use and lower prostate cancer detection. Health Aff (Millwood). 2012;31:741–749.
17. Aizer AA, Paly JJ, Zietman AL, et al. Multidisciplinary care and pur-suit of active surveillance in low-risk prostate cancer. J Clin Oncol. 2012;30:3071–3076.