The Pitfalls of Deducing Ethics From Behavioral Economics: Why the Association of American Medical Colleges Is Wrong About Pharmaceutical Detailing

ethical principles are by themselves inadequate to fully specify the ethics of human practices (although they may certainly point the way). It is the nature of practices themselves and the constraints that given practices impose on practitioners that determine the specific ethics of decision making in those practices. The AAMC experts and the AAMC itself presume that the constraints on physician decision making among those who subject themselves to pharmaceutical detailing have been adequately elucidated by the psychological experiments cited at the AAMC Symposium. Is this likely to be correct? January, Volume 10, Number 1, 2010 ajob 3 D ow nl oa de d by [ U ni ve rs ity o f C al if or ni a D av is ] at 1 6: 51 1 1 Ja nu ar y 20 16 The American Journal of Bioethics Generalizing from the laboratory to the field in psychology is not quite as straightforward as it can be in physiology. The phenomena of interest in pharmaceutical detailing are, as with other social activities, participant-relative rather than simply given. That is, they are socially constructed and hence dependent upon the governing norms, skills, and perceptions of participants, in this case pharmaceutical representatives and physicians (Searle 1995). That being the case, the generalizability of laboratory experiments to the field setting of detailing will depend upon the extent to which we can be assured that the laboratory setting has recreated the essential features of the detailing interaction as perceived by physicians and pharmaceutical representatives. Laboratory experiments do not typically offer such “external validity”—that is, generalizability to real-world contexts. Their forte is “internal validity”—a higher likelihood of legitimate inferences about cause–effect relationships and hence of adequate models of aggregate behavior (Loewenstein 1999). The AAMC experts were, of course, aware that their experiments did not provide formal external validity in the case of pharmaceutical detailing. But they were clearly confident that these experiments had captured the important aspects of detailing in spite of a lack of corresponding field work to ensure such capture. Such confidence presumably followed from a belief that the cause–effect relationships demonstrated in the laboratory would account for the social phenomena of detailing without a need for further modification. Of course, to the extent that social phenomena can be adequately conceived in terms of the sort of cause-effect relationships that emerge in laboratory work, generalizations to the field from the laboratory data of behavioral and neuroeconomics are likely to succeed. But there are many grounds for supposing that social phenomena are not amenable to analysis in terms of such cause–effect models, or are so only to a very limited extent: that what is called for to understand pharmaceutical detailing or any other reallife activity are the methods not of laboratory psychology alone but in addition those of sociology, anthropology, and economics. As Clifford Geertz observes: Human beings, gifted with language and living in history, are, for better or worse, possessed of intentions, visions, memories, hopes, and moods, as well as of passions and judgments, and these have more than a little to do with what they do and why they do it. An attempt to understand their social and cultural life in terms of forces, mechanisms, and drives alone, objectivized variables set in systems of closed causality, seems unlikely of success. (1995, 127) This is not to say that the cognitive errors and biases uncovered in the laboratory experiments of behavioral decision research are not relevant to real life; they clearly are. The laboratory may usefully elucidate such errors and, by varying experimental treatments, the conditions under which they are likely to emerge (Bardsley 2005). But the bearing of such research on particular human activities must be demonstrated rather than presumed. That laboratory findings do not generalize to the field has been shown for other kinds of activities. The laboratory may both underestimate or overestimate pro-social behavior for different field contexts (List 2006; Levitt 2007). To assess whether physicians routinely make the kinds of errors suggested by laboratory study would require field research. As such fieldwork is lacking, the AAMC was premature to draw conclusions about physician bias and irrationality in response to detailing without it. It is the more general lack of such work that has so far left prominent economists unmoved by the critique leveled against neoclassical economics by their behavioral economics colleagues (Clement 2002; Levitt 2008). Even more troubling than this faulty use of science, however, are the ethical conclusions that the AAMC has drawn from it. IMPLICATIONS OF COGNITIVE ERROR IN DETAILING FOR ETHICS AND POLICY While the policy of academic medicine toward detailing should be determined by what we know of its effects, such policy ought not to be couched in terms of unqualified ethical declarations, whether of approval or prohibition. Because we experience the demands of morality individually, unqualified ethical imperatives are warranted when the moral valence of an activity enjoined or prohibited is clear for any person at whom the imperative is directed. The usual prohibitions associated with medical ethics have this character; betraying patient confidentiality, taking advantage of patient vulnerability, and killing patients (the latter until lately, at least) are always wrong for all physicians. The immorality of the prohibited act affixes unconditionally to the concept of the prohibited act. It is that kind of categorical wrongness that justifies our view of such transgressions as breaches of medical ethics and professionalism. We could have that kind of moral clarity about pharmaceutical detailing only if physician engagement with it was known to be inevitably and always harmful. Certainly the empirical study of detailing apart from behavioral economics offers no reason to draw that conclusion. Neither the work analyzed by Wazana nor the economics literature is decisive as to the aggregate effects of detailing on physician prescribing, health care costs, or patient health outcomes. The Wazana studies generally were not designed to assess these outcomes directly and paid insufficient attention to possible useful learning from detailing. The economics literature makes more use of actual data regarding prescribing, pharmaceutical sales, and detailing visits, but its models are necessarily suggestive rather than conclusive as to detailing’s effects. We might conclude from both the Wazana studies and the economics literature that detailing is morally significant according to its uses and consequences, which may be either good or bad according to its content and the use made of that content by physicians who engage with it (I do not here address the position that any physician engagement with commercial advertising having to do with medicine or its practice is of itself a breach of professional ethics). Policy responses to detailing 4 ajob January, Volume 10, Number 1, 2010 D ow nl oa de d by [ U ni ve rs ity o f C al if or ni a D av is ] at 1 6: 51 1 1 Ja nu ar y 20 16 Pitfalls of Deducing Ethics From Behavioral Economics insofar as they are based upon either the Wazana studies or the economics work should therefore take the form of prudential determinations rather than ethical pronouncements, whatever is decided as to whether physicians should be permitted to participate. If we turn our attention from the Wazana studies and the economics literature to behavioral economics, we will not find the state of our knowledge about pharmaceutical detailing to be transformed. When fieldwork has been done to explore the actual importance of the cognitive and motivational biases detected in the laboratory that might affect detailing interactions, we are likely to find that such biases affect some but not all physician decision making that is responsive to detailing. Such results will continue to warrant a prudential policy response but not outright ethical condemnation of detailing such as the AAMC has delivered. ALTERNATIVE POLICY RESPONSES TO DETAILING Given some prevalence of physician irrationality in response to detailing, as is likely to be the case, the policy response might to restrict detailing and to discourage or prohibit physician participation, or it might be to seek improvement in physician processing of detailing. Which of these options we prefer will likely depend upon our beliefs, both empirical and normative, about human nature and society—in particular, our beliefs as to the advantages and disadvantages of commercialism in given contexts and as to what potential individuals have for overcoming irrationality in those contexts. It seems likely that differing beliefs on such basic issues inform the two sides of the debate presently taking place among academic physicians about medicine’s relations with industry. On one side are industry skeptics, who seek to better police our interactions with the drug companies, which, in their view, have been far too free and easy. Academics in bed with these companies are responsible, it is claimed, for biased research, distorted education, and biased practice guidelines that have altered medical practice to further industry interests at the expense of patients and their insurers. Academics thus need to more carefully regulate their relations with industry; conflict of interest must be eliminated rather than merely disclosed; physicians with industry ties ought not to be authoring guidelines; industry funding for education must be eliminated; and pharmaceutical marketing should be severely curtailed (Brody 2007; Relman 2007; Brennan et al. 2006; Katz et al. 2003). Those resisting such measures point to the benefits of the industry–academic relationship and suggest that these will be sacrificed by limiting that relationship as suggested by industry skeptics. Industry ties, far from being grounds for suspicion, ought to be encouraged among academics. Bias in research comes from multiple sources and financial bias from industry support is surmountable. Guidelines ought to be written by those who are the most knowledgeable about the science informing them—which will include, among others, physicians with industry ties. Restrictions on industry advertising and support for education will lead not to more impartial but to less well informed physicians. Stringent conflict of interest rules are likely to stifle the production of new drugs as those scientists best able to advise pharmaceutical companies seeking to develop them are inhibited from doing so (Stossel 2005; Epstein 2007; Huddle 2008). This debate is clearly political and ideological; it is reflected in similar debates occurring in other disciplines, most notably economics and policy. Those who have been most receptive to the research program of behavioral economics have been economists and others on the left who advocate more regulation of markets and other choice settings in which people might do better if protected from their own irrationality (Tetlock and Mellers 2002). The approach to conflict of interest that accompanies this regulatory prescription is generally to seek its elimination by the structural reform of choice settings. The opposing set of preferences, held by economists and others on the right, follows from a greater confidence that people can overcome irrationality and a greater reluctance to transfer individual decision making to groups of experts, who may be no less prone to irrationality than those whom they seek to protect. To this way of thinking, behavioral economics looks like one more way to justify government micromanagement of individual decision making. Those on this side of the policy divide see conflict of interest as an important problem but are more inclined to manage such conflicts than to seek their removal—believing that the cure may in this case be worse than the disease, as removable conflicts of interest are often accompanied by the compensating advantage of having problems addressed by those most qualified to solve them. This debate is played out on a range of issues, from consumer protection to campaign finance reform. The error of the AAMC is in mistaking an inevitable political debate for a contest of good and evil, and in thus supposing that professional ethics mandates endorsement of one side of that contest. It comes to this conclusion only by presuming what very much remains to be proven, that the bounded rationality seen in the laboratories of behavioral economists is both pervasive in detailing interactions and insurmountable. The latter point is important. Whatever the type, prevalence, and severity of cognitive and motivational error that may be demonstrated by fieldwork on pharmaceutical detailing, physician susceptibility to such error is both unlikely to be uniform and likely to be improvable through education. Given the other advantages of detailing, such as usable knowledge of new drugs getting to physicians more rapidly than it otherwise would, improving physician processing of detailing may be a better policy response than prohibition. Promoting the effective processing of detailing would also fit well with an important contemporary health policy priority, that of protecting individual physician discretion in clinical work. January, Volume 10, Number 1, 2010 ajob 5 D ow nl oa de d by [ U ni ve rs ity o f C al if or ni a D av is ] at 1 6: 51 1 1 Ja nu ar y 20 16 The American Journal of Bioethics DISCRETION AND CONTROL IN STRUCTURING