Harsh Criticism of Study Showing Complementary Medicine for Cancer Lowers Survival

Roxanne Nelson, RN, BSN

January 31, 2019

Harsh criticism is being directed at a study that concluded cancer patients who used complementary medicine (CM) were more likely to refuse standard care, and as a result, faced a higher risk of death.

The use of CM in this study was associated with worse 5-year overall survival compared with patients who didn't use CM (82.2% vs 86.6%; P = .001) and a twofold greater risk of death. The study was published last year in JAMA Oncology, and reported at the time by Medscape Medical News.

The authors now are facing a firestorm of criticism, detailed in letters to the journal from five separate groups, and the criticism has been amplified on Twitter.

One of the critics, Lars Haakon Soraas, MSc, from the Norwegian University of Life Sciences and Nordic Cochrane Centre, Copenhagen, Denmark, says in a long thread on his Twitter feed that the five letters to the journal are "completely destroying the pseudoscience article."

He also hopes that the journal and/or authors retract the study "as its presence online will continue to misinform cancer patients worldwide."

Lead author of the study Skyler Johnson, MD, Yale School of Medicine, New Haven, Connecticut, argues on Twitter that these letters to the editor "do not 'completely destroy' our work, as any thoughtful, objective reader or scientist familiar with these processes can attest." He points out that concerns about the study have been previously addressed on Twitter, in the discussion section of the article, and again in their response to the letters to the journal.

"There is a HUGE difference between unproven anticancer therapies noted by docs (our study) and the most common complementary medicines self-reported by patients," he adds.

Study Details

For the study, Johnson and colleagues conducted a retrospective review of data on 1,901,815 patients from the National Cancer Database. The cohort included patients diagnosed with nonmetastatic breast, prostate, lung, or colorectal cancer.

From this large cohort, 258 patients (0.01%) were identified as using CM, loosely defined as "other-unproven: cancer treatments administered by nonmedical personnel" that were given in addition to at least one conventional modality, defined as surgery, radiotherapy, chemotherapy, and/or hormone therapy.

Patients who received CM were then matched with 1032 patients who had received conventional therapy.

However, the association of CM with increased mortality was largely based on treatment delay or refusal. After adjustment for treatment refusal and delay from diagnosis to treatment, CM was no longer significantly associated with the risk for death (HR, 1.39).

At the time, Medscape Medical News solicited comments from Donald Abrams, MD, professor of clinical medicine at the University of California San Francisco and a general oncologist at Zuckerberg San Francisco General Hospital. He noted that because the study is retrospective, it is not the "highest level of evidence" and has many limitations related to its design.

Controlling for delayed or refusal of treatment eliminated survival disparities between the groups, he pointed out. But importantly, "there is no information available on why patients refused therapy, and no information on what complementary therapies were used by the patients or how many types of therapies they may have used."

Some of these same limitations are now outlined in the letters published in JAMA Oncology, but there are also other concerns about methodological flaws.

In one letter, Ana Muñoz van den Eynde, PhD, Research Unit on Scientific Culture, CIEMAT, Madrid, Spain, writes that the study has three problems: "Two are genuinely methodological, whereas the third is more related with the ethics of research. The three problems severely undermine the study's scientific rigor."

The authors found that CM is an independent variable associated with a greater risk of death, but the "problem arises when, after also adjusting for treatment refusal and delay from diagnosis to treatment, CM no longer has a statistically significant association with the risk of death," she writes. "This means that the supposed association between CM use and survival is a spurious one, ie, a false correlation between two variables that is caused by a third variable."

She also highlights the small number of patients using CM identified in the database (only 258 patients [0.01%] of nearly 2 million). This is not a representative sample, she argues, as other studies have found that an estimated 48% to 88% of cancer patients report the use of complementary and alternative medicine as part of their therapy. 

The third problem was the lack of objectivity on the part of the authors. "Ignoring that the relationship between CM use and risk of death is spurious, they establish a causal association between the higher rate of conventional cancer therapy refusals in the CM group and a higher risk of death," writes Muñoz van den Eynde. 

She adds that when she looked at the "real figures of survival that were provided by the authors, instead of an estimate obtained by an equation lacking the relevant variables," there was no difference in survival rates between patients using and not using CM.

In another letter, Kevin Lee, MD, PhD, and Nathan Douthit, MD, both from Brookwood Baptist Health, Birmingham, Alabama, also highlight the small number of patients using CM and argue that this "conservatively represents a 1000-fold under-ascertainment."

In addition, "the current definition of CM is so broad, including prayer, diets, vitamins, and supplements, as to be ungeneralizable."

They also worry that studies such as this will be "used to further drive a wedge between traditional healthcare practitioners and patients interested in CM."

Some healthcare practitioners are already skeptical and even hostile towards the use of CM, and this study "could further reinforce the cognitive biases of these physicians, leading to decreased understanding and openness of communication with patients with cancer."

In another letter, Linda E. Carlson, PhD, RPsych, Cumming School of Medicine, University of Calgary, Alberta, Canada, and colleagues, point out that for a cancer patient to be coded as a CM user in the study the treating oncologist had to check a box titled, "Other-unproven: Cancer treatments administered by nonmedical personnel."

Thus, it is likely that only the "most extreme examples of treatment refusal or use of unrecommended approaches were included in the 258 people who were classified as CM users," they write. Not only is this an under-representation of the target population, it is also likely that half or more of those classified as "nonusers" were in fact using CM. This renders the study results invalid, they argue.

Carlson and colleagues also take issue with the study definition of CM, which is said to be "used in addition to conventional cancer therapy and may be used as a substitute for adjuvant therapies."

"Within the specialty of integrative oncology, there is a clear distinction between therapies that are 'complementary' — used along with conventional care — and those that are 'alternative' — used instead of conventional care," they write. The definition used in the study differs from the one used by the Society for Integrative Oncology, in which there is no suggestion "that patients should delay or refuse conventional cancer care; rather, patients are encouraged to seek care that draws on both conventional and complementary approaches," they note.

In yet another letter, Ozan Bahcivan, MSc, University of Barcelona, Spain, reiterates the point that CM was not clearly defined, but also highlights the fact that the finding on survival outcomes with CM were significant only for a subgroup of patients with breast cancer. This finding was of "borderline significance for patients with colorectal cancer" and was "not true for patients with prostate and lung cancer."

Also homing in on this issue in a letter is Soraas, who on Twitter highlighted the letters to the journal, as mentioned earlier. Writing with colleagues from the Nordic Cochrane Centre, they argue that the study was influenced by selection bias.

"In our opinion, a sample of only 258 patients is unlikely to be representative of the large population of patients with cancer using CM," they write. This small group was dominated by breast cancer patients (n = 186), so the conclusions for some of the other cancer types were based on "subgroups containing as few as 15 patients."

Authors' Response

Johnson and colleagues respond to the criticism of their study in a short letter, and have also responded on Twitter to some of the comments made by Soraas.

The authors point out that in their study CM had a very precise definition: "cancer treatment administered by nonmedical personnel." Hence, it is likely that CM defined as such includes treatments for cancer with a proposed, albeit unproven, biological mechanism, safety, and effectiveness, they note in their letter. This is quite distinct from therapies used for improvement of quality of life, such as mind-body therapies (yoga, meditation) or acupuncture.

On Twitter, lead author Johnson points out their study did not, as Soraas claims, show "that cancer patients using CM had twice the risk of dying as non-users."

The study conclusion was "...patients who received CM were more likely to refuse additional conventional cancer therapy and had a higher risk of death. The results suggest that mortality risk associated with CM was mediated by the refusal of conventional therapy," Johnson points out.

JAMA Oncol. Published online January 24, 2019. Abstract

For more from Medscape Oncology, follow us on Twitter


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.