Wednesday, June 13, 2012
Guest Blogger Leslie Francis - Approving Research with Large Sets of Patient Data: the Puzzling Use of Paradigms from Contract Law
As a long-term IRB member at my institution, I have observed many IRB members stumble over a perplexing word in the Federal Regulations governing research with human subjects: “practicably.” According to the Regulations, one of the criteria for approving a waiver or alteration of informed consent is that the research could not be “practicably” carried out if IC were required. (45 C.F.R. § 46.116(c) (2012). “Impractical,” IRB members think the word should be, and that’s how my IRB anyway tends to understand it in discussing whether proposed research studies come within the purview of the rule. But I’m a contracts professor, too—familiar with explaining to my students the difference between “impractical” and “impracticability” as a defense to contractual obligations. That keeping contractual obligations would be hard, or even very hard—very, very clearly damaging, expensive, and certainly not prudential or cost-effective—is not enough for the defense. To meet the standards for the impracticability defense, the party claiming it must show that circumstances have changed, in a manner that could not have been anticipated at the time the contract was made, and that the change goes to a fundamental assumption of the contract.
Here a typical kind of example for which IRBs are asked to consider waiver of informed consent. A researcher wishes to engage in retrospective analysis of data collected in patient care, to ascertain the comparative (or cost) effectiveness of a particular form of patient management. The research itself might be quite benign—for example, effect on patient outcome of time to the administration of an antimicrobial to a febrile infant presenting in an emergency department. Patient records are readily available—and for some reason connected with the research question (e.g. interest in linking outpatient and inpatient data), the researcher needs to use data that have not been de-identified to HIPAA standards. The data set needed for analysis may be quite large: hundreds, or even thousands of patient records. The patients themselves are not readily available: they may have moved, been lost to follow up, or even died. In any event, seeking informed consent from each of them would be very, very expensive and time-consuming—far more so than the benefits of the research seem likely to be. And studies might be biased if certain sub-groups are more difficult to contact or more likely to refuse to allow information about them to be used. The researcher describes security and confidentiality protections for the data that, if followed, would be adequate; and there is little reason to think that people might have principled reasons for opposing the research. So, the IRB reasons a waiver of IC is permissible for this research: it would not be practical for the researcher to obtain IC, and if an IC requirement were imposed, the research (which might after all yield important results) would not be carried out.
This is practicality, yes—but is it “practicability”? In contract law terms, probably not. For purposes of the contract law defense, the questions put to the IRB would be quite different. They would not just be how difficult, or expensive, or time-consuming it would be to obtain IC in order for the research to take place. If contract law “impracticability” is the issue, IRBs should also be asking what the parties—patient and provider—understood about data use at the time of treatment, whether the need for use of data in research was or could have been understood at the time, whether circumstances have changed since the time of the treatment, and whether these changes go to basic assumptions of the contractual/treatment relationship. Perhaps answers to these questions would favor an impracticability finding for the use of data obtained in care many years ago, but it seems unlikely that the possibility of using data in research would not be anticipated today, at least by contemporary health care providers. At least, this is so if a relatively general understanding such as the following is sufficient: “we might wish to use information gained in your treatment for research about treatments for your condition, including comparative effectiveness or cost-effectiveness studies.” So it would seem that there will be few cases of research for which the contract law impracticability standard can be met. The most likely cases would be ones in which radically different types of research emerge, ones that could not have been anticipated at the time of treatment (and ones that challenge basic assumptions about the treatment relationship at the time it occurred). Several possible candidates come to mind: the role of genetic (and epigenetic information) and the possibilities of stem-cell research. But these are types of research given particular scrutiny on other grounds—not the ordinary types of informational research for which IC waivers are sought.
Another disanalogy with contract law is also apparent here. “Waivers” of contractual conditions or terms are for the parties themselves in the first instance. For someone else to exercise a waiver, there would need to be some basis—for example, actual, implied, or apparent authority to act on behalf of the principal. Under the Federal Regulations, however, waiver authority is given to the IRB on the theory that the IRB will adequately protect the interests of research subjects. IRBs may indeed provide such protection; my only point here is that this is not how contract law would approach the problem at all. I do not know why “practicability” was the term chosen in drafting the Federal Regulations. (I’ve thought about this as a research project, and made some preliminary inquiries without success as yet.) I do know, however, that use of the models of impracticability and waiver from contract law draws on a paradigm that is potentially significantly misleading in the context in which it is used. Contract law is, first and foremost, about individual voluntary arrangements, and the protection of expectations created therein—at least, that’s the opening line I offer my students, with many criticisms to come about the importance of protection from coercion or undue influence, the importance of reliance, and the relevance of economic inequality. If so, use of the paradigms from contract law suggests an approach to informational research in terms of the voluntary arrangements of the parties. The doctrine of waiver is initially in line with this—it rests on a background assumption of the need for individual IC—but departs from it by permitting a third party (however trustworthy) to perform the waiver on anyone’s behalf. I don’t, actually, think that the appropriate paradigm for approaching informational research is a model drawn from the law of voluntary agreements. In this, I am sympathetic to the concern in last fall’s ANPRM about the Common Rule that the current regulations are not a good fit for much informational research. The approach suggested in the ANPRM, however, of a separate review process simply emphasizing security and confidentiality protection and dispensing with IC altogether, also seems misguided. There are important ethical questions to be asked beyond security and confidentiality protections, about permissible uses of patient information, discussions with patients and patient expectations, community benefits from data use, and just distribution of these benefits, among other questions, that so far have not been fully explored in discussion of the use of large sets of patient data for informational research.