Improving outcomes in breast cancer: where should we target our efforts?

Abstract

Rural-urban differences in health outcomes, including breast cancer, in the US have been studied for decades, but often with inconsistent findings. Possible reasons include methodological differences, lack of prospective investigations, small number of studies overall, and the tendency to measure rurality as a simple patient-level predictor variable. Studies have tended to assume that the same racial/ethnic cancer disparities found in the general population exist in rural regions, but this conclusion may not always be warranted. Needed are better definitions of rurality; the capability to define important predictor variables such as race, ethnicity, education, and income with greater precision than at present; and data revealing the patient’s own perspective regarding care decisions. Future studies should examine whether the impact of rurality status on outcomes varies with geographic location by including the appropriate interaction terms in the outcome prediction models, as well as patient-reported reasons that might explain the outcomes observed.

Keywords::

• breast cancer
• cancer
• disparities
• outcomes
• rural health

Defining the problem

Rural–urban differences in health outcomes in the USA have been studied for >40 years, often with inconsistent findings [1]. When studying rural–urban differences in cancer outcomes, investigators generally focus on findings of ‘disparities’ in rural settings. The National Cancer Institute (NCI) defines cancer disparities as ‘adverse differences in cancer incidence, cancer prevalence, cancer death, cancer survivorship, and burden of cancer or related health conditions that exist among specific population groups in the United States.’ In addition to disparities based on age, socioeconomic status, race, ethnicity and gender, the NCI definition includes individuals ‘who are … medically underserved (have limited or no access to effective health care) – regardless of ethnic and racial background [2].’ The latter reference to ‘medically underserved’ often takes into account rural regions.

A review by Meilleur et al. [3] suggests that the varying results from rural–urban comparisons may reflect methodological differences across studies. A key methodological issue may be the inclination in many studies to measure rurality as a simple patient-level predictor variable, without accounting for the complexities of the geographical areas evaluated. For example, the same US Census definition of ‘rural’ may apply to areas of the Far West, Mississippi Delta and Appalachia, yet considerable diversity exists not only across but also within each of these regions. Other definitions of geography may be based on the classification using the US Department of Agriculture’s Rural–Urban Continuum Area (RUCA) code [4]. Studies using different rurality definitions may evaluate the same population from distinct perspectives, thus arriving at contradictory conclusions.

Whatever definition of rurality a study selects, sorting out the impact on outcomes of rural/urban status relative to other important factors, including insurance status, poverty, race, ethnicity and health literacy, remains a challenge. In a study of both all-cause and cause-specific mortality comparing rural–urban differences, poverty was found to account for only part of these differences, but the disparities between urban and rural mortality rates were greatest in geographical areas with the highest poverty rates [5]. Much of the research comparing urban and rural cancer outcomes has been retrospective and observational in nature, using available demographic and clinical data, including administrative data sets [6]. Since most cancer clinical trials are conducted at urban-based academic centers, where expectations of including a balance of urban and rural patients are uncommon, rural patients are frequently underrepresented in many prospective trials. Clinical trial results are rarely, if ever, analyzed based on the geographical area of residence, and rurality is not a required demographic in reporting trial accrual statistics [7]. Perhaps as an indirect consequence due to lack of data, rural health care may be assumed to be delivered at a lower standard, thought to be less evidence based and hypothesized to be of poorer quality compared with more urban settings [8].

What outcomes can or should be compared?

While the NCI and other agencies have historically investigated urban–rural differences in cancer incidence, prevalence, mortality and survival, evaluating varying patterns of care based on geographical residence of distinct populations is becoming increasingly emphasized. In breast cancer, investigators have distinguished between urban and rural settings in examining rates of screening mammography, receipt of first course of therapy and type of treatment, for example, breast-conserving surgery (BCS) versus mastectomy. Screening may be seen both as a marker for access to care and as a measure of cancer care quality [9]. A national cross-sectional study examining rates of screening mammography using the Behavioral Risk Factor Surveillance System, a population-based telephone survey, looked at the receipt of mammography by rural residence and sociodemographic factors including race [10]. In unadjusted analyses, white women from both smaller and larger rural areas reported fewer mammograms than their urban counterparts. Yet, African American (AA) women from urban and large rural areas actually received mammograms – indeed, all screening services examined – more often compared with white urban women. When these analyses were adjusted for provider availability and other factors, AA in both urban and rural settings reported higher rates of screening than whites [10]. These findings were deemed ‘surprising’ by the authors, and indeed, race/ethnicity is often difficult to separate from other aspects of geographical settings. The disparities literature sometimes assumes that similar racial/ethnic differences in cancer screening rates found in the general population exist in rural regions, but data to support this conclusion are limited.

To this point, in a population-based study of breast cancer treatment in the largely rural region of southwest Georgia, Lipscomb et al. [11] examined racial differences in receipt and completion of adjuvant chemotherapy. In multivariable regression analyses, they found no significant AA–white difference in initiation of recommended therapy, but a significant AA–white difference in completion of therapy – with blacks staying the course at a higher rate than whites (with odds ratios >2).

Similarly, using data from the NCI’s Surveillance, Epidemiology and End Results (SEER) registry-based data system and focusing also on the state of Georgia, Markossian and colleagues [12] evaluated disparities among women diagnosed with late stage breast cancer. Rural residency was not found to be significantly associated with risk of death from breast cancer. Late-stage diagnosis of breast cancer was also neither significantly associated with rural residency, nor was receipt of surgery. Urban women were more likely to undergo BCS (54.9%), while slightly more rural women chose mastectomy (53.1%). This finding was statistically significantly different (p < 0.0001), but arguably not clinically meaningful. Many factors, including personal preference, are integrated into treatment decision-making, and BCS (with radiation) and mastectomy are clinically equivalent choices. However, women residing in rural counties did receive significantly less radiation compared with urban women (p < 0.0001) although the appropriateness of receipt of radiation was not fully explored based on stage, nodal status, tumor size or other factors.

Interestingly, very recent data [13] from a randomized controlled clinical trial indicated that many women, particularly those who are older, with smaller (<3 cm) tumors, node negative and hormone receptor positive, will do exceedingly well without adjuvant radiation therapy post-BCS if they are receiving endocrine therapy. The trial also demonstrated the lower risk of radiation-associated adverse events and greater convenience of not having to undergo daily radiation treatments. So while the SEER-based study found that rural breast cancer patients received adjuvant radiation at lower rates than urban patients, the lack of impact on cancer-related mortality may be in alignment with newer data about radiation treatment effectiveness as well as an increasing understanding that more is not always better in cancer care.

Where should we target our efforts to improve rural outcomes in breast cancer?

The recent review of Meilleur et al. [3] that focused on cancer outcomes in rural areas highlights the limitations of studies conducted to date: variable methodologies, gaps in the literature based on small numbers of studies overall and a lack of sufficiently well-designed, prospective investigations. Urgently needed are better definitions of rurality and the capability to define such important predictor variables as race, ethnicity, age, education, income and insurance status with greater detail and precision than is currently typically possible when working with data from census tract or RUCA systems. Both reporting and analyzing rural status would be facilitated with the addition of geographical residency to the ‘Targeted/Planned Enrollment Table’ forms, which include gender, race and ethnicity of subjects in clinical research studies, as required by the NIH [14]. In addition, improved granularity related to the exact geographical setting being studied is needed.

A final point is that none of the studies cited as examples benefited from one additional source of data that could illuminate the causal mechanisms leading to the outcomes observed: the patient’s own ‘voice’, as might be encoded through surveys or other reporting vehicles. The findings from Behavioral Risk Factor Surveillance System or SEER data might be interpreted quite differently if more patient-level data were available to explain why rural AA received more screening services. Access to patient reports of superior social networks or community-based support could well explain the results from Lipscomb et al. [11], indicating greater completion of adjuvant chemotherapy among rural AA women with breast cancer. If the rural women from the SEER-based study who received a mastectomy rather than BCS were questioned as to their treatment decision making, this could shed light on whether they actually preferred a mastectomy for multiple appropriate reasons, or if they would have liked to have pursued BCS but distance from a radiation center or other barriers to care prevented them from pursuing their preferred treatment choice. Lack of patient-level data in all these instances inhibits the ability to draw accurate conclusions.

Thus, to sharpen our understanding of the impact of rurality on breast cancer treatment and outcomes, well-designed studies are needed that include adequately defined predictor variables, including rurality separate from socio-economic status, race, and so on, as well as additional prospective studies that permit collection of patient-reported data. The latter should include patient preferences regarding decision-making styles, patient-reported outcomes (including health-related quality of life and symptom burden) and patient perceptions of barriers and facilitators to care. Such data could contribute greatly to address the gaps in our understanding of rural breast cancer care, its quality and meaningful patient outcomes.

Financial & competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.