The Hydra

Almost nobody favors long consent forms for prospective research subjects. Almost everybody thinks they interfere with informed consent’s purpose—good decisions. Nevertheless, almost everybody believes consent forms have long been getting longer.

Years ago, Paul Appelbaum lamented the “tendency to cram ever more information into consent forms.”¹ Weeks ago, Ilene Albala and her colleagues (one of them Appelbaum) reported in IRB: Ethics & Human Research that the length of one institutional review board’s forms “increased roughly linearly by an average of 1.5 pages per decade. In the 1970s, the average consent form was less than one page long and often only a paragraph or two, but by the mid-1990s the average form had increased in length to over 4.5 pages.” Similarly, “Baker and Taub demonstrated that the mean length of consent forms nearly doubled between 1975 and 1982. More recently, Beardsley and colleagues in Australia found that the median length of consent forms increased from seven to 11 pages between 2000 and 2005.”²

Long forms are deplored on several grounds: That people will not read them. That even if people read them, they would not understand them. That even if people understood them, the forms would not promote better and might promote worse decisions.

For example, in a study of principal investigators, over half thought contemporary consent forms “unlikely to be read” or “incomprehensible.” Onerous length and detail that reduced subjects’ understanding were bemoaned. One researcher thought “most subjects skim through the incredibly long informed consents, believing that most of it is simply bureaucracy.”

These concerns are well founded. First, long forms repel, confuse, bore, and distract. (It appears “that, in an educational context, people are unlikely to read entire documents that contain more than 1000 words, or about 4 pages.”³) Furthermore, these forms compete for your attention with hundreds of other disclosures about innumerable matters. Who could—who would—study them all? Who reads credit card contracts, mortgage agreements, retirement account prospectuses, bank privacy statements, online purchase conditions, HIPAA warnings, insurance provisions, or rental car forms?⁴ Not I. Not you. So, for example, the patients HIPAA blesses do “not appear to recognize, understand, or care about this complex law as it applies to research.” They ask about disclosures “exceedingly rarely. For example, one academic health center reported that between 2003 and 2007, the institution received only 23 requests for accounting of disclosures, and none were from research.”⁵

Second, in both research and clinical medicine even good forms and processes fail to achieve their educational goals. Even tested after optimal “consenting,” patients correctly answer only a third to a half of the questions asked.⁶ And even patients with “a relatively large variety of information sources” use incorrect information, so that fewer than half the breast cancer patients in one study understood treatments’ survival rates, and fewer than a fifth understood recurrence rates.⁷ The causes are many. For example, illiteracy and innumeracy prevent many people from reading many forms. (Roughly ninety percent of the people in one study had at least some college education, but 40 percent “could not solve a basic probability problem or convert a percentage to a proportion.”⁸) Yet the simpler your language, the more words you need to explain your ideas, and the longer forms get.

Third, people can keep only a few things in mind when analyzing a problem. Miller’s “magical number seven” is the classic estimate, and it is easily exceeded. For example, Miranda warnings are familiar, few, and short. Yet even with “verbal chunking” (combining data for easier storage) “the upper limit of information processing for Miranda warnings is likely less than 75 words,”⁹ considerably fewer, that is, than the usual Miranda warning. And when anesthesiologists and nurse practitioners tried to educate their patients, they “vastly exceeded patients’ short-term memory capacity.”¹⁰ Furthermore, integrating “different types of information and values into a decision is a very difficult cognitive process.” Indeed, information can decrease the reliability of decisions. For example, the “reliability of the choices [of horse-racing handicappers] decreased as more information was made available.”¹¹

If long forms are widely and rightly...