SEARCH

SEARCH BY CITATION

Keywords:

  • clinical trials;
  • Phase I;
  • decision making;
  • informed consent;
  • quality of life

Abstract

  1. Top of page
  2. Abstract
  3. A Typology of Patients' Expectations
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

BACKGROUND

Patients in Phase I clinical trials sometimes report high expectations regarding the benefit of treatment. The authors examined a range of patient characteristics to determine which factors were associated with greater expectations of benefit from Phase I trials.

METHODS

Participants were adult patients with cancer who had been offered participation in Phase I studies and had decided to participate. Patients completed interviewer-administered surveys before initiation of treatment. Physicians assessed Eastern Cooperative Oncology Group performance status for each patient. Statistical analyses (Pearson product moment correlation and t tests) used multiple imputation to account for missing data.

RESULTS

Overall, 593 patients who were offered participation in Phase I trials were contacted, and 328 patients agreed to participate in a study of decision making by cancer patients. Of these, 260 patients (79%) enrolled in a Phase I trial. Patients' expectations regarding the chance that their disease would be controlled with experimental therapy were unrelated to age, gender, living situation, education level, or functional status. Expectations were correlated positively with beliefs about the benefit of standard therapy and the maximum benefit patients may experience from experimental therapy. Greater expectations of benefit were associated with better health-related quality of life, stronger religious faith, optimism, relative health stock, monetary risk seeking, and poorer numeracy.

CONCLUSIONS

Expectations expressed as beliefs in personal outcomes may be related more to quality of life and personality variables than to patients' knowledge or functional status. Whether such expectations are accurate reflections of knowledge has important implications for evaluating the informed consent process. Cancer 2003;98:166–75. © 2003 American Cancer Society.

DOI 10.1002/cncr.11483

A key element of the informed consent process is the patient's correct understanding of the treatment or clinical trial in which he or she is participating. 1 For this reason, research on patients' knowledge about treatment is critical for ensuring the adequacy of informed consent. Patients' perceptions of their chances of benefit from treatment or from participation in a clinical trial has been a source of special concern. Patients with cancer who are making treatment decisions frequently overestimate the chance of benefit from particular treatments. 2 For example, this was found to be true for many patients who were considering whether to receive adjuvant therapy for breast carcinoma. Siminoff and colleagues 3 surveyed patients and their physicians regarding the risk of recurrence with adjuvant therapy. Sixty percent of the patients provided estimates of their chance of cure that were at least 20% greater compared with the estimates provided by physicians. Even among patients whose physicians provided numerical estimates of the chance of cure, close to half of the patients overestimated the chance of benefit by 20% or more.

Lee and colleagues 4 recently examined discrepancies between patients' and physicians' estimates of the success of stem cell transplantation. Patients who agreed to undergo transplantation were surveyed prior to treatment, as were their physicians. Patients with less severe disease provided estimates that agreed reasonably well with those of their physicians. In contrast, patients with moderate to severe disease provided estimates of the chance of cure from stem cell transplantation that were markedly higher compared with the estimates provided by physicians.

For patients considering participation in a trial of a new experimental agent, there is great potential for overestimating benefit. The primary goal of a Phase I trial frequently is the determination of toxicity or a biologic endpoint other than therapeutic antitumor effect. Historically, the response rate of Phase I trials has been ≤ 5%. 5 Presumably, if the consent process for Phase I trials is successful, then patients should understand the very small chance of personal benefit and the significant potential for experiencing toxicity. Given such an understanding, one would expect that the chief motivation for participation in Phase I trials would be altruistic. A review of the literature 5 suggests, however, that most patients say they are motivated primarily by the hope of personal benefit. In a recent study by our group, 6 patients who were offered participation in a Phase I trial were asked about their chances that the treatment would control their cancer. The median response was 60%, with three-fourths of the patients reporting a chance of benefit of at least 50%—a value 10 times greater than the frequency of benefit typically resulting from Phase I trials. Other studies have found similarly high expectations of benefit. 2, 7

Studies like these present clear evidence that patients sometimes report expectations regarding the benefit of treatment that are regarded as unrealistically high by the medical community. It is unclear, however, what these reported expectations mean to the patients. In a preliminary step toward answering this question, we examined a broad range of patient characteristics to determine which factors were associated with greater expectations of benefit from a Phase I trial. In doing so, we hope to understand better the meaning of patients' reports of expectations of benefit.

A Typology of Patients' Expectations

  1. Top of page
  2. Abstract
  3. A Typology of Patients' Expectations
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Patients' expectations take the form of statements about uncertain events (i.e., positive outcome from treatment). In attempting to understand these statements, it is important to be sensitive to the fact that there are different ways of expressing uncertainty, corresponding to different philosophical interpretations of the idea of probability. In the history of the philosophy of probability, a fundamental distinction has been made between two types of probability. Whereas the two types go by various names, 8 we adopt Hacking's scheme of dividing statements of uncertainty into frequency-type and belief-type statements. 9, 10 A frequency-type statement refers to the relative frequency of outcomes for a group of patients. An example of such a statement might be, “Eighty out of every 100 patients will benefit from the treatment I am receiving.” A belief-type statement, conversely, refers to how confident a person is that a particular outcome will occur for a particular patient. An example might be, “I am 80% confident that I will benefit from this treatment.”

These two types of probability differ greatly from one another. To understand this, imagine a patient who says, “I understand that 10 out of 100 patients will benefit, but I am 80% confident that I will be 1 of the 10.” This statement is not illogical or contradictory, because frequency-type ideas (e.g., “X out of 100 will benefit”) do not mean the same thing as belief-type ideas (e.g., “I am Y% confident that I will be 1 of the X”). Understanding what patients mean requires identifying the type of probability they are using when communicating their chances of benefit. From an ethical standpoint, whether there should be concern about a patient who expresses 80% confidence in a good outcome depends entirely on how such statements are generated and what they mean to the patient (our unpublished data).

In the current study, we asked patients with cancer about their expectations regarding the chance that they would benefit personally from an experimental treatment. Therefore, our inquiry was more likely to encourage a belief-type response than a frequency-type response. The latter may be encouraged by asking patients, for example, to indicate how many patients out of 100 will benefit from the treatment. Although patients may have understood our query differently than we intended, we assumed that most patients would respond with belief-type expectations. The current study constitutes a first step toward understanding those expectations by examining the correlates of belief-type expectations about benefit from an experimental therapy among patients who agreed to participate in Phase I trials. The variables we examined included demographic information, health-related quality of life (HRQOL), numeracy, optimism, monetary risk preference, spirituality, and other variables that have been related to patients' decision making.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. A Typology of Patients' Expectations
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Patients and Procedures

Eligible patients were adults with cancer who had been offered participation in Phase I studies and had decided to participate. Patients who already had initiated treatment were excluded. Eligibility criteria included 1) advanced malignancy for which either there was no standard effective therapy or standard therapy had failed, 2) age 18 years or older, 3) life expectancy ≥ 3 months, and 4) an Eastern Cooperative Oncology Group (ECOG) performance status of 0–2 (i.e., the patient was ambulatory at least 50% of the time). Completion of ascertainment of patients who chose to participate in a Phase I study was confirmed by comparison with protocol office enrollment lists. Physician consent was obtained prior to patient contact. Written consent to participate in a study was then obtained from the patients. Patient baseline surveys were conducted either in person or by telephone. The study design and survey instruments were approved by the Institutional Review Boards at each participating study site and at the data center (Fox Chase Cancer Center, Philadelphia, PA; Duke University Medical Center, Durham, NC; University of Maryland, Baltimore; Georgetown University, Washington, DC; and Northwestern University, Chicago, IL). Recruitment took place over 18 months.

Measures

A multidisciplinary team that included medical oncologists, nurses, psychologists, clinical economists, and a medical ethicist developed the questionnaires. Domains were developed based on study objectives, literature review, pilot tests, and relevant past studies. 7, 11–21 The survey measures used for this analysis are described below.

Demographic characteristics

Patients reported their own age (in years), gender, level of education, marital/partnered status, and living situation. Education was recoded as a dichotomous variable, with 1 indicating a college education and 0 indicating no college education. Living situation also was recoded as a dichotomous variable (with 1 indicating living alone and 0 indicating other), as was marital/partnered status (with 1 indicating married/partnered and 0 indicating other).

Expectations of benefit and toxicity

To measure patients' expectations of experimental or standard treatments, we asked respondents to place a mark on a horizontal bar ranging from 0% to 100% to describe the likelihood of health outcomes associated with the treatments. Questions about benefit asked the patients to indicate the probability that a treatment would control their cancer. Questions about harm asked the patients to indicate the probability that a treatment would cause them to experience severe adverse reactions. Patients also were asked to indicate what they perceived as the maximum benefit that any patient may experience as a result of experimental cancer therapy. Possible responses to this question were “cancer totally cured;” “reduction of the number or size of tumors;” “cancer is controlled (i.e., will not get any worse; stable disease);” “will not shrink cancer, but will improve symptoms;” and “will not shrink cancer or improve symptoms;” as well as “not applicable,” “don't know,” and “unsure.” This variable was coded as a continuous variable ranging from 1 (greatest expectation of benefit or “cancer totally cured”) to 5 (no expectation of benefit or “will not shrink cancer or improve symptoms”).

Numeracy

A single item assessed how well patients understood a statement about the relative frequency of benefit from a hypothetical treatment. Patients were asked to imagine that their physician described a treatment that controlled cancer in “40% of cases like yours.” Using a multiple-choice format, patients were asked to select the statement that best described what their physician meant. The correct response was, “For every 100 patients like me, the treatment will work for 40 patients.” Other response options included, “the doctor is 40% confident that the treatment will control my cancer;” “the new treatment will reduce my disease by 40%;” and “I am not sure what this information means.”

HRQOL

We assessed three measures of HRQOL. To measure patients' overall health perceptions, we used the General Health Perceptions scale of the Medical Outcomes Study Short Form 36 (SF-36). 21 Overall physical and mental health were measured using the Physical Component Summary (PCS) score and the Mental Component Summary (MCS) score of the Medical Outcomes Study Short Form 12 (SF-12). 20 All 3 measures of HRQOL produce scores ranging from 0 to 100, with higher scores reflecting better health.

Performance status

To index the functional status of the patients, we obtained the physician-rated ECOG performance status for all patients whose physician completed a physician survey. ECOG performance status ranges from 0 (no symptoms) to 4 (bed-bound). To be eligible for the current study, patients were required to have an ECOG performance status of 0–2. Due to the sparse frequencies, we recoded ECOG status as no symptoms (ECOG performance status, 0) or symptoms (ECOG performance status, 1 or 2).

Decisional Conflict Scale

To measure several aspects of decisional conflict, we used the previously validated Decisional Conflict Scale, 15 which uses five-point Likert scales. The Uncertainty subscale (three items) measures the extent of uncertainty the person is experiencing regarding the decision. A second subscale (nine items) addresses feelings of uncertainty. The Effective Decision-Making subscale (four items) measures perceptions that a decision is well-informed and likely to be adhered to. Finally, the Overall Decisional Conflict score sums the items in the subscales for an aggregate score. For all subscales, higher scores indicate greater decisional conflict.

Spirituality

Strength of faith was measured by asking patients, “How strong would you say your religious beliefs are?” Responses ranged from 1 (not very strong) to 4 (very strong). Comfort of faith was measured by asking patients, “How comforting to you are your religious or spiritual beliefs?” Responses ranged from 1 (not at all comforting) to 4 (very comforting). For both questions, responses of not applicable were considered to reflect a level of strength or comfort lower than not very strong or not at all comforting. Responses were coded such that higher scores reflected greater strength or comfort of faith.

Optimism

We assessed self-reported dispositional optimism using a single item, which asked patients to rate themselves on a 7-point Likert scale ranging from 1 (very optimistic) to 7 (very pessimistic). For the purposes of the analysis, we reverse-keyed this item, so that higher scores reflected greater optimism.

Monetary risk preference

To measure monetary risk preference, the interviewers asked patients to identify the lowest dollar amount—ranging from $100 to $800—they would be willing to accept instead of a lottery ticket that offered a 50% chance of winning $1000. The expected value of this gamble is $500. Thus, we recoded responses into risk-averse (≤ $450), risk-neutral (> $450 and < $550), and risk-seeking (≥ $550).

Control Preferences Scale

The extensively studied Control Preferences Scale 11 was used to assess patients' wishes regarding shared decision making. Patients selected from among five descriptions of the relative roles of patient and physician, ranging from “I prefer to make the final selection about which treatment I will receive” (option A), to “I prefer that my doctor and I share responsibility for deciding which treatment is best for me” (option C), to “I prefer to leave all decisions regarding my treatment to my doctor” (option E). Responses were coded from 1 to 5, with higher values reflecting a preference for less participation in the decision.

Relative health stock

Health stock is a concept from the health economics literature pioneered by Grossman. 22 In the Grossman model, health stock is an individual's current health, including her expected longevity and quality of life. With each day of life, an individual reduces her health stock by 1 day. Barring disease, accident, and injury, an individual will draw down her natural endowment of health stock until she reaches old age and dies of natural causes. In other words, health stock represents quality-adjusted life expectancy. Illness reduces an individual's health stock substantially. Our concept of relative health stock, elaborated elsewhere by Gaskin et al., 14 refers to the patient's health stock due to her current cancer diagnosis in relation to her health stock prior to the current diagnosis of cancer or change in condition. In our questionnaire for the current study, we asked patients to think of the fullness of their life—prior to their most recent diagnosis—as a whole pie. Using a diagram containing eight pies ranging from a full pie to an empty pie, we then asked patients to identify the portion of pie that best represented how much of the fullness of their life was lost due to their recent cancer diagnosis or change in condition. Responses were converted to a 0–100 scale, with 0 denoting that all of life's fullness had been lost, and 100 denoting that none of life's fullness had been lost.

Importance of quality versus length of life

A five-point, Likert-type scale was used to determine the extent to which patients valued quality of life versus length of life, with choices including quality of life is all that matters, quality of life is more important, both are equally important, length of life is more important, and length of life is all that matters.

Previous experimental therapy

Patients indicated that they received prior experimental therapy (e.g., clinical trial) by placing a tick mark next to the item on a list of types of cancer treatments.

Statistical Analysis

Missing data were handled using multiple imputation by means of a full Bayes multivariate normal imputation model containing all study variables. 23 That is, a full Bayes multivariate normal model of the data with noninformative priors was used to generate posterior densities from which values were drawn to replace missing values. For any missing value, five values were drawn from the posterior distribution of the imputation model, such that five different versions of the data set were created. All statistical analyses (e.g., t tests, chi-square tests) were conducted separately by imputation sample, and the results were combined using rules described by Schafer. 23

Demographic characteristics were calculated for the original (nonimputed) sample. Means and standard errors of key study variables were calculated for complete data only as well as for the multiply imputed sample. The fraction of missing information 23 also was computed for each of the key study variables. Missing information is more informative than the percentage of respondents who miss a variable, because the former refers to how much information about a variable has been lost due to the missing data. For example, 30% of the sample may be missing variable Z, but Z is associated strongly with the other variables in the study. So long as some other variables are observed, variance in Z will be recoverable from variance in variables that are correlated with Z; thus, the fraction of missing information on Z will be < 30%.

The correlations between expectations of benefit from experimental therapy and other study variables were ascertained using the Pearson product-moment correlation and t tests. Nonparametric tests were conducted on the nonimputed data set to confirm the direction and significance of all correlations found using the parametric tests. Because parametric analyses are required for analyzing the multiply imputed data set, we report only the parametric results (Pearson correlation coefficients [r] and t tests). Potential correlates of expectations of benefit were selected on the basis of a priori theory; thus α = 0.05 was the criterion used for statistical significance in all t tests, and 95% confidence intervals were used in the case of all Pearson correlation coefficients (r). 24

RESULTS

  1. Top of page
  2. Abstract
  3. A Typology of Patients' Expectations
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Overall, 593 patients who were offered participation in Phase I trials were contacted, and 328 patients agreed to participate in a study of cancer patients' decision making. Of these, 260 patients (79%) chose to enroll in a Phase I trial (144 men [55%] and 116 women [45%]). This group constituted the sample for the study. Physician-rated ECOG performance status was available for 199 of 260 patients in the sample (76%).

Table 1 shows the demographic and clinical characteristics of the sample. The mean ± standard deviation age was 57.6 years ± 12.7 years. The majority of the patients (85%) were Caucasian. Sixty-four percent of patients had attended at least some college. The majority of patients (88%) did not live alone. Sixteen percent of patients reported that they already had received experimental therapy.

Table 1. Characteristics of Patients Choosing to Participate in Phase I Clinical Trialsa
CharacteristicNo. of patientsParticipating patients (%)
  • SD: standard deviation.

  • a

    Values are expressed as number of patients (percentage) unless otherwise indicated.

Age (yrs)255
 Mean ± SD57.6 ± 12.7
 Median (interquartile range)58.0 (49.0–68.0)
 Minimum-maximum25.0–80.0
Female gender260116 (44.6)
Caucasian258219 (84.9)
Some college education256164 (64.1)
Living situation257
 Alone 32 (12.5)
 Not alone225 (87.5)
Married/partnered255197 (77.3)
Previous experimental therapy260 42 (16.2)

A total of 54 Phase I trials of different varieties were active during the study period. These included studies of investigational agents alone (37%), investigational and standard agents (28%), and standard agents in combination (35%). Of the 20 studies of investigational agents alone, 16 studies (80%) involved cytostatic or biologic therapies.

Table 2 shows the descriptive statistics for the main study variables for both the original sample and the multiply imputed sample on which all subsequent analyses were performed. A comparison of the original and imputed numbers indicates that the multiple imputation did not distort the means or variances of the original variables. The fraction of missing information created by missing data generally was low, with the exception of the ECOG performance status, a variable that was obtained only for patients whose physicians completed the physician survey. Note that the multiply imputed sample retained the same mean and standard error for ECOG performance status as the original sample.

Table 2. Descriptive Statistics for the Primary Study Variablesa
VariableNo. of patientsOriginal dataFraction of missing informationImputed data
  • SF-36: Short Form 36; SF-12: Short Form 12; PCS: Physical Component Summary; MCS: Mental Component Summary; ECOG: Eastern Cooperative Oncology Group.

  • a

    Values are expressed as mean (standard error) unless otherwise indicated.

  • b

    Values are expressed as the mean proportion (standard error).

  • b, c

    The variable was coded as follows: 1, 100–400 (risk-averse); 2, 450–550 (risk-neutral); and 3, 600–800 (risk-seeking). Raw scores ranged from 100 to 800 in increments of 50.

Demographic characteristics    
 Age (yrs)25557.62 (0.80)0.0257.61 (0.79)
 Female genderb260 0.45 (0.03)0.00 0.00 (0.00)
 Some college educationb256 0.64 (0.03)0.02 0.64 (0.03)
 Percent not living aloneb257 0.88 (0.02)0.01 0.88 (0.02)
Treatment expectations    
 Benefit from experimental therapy23464.58 (1.66)0.1064.52 (1.57)
 Harm from experimental therapy23336.28 (1.75)0.1036.82 (1.71)
 Benefit from standard therapy22338.79 (1.91)0.1439.73 (1.86)
 Harm from standard therapy22856.18 (2.01)0.1255.40 (1.98)
 Maximum potential benefit from experimental therapy225 1.98 (0.06)0.13 2.04 (0.07)
Numeracy query answered correctly250 0.72 (0.03)0.04 0.72 (0.03)
Quality of life    
 SF-36 General Health Perceptions24853.97 (1.41)0.0553.87 (1.40)
 SF-12 PCS score23940.06 (0.73)0.0840.17 (0.70)
 SF-12 MCS score23948.87 (0.64)0.0848.82 (0.66)
ECOG performance status199 1.62 (0.04)0.23 1.63 (0.04)
Decisional Conflict Scale    
 Decisional Uncertainty253 7.36 (0.18)0.03 7.33 (0.19)
 Factors in Decision-Making253 6.09 (0.13)0.03 6.06 (0.13)
 Effectiveness of Decision-Making248 5.17 (0.10)0.05 5.05 (0.12)
 Overall Decisional Conflict25518.39 (0.32)0.0218.43 (0.32)
Spirituality    
 Strength of faith236 3.29 (0.06)0.09 3.25 (0.06)
 Comfort of faith232 3.37 (0.06)0.11 3.32 (0.06)
Optimism247 2.15 (0.08)0.05 2.15 (0.08)
Monetary risk preferencec245 1.67 (0.05)0.06 1.68 (0.05)
Control preference255 1.72 (0.05)0.02 1.72 (0.05)
Relative health stock23561.06 (1.77)0.1060.65 (1.67)
Quality vs. length of life253 2.79 (0.04)0.03 2.79 (0.04)
Previous experimental therapy260 0.16 (0.02)0.00 0.16 (0.02)

The distribution of expectations regarding benefit from experimental therapy is shown in Figure 1. Two aspects of the distribution are noteworthy. First, the distribution was skewed toward higher expectations, with nontrivial numbers of patients indicating an 80–100% chance of benefit. The other striking observation is the large number of patients who gave a response of 50%.

thumbnail image

Figure 1. Distribution of expectations regarding benefit from experimental therapy.

Download figure to PowerPoint

The correlations between expectations of benefit from experimental therapy and other patient characteristics are shown in Tables 3 and 4. Patients who provided higher estimates of the chance of benefit from experimental therapy had higher expectations of benefit from standard therapy and expected a greater degree of maximum benefit (e.g., “cancer totally cured”). Patients who provided higher estimates of the chance of benefit also tended to have better SF-12 PCS and SF-36 General Health Perception scores as well as higher levels of faith, optimism, and relative health stock. They also tended to be risk-seeking more than patients with lower expectations of experimental therapy. Finally, patients with higher expectations of benefit from experimental therapy were less likely to answer the numeracy question correctly.

Table 3. Correlation of Patient Characteristics with the Probability of Benefiting from Experimental Therapy: Continuous Variables
CharacteristicContinuous variable
rLower limitUpper limit
  1. r: correlation coefficient; SF-36: Short Form 36; SF-12: Short Form 12; PCS: Physical, Component Summary; MCS: Mental Component Summary.

Age−0.02−0.140.10
Treatment expectations   
 Harm from experimental therapy−0.07−0.190.06
 Benefit from standard therapy0.380.270.48
 Harm from standard therapy−0.06−0.180.07
 Maximum potential benefit from experimental therapy0.350.190.49
Quality of life   
 SF-36 General Health Perceptions0.280.160.39
 SF-12 PCS score0.150.020.27
 SF-12 MCS score0.06−0.080.19
Decisional Conflict Scale   
 Decisional Uncertainty−0.09−0.210.04
 Factors in Decision-Making−0.03−0.160.09
 Effectiveness of Decision-Making−0.08−0.210.06
 Overall Decisional Conflict−0.09−0.220.04
Spirituality   
 Strength of faith0.170.040.29
 Comfort of faith0.10−0.030.23
Optimism0.260.140.37
Relative health stock0.150.010.28
Quality vs. length of life0.06−0.070.19
Table 4. Correlation of Patient Characteristics with the Probability of Benefiting from Experimental Therapy: Categorical Variables
CharacteristicMean expectation (SE)P value
  1. SE: standard error; ECOG: Eastern Cooperative Oncology Group.

Gender  
 Female63.41 (2.38)0.53
 Male65.41 (2.10)
Living situation  
 Not alone64.89 (1.69)0.53
 Alone61.87 (4.50)
Previous experimental therapy  
 Yes58.90 (4.05)0.15
 No65.60 (1.72)
ECOG status  
 No symptoms65.90 (2.43)0.96
 Symptoms63.83 (2.07)
Monetary risk preference  
 Risk-averse55.26 (3.85)0.014
 Risk-neutral56.55 (4.49)
 Risk-seeking77.50 (4.98)
Control preference  
 Prefers to take more control60.36 (3.94)0.26
 Prefers to share responsibility with physician60.46 (4.52)
 Prefers that physician take more control69.97 (5.06)
Numeracy question answered correctly  
 Yes62.35 (1.86)0.038
 No70.20 (3.00)
Some college education  
 Yes63.45 (2.01)0.48
 No66.00 (2.60)

DISCUSSION

  1. Top of page
  2. Abstract
  3. A Typology of Patients' Expectations
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Our overall goal was to identify characteristics of patients that were related to patients' expectations about the benefit of participating in a Phase I clinical trial. In particular, we examined factors that were associated with responses to belief-type questions about the chance of benefit, in which patients are likely to express their confidence that they will experience benefit (e.g., “There is a 60% chance that an experimental therapy will control my cancer.”). Belief-type statements contrast with frequency-type statements, which are statements about the relative frequency of some event in the world (e.g., “An experimental therapy will control cancer in 60 out of every 100 patients.”).

Consistent with past research, patients who agreed to participate in Phase I trials reported high expectations regarding the chance that their cancer would be controlled. 2, 7 This was true regardless of age, gender, living situation, and level of education. The shape of the distribution of patients' expectations (Fig. 1) demonstrated an interesting finding in addition to the generally high values. That is, the most frequently reported single value was 50%. There are several reasons why this may have been the modal value. First, some patients may not have thought about their chances in such numerical terms and simply guessed the middle value. Another possibility is that some patients did have a rationale for selecting 50%, corresponding to the idea that “it will either work or not work, so I'd say my chances are 50:50.” Future qualitative work must be conducted to explore the justifications people give for selecting 50%. At the very least, this finding suggests that the selection by these patients of a quantitative rating of confidence is not a straightforward process. This general conclusion is supported by our other work, 2 in which greater than one-third of patients who agreed to participate in Phase I trials were unable to provide numerical estimates of their life expectancy.

One possible reason why some patients estimate their chances of benefit to be so high is that those patients have a milder benefit in mind. For example, one patient may interpret benefit as minimizing the symptoms of cancer, whereas another patient may interpret benefit as curing the cancer. On the assumption that less dramatic effects are more probable than very dramatic effects, the first patient would provide a higher likelihood of benefit than the second patient. If this is true, then expectations of the chance of benefit would correlate negatively with the extent of benefit the patients believe one may receive from treatment. In fact, we found the opposite. Patients who thought the chance of benefit was higher were inclined to believe that the type of benefit would be greater (e.g., curing cancer rather than simply controlling symptoms), although the size of the correlation (r = 0.21) was relatively small. This is an important finding, because it eliminates an alternative explanation for high patient expectations—that of differing interpretations of benefit. Moreover, it suggests that patients who are most confident that they will experience a benefit may have more dramatic benefits in mind.

Another alternative explanation for the high ratings is that there was likely some selection bias. Just over half of the patients who were approached actually agreed to participate. Perhaps the patients who agreed to participate in our study were more optimistic compared with patients who declined participation. Two points should be kept in mind when considering the possible influence of selection bias. First, although selection bias may have affected the shape of the overall distribution, it would not alter the fact that many of the patients had high expectations. Second, regardless of whether the distribution we obtained is subject to selection bias, we documented a wide range of expectations (0–100%) and, thus, still are able to describe the correlations between patient characteristics and expectations, which are discussed below.

Differentiating between belief-type and frequency-type probabilities allows consideration of how some factors may be related differentially with patients' expectations. We expected that knowledge-related factors would not be associated strongly with the belief-type probabilities assessed in our study. We found that a numeracy item measuring the ability to understand an aggregate statement about treatment outcomes demonstrated a small correlation with belief-type expectations, amounting to a difference in mean expectations of only 8 percentile points. Another knowledge-related variable—level of education—was unrelated to patients' expectations of benefit. The lack of strong influences of knowledge-related factors is understandable. A patient may have perfect knowledge about the frequency-type probability of benefit—for example, that 5 of every 100 patients will benefit—but the patient's confidence that she will be 1 of the 5 beneficiaries likely is influenced by other factors. Although we did not hold strong hypotheses about which factors would be correlated, we expected they would have more to do with the patient's religious convictions and personality. Indeed, we found that both the strength of patients' religious beliefs and dispositional optimism were associated positively with patients' confidence in a good outcome. Similarly, patients' monetary risk preferences and relative health stock were associated with patients' belief-type expectations.

Patients' expectations of benefit from experimental therapy demonstrated the strongest correlation with patients' expectations of benefit from standard therapy. Together with the fact that self-reported HRQOL was related to expectations of benefit from experimental therapy, the correlation suggests that patients with high expectations see themselves as generally strong and healthy and likely to benefit from different treatments. Similar correlations were found by Cheng et al. 2 The self-perception of good health in our study, however, did not correspond to physical health, as evaluated by patients' physicians; physician-rated ECOG performance status was unrelated to patients' expectations. ECOG performance status—especially the dichotomized version that we used—may not be as sensitive to differences in health status. Still, if expectations of benefit were associated with actual physical health status, then the expectation would be to observe more than the 2 percentile difference that we found between active and restricted ECOG status.

The current study had several limitations. First, we assessed a broad array of constructs relevant to patient decision making. In doing so, however, we often had to use brief, single-item measures. Such measures can be unreliable and, thus, may have limited our ability to find substantial associations between constructs. 25 This limitation likely explains the relatively modest correlations observed in our study. Now that some key patient factors have been identified, a more focused data collection effort is needed to assess the constructs with more sensitive measures that will improve the ability to find correlations. Second, we were confined to quantitative measures. Future qualitative research is needed to understand the meaning of patient expectations to the patients and how they arrive at the responses they provide in quantitative studies like these. Despite the broad measurement strategy, there remain important aspects of the patients' experiences that we did not capture, including the recent clinical history of the patient (e.g., recent increases in symptoms, failure of a standard treatment regimen, etc.) that may be related to patients' expectations of benefit. The influence of other people in the patient's environment, such as the recruiting clinical investigator or even the research assistant who administered our questionnaire, also may have affected patients' expectations. Although our emphasis has been on patient characteristics, the importance of others' remarks cannot be underestimated. Another limitation is that we queried only belief-type expectations and, hence, could not compare the correlates of belief-type expectations with frequency-type expectations.

Finally, because all of the patients studied were from large cancer centers located in major cities, it is possible that our results may have been different if patients had been recruited from other geographic regions or from smaller cancer research centers. However, because of the complex and specialized clinical, laboratory, and regulatory aspects of Phase I clinical investigations, such studies commonly are conducted at cancer research centers of excellence. In this sense, all sites that conduct Phase I trials will share certain characteristics in terms of resources, personnel, and mission.

The results reported here have helped to characterize patient factors associated with belief-type expectations. To determine when and how patient expectations may be harmful, further studies are needed to understand the psychosocial consequences of having high expectations. This is difficult to study in patients participating in Phase I trials, because such patients, including those in the current study, tend not to live long enough for researchers to obtain adequate follow-up assessments of mental health. Future research also is needed to understand the nature of frequency-type expectations of benefit. Frequency-type expectations, as we have suggested, may be more likely than belief-type expectations to reflect patients' knowledge in the context of the informed consent process. Such information would be critical for evaluating the adequacy of informed consent.

Acknowledgements

  1. Top of page
  2. Abstract
  3. A Typology of Patients' Expectations
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

The sites, principal investigators, and project coordinators for this study included the following: Fox Chase Cancer Center, Philadelphia, PA (principal investigator, Neal J. Meropol, M.D.; study coordinators, Joanne Buzaglo, Ph.D., Sandra Corbett, M.S., and Jennifer Driscoll, M.S.); Georgetown University, Washington, DC (principal investigator, Caroline B. Burnett, Sc.D., R.N.; study coordinator, Shakira Washington, M.H.A.); Lurie Cancer Center, Northwestern University, Chicago, IL (principal investigator, Al B. Benson III, M.D.; study coordinator, Kimberly Smith, M.S.); University of Maryland, Baltimore, MD (principal investigator, David Van Echo, M.D.; study coordinator, Jennifer DeSanto, R.N., M.S.); and Duke University Medical Center, Durham, NC (principal investigator, Kevin A. Schulman, M.D.; study coordinator, Liana D. Castel, M.S.P.H.).

The authors thank Damon Seils for general advice and editorial assistance.

The views expressed herein do not necessarily reflect the views of the National Cancer Institute.

Dr. Marshall has received speaker's honoraria from Roche (Basel, Switzerland), Pharmacia (Gaithersburg, MD), and Sanofi-Synthelabo (Paris, France). Dr. Marshall also has worked as a consultant for Boehringer Ingelheim.

REFERENCES

  1. Top of page
  2. Abstract
  3. A Typology of Patients' Expectations
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES
  • 1
    Beauchamp TL, Childress JF. Principles of biomedical ethics, 5th edition. New York: Oxford University Press, 2001.
  • 2
    Cheng JD, Hitt J, Koczwara B, et al. Impact of quality of life on patient expectations regarding Phase I clinical trials. J Clin Oncol. 2000; 18: 421428.
  • 3
    Siminoff LA, Fetting JH, Abeloff MD. Doctor-patient communication about breast cancer adjuvant therapy. J Clin Oncol. 1989; 7: 11921200.
  • 4
    Lee SJ, Fairclough D, Antin JH, Weeks JC. Discrepancies between patient and physician estimates for the success of stem cell transplantation. JAMA. 2001; 285: 10341038.
  • 5
    Daugherty CK. Impact of therapeutic research on informed consent and the ethics of clinical trials: a medical oncology perspective. J Clin Oncol. 1999; 17: 16011617.
  • 6
    Meropol NJ., Schulman KA, Weinfurt K, et al. Discordant perceptions of patients and their physicians regarding Phase I trials [abstract]. Proc Am Soc Clin Oncol. 2002; 21: 245a.
  • 7
    Daugherty C, Ratain MJ, Grochowski E, et al. Perceptions of cancer patients and their physicians involved in Phase I trials. J Clin Oncol. 1995; 13: 10621072.
  • 8
    Kahneman D, Tversky A. Variants of uncertainty. Cognition. 1982; 11: 143157.
  • 9
    Hacking I. The emergence of probability: a philosophical study of early ideas about probability, induction, and statistical inference. New York: Cambridge University Press, 1975.
  • 10
    Hacking I. An introduction to probability and inductive logic. Cambridge: Cambridge University Press, 2001.
  • 11
    Degner LF, Kristjanson LJ, Bowman D, et al. Information needs and decisional preferences in women with breast cancer. JAMA. 1997; 277: 14851492.
  • 12
    The EuroQol Group. EuroQol—a new facility for the measurement of health-related quality of life. Health Policy. 1990; 16: 199208.
  • 13
    Ganz PA, Lee JJ, Siau J. Quality of life assessment: an independent prognostic variable for survival in lung cancer. Cancer. 1991; 67: 31313135.
  • 14
    Gaskin DJ, Kong J, Meropol NJ, Yabroff KR, Weaver C, Schulman KA. Treatment choices by seriously ill patients: the Health Stock Risk Adjustment model. Med Decis Making. 1998; 18: 8494.
  • 15
    O'Connor AM. Validation of a decisional conflict scale. Med Decis Making. 1995; 15: 2530.
  • 16
    Stiggelbout AM, Kiebert GM, Kievit J, Leer JW, Stoter G, de Haes JC. Utility assessment in cancer patients: adjustment of time tradeoff scores for the utility of life years and comparison with standard gamble scores. Med Decis Making. 1994; 14: 8290.
  • 17
    Verhoef LC, de Haan AF, van Daal WA. Risk attitude in gambles with years of life: empirical support for prospect theory. Med Decis Making. 1994; 14: 194200.
  • 18
    Von Neuman J, Morgenstern O. Theory of games and economic behavior, 2nd edition. Princeton: Princeton University Press, 1947.
  • 19
    Ware J Jr., Kosinski M, Keller SD. SF-12: how to score the SF-12 Physical and Mental Health Summary Scales. Boston: The Health Institute, New England Medical Center, 1995.
  • 20
    Ware J Jr., Kosinski M, Keller SD. A 12-item short- form health survey: construction of scales and preliminary tests of reliability and validity. Med Care. 1996; 34: 220233.
  • 21
    Ware JE, Snow KK, Kosinski M, Gandek B. SF-36 health survey: manual and interpretation guide. Boston: The Health Institute, New England Medical Center, 1993.
  • 22
    Grossman M. On the concept of health capital and the demand for health. J Pol Econ. 1972; 80: 168199.
  • 23
    Schafer JL. Analysis of incomplete multivariate data. Boca Raton: Chapman & Hall/CRC, 1997.
  • 24
    Savitz DA, Olshan AF. Multiple comparisons and related issues in the interpretation of epidemiologic data. Am J Epidemiol. 1995; 142: 904908.
  • 25
    Ghiselli EE, Campbell JP, Zedeck S. Measurement theory for the behavioral sciences. San Francisco: WH Freeman and Company, 1981.