Preferences of older cancer patients: can you judge a book by its cover?

Abstract

In recent years, studies of databases have yielded a number of interesting insights into patterns of cancer care. These databases, collected for administrative purposes, such as Medicare billing, have been mined as a source of populationbased data on the nonmedical determinants of treatment choice. Many of the findings have been unsettling. For example, recent large database studies have found that the proportion of women with breast cancer who receive mastectomy rather than breastconserving surgery is up to three times higher in the central United States than on both coasts (7,2), and that insurance status predicts for stage at diagnosis and survival, stage for stage, in breast cancer (5). These studies {1-3) have also confirmed what smaller case series have suggested for some time, that older cancer patients receive less aggressive treatment than younger patients. Although large database studies have proven to be extremely useful in demonstrating that nonmedical factors may be driving some of the variability in cancer care, they are much less helpful in identifying where the problem lies or even if there is a problem. Treatment choice is influenced by a number of factors: physician knowledge, physician bias, geographic access to specialized treatment facilities, financial access to care, comorbid conditions, and patient preferences. In contrast to the other determinants, patient preferences are rarely, if ever, evaluated in patterns of care studies. In some cases, there is little need to assess preferences. Threefold differences between mastectomy rates in Massachusetts and Minnesota are almost certainly not due to marked differences in the preferences between women in those two states. However, an assessment of patient preferences is essential in determining whether the less aggressive cancer treatment received by older Americans represents age discrimination or a response to patients' preferences. A previous study (4) has suggested that, although the primary determinant of cancer treatment for the elderly is physicians' recommendations, older patients may give greater weight to considerations of toxicity than younger patients. However, studies of actual decision making cannot disentangle patients' innate preferences about treatment outcomes from the manner in which physicians choose to present those outcomes. The report by Yellen et al. (5) in this issue of the Journal makes an important contribution to our understanding of the role of elderly patients' preferences in determining cancer treatment choice. Using structured vignettes to assess cancer patients' willingness to accept toxic therapy to enhance survival, Yellen et al. found that older patients were as willing to choose chemotherapy as were younger patients, but that older patients required a greater survival advantage before they would choose a toxic regimen over a less toxic alternative. One could argue that preferences expressed about hypothetical choices might differ from decisions patients would make if they faced the same choice in real life. However, the value of using scenarios in this setting is that it eliminates the confounding effect of physicians' biases. What are the implications of this work for the practicing oncologist? One might conclude that treatment should be recommended with equal enthusiasm to all patients regardless of age, but that special attention should be paid to the severity of expected toxic effects when presenting alternatives to older patients. However, an age-based approach to patient counseling should be accepted with caution. Perhaps the most striking feature of the data presented by Yellen et al. (5) is not the differences in preferences across age groups, but rather the wide variability in preferences within age groups. Examinations of patient preferences in a variety of diseases have repeatedly demonstrated that individuals differ widely in their values for different aspects of health (6). Factors such as age, socioeconomic status, religion, and health status explain relatively little of this variability (7). Individuals' goals in life and the sacrifices they are willing to make to attain them are highly personalized and cannot be intuited from a sociodemograph ic profile.