In praise of undercover research
================================

* Chris Herrera, PhD

*   © 2008 Canadian Medical Association

It is right to be wary of research conducted without the participants' consent. There is especially good reason to look closely when scientists try to answer questions that seem trivial or that involve substantial risk to the unaware participants. In the worst cases, we could imagine researchers skirting informed-consent guidelines with no clear sense of purpose and even fewer incentives to show that their work is safe. To many critics, this evokes images too similar to textbook examples, such as the Nazi medical experiments or the Tuskegee syphilis studies.

Alternatively, it is important not to exaggerate such concerns, even when physicians are unaware they are being studied. Arguments in favour of undercover, unannounced research deserve our attention when there is a chance of improving our understanding of health care practice and patient safety. In the study by Borkhoff and colleagues in this issue of *CMAJ*,1 there was a reasonable assumption that “there may be a difference between what is reported in a survey and what occurs in clinical practice,” as has been demonstrated previously in a context with more immediate public health consequences.2 Rather than stop at asking the physicians about their treatment preferences, the authors used standardized patients as undercover researchers. The results of this study validated the initial assumption, without any appearance of undue risk. Other researchers have successfully used a similar rationale.3 One hopes that methodologic explorations of this kind can help reform the perception of undercover research.

It is worth asking whether some aspects of the protocol followed by Borkhoff and colleagues were overly cautious. For example, the participating physicians visited by the undercover patient-researchers weren't completely uninformed or unwilling; the physicians had been advised of the research a few months before the visit and were given the option not to participate. This provision is enough to make the analogies that we might be tempted to draw from the more notorious cases in history seem very inappropriate.4 Nevertheless, it is unclear whether this provision was necessary, given the likely risks.5 Such precautions are often not taken when other groups are studied; for example, street gangs are not told to expect new “recruits” with recorders and notebooks.

Second, the research team held to a fairly rigid script that governed how the patient-researchers were to behave. This too was probably an effort to err on the side of safety, and there might also have been sound methodologic justification for it. But would giving the patient-researchers more interpretive license have negatively skewed the risk–benefit ratio? Of course, it is one thing to test a physician's readiness to perform CPR if a patient pretends to go into cardiac arrest and another thing to calmly present the doctor with radiographs and to seek advice on the next stage of treatment, as was done in this study. Yet few would argue that each genuine patient acts the same as the next patient. Thus, the drive to minimize risks could have unnecessarily injected an element of artificiality into the setting. To the extent that risk–benefit calculations require meaningful results, ethical justification can also be more complex than is initially thought.

Many researchers might see the unique doctor's role and the physician–patient relationship as a reason to avoid undercover research completely. Methods that work when studying union organizers or cult members might strike some as disrespectful, if not counterproductive, in the health care clinic. Unfortunately, critics rarely explain what makes the medical profession or the relationships we associate with it so unique. References to trust inevitably find their way into such discussions. Yet trust should be distinguished from blind faith. The fact that the physician–patient relationship relies on trust does not imply that a patient does not have the right to know whether his or her physician is conducting what is, for all intents, a business transaction in an open and competent manner.6

The study by Borkhoff and colleagues revealed that a female patient who trusted her doctor to treat her knee problems the same as a male patient's would, in many instances, have been naive. Far from allowing this trust to be damaged, the oversight committee that gave its approval to the undercover research could have insisted that the protocol show promise of actually placing this trust on a more secure footing, such as if the study provides evidence that can improve medical education. If research reveals that a gender bias affects treatment patterns, the prospect of trust between doctors and patients should be contingent on the former doing something to correct or explain such disparities. A similar argument could apply to the trust that would ideally exist between insurers, health care workers, patients, educators, students and so on.7

Oversight committees might find that those who advocate undercover research often have no way to predict, much less control for, the risks that might follow from their research. But we can deal with those cases without overstating the risks to the physicians' interests, especially compared with the risks that might be expected during a visit by ordinary patients. After all, there has been a proliferation of public websites that allow patients to voice their views about specific physicians, their staff and the quality of magazines in the waiting room. The notion that a patient might immediately share a detailed, and perhaps unflattering, opinion of his or her interaction with a physician is no longer far-fetched. In addition, even where researchers are given more leeway in how they are to act, oversight guidelines typically bar them from identifying physicians by name or institution. Ordinary patients, in contrast, are under no such restraint any more than undercover television journalists are.

With that thought, the debate comes full circle, to the claim that there is something that morally separates journalists from ordinary patients and separates them both from scientists. Whereas there may be several distinctions that are worth making, reflection on what the differences are could help make undercover research more mainstream. To dismiss such research without delving into the details would be a poor way to take the moral high ground. On the contrary, we display a lack of moral imagination if we assume that something must be wrong with research that peers into the doctor–patient relationship without including full and informed consent.

@ See related article page [681](http://www.cmaj.ca/lookup/volpage/178/681?iss=6)

**Key points of the article** 

• Undercover research can be assessed similarly to other research methods, such as those used in a clinical trial, in terms of likely risks and benefits.

• The use of undercover research can clarify and enhance, rather than diminish, the trust that ought to exist between patients and physicians.

• Undercover research should be seen as one method among several to collect evidence about the nature and quality of routine medical care.

## Footnotes

*   **Competing interests:** None declared.

## REFERENCES

1.  1. Borkhoff CM, Hawker GA, Kreder HJ, et al. The effect of patients' sex on physicians' recommendations for total knee arthroplasty CMAJ 2008;178:681-7.
    
    [Abstract/FREE Full Text](http://www.cmaj.ca/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoiY21haiI7czo1OiJyZXNpZCI7czo5OiIxNzgvNi82ODEiO3M6NDoiYXRvbSI7czoyMDoiL2NtYWovMTc4LzYvNzIzLmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 

2.  2. Smith GD, Mertens T. What's said and what's done: the reality of sexually transmitted disease consultations. Public Health 2004;118:96-103.
    
    [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.1016/j.puhe.2003.05.004&link_type=DOI) 
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=15037038&link_type=MED&atom=%2Fcmaj%2F178%2F6%2F723.atom) 
    
    [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=000189109100004&link_type=ISI) 

3.  3. Jung B, Reidenberg M. Physicians being deceived. Pain Med 2007;8:433-7.
    
    [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.1111/j.1526-4637.2007.00315.x&link_type=DOI) 
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=17661857&link_type=MED&atom=%2Fcmaj%2F178%2F6%2F723.atom) 
    
    [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=000248575600005&link_type=ISI) 

4.  4. Cave E, Holm S. Milgram and Tuskegee: paradigm research projects in bioethics. Health Care Anal 2003;11:27-40.
    
    [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.1023/A:1025333912720&link_type=DOI) 
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=14510307&link_type=MED&atom=%2Fcmaj%2F178%2F6%2F723.atom) 
    
    [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=000184811900004&link_type=ISI) 

5.  5. Madden JM, Quick JD, Ross-Degnan D, et al. Undercover careseekers: simulated clients in the study of health provider behavior in developing countries. Soc Sci Med 1997;45:1465-82.
    
    [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.1016/S0277-9536(97)00076-2&link_type=DOI) 
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=9351137&link_type=MED&atom=%2Fcmaj%2F178%2F6%2F723.atom) 
    
    [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=A1997YB56100001&link_type=ISI) 

6.  6. VanGeest J, Weiner S, Johnson T, et al. Impact of managed care on physicians' decisions to manipulate reimbursement rules: an explanatory model. J Health Serv Res Policy 2007;12:147-52.
    
    [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.1258/135581907781543102&link_type=DOI) 
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=17716417&link_type=MED&atom=%2Fcmaj%2F178%2F6%2F723.atom) 

7.  7. Walsh K. You couldn't make it up. Or could you? Med Teach 2006;28:81-2.
    
    [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.1080/01421590500441752&link_type=DOI) 
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=16627330&link_type=MED&atom=%2Fcmaj%2F178%2F6%2F723.atom)