Information presentation and decisions to enter clinical trials: a hypothetical trial of hormone replacement therapy

Abstract

We examined recruitment to an imaginary trial of hormone replacement therapy (HRT) following two different styles of information about HRT. We predicted that for treatments which, like HRT, are available outside a trial, people offered the facts as currently known would be less likely to remain unsure about the relative costs and benefits, and so less likely to agree to enter a randomised trial. In contrast, when the information provided reflected the current state of uncertainty which justified the trial, we predicted that people would be less likely to form a preference for one treatment arm over the other, and so more likely to agree to enter a trial. One hundred women aged 25–40 years were informed about HRT via a video and an information leaflet. For half the participants the information was framed in a way which emphasised the current state of uncertainty about the relative costs and benefits of HRT, and in that respect it reflected the justification for a trial. This version was considered to be similar in style to information commonly provided to potential trial participants. For half the participants the same information was framed in a way which offered explicit numerical detail about currently known facts, and in that respect it was considered to be similar in style to information commonly available to doctors prior to a trial. Women learned as much about HRT in the two conditions, but women given the explicit versions were more likely (i) to hold a stronger view about whether or not they would take HRT (ratings were not elicited from the first 30 participants in this condition. N=20, p<0.05 1 tailed) and (ii) to refuse entry to the trial (N=50, p<0.05 2 tailed). Those who, given the explicit version, agreed rather than refused to enter the trial, scored higher on believing that others control their health (p<0.01 2 tailed).

Introduction

With the increasing demand for cost effective health care, there is pressure to produce high quality research into the best medical treatments. Scientifically, randomised clinical control trials (RCTs) are the best method of establishing accurate evidence for the prevention, treatment and cure of illness (Sacks, Chalmers & Smith, 1982). However, a number of authors have reported difficulty in recruitment and adherence to RCTs (Benson et al., 1991, Mansour, 1994, Rimer, 1992, Schain, 1994). This could bias the research results (Kotwall et al., 1992, McCusker et al., 1982). Concern has also been expressed about the ethics of RCTs and informed consent procedures (Baum, 1990, Botros, 1990, DeLuca et al., 1995, Williams and Zwitter, 1994). For a recent review see Edwards, Lilford, Thornton and Hewison (1998). Researchers are faced with the dilemma of how to satisfy legal and ethical constraints whilst still recruiting the maximum number of participants to fulfil public and political demand for scientific evidence.

To date, there has been little investigation of how participants make the decision whether or not to enter a clinical trial or the variables which may affect this decision. A few studies have looked at the effect of the way in which information is framed on patients’ decision making, drawing on work by Kahneman and Tversky (1984). The main premise is that decisions are based on a reference point which is influenced by an individual’s perception of whether a particular outcome is a gain or a loss. It has been suggested that the location of the reference point can be altered by the way in which information is presented or framed (Kahneman & Tversky, 1979). This idea has also been applied to health related decisions (Meyerowitz & Chaiken, 1987). Marteau (1989) demonstrated how the framing of information altered the reference points of medical students. In her study involving imagined surgery, significantly more participants chose to undergo surgery when the chance of survival was framed positively (a 10% chance of survival) rather than negatively (a 90% chance of death). The survival percentage was seen as a gain and the death percentage as a loss. These findings suggest that the manner in which information is presented may affect health related decisions.

In contrast to this, Llewellyn-Thomas, McGreal and Thiel (1995), found that the manner of framing probabilistic treatment information (probabilities about the side effects of treatment were presented either in a neutral, positive or negative manner) did not affect willingness to participate in a hypothetical cancer trial. However, the investigators suggest that since participants had a real diagnosis of cancer, they may already have considered various treatment options, including the possibility of entering a trial. As a consequence, any framing effect could have been masked in this study.

It may not simply be the positive or negative framing of probabilistic information which is important but the certainty with which it is presented. For a trial to receive approval there must be no clear evidence that one treatment is better than the other (Lilford & Jackson, 1995): there should be a state of equipoise about the risks or benefits of trial treatments. However, Botros (1990), argued that individual doctors are rarely in a state of equipoise concerning the treatments under investigation. Doctors may have slight preferences for one treatment over the other, based on their individual interpretations of the limited evidence available prior to the trial. If so, this raises the question of how information about treatments should be presented to patients. If the information provided emphasises the current state of uncertainty about the relative costs and benefits of the treatments under trial, then patients may have little basis on which to form a preference for one treatment over the other, and as a consequence they may be prepared to be randomised. In contrast, if the information provided makes explicit the evidence so far available, providing numerical detail, then patients, like doctors, may be more likely to form a preference for one treatment over the other. Given the limited evidence, the preferences formed would not be expected to be consistent across patients, but unless the preferred treatment is available only within the trial then such patients may be unwilling to be randomised.

For our study, hormone replacement therapy (HRT) was selected as the treatment under examination because although there is already a good deal of evidence about its effects, it is also currently the subject of clinical trials. We could therefore provide people with summaries of the evidence as it stands at present, either in an “ambiguous” manner emphasising the current uncertainty about the relative costs and benefits, or in an “explicit” manner providing the current best estimates of the effect of HRT, but explaining that these figures are based on non-experimental evidence and hence are potentially inaccurate. Our aim was to find out (i) whether people were less likely to form a preference for or against HRT when information was framed in an “ambiguous” rather than an “explicit” manner, and (ii) whether people were more likely to be prepared to enter a hypothetical trial when information was framed in an “ambiguous” rather than an “explicit” manner. In this way we were able not only to examine the relative effects on trial entry of our two ways of framing information, but we were also able indirectly to check on our assumption about why any effect occurred. Our assumption was that people are more willing to enter a trial if they have no preference for one treatment over the other, and less willing if they have a preference (provided, as in this case, the preferred treatment is available outside the trial). People may of course agree to enter a trial despite having a preference for a particular treatment, perhaps through an altruistic desire to further scientific knowledge or because they failed to understand randomisation (Howard and DeMets, 1981, Jensen et al., 1993, Simes et al., 1986).

We were also interested in identifying aspects of individual differences which related to decisions about whether or not to enter the hypothetical clinical trial. Verheggen and van Wijmen (1996) noted that very few empirical studies have looked at the psychological aspects of entering clinical trials. Debate has occurred about a number of individual difference variables such as education, knowledge levels and the desire of patients to participate in treatment decision making, on accrual to RCTs (Llewellyn-Thomas et al., 1995, Strull et al., 1984). It has been suggested that some patients prefer to leave decision making to their doctors and require very little information or positively avoid information and others demand a large role in decision making (Miller et al., 1988, Llewellyn-Thomas et al., 1991). It is possible that those with an internal locus of control (people who consider themselves in control of their own destinies) would want to make their own decisions. Conversely those with an external locus of control (individuals who believe that external factors such as chance or powerful others determine the outcome of their lives) may rely on clinicians to make decisions for them. Such factors could potentially influence decision making and may even override framing effects. Therefore, these variables were also investigated in this study.

In summary, our study had two main aims. The first was to determine the effect of framing information about HRT in an “explicit” or “ambiguous” manner on formation of preference for or against taking HRT, and on decisions whether or not to participate in a hypothetical clinical trial. The second was to investigate the relationship between individual differences (such as desire for involvement in own medical care and locus of control) on the decision about whether or not to enter a clinical trial.