This is not about global warming—- ;-) ! But how many times have you thought about ecologic fallacy? Not many of us think about it and would wonder what it has to do with cancer care. This problem has been on my mind for years as I have watched cancer patients and health care professionals struggle with the difference between their real-life experience of individual patients and the result of medical research literature.
Because cancer does not have one “cause,” literature is replete with risk factors for cancers. For example, breast cancer has been consistently associated with early onset of menarche, late onset of menopause, low and nulliparity, late age at first full-term pregnancy and high total lifetime breastfeeding duration. Yet many cancer care professionals, patients, and patients’ relatives have often wondered why their experience is so different from this litany of risk factors.
Many times, the individual patient with breast cancer in Africa is young, multiparous, has had late age at onset of menarche, and has had high total lifetime duration of breastfeeding. This causes great distress because apparently the patient’s profile does not fit that of someone “who should have cancer.” This distress derives directly from drawing conclusions about individuals—in this case, patients with breast cancer—from aggregate—in this case, epidemiology—data.
So next time you see patients whose characteristics do not fit that of the general population of individuals with that kind of cancer, think about ecological fallacy and remember that this one case does not make a difference. There will always be patients in the tails of the distribution whose characteristics are different from that of the majority of patients and this should not lead us to erroneous conclusions.
Now, having said this, such outliers may, in some circumstances, be informative. They may represent a different pattern of disease with different etiology, requiring different treatments and having different outcome. How can one tell which of the aberrant presentations is an interesting phenomenon and which is not? Now that is where the challenge lies.