Variance in Outcomes by Physician Providers for Patients with Non-Hodgkin's Lymphoma.

Abstract

We previously documented an improvement in survival for patients diagnosed with non-Hodgkin's lymphoma at our institution during the years 1998–2003 compared to those diagnosed during 1990–1997. [Blood 2005; 106:242b] Because patients were being treated by several community-based practitioners, we examined the variance in lymphoma-specific survival by physician provider. The Hoag Cancer Registry data base was used to identify patients with a diagnosis of lymphoma. Observed and relative 5-year survival rates were calculated using a computer software program designed specifically for this purpose (Electronic Registry Systems, Inc., Cincinnati, OH). Relative survival was used to estimate lymphoma specific survival. The 802 patients from all practitioners had a median age of 64 years and a relative 5-year survival of 66% and observed 5-year survival of 54%. Data was reviewed for the 6 individual physicians who had managed at least 40 lymphoma patients during 1990–2003 (range 48–189), who accounted for 77% of all lymphoma patients. The observed and relative 5-year survival by these practitioners were: A 62% & 78%, B 67% & 77%, C 54% & 69%, D 51% & 65%, E 58% & 64%, and F 45% & 54%. Patient populations were dissimilar: practitioners D & E had the highest numbers of patients, B & F had the youngest patients, A & C the oldest, A the highest proportion with indolent lymphoma, and B the highest proportion with stage I disease. Combining data for physicians ABC (highest survival rates) (n=215) vs DEF (n=404) revealed a superior survival rate for ABC (p=.020), and a difference by whether systemic chemotherapy was initiated at the time of initial diagnosis (p=.016), as opposed to no therapy or local therapy only. This study demonstrated a 24 percentage point difference in relative 5-year survival rates among practitioners. There were differences in prognostic variables for specific populations of patients managed by individual physicians, but such differences were not seen for ABC vs DEF. Collectively, the three physicians whose patients had the best survival were more likely to use systemic therapy within four months of initial diagnosis rather than local therapy or no treatment. It is important to understand the impact of prognostic variables and physician-specific treatment differences on the variance of outcomes among different practitioners.