In this large, population-based study in two metropolitan areas, we found high rates of RT receipt among patients undergoing BCS but lower rates among patients undergoing mastectomy. For patients with strong indications for treatment on the basis of guidelines adopted before the study period, 95.4% of patients undergoing BCS received radiotherapy compared with only 77.6% of patients undergoing mastectomy. Despite deliberate oversampling for minority patients, we did not observe significant differences by ethnicity or education, although we did observe a modest income gradient, as lowest-income patients were somewhat less likely to receive treatment. Patient preferences and provider interactions were strongly associated with likelihood of RT receipt across surgery types and indication strengths. Most notably, among patients who received mastectomy and had strong indications for RT, only 63.2% received RT when desire to avoid radiation was high and reported surgeon participation in the decision was low, compared with greater than 90% when surgeon participation was high. Most patients with strong indications for RT after mastectomy who failed to receive treatment reported that their doctor either did not discuss RT or said it was not needed.
In recent years, attention has been focused on concerns about underutilization of adjuvant RT after BCS. RT receipt after BCS has even been used as a quality indicator.32
The high rate of RT we observed after BCS among patients with strong indications is encouraging, especially because this constitutes the majority of patients with breast cancer who have strong indications for RT. The rate of post-BCS RT in this population-based study was similar to that recently reported within National Comprehensive Cancer Network institutions.11
These results differed, however, from other population-based studies of RT utilization. Not only older2–4
but also recent studies5,6
that used SEER data alone have suggested substantial underutilization of RT after BCS. For example, a recent analysis of SEER data suggested that rates of RT after BCS decreased from 79.4% in 1988 to 66.4% in 2004.6
However, the validity of radiation treatment information in registry data is limited.8
For example, in one study comparing SEER registry data to medical record review, RT use was accurately captured by registry data in only 72% of cases.7
Some researchers have responded to concerns about the limitations of registry data by focusing on the linked SEER-Medicare data set,33,34
but these studies generally include women younger than 65 years old and so have limited generalizability.13,35–39
The few studies that have obtained more complete data suggested a higher rate of RT receipt after BCS than registry studies (84% in a comprehensive chart review study of women treated in New York City in 1999 to 200010
and 86% in a study surveying hospital cancer registrars about patients diagnosed in 19949
), but these estimates are now dated. Studies that are dated or that rely on incomplete data cannot provide information to evaluate whether quality initiatives have been successful, nor are they helpful in guiding ongoing quality improvement efforts.
Less attention has been paid to the receipt of radiation after mastectomy for patients who need it. The moderate rate of RT after mastectomy observed even in patients with strong indications suggests that a substantial minority of patients with breast cancer remain undertreated. In our study, we defined a subgroup of patients for whom RT was strongly indicated on the basis of clinical guidelines.25
In these patients, the absolute risk of locoregional recurrence exceeds 30% in the absence of RT and is reduced by two thirds with RT, yielding a survival benefit. Yet, even among patients with strong indications for treatment, those undergoing mastectomy were substantially less likely to receive RT than those undergoing BCS, despite similar expected benefit. This gap was particularly pronounced when provider participation in the radiation decision was reported to be low and patient desire to avoid radiation was high.
Among patients with weaker indications, we also observed higher RT receipt among those undergoing BCS than those undergoing mastectomy. Our finding that receipt of RT for patients with weaker indications was lower in LA than Detroit underscores that there may be controversy among clinicians for this group (in contrast to the group with strong indications, who had similar rates of RT receipt regardless of site).40
In the Cancer and Leukemia Group B (CALGB) trial that defined a group of patients who might consider RT omission after BCS (ie, women age 70 years and older with stage I, estrogen receptor–positive disease), the 5-year risk of local recurrence was only 4% after BCS and tamoxifen alone.22
In contrast, in the largest American series of patients undergoing mastectomy with involvement of one to three axillary nodes (who constitute the majority of patients with mastectomy with weaker indications), the risk of locoregional failure in the absence of RT was 13%.29,30
Thus, although legitimate clinical uncertainty influences decision-making in the group of patients with weaker indications, our finding of substantially higher rates of RT among patients undergoing BCS with weaker indications—rates that exceed even the rates of RT received by patients who received mastectomy with strong indications—seems remarkable, especially because RT for patients undergoing BCS with weaker indications is unlikely to yield a survival benefit, whereas RT after mastectomy yields a survival advantage that could be as high as 10% in patients with strong indications.
In the population in which we observed substantial underutilization—patients with mastectomy—provider involvement was an important correlate of RT receipt. Even patients who expressed preferences to avoid RT were highly likely to receive it if their surgeons were highly involved in the decision process. Patients who did not receive radiation after mastectomy were most likely to report that their providers had either failed to discuss RT or had failed to recommend it. Because most patients with lymph node involvement receive adjuvant chemotherapy before RT, discussions between patients and surgeons may take place many months before the delivery of RT. Therefore, educational efforts targeting both surgeons and medical oncologists may be important in improving rates of RT receipt in this population, as may be the development of tailored decision aids that encourage communication between patients and providers.41,42
This study has several strengths, including a large, diverse patient sample and access to both clinical data and patient reports of treatment receipt, individual socioeconomic characteristics, patient preferences, and provider interactions. Nevertheless, several limitations merit comment. First, the location of the study in the greater Detroit and LA metropolitan areas may limit the generalizability of the findings, particularly to more rural areas. Second, our measures were drawn from patient self-report, which may be prone to bias. Although there is no clear gold standard for comparison, our measure of self-reported RT receipt has strong face validity, because patients surveyed months after diagnosis should accurately recall whether or not they received radiation treatment. Furthermore, self-reported RT receipt was highly and logically correlated with clinical and treatment factors that direct radiation treatment recommendations. Patient recall of communication issues may be more prone to bias, however. Third, although the sample size was adequate to detect substantial differences, power to detect modest differences was more limited. Finally, although the response rate to this survey was high, it is possible that selection bias may also have influenced our results.
Our findings have important implications for physician behavior and clinical policy. The results suggest that we have largely achieved success in the appropriate use of RT after BCS in metropolitan areas like those we studied, but more attention needs to be paid to the use of RT after mastectomy. We found that surgeon participation in the RT decision was a powerful correlate of use. This underscores the need to focus physician attention on potential gaps in treatment delivery. We also found that patient concerns about radiation were negatively associated with RT use. This is important because patients with these concerns may choose mastectomy with the intention of avoiding radiation, resulting in a higher prevalence of these concerns in patients undergoing mastectomy than undergoing BCS.43
Thus, it is important to consider these concerns when informing these patients about RT. Our findings suggest that initiatives to ensure that surgeons are informed about the role of RT after mastectomy, to encourage provider participation in the postmastectomy RT decision, and to improve patient education in this setting, would further optimize care for patients with breast cancer.