Main

The right of self-determination is a fundamental human right and an established tenet of ethical medical research and clinical practice. The untenable effects of ignoring informed consent have been thoroughly described elsewhere (Pincus et al. 1999). The recent National Bioethics Advisory Commission (NBAC) report has provoked controversy because of its specific focus on psychiatric patients as being potentially vulnerable to impaired decision making, as well as for other perceived deficits (Carpenter and Vasi 1999; Michels 1999; National Bioethics Advisory Commission 1998). Although a subject of debate, several commentators criticize the recommendations of the commission for their potentially stigmatizing effects. Thus, it has been pointed out that not just patients with psychiatric disorders, but also patients with a variety of other medical conditions, have been shown to be potentially vulnerable to impaired decision making (Oldham et al. 1999; Roberts 1998; Roberts and Roberts 1999). Furthermore, it has been argued that the NBAC report did not take into consideration the available research on decision making by psychiatric patients, which indicates the variable nature of performance by patients with psychiatric illnesses as well as patients with medical or surgical conditions (Roberts 1998; Roberts and Roberts 1999). In addition, several studies recently found that consent-related capacity of patients with schizophrenia was enhanced with straightforward educational interventions (Wirshing et al. 1998; Carpenter et al. 2000).

The importance of obtaining proper informed consent from research subjects and patients cannot be overemphasized, as noted by Lieberman and Aghajanian (1999). Adequate informed consent consists of three required elements: full information; voluntary participation; and capacity to make a decision (Appelbaum and Roth 1982; Christensen et al. 1995). Capacity for decision making, in turn, is composed of four functional abilities: the ability to understand relevant information; the ability to appreciate the nature of a situation and its likely consequences; the ability to reason with the information and weigh options logically; and the ability to communicate the choice (Appelbaum and Grisso 1988; Grisso and Appelbaum 1998; Pincus et al. 1999).

Researchers have studied decision-making capacity using a variety of instruments, including the MacArthur Competence Assessment Tools for treatment and research (Appelbaum et al. 1999; Grisso and Appelbaum 1998), among others (Bean et al. 1994; Janofsky et al. 1992; Saks and Behnke 1999; Miller et al. 1996).

An extensive literature consistently reports a variety of problems related to the understanding component of decision making (Taub and Baker 1984; Taub 1986; Verheggen and van Wijmen 1996; Edwards et al. 1998; Silva and Sorrell 1988; Sugarman et al. 1998; Gotay 1991). These deficits are, however, not restricted to patients with psychiatric diagnoses or cognitive impairments. Many physical illnesses, as well as medications, can place patients at risk for impaired understanding of research or treatment (Sugarman et al. 1999; Grisso and Appelbaum 1998). In fact, studies comparing the abilities of patients with and without psychiatric illness have shown that the presence of a psychiatric disorder, albeit a risk factor, does not predetermine whether a patient can understand key information (Appelbaum and Roth 1982; Appelbaum et al. 1999; Grisso and Appelbaum 1995; Appelbaum and Grisso 1995; Lidz et al. 1984; Meisel and Roth 1981; Roth et al. 1987; Stanley et al. 1987; Sugarman et al. 1999; Carpenter et al. 2000). Several investigators have demonstrated that individuals with psychiatric disorders do possess definite strengths related to decision-making capacity (Grisso and Appelbaum 1998; Carpenter et al. 2000; Kleinman et al. 1993; Stanley et al. 1981; Soskis 1978).

No generalized agreement dictates how to measure understanding, despite some attempts to devise assessment tools (Miller et al. 1996; Grisso and Appelbaum 1998). One obstacle is the definition of “understanding.” To understand a treatment or research protocol, a patient must receive, encode, retain, and process the information. This necessarily involves sensory modalities, attention, memory, and cognition. Terms used in the literature to describe this complex process include “understanding,” “comprehension,” “knowledge,” and “recall.” Recall alone does not imply understanding. Furthermore, long-term recall is not always necessary, for example, in immediate treatment decisions (Grisso and Appelbaum 1998). Knowledge also does not always imply understanding.

Despite this lack of agreement on how to define and measure understanding, problems with understanding of both research and treatment protocols have been widely reported. Deficiencies in patients’ understanding include lack of awareness of being a subject in a research study; poor recall of supplied information; lack of understanding of randomization procedures and placebo treatments; inadequate recall of important risks of procedures or treatments; lack of awareness of the ability to withdraw from a research study at any time; the “therapeutic misconception” (i.e., the belief that treatment decisions are being made solely with the individual subject's benefit in mind); and confusion about the dual roles of physician/researchers (Levine 1992; Silva and Sorrell 1988; Verheggen and van Wijmen 1996; Edwards et al. 1998; Appelbaum et al. 1982; Sugarman et al. 1998; Robinson and Merav 1976; Muss et al. 1979; Cassileth et al. 1980).

Clearly, no single method to improve understanding and/or recall offers a panacea for the array of potential pitfalls listed above. A critical review of methods to improve understanding should reveal effective strategies and suggest potential areas for further investigation. Although the number of articles published on informed consent has increased substantially over the last 30 years (Kaufmann 1983; Sugarman et al. 1999), the number of studies that actually test methods to improve the informed consent process has been limited.

One positive outgrowth of the controversy surrounding the NBAC report has been a call for more rigorous research into informed consent procedures and decision-making capacity in patients at potential risk for impairments (Hyman 1999; Charney et al. 1999; Jeste et al. 1999). As an example, further research is needed into the various factors (i.e., cognitive as well as subjective influences) that affect decision making by psychiatric patients (Childress and Shapiro 1999; Roberts and Roberts 1999; Oldham et al. 1999). This is crucial, because even in patients with presumably intact decision-making capacity, poor understanding of information about proposed treatments or research protocols is common (Edwards et al. 1998; Sugarman et al. 1998).

In this article, we focus on whether understanding of informed consent can be improved with educational interventions. As a starting point, a review of the extant literature on this topic will help clarify what is currently known and what remains to be understood about how researchers and clinicians can improve understanding of informed consent. Furthermore, the limitations of studies reviewed here underscore the need for an increase in the amount and quality of the research conducted in this area. Because much of this research has been conducted with nonpsychiatric patients, we include studies from different disciplines in the hope of discerning patterns regarding what kinds of interventions may be helpful, regardless of a patient's diagnosis.

We conducted a literature search for empirical studies of informed consent using the Medline (from 1966 to present) and PsycINFO (from 1967 to present) databases. We searched for articles using the major subject term “informed consent,” reviewed these by title and abstract, and retrieved studies that fit our inclusion criteria. Review of these articles’ texts and references led to identification of other relevant sources. We also carefully searched a recent annotated bibliography on empirical research on informed consent (Sugarman et al. 1999), as well as several reviews of this topic (Taub and Baker 1984; Taub 1986; Verheggen and van Wijmen 1996; Edwards et al. 1998; Silva and Sorrell 1988; Sugarman et al. 1998; Gotay 1991).

For the final group of articles, we included English language studies of adults that evaluated an intervention(s) designed to improve subjects’ understanding of consent information. We included studies from a variety of disciplines. We excluded studies that were purely descriptive (e.g., studies of the readability of consent forms without an intervention to improve readability). We also excluded studies of other aspects of consent (e.g., willingness to consent, associated anxiety levels, satisfaction with consent) that did not include an intervention to improve understanding. Intervention studies that examined other aspects of informed consent were included only if they also studied a measure of patients’ understanding or recall.

The studies reported on a wide variety of approaches to improving patients’ understanding of informed consent. The patient populations, sample sizes, methods, and outcome measures varied widely. In addition, the studies examined either consent for research or consent for treatment and were conducted in a variety of disciplines (i.e., psychiatry, surgery, and medicine). For these reasons, we chose not to conduct a meta-analysis.

Study Characteristics

Of the 34 studies, 12 examined patients scheduled for a surgical or radiological procedure, eight included healthy volunteers, five examined patients with psychiatric disorders, four looked at patients with cancer, three included frail geriatric patients or long-term care residents, one enrolled pregnant women, one studied injection drug users, and one examined patients with advanced HIV disease.

Outcome measures varied widely, from multiple-choice and true/false questions to structured interviews. Some tested understanding immediately after the consent procedure; whereas, others used both immediate and delayed measures of understanding and recall. Other authors have commented that, although many studies purport to measure understanding or comprehension, what they actually measure is memory for presented information (DeRenzo et al. 1998; Lavori et al. 1999; Silva and Sorrell 1984). Often, the terms “comprehension,” “knowledge,” “understanding,” and “recall” are used interchangeably.

We identified a number of studies that attempted to measure the effectiveness of various interventions over time, thereby providing some information, not merely about immediate recall but also about the durability of information provided (Simes et al. 1986; Taub et al. 1981; Taub and Baker 1983; Tindall et al. 1994; Tymchuk et al. 1988). Because the timing of measurement of understanding varied greatly among studies, caution is indicated when comparing results. Poor performance on outcome measures assessed after a delay of 1 to several weeks may indicate only that memory is fallible, not that the subjects failed to understand information when it was initially provided (Silva and Sorrell 1984). Nevertheless, certain strategies may help subjects remember the material beyond the initial testing period (Kleinman et al. 1996; Taub et al. 1981; Tindall et al. 1994; Wirshing et al. 1998). Interventions that showed a benefit when delayed recall (generally several weeks later) was tested included an additional interaction (telephone call or information visit) with a nurse, (Dodd and Mood 1981; Aaronson et al. 1996) informational videotape, (Weston et al. 1997) written (vs. or in addition to oral) preoperative information, (Armstrong et al. 1997; Askew et al. 1990; Morrow et al. 1978) simplified and illustrated presentations, (Krynski et al. 1994; Tymchuk et al. 1988; Tymchuk and Ouslander 1991), and corrected feedback or multiple learning trials (Kleinman et al. 1993; Kleinman et al. 1996; Taub et al. 1981; Wadey and Frank 1997; White et al. 1995; Wirshing et al. 1998).

Consent for Research

We categorized the studies by consent for research or for a treatment or procedure (including hypothetical treatments) to make the tables more accessible to researchers or clinicians who wanted to incorporate the findings of these studies into their own efforts to improve subjects’ or patients’ understanding of informed consent.

Several authors have noted the importance of distinguishing between informed consent for research, which differs fundamentally from consent for treatment (Appelbaum et al. 1982; Taub et al. 1986). Participants in clinical research need to understand the difference between individualized reatment and research protocols. For example, in research, randomized assignment to treatment arms may occur, placebos may be given, and—in many instances—there is a possibility that subjects will not benefit directly from the experimental intervention. When subjects do not understand these distinctions and assume that decisions about their care will be made only for these distinctions and assume that decisions about their care will be made only for their individual benefit, this is termed “the therapeutic misconception” (Appelbaum et al. 1982). Because of these important distinctions, we separated these categories of consent for research and consent for treatment in our analysis.

We found that in both categories, educational strategies to improve understanding were generally successful. Several authors noted that patients had more than a passing interest in being informed of the purpose, procedures, and risks of the actual treatment recommended for them—for example, when chemotherapy drugs had potentially lethal side effects (Dodd and Mood 1981) or when the risks of intravascular contrast material were not previously known to patients undergoing radiological procedures (Hopper and Tyler 1989). We found, however, that even in studies of consent for hypothetical treatments, patients showed improved understanding with the interventions described (Epstein and Lasagna 1969; Tymchuk et al. 1988; Krynski et al. 1994; Tymchuk and Ouslander 1991).

Table 1 lists 16 studies of consent for research. Eleven of these demonstrated greater understanding among patients in the experimental condition.

Table 1 Studies on Enhancing Informed Consent for Research
Full size table

Positive Studies

The 11 positive studies included comparisons of simplified with the standard (complex) consent forms (Bjorn et al. 1999; Young et al. 1990); comparison of uniform total disclosures with individualized disclosures personalized by the physician (Simes et al. 1986); a telephone-based intervention to inform patients more thoroughly about a clinical trial (Aaronson et al. 1996); a step-wise consent process (Rikkert et al. 1997); the use of interviews, repetition, corrected feedback, and multiple learning trials (Carpenter et al. 2000; Taub et al 1981; Wirshing et al. 1998), and having patients speak or write about the consent information (Sorrell 1991). In two studies, patients served as their own controls, with change over time being the reported outcome measure (Rikkert et al. 1997; Wirshing et al. 1998). Two of the positive investigations used videotape to enhance the informed consent process (Fureman et al. 1997; Weston et al. 1997). Both videotape studies reported that patients in the intervention group demonstrated less decrease in knowledge over time (2 weeks to 2 months later).

Negative or Inconclusive Studies

Of the five negative studies, two were inconclusive regarding the effect of simplified, more readable consent forms on patient understanding (Taub et al. 1986; Taub et al. 1987). Two other reports on corrected feedback (Taub and Baker 1983) and the addition of an oral discussion to written information (Tindall et al. 1994) produced mixed results. The latter study demonstrated improvement from pre- to postconsent, but the oral discussion did not add significantly to this improvement. Benson et al. (1985) studied “improved” versus “natural” disclosures and reported mixed results; that is, patients with schizophrenia did not benefit from the improved disclosures; whereas depressed patients benefited somewhat. In this study, however, over all patient understanding remained low in both groups, despite attempts to improve the disclosure process.

Consent for Treatment

Table 2 summarizes 18 studies of consent for treatment; 14 of them found the experimental condition to be of some benefit.

Table 2 Studies on Enhancing Informed Consent for Treatment
Full size table

Positive Studies

Of the 14 positive reports, three compared varying degrees of oral and written consent (Armstrong et al. 1997; Dawes et al. 1992; Inglis and Farnill 1993); another compared short, medium, and long forms of written information (Epstein and Lasagna 1969). Two other positive reports examined the effects of providing illustrated, highly readable, large-print versions of clinical vignettes (Tymchuk and Ouslander 1991; Krynski et al. 1994; Tymchuk et al. 1988). Other positive interventions provided take-home information sheets 1 to 3 days before a scheduled procedure (Askew et al. 1990; Morrow et al. 1978), had a nurse review the consent information with the patient (Dodd and Mood 1981), gave corrected feedback to patients (Kleinman et al. 1993), and had patients orally repeat the consent information (Wadey and Frank 1997; White et al. 1995). A videotape intervention (which included “advance organizers” that alerted patients to the material to be presented, on-screen graphics, and summaries of key points) helped patients score better on a knowledge test (Agre et al. 1994). Another study used an interactive video that allowed patients to receive more information if they chose to (Hopper et al. 1994).

Negative or Inconclusive Studies

The four negative reports involved using oral and/or written consent of varying levels of detail (Hopper and Tyler 1989; Stanley et al. 1998), provision of a written consent form at varying times (24 to 72 h vs. 15 to 60 min) before a radiological examination (Neptune et al. 1996), and the use of a professionally prepared video about ECT that augmented written material (Westreich et al. 1995).

Role of Psychiatric Illness or Cognitive Impairment

Several researchers examined degree of cognitive impairment and psychiatric symptoms as covariates. Rikkert et al. (1997) noted that mildly demented patients demonstrated poorer understanding that nondemented patients; patients with major depression, however, did not differ from the nondepressed ones. Notably, in this study, understanding increased significantly in both mildly demented and nondemented patients after a 1-week consent “try out.” Wirshing et al. (1998) found delayed understanding was negatively correlated with scores on the conceptual disorganization subscale of the Brief Psychiatric Rating Scale (BPRS) (Overall and Gorham 1986); however, patients with schizophrenia did show improvement in understanding and retention of information with the more rigorous, educational consent process that was used. Carpenter et al. (2000) found that although poor performance on a decisional capacity instrument was moderately correlated with severity of psychiatric symptoms (assessed by the BPRS), performance was even more strongly influenced by the degree of cognitive impairment. Nonetheless, patients with schizophrenia who received the educational intervention improved their performance to the same level as non-ill comparison subjects.

Predictors of Performance

The variables most often examined were age, education, and vocabulary level. Gender was infrequently examined as a variable associated with understanding, with inconsistent results (Hopper et al. 1994; Morrow et al. 1978; White et al. 1995). Of 10 studies that evaluated an age effect, six found that performance was significantly and inversely related to age (Taub et al. 1986; Taub et al. 1987; Krynski et al. 1994; Neptune et al. 1996; Aaronson et al. 1996; Agre et al. 1994), another found a similar, but nonsignificant, trend toward this association (Morrow et al. 1978), and three investigators found no correlation between age and understanding (Tymchuk et al. 1988; Sorrell 1991; White et al. 1995). One of the three studies that found no age effect examined patients within a fairly narrow age range: Tymchuk et al. (1988) enrolled long-term care residents (mean age 84 years, SD 5 years). Sorrell (1991) studied 80 healthy women consenting to a breast self-examination teaching program (mean age 40 years, standard deviation and range not given), raising the question of whether there was a sufficient age range to detect an age effect. Interestingly, despite the negative association of age with understanding, even older patients derived significant benefits from education interventions in a number of studies (Krynski et al. 1994; Rikkert et al. 1997; Taub et al. 1981; Taub et al. 1987; Tymchuk et al. 1988; Tymchuk and Ouslander 1991).

Nine studies examined the relationship between education and understanding. Eight reported a positive association (Bjorn et al. 1999; Young et al. 1990; Taub et al. 1986; Taub et al 1987; Sorrell 1991; Neptune et al. 1996; Aaronson et al. 1996; Agre et al. 1994). Two studies that evaluated patients’ vocabulary levels (as measured by the vocabulary subtest of the Wechsler Adult Intelligence Scale [WAIS]) found that performance on immediate as well as delayed tests of understanding varied directly with vocabulary level (Taub and Baker 1983; Taub et al. 1981).

In this review of interventions to improve patients’ understanding of information given during informed consent, we were particularly interested in studies of patients with psychiatric illnesses or cognitive impairments. Several studies found that specific diagnoses per se do not render patients vulnerable to impaired decision making; on the other hand, cognitive impairments associated with diverse conditions can place medical, surgical, and psychiatric patients at risk for impaired understanding, thus potentially impairing their decision-making capacity (Carpenter et al. 2000; Krynski et al. 1994; Tymchuk et al. 1988; Tymchuk and Ouslander 1991; Wirshing et al. 1998; Jaffe 1986). Previous reports have also illustrated this point (Grisso and Appelbaum 1991; Holzer et al. 1997; Grisso and Appelbaum 1998). Patients with schizophrenia, as well as other populations, show a great deal of variance in performance. More important, patients in the studies reviewed have demonstrated substantial and clinically significant improvements in their performance on informed consent assessments when educational interventions were utilized (Rikkert et al. 1997; Carpenter et al. 2000; Kleinman et al. 1993; Wirshing et al. 1998). These observations suggest that clinicians may employ specific methods to identify those patients who may be at risk for inadequate understanding. More research is needed to ascertain what specific factors place individuals at risk for impaired decision making and how best to modify these risk factors (Roberts and Roberts 1999).

Summary of Findings

Twenty-five out of the 34 studies reviewed found that patients’ understanding or recall showed improvement with a wide variety of interventions. Some strategies, however, were consistently more effective than others. More highly structured and more uniform consent processes, better organized, shorter and more readable consent forms, and simplified and illustrated formats all improved patients’ understanding. Corrected feedback, multiple learning trials, “advance organizers” (which alert patients to information about to be presented), and summaries of information also enhanced understanding. Highly detailed information, however, was not consistently associated with better understanding. Augmenting or replacing the consent form (e.g., with a videotape) showed some promise, but these studies were limited in number, and results were inconsistent.

Certain types of strategies may help patients remember the material beyond the initial testing period. These strategies included an additional telephone discussion or informational visit from a nurse (Aaronson et al. 1996; Dodd and Mood 1981), an informational videotape (Weston et al. 1997), written (versus or in addition to oral) preoperative information (Armstrong et al. 1997; Askew et al. 1990; Morrow et al. 1978), simplified and illustrated presentations (Krynski et al. 1994; Tymchuk et al. 1988; Tymchuk and Ouslander 1991), and corrected feedback and multiple learning trials (Kleinman et al. 1993; Kleinman et al. 1996; Taub et al. 1981; Wadey and Frank 1997; White et al. 1995; Wirshing et al. 1998).

In examining whether understanding was correlated with specific demographic variables, we found that educational and vocabulary levels of patients were consistently and positively associated with measures of understanding in a variety of studies. Older patients with less education seemed to be more vulnerable to poor understanding (Taub et al. 1986; Taub et al. 1987). Age alone, however, was not consistently associated with decrements in performance, and, in fact, older patients benefited from strategies to improve their understanding (Krynski et al. 1994; Rikkert et al. 1997; Taub et al. 1981; Taub et al. 1987; Tymchuk et al. 1988; Tymchuk and Ouslander 1991).

We should point out limitations of this review. As in many reviews of the literature, we most likely missed a few articles, and we did not include those published in languages other than English. In addition, a tendency among authors and journals to publish positive rather then negative studies might have resulted in a bias toward positive data. There may also have been subject selection biases in individual studies. For example, patients who were very ill or who were thought to lack the cognitive capacity to participate in research were likely excluded. In research on informed consent, this is to some degree inevitable, because patients must be able (or must be thought to be able) to provide adequate consent.

Another limitation of this review pertains to the studies themselves. The reports were heterogeneous in terms of both patients and interventions. In addition, individual studies had several common limiting aspects. For example, some investigators did not describe particular features of the study populations, such as age range or education level. A few studies did not report the readability level of consent forms used, despite stating that the consent documents were simplified or revised. Some reports did not provide full details of the interventions or outcome measures used. In some cases, small sample sizes limit the generalizability of the results. Finally, some studies did not analyze possible confounding variables, such as correlations between age or education and understanding.

Directions for Future Research:

A growing body of literature supports the need to enhance the informed consent process. Below are a few areas for further research.

Use of New Technologies

The emergence of multimedia, interactive, and web-based technologies should help further the exploration of novel methods for enhancing informed consent (Rosoff 1999). Interactive technology, for example, can create information tailored to specific patients’ learning styles and preferences. Some groups are already using multimedia tools to inform potential research participant about key components of research (Jimison et al. 1998). Computers also can be used to test patients before, during, and after a consent procedure and provide feedback, review, and retesting.

Controlling for Effects of Memory on Understanding

The results of numerous studies reviewed here are confounded by the confusing between understanding and recall. Taub et al. (1981) suggest that allowing patients to refer to consent materials during assessments of understanding is one way to control for the effects of memory. Few of the studies reviewed here have adequately separated out the effects of memory from the outcome measure variously referred to as “comprehension,” “understanding,” or “knowledge.” More rigorous techniques should be applied to study this aspect of informed consent, because patients’ involvement in research or treatment does not occur only at one discrete point in time. Several groups are actively devising and testing methods to educate patients over time about research participation (Wirshing et al. 1998; Carpenter et al. 2000).

Risk Factors Related to Decision-Making Capacity

Studies of vulnerabilities of patients with psychiatric disorders represent a fruitful area for research. Some of the most important tasks will be to clarify individual variation in abilities to assimilate consent information, test for cognitive or other risk factors for poor understanding, and find ways to bring at-risk patients up to an adequate level of understanding (Wirshing et al. 1998; Carpenter et al. 2000). Incorporating cognitive testing into preconsent procedures may clarify associations between these functions and understanding of informed consent materials. Measuring other aspects of decision-making capacity, such as reasoning and appreciation, as well as risk factors for impairments in these domains, is also important. A crucial task is to explore methods to enhance understanding, appreciation, and reasoning related to the informed consent process. The limited nature of our conclusions in this article reflects the paucity of research in this area, particular pertaining to patients with psychiatric disorders.

Capacity for decision making should be viewed, not as an all-or-nothing state, but as multiple functional abilities along a continuum that can change over time. Furthermore, alleged deficits in patients’ understanding of informed consent may, in fact, be related, at least partly, to poorly conceived, written, designed, or organized informed consent documents or processes. Hence, deficits in understanding or appreciation may be remediable with well-designed educational interventions. It is a truism that not everyone learns best in the same way. Therefore, we hope that clinicians and researchers will make increasing use of targeted interventions that take into account individual variation as well as potential risk factors for impaired understanding.