Were these women irrational when they rejected adjuvant breast cancer treatment?

Thirty percent of surveyed women with breast cancer reported skipping additional treatment following surgery. 

The implication: these women were irrational.

Although the research paper was released in November, these women were diagnosed with cancer a decade ago, 2005-2007, and the data were collected almost a decade ago, 2007-2009.

Perhaps the most egregious part of this research is that it treats “breast cancer” as a singular disease rather than the extremely heterogenous entity that it is. And how relevant are the findings to today’s patient given the age of the data?

This research crossed my Facebook feed earlier in the month. The report lamented that one-in-three women did not follow “at least one treatment recommendation.” The authors – both of the academic study and reporters – attributed these decisions to “distrust in the healthcare system.”

Who ignored physician recommendations?

Survey participants (limited to breast cancer patients from Florida and Pennsylvania) who did not follow all breast cancer adjuvant treatment recommendations were in the top third of respondents who did not trust the healthcare system, according to survey results. They were more likely

  • to be older
  • to have been diagnosed at Stage 1
  • to live in Florida than in Pennsylvania

We know that breast cancer diagnosed in younger women may be more aggressive, so that might explain why it was “older” women who did not follow physician recommendations.

We know that over-treatment at Stage 1, especially when it comes to chemotherapy, is a reality (see below for details). What we don’t know is if patients were making a rational decision to reject recommendations of chemo, for example.

However, researchers note that “distrust was not significantly associated with [chemo, radiation and hormone therapy] adjuvant treatments.” And even though all three treatment types had a “relatively small amount of discordance” it was this “relatively small” set of behavior that is the centerpiece of the research report.

Although most respondents had insurance, fewer Floridians (93%) had insurance compared with Pennsylvanians (97%, a statistically significant difference). Researchers conceded that a lack of insurance “may have increased the likelihood that Florida patients” would not follow physician recommendations. Given the cost of treatment, that seems as likely an explanation as “distrust.” 

Relevance to today’s breast cancer patient

The study’s 2005-2007 cohort does not reflect the demographics of today’s newly diagnosed breast cancer cohort, either in age, race or stage.

  • All study participants were younger than 65 at diagnosis with an average age of 52 (standard deviation was 10); the median age at breast cancer diagnosis today is 62. (Why didn’t researchers report the median age of their cohort?)
  • Breast cancer incidence was lower in both Florida and Pennsylvania than the national average for both white and black women based on the 2007-2008 American Cancer Society report.
  • Approximately half (49.6%) of the study participants had Stage 1 cancer at diagnosis; 44.6% were diagnosed at Stage 2; and 5.9% at Stage 3. Today’s distribution: 65% is localized (which may be Stage 1 or 2 depending on tumor size); 33% is regional or distant (Stage 2, 3, or 4); and Stage 4 is about 10%. A small percentage of case in the database are not staged.

Assumptions about adjuvant treatment

Researchers acknowledge that they are assuming “that greater cancer treatment discordance would lead to worse patient outcomes.” Although researchers reference a separate model that “suggested a 40% increased odds of cancer recurrence for those who reported treatment discordance” in the narrative, they do not provide a citation for this model.

However, even in 2001, researchers were highlighting the challenges in thinking of adjuvant treatment as a one-size-fits-all equation for breast cancer:

To answer [the question, who benefits from adjuvant treatment], the prognosis of the patient without further treatment must be assessed… An increased risk of relapse and death is associated with greater lymph node involvement (not Stage 1), larger tumor size (not Stage 1), and more aggressive histologic features (independent of stage). ER-positive tumors, in contrast, are associated with a 10% lower risk of relapse at 5 years compared with ER-negative tumors… a woman with an ER-positive well-differentiated node-negative tumor smaller than 1 cm has a mortality of only 10% over 10 years (parenthetical comments added).

Researchers also assume that patients who start treatment but later discontinue it are doing so based on system distrust. It is equally – perhaps more – likely that discontinuing treatment, especially chemo and hormone therapy, is a function of reduced quality of life. Researchers do not replicate the survey questions in an appendix, so we don’t know what they asked the subjects. 

These researchers have treated “breast cancer” as a singular disease rather than an extremely heterogenous entity. This is the achilles heel of this research, for without a discussion of the type of breast cancer, it is impossible to “pin the blame” for rejecting recommendations on “mistrust” unless “mistrust” also means “recommendations disconnected from current research and risk.”

Treatment recommendations

Researchers do not identify what percentage of breast cancer patients were prescribed adjuvant treatment, whether radiation, chemotherapy or hormonal therapy. We also do not know which women rejected all three treatments, two treatments, or only one treatment.


According to the researchers, “11% were discordant with chemotherapy.”

The world of cancer in general, and breast cancer in particular, has changed radically since 2007.  Even in 2007, doctors and researchers were publicly talking about the challenges of chemotherapy over-treatment:

If you take 100 postmenopausal women with a small tumor that has estrogen receptors – most do – and no sign the disease has spread to the lymph nodes, probably 12 to 15 of the women need chemotherapy to reduce recurrence, Dr. Samuel says.

To save those patients’ lives, all 100 have to be treated with chemotherapy, because no definitive test indicates who really needs it, says Dr. Samuel. Yet the vast majority will do well with lumpectomy, radiation and hormone therapy that keeps cells from being refueled by estrogen.

And because of that 12-15 percent, chemo was (almost) always recommended. Would a woman be “wrong” to choose to forego chemo?

Even in the early 2000s, researchers were reporting that the most common form of breast cancer, estrogen-positive cancer, does not respond well to chemotherapy (on average). Also, researchers had uncovered that the benefits of chemo seemed to accrue more to women who were younger than 50.

Yet even today, a decade later, chemotherapy is a recommended course of treatment for lobular invasive cancer, one that research shows over-and-over as unresponsive to chemo.

If the “discordant” women in this study were over 50, menopausal and had estrogen-receptor-positive breast cancer, would she have been “negligent” for choosing not to pursue chemo? I don’t think so.


According to the researchers, “10% were discordant with radiation.”

A 2004 meta-analysis of 15 randomized clinical trials noted that omitting radiation treatment increases the risk of breast cancer recurrence but has only a small increase in mortality risk.

Hormone therapy

According to the researchers, “18% were discordant with hormone therapy.”

First, hormone therapy is a recommended course of treatment for breast cancer that is estrogen-receptor-positive. The treatment for pre- and post-menopausal women is different.

Although the average age in this study was 52, we do not know the menopausal distribution. The lack of this data continues the researcher’s treatment of breast cancer as a singular disease, both in cause and treatment.

We know that some breast cancer tumors that respond to aromatase inhibitors (such as invasive lobular cancer) but not to tamoxifen. However, tamoxifen is the recommended course of treatment for pre-menopausal women.

Moreover, only about half of estrogen-receptor-positive breast cancer responds to anti-estrogen treatment.

However, hormone therapy has its own set of side-effects, as this 2004 analysis details. BreastCancer.org summarizes some of the most common side effects of today’s hormonal therapies (a much more extensive set of options than in 2005-2007).

The Florida-Pennsylvania survey was administered up to four years after diagnosis; at that time, hormonal therapy was recommended as a five-year course of treatment. We don’t know what percentage of women were prescribed hormonal therapy nor at what date they started/stopped/ignored the recommendation, additional study weaknesses.

In 2011, researchers from the Northwestern University Feinberg School of Medicine reported that 10% of patients prescribed aromatase inhibitors had abandoned treatment within 24 months; 36% dropped out of treatment within four years “because the side effects linked to them were unbearable.”

A Brazilian study of  5,861, released in 2014, found that only 79.0% of patients were still undergoing hormonal treatment at the end of the first year; only 31.0%, at five years of treatment. Why did these women, diagnosed from 2002 to 2008, stop taking tamoxifen and/or the aromatase inhibitors anastrozole or letrozole?

The most consistent results indicate an association between the worst rates of adherence and persistence with extremes of age (the oldest and youngest patients), increased budgetary costs, monitoring by general practitioner (versus by an oncologist), treatment in which the medication changes (switching to the aromatase inhibitor after treatment with tamoxifen) and side effects of the therapy.

Notice that nothing here has to do with “system trust” and much has to do with quality of life.


Researchers reported that most survey participants were White (69%) and Non-Hispanic (94%); about one-quarter were Black (27%).  The mean age of the 2,754 respondents was 52. Most survey respondents were married (66%), employed (58%), and had health insurance (95%).

The median age at diagnosis today, according to the SEER database, is 62 years of age. 

The American Cancer Society report from 2007-2008 noted the different breast cancer incidence rates among population sub-groups:

breast cancer data 2007-2008
American Cancer Society chart

Breast cancer incidence rate (per 100,000) among white women versus African-American women suggest both states have lower-than average rates for both both white and African-American women. Pennsylvania comes considerably closer to the national average for African-American women than Florida (2000-2004, reported in 2007):

  • US: 132.5 versus 118.3
  • Florida:  122.0 versus 101.7
  • Pennsylvania: 127.1 versus 117.3


Researchers mailed surveys to women listed in Pennsylvania and Florida cancer registries between January 1, 2007 and December 31, 2009.  Why these states? The researchers state:

As a study on racial differences in cancer treatment, these states were ideal due to the size, racial diversity of their populations, and the ability to recruit patients directly from cancer registry files. Study inclusion criteria included: biological female; localized invasive breast cancer; lobular or ductal histology; age under 65 at the time of diagnosis; lived in one of the two states (i.e., PA or FL); and diagnosed between January 1, 2005 and December 31, 2007 (emphasis added).

Lets examine the demographics claim, using 2010 data from the U.S. Census:

census data
US Census Quick Facts, 2010

Both states are “whiter” than the national average, and Florida also has a larger percentage of residents with Hispanic or Latino origins. Neither are close to being representative of residents with Native American/Alaskan or Asian heritage, but these groups tend to have lower breast cancer incidence.

Florida over-represents residents of Black or African American heritage and Foreign born persons while Pennsylvania under-represents both groups. When the two states data are averaged, perhaps the gross population comes close to representing the national average in these two categories.

These data do not show us if the population of women under age 65 is mirrored in the two states. We know that the population pool skews much older in Florida, with 17.3% older than 65 compared to a national average of 13.0%, but we don’t know what the implications of that might be for the cohort under age 65.


From the outset, this research treats “breast cancer” as a singular disease rather than an extremely heterogenous entity. Moreover, it tacitly links correlation with causation in its discussion of patients who “mistrust” the health care system with rates of rejection of breast cancer adjuvant treatment recommendations.

And then the small print … does this mean that the article is not peer-reviewed?

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.



Healthcare System Distrust, Physician Trust, and Patient Discordance with Adjuvant Breast Cancer Treatment Recommendations. DOI: 10.1158/1055-9965.EPI-17-0479 (full paper, pdf)

Image source: HealthLine


Leave a Reply