- Research Article
3- 10.1093/jnci/djq511
Random Physician Effect and Comparative Effectiveness of Treatment for Ductal Carcinoma In Situ
- Jan 03, 2011
- JNCI Journal of the National Cancer Institute
- B A Virnig + 1 more +1
In this issue of the Journal, Dick et al. (1) report on the comparative effectiveness of surgical and radiation therapy (RT) for ductal carcinoma in situ (DCIS). The authors found that rates of ipsilateral recurrence (both invasive and in situ) are lower for women undergoing mastectomy, highest for women undergoing breast-conserving surgery (BCS) without RT, and intermediate for women receiving BCS with RT. The study provides the same rank order for treatment effectiveness as other observational studies [see (2) for a summary] and does an excellent job of controlling for clinical factors such as multifocal disease and histology that should necessarily influence treatment choice. The importance of these patterns must be emphasized because the tumor factors associated with highest risk of recurrence were disproportionately concentrated in the mastectomy group (1). The study is innovative because it also quantifies the contribu tion of physician-level variation in treatment choices and outcomes. Although many studies have shown a strong association between close or positive margins and cancer outcomes [see (2) for summary], the work of Dick et al. (1) suggests that some physicians are more likely to have patients with positive margins than others and that there is unexplained variation among surgeons in choice of treatment. For example, although randomized trials have demonstrated that RT decreases ipsilateral breast tumor recurrence rates after BCS (3,4), there is growing controversy about whether RT is beneficial for all women (5), and current National Comprehensive Cancer Network guidelines suggest that RT can be omitted after BCS for some patients deemed to be at low risk of recurrence (6). Studies also show wide variation in the accepted width of negative surgical margins for DCIS (7) even though attainment of negative surgical margins (no ink on tumor) is considered to be the minimum acceptable margin for patients with DCIS (6). Dick et al. (1) found that after accounting for the role of treatments, patient and tumor characteristics, and positive margins, between 15% and 30% of the variation in outcomes, that is, DCIS recurrence or invasive breast cancer, could be uniquely attributed to the choice of physician. This estimate is critical for policy implementation but most likely represents an underestimate of the impact of physicians because it removes the effects of physician choice of treatments and acceptance of positive margin status. Thus, although some physicians may be more or less likely to re-excise tumors with positive margins, the estimate of physician effect reflects the influence remaining after taking margin status and treatment into account. The literature addressing the impact of physicians on treatment choice and outcomes has thus far failed to completely merge with the growing field of comparative effectiveness research. One part of the literature aims to use scientific evidence to identify the best treatment or to quantify the advantages and disadvantages of treatment strategies, whereas another part documents unexplained variability that is associated with provider choice. In their analysis, Dick et al. (1) bring these two approaches together. In doing so, their work underscores the challenge that patients face—how exactly does one choose providers? Dick et al. (1) report a statistically significant physician random effect. That is to say, their model as sumes that the benefit of treatment is the same regardless of treat ing physician. It also assumes that the benefit of negative margins is the same. With those assumptions in place, they found that the physician providing the treatment matters, even after taking into account all other risk factors. So what should women be advised to consider when choosing a provider? From this article (1) and the general DCIS literature, it seems that having some sense of whether the provider tolerates positive margins or not is important, as is some knowledge about whether she or he generally prefers BCS or mastectomy or recommends RT following BCS. But, how should women select a provider knowing that up to 35% of the variation in outcomes is based on their choice of physician but that there are no actionable characteristics that can be taken into account? One option for incorporating physician-level variability into actionable policy is to publish scores for all physicians performing breast cancer surgery in a particular area. With this approach, it would not matter why one physician had higher or lower recurrence rates or positive margin rates, the rates would simply be reported so that women could take this into account when selecting
Read more
