ASSESSMENT OF PATIENT PREFERENCES AMONG MEN WITH PROSTATE CANCER

ASSESSMENT OF PATIENT PREFERENCES AMONG MEN WITH PROSTATE CANCER

Vol. 159,158-163,January 1998 Printed in U . S d ASSESSMENT OF PATIENT PREFERENCES AMONG MEN WITH PROSTATE CANCER PETER C. ALBERTSEN, ROBERT F. NEASE...

754KB Sizes 0 Downloads 2 Views

Vol. 159,158-163,January 1998 Printed in U . S d

ASSESSMENT OF PATIENT PREFERENCES AMONG MEN WITH PROSTATE CANCER PETER C. ALBERTSEN, ROBERT F. NEASE, JR. AND ARNOLD L. POTOSKY From the Division of Urology,Zkpartmnt of Surgery, University of Connecticut Health Center, Farmington, Connecticut, Division of Geneml Medical Sciences, Department of Internal Medicine, Washington University School of Medicine, St. Louis, Missouri, and Applied Research Bmnch, Division of Cancer Prevention and Control, National Cancer Institute, Bethesda, Maryland

ABSTRACT

Purpose: We developed a self-administered paper based instrument to assess patient preferences quantified as utilities for common outcomes associated with the management of prostate cancer. Materials and Methods: A total of 50 patients was invited to test a self-administered paper based instrument designed to assess preferences for health outcomes associated with the management of localized prostate cancer. The 50 patients were selected from a group of 625 randomly identified men with prostate cancer who responded to a survey instrument designed to assess health related quality of life. The 50 patients selected for this pilot project were chosen because of the wide range of responses to the quality of life survey. Patient utilities were assessed for the 5 health states of overall quality of life, problems related to prostate cancer, and problems related to urinary, bowel and sexual dysfunction. Results: Patients were able to complete the assigned tasks. The self-administered instrument had high test-retest reliability. In addition results obtained from this instrument showed a correlation with results obtained from assessments using other instruments, including an analog scale, a computer based system known as U-Titer, a quality of life survey and the Health Utility Index:3. Conclusions: A self-administered paper based instrument can be used to assess patient utilities for health states associated with prostate cancer management. Results from the instrument tested appear to be reliable and valid, and are comparable to those obtained from other assessment techniques. A self-administered paper based instrument has distinct advantages when conducting large survey studies because it can be incorporated at relatively low cost. Kcr WORDS:prostate, prostatic neoplasms, quality of life, questionnaires

Patients with newly diagnosed, clinically localized prostate cancer face difficult decisions regarding management of the disease.1.2 Young men and men with high grade prostate cancer are likely to suffer considerable morbidity and excess mortality, and often seek aggressive therapy. Older men, especially those with low to moderate grade disease, are much less likely to suffer from disease progression and, therefore, are less likely to benefit fmm aggressive treatment. For these men the risks associated with treatment may outweigh any potential gain in longevity or deferred morbidity from prostate cancer progression. One way to explore this risk-benefit trade-off entails an assessment of patient preferences for key outcomes associated with prostate cancer management. The preferences can be expressed quantitatively as “utilities.” Health care researchers recommend that estimates of patient utilities should be included as part of cost-effectivenessanalyses that measure the impact of health care interventions on patient quality of life.3 Utilities are quantitative expressions of preference for potential health states and usually fall on a scale between 0 (death) and 1 (optimal health). Optimal health is an abstract concept that can be interpreted to mean “normal good health,” %freedom from disease, symptoms or dysfunction,” or “health as good as can be imagined.”3 While some individuals may attain or even exceed this definition, a population of average patients will always fall short. Few interventions can restore patients to a utility of 1.0. Researchers have

measured patient utilities in a number of medical disciplines, including cardiology, dermatology, renal medicine and transplantation, but to our knowledge have never formally assessed utilities among patients with localized prostate cancer.4-9 The primary objective of this pilot study was to develop a self-administered, paper based instrument to assess patient preferences, quantified as utilities, for common outcomes associated with the management of prostate cancer, includ- I ing overall health, a general health state encompassing problems associated with prostate cancer and specific health states encompassing problems associated with bowel, bladder and sexual dysfunction. Our goal was to measure the central tendency and the distribution of utilities for health states associated with men undergoing treatment for prostate cancer. In the absence of a gold standard, validation of any preference assessment instrument is difficult. We evaluated the performance characteristics of our instrument in 2 broad ways. We measured its reliability, defined as the degree to which the instrument produced similar responses in stable patients over time, and its validity, defined as the degree to which patient responses compared to other measures believed to be related to patient preferences for health states. Specifically we determined the association between responses to the self-administered paper based instrument with a computer assisted utility assessment technique, a Accepted for publication July 25, 1997. Supported by Grants NOl-CN05226 and N01-CN67005 from the previously validated general health status questionnaire National Cancer Institute. that relies on preference weights derived from the general 158

ASSESSMENT OF PATIENT PREFERENCES AMONG MEN WITH PROSTATE CANCER

169

population, patient self-ratings and patient responses to a disease-specific quality of life questionnaire. A self-administered instrument has distinct advantages when conducting large survey studies because it can be incorporated a t relatively low cost. Furthermore, a selfadministered paper based instrument does not require specialized software or a trained interviewer. Successful development of a simple self-administered, paper based instrument would allow researchers and clinicians to assess patient utilities for health states associated with prostate cancer using an instrument similar to more complex inperson time trade-off or standard “gamble”techniques.

years of his remaining life to be free of impotence) has a utility for impotence of 0.8 (8/10). The last portion of each assessment module asks patients to rate their current health using a different technique known as an analog scale. Patients are asked to rate their health on a scale ranging from 0 to 100 where 0 represents the worst health imaginable and 100 ideal health. In addition to the self-administered, paper based instrument we also assessed patient utilities using a computerized interviewing program known as U-Titer. U-Titer is an interactive computer program designed to provide a standard instrument for preference and utility assessment using a number of metrics, including time trade-off and analog techniques. To use U-Titer a patient sits a t the computer and answers a series of questions presented on the &n using a MATERIALS AND METHODS track ball. Graphical aids are available to ease the utility study popuzation. The study population comprised 50 assessment task. A trained researcher wm present at each patients participating in a larger population based study of session to provide help with use of the computer but wm 625 men designed to assess health related quality of life insm,&d to minimize patient interaction during the interamong men with prostate cancer. The larger population view. based cohort was identified using a rapid case ascertainti^^^^ ofutilitiesassociated with cancer ment system developed in Connecticut to identify patients agement can be obtained from the 2 sources of patientswho with newly diagnosed cancers. All men residing in Con- have been diagnosed with the disease or a general necticut who were diagnosed with prostate cancer from of individuals who may or may not have cancer. T~ October 1, 1994 to October 31, 1995 were identified, and a the results frorn interviewing patients sample consisting of all men under age 60 years, all black with cancer with preference weights frorn a men and a random sample of white men Over age 60 Years general population, we asked the 50 patients in our study to was identified for possible inclusion in the quality of life complete a self-ad-stered paper based version of the study. After obtaining informed consent from treating phy- Health utility Index: 3 (HUI:~)developed by ~~~l~ et d.10 sicians, identified men were asked to complete a question- me m : 3 is a health state classification design& to naire assessing quality of life associated with prostate estimate utilities for patientswith diverse health problems cancer management. A total of 625 men responded to the using an approach known as multi-attribute thmry.ii T~ survey. estimate these utilities patients completed the HUI:3, a selfThe 50 patients invited to participate in our study were administered 15-item survey, and responses were converted identified from this larger cohort. All patients participating to utility estimates using the p r o m 1 provided by the develin the Pilot Project had been diagnosed with Prostate Cancer opers of the HUI:3. This process involved 2 steps. Based upon at least 15 months before completing the pilot study instru- responses to the 15 items we initially assigned each patient ment and 6 to 12 months after completing the health related to a functional level within each of 8 attributes, including quality of life Survey. As a msult all Utility assessments Were vision (6 levels), hearing (6), speech (5), emotion (5), pain (5), performed at least 1 Yea after treatment completion. Pri- ambulation (6), dexterity (6) and cognition (6). Each selfm a d y for the convenience of participants we chose to invite reported health state was represented by a set of 8 numbers only Patients residing in the greater Hartford metropolitan (1 level value for each of 8 attributes). We then applied a area. To ensure inclusion of a wide range of patient prefer- multi-attribute utility function to these 8 level values to ences we grouped s u b j e e from the quality of life study into generate a single utility value for overall health state. To the 2 categories for each health state assessed in the quality determine the multi-attribute utility function, the developers of life questionnaire of those reporting “no,” ”very Small,’’ or of HUI:3 assessed time trade-off utilities for various health “small” problems and those reporting “moderate” or %if state scenarios among a cohort of healthy Canadians. problems. Thirty attending physicians granted permission to Patients required from 20 to 80 minutes to complete the contact a total of 84 patients to participate in the project, 50 assessments. This time reflects the time needed to complete of whom completed the initial study interview and question- the self-administered, paper based instrument, the time naire, and 47 who completed the followup questionnaire trade-off analysis using U-Titer and the HUI:3. All 50 patients completed the overall health and prostate cancer modmailed 1week later. Study instruments. The instrument developed for this ules. Only patients who acknowledged urinary, bowel or sexproject consists of 5 assessment modules and a question. ual problems were asked to complete time trade-off questions Each assessment module focuses on 1of 5 outcome measures for specific modules. Men without complaints were asked to of interest, including overall health, overall problems related skip to the next module. Thirteen men noted urinary probto prostate cancer, bladder problems, bowel problems and lems, 9 bowel problems and 32 sexual problems while comsexual dysfunction. The question asks men how they would pleting the time trade-off instrument. Statistical methods. We used nonparametric tests for most feel about spending the rest of their life with all of their of our statistical analyses. To assess test-retest reliability we current prostate related problems. Each of the 5 assessment modules includes a series of determined the Spearman rank correlation coefficient bequestions that ask a patient to imagine that he has 10 tween paper based utilities at test and retest during a 1-week years to live. The patient is then asked how many years, period, as well as the intraclass correlation coefficient. The ranging from no time to 10 years, that he is willing to intraclass correlation coefficient represents the proportion of sacrifice to achieve “ideal health” compared with his “cur- observed variation due to variation between rather than rent health.” This approach i s known as a time trade-off within subjecta.12 We also report mean change in utilities utility and is defined as the number of symptom-free years and 95% confidence intervals (CI) of mean changes. When divided by the number of years with symptoms at the point comparing utilities among ordered categories (for example of indifference. For example, a patient who is indifferent satisfaction with current health) we used S p e m ’ s rank between living with impotence for 10 years and living correlation coefficient. For comparieons between 2 groups we without impotence for 8 years (that is willing to give u p 2 used the Mm-Whitney U test. To asseas agreement be-

--

160

ASSESSMENT OF PATIENT PREFERENCES AMONG MEN WITH PROSTATE CANCER

tween different utility measures (paper based utilities and utilities assessed by U-Titer) we report Spearman’s rank correlation coefficient for the 2 measures. We also report mean difference between measures as well as 95% CI. RESULTS

Patient chumcteristics. Table 1 shows study population characteristics. Half of the study patients (25)reported concern or bother surrounding overall prostate cancer while the other half reported no concern or little concern surrounding prostate cancer management. Assessments were obtained from the health related quality of life questionnaire completed by the subjects 6 to 12 months before the interview that assessed patient utilities. All 50 patients were able to complete the 3 instruments. Three patients did not complete the self-administered, paper based retest. Utilities for symptoms. Table 2 displays the assessed time trade-off utilities for overall health, all current prostate troubles, current urinary troubles, current bowel troubles and current sexual dysfunction. Patients demonstrated a high median utility for overall health, overall prostate troubles, urinary troubles, bowel troubles and sexual dysfunction ranging from 0.950to 1.0. Reliability. To determine the stability of utilities over time we retested all patients 1week after the original interview using the same assessment tool. Table 3 lists changes and associated 95% CI for each of the 5 health states tested. Unfortunately paired observations are missing for some patients. Three men who did not acknowledge bladder or bowel dysfunction on the initial test did not complete the entire retest. Four men who completed the initial urinary section failed to complete that &ion on retest. One man who completed the bowel section did not complete that section on retest, while 6 men who initially completed the sexual dysfunction section did not complete that section on retest. Despite relatively small sample sizes for some sections the overall range of mean changes was minimal, ranging from -0.038 to +0.002. In each case the 95% CI includes 0.0, suggesting that the paper based instrument is reliable for estimating mean patient utilities. Validity. Assessing the validity of any utility assessment instrument is hampered by the absence of a gold standard. We approached this problem by assessing patient utilities using multiple different metrics and then correlating responses. Our instrument used time trade-off metrics and an analog scale. The computer based assessment instrument, U-Titer, uses time trade-off and analog scales. The HUL3 is a more easily administered, general health status questionnaire that asks patients to rate w e n t function. Utility estimates derived h m W : 3 are based on preferences of a general healthy population rather than a population of men

TABLE2. Time trade-off utilities for overall health and current prostate troubles, urinary troubles, bowel troubles and sexual dysfunction Overall health, relative to ideal health AU current prostate troubles, relative to no prostate troubles Current urinary troubles, relative to no urinary troubles Current bowel troubles, relative to no bowel troubles Current sexual dysfunction, relative to no sexual dvafunction

No.Pts. Mean M50 0.910 0.950 50 0.918 0.950 13

0.892

0.950

9

0.978

1.ooO

32

0.898

0.950 -

with prostate cancer. Utility estimates obtained from the self-administered instrument were compared with patient responses to questions designed to assess current functional status. In general patient responses were sensitive to overall degree of satisfaction with their treatment decision (see figure). Patients who were delighted, pleased or mostly satisfied with the treatment decision showed much higher median values and a narrower range of values compared with patients who described their degree of satisfaction as mixed, mostly dissatisfied, unhappy or terrible (p = 0.0006). The substantial variation in utilities among patients who described overall satisfaction with the treatment decision as mixed, mostly dissatisfied, unhappy or terrible was sufficiently large to cause overlaps with patients who described more positively their satisfaction with their prostate troubles. Correlation between different utility assessment techniques. Table 4 lists associations between estimates of patient utilities using the 2 different assessment techniques of time trade-off and an analog scale. One subject did not complete the analog scale for overall health, while a different subject did not complete the analog scale for urinary problems. Results show a correlation between techniques when assessing overall health and prostate troubles (Spearman’s rho = 0.48 and 0.5, respectively). No statistically significant associations were noted between the 2 techniques when assessing bladder, bowel or sexual dysfunction but this may have been the result of the small sample size assessed or the 6 to 12-month lag between the assessment of bowel, bladder or sexual dysfunction and the utility assessment of the health states. Correlation with a computer assessment tool. To determine whether the paper based instrument provides estimates similar to those obtained using a computer based approach we calculated the difference between utilities (U-Titer minus paper) for each outcome for each subject. Overall the average differences in mean values were minimal, ranging from - 0.023 to 0.026. Table 5 lists changes and associated 95% CI for each of the health states tested. In each case the 95% CI includes 0.0, suggesting little difference in mean utilities obtained using the 2 assessment tools. The TABLE1. Characteristies of the studv mDulatwn paper based instrument is not suficiently precise, howMedian we,yrs. (range) 66 (48-81) ever, to measure individual patient utilities and, therefore, No.race: white 45 should be used only in the context of research studies BlacL 5 designed to measure median utilities for large patient No. treatments received? cohorts. 27 Wd.¶lwry Correlation between patient utility estimates and the Radiation therapy 17 Hormonal therapy 12 HUZ:3. The mean utility for overall health based on rewatchful waiting 8 sponses to the HUI:3 was 0.885, while the median was 0.917. No. degree of bother (no, very small,s d m o d e r a t e , big): This score compares with 0.91 and 0.95, respectively, for the Worrvlmarrn over O I U S ~cancer ~ ~ 25 /25 self-administered paper based instrument. Although estimates of central tendency of utilities were similar, the asso23 n 5 ciation between the HUI:3 utility and the paper based utility * Twelve patienb had more than 1 treatment, 5 radiation therapy D ~ U S was modest and not statistically significant (Spearman’s hormones, 4 surgery plus hormones, 1 surgery and radiation, and 3 &r&ry, rank correlation coefficient 0.19, p = 0.18). radiation and hormones. Correlation with self-reported health related quality of life. t Two patients did not respond to the sexual dysfunction question on the health related quality of life survey. TO determine whether patient utility estimates correlated

ASSESSMENT OF PATIENT PREFERENCES AMONG MEN WITH PROSTATE CANCER TABLE3. Test-retest reliability

of

No, pts. Overall health, relative to ideal healtb All current prostate troubles, relative to no prostate troubles Current urinary troubles, relative to no urinary troubles Current bowel troubles, relative to no bowel troubles Current sexual dysfunction, relative to no sexual dysfunction

161

the self-administered paper based utility assessment during 2 weeks Intraclass Correlation Coefficient

Change

95%CI

0.83 0.91

0.002 -0.013

-0.026 to 0.030 -0.031 to 0.006

47 47

Rho 0.56 0.54

9

0.57

0.99

-0.017

-0.055 to 0.022

8

0.68

0.22

-0.038

-0.096 to 0.021

26

0.71

0.96

-0.004

-0.024 to 0.017

tive treatments for prostate cancer, researchers and clinicians must have information concerning 3 critical areas, including outcomes important to patients, frequency and severity of these outcomes and the value patients place on 1.0 these outcomes. Actuarial survival has been the traditional measure of success aRer cancer therapy but there is growing interest in measuring quality of life among men receiving 0.9 treatment for prostate cancer.13-15 Because prostate cancer typically strikes older men researchers have questioned 0.8 whether the marginal increase in longevity potentially achieved by aggressive therapy is offset by a decrease in quality of life.16-1s Formal assessment of patient preferences provides important insights when interpreting quality of life instruments. This information is especially critical when con0.6 structing computer simulations of patient outcomes or costO” utility assessments. 0.6 JOur pilot project was designed to test the feasibility of Mostly Mi.Mostly Delighted or obtaining estimates for patient utilities from patient perspecSalisfkJd d-ed. Pleased tive using a self-administered paper based instrument. We (n=W Unhappy, or (n = 17) assessed patient utilities using a time trade-off technique Temblr (n = 11) and a 0 to 100 analog rating scale to obtain estimates among men with prostate cancer. Our results indicate that patients p=0 . m are capable of completing an instrument containing up to 6 Time trade-offutility estimates by level of patient satisfactionwith modules designed to assess 5 health states associated with treatment. Box plot demonstrates median and 95% CI for utilities prostate cancer management, including overall quality of assessed. life, overall problems associated with the prostate and problems associated with bowel, urinary or sexual dysfunction. Despite small sample sizes for some modules, the modest TABLE4. Association between responses to time trade-off questions score changes on test-retest comparisons suggest that utilitywith the analog scale used in the self-administered paper based measurements for each of the 5 health states assessed are assessment tool reliable measures of patient preferences for different outOutcome NO.Reswonses Spearman’srho p Value comes associated with prostate cancer management. 49 0.48 o.Ooo9 Overall health To assess validity we compared results from the self0.0005 Prostate problems 50 0.50 administered paper based instrument with several other 0.35 Bladder problems 12 0.28 tools. With the exception of bowel problems results obtained 0.80 Bowel problems 9 -0.089 32 using time trade-off techniques were similar to those ob0.04 0.84 Sexual dysfunction tained using an analog scale. Furthermore, our findingsdemonstrated an association between patient utility estimates and self-reported descriptions of health related quality of life with self-reported evaluations of health related quality of life obtained from a survey administered 6 to 12 months previwe divided the pilot cohort into 2 groups according to their ously. The relatively poor correlations between time trade-off responses to the health related quality of life questionnaire. questions and an analog scale for urinary, bowel and sexual Patients describing overall quality of life problems or pros- dysfunction may be related to small sample size. Our results showed only minimal differences from those tate problems as “none at all” or “only a little” were placed in 1 group while those reporting “some” or “a lot” were placed obtained with U-Titer, a computer based utility assessment in another group. Table 6 reports median patient utility technique. Statistically significant correlations were noted estimates for overall health and problems related to prostate, for each of the assessment modules with the exception of urinary, bowel or sexual dysfunction. Significant differences bladder dysfunction and bowel dysfunction. Failure to in utility estimates were noted between the 2 groups for achieve statistical significance for these 2 modules may be a overall health (p = 0.048) and prostate problems (p = function of the small sample size tested. U-Titer has been 0.0021). No statistically significant differences were seen for shown to be a valid measure in many clinical domains, inurinary, bowel or sexual dysfunction, which may be related to cluding ischemic heart disease, psoriasis, breast cancer prethe small sample size of these groups. vention, coronary artery bypass grafting, low back pain, 08teoporosis prevention, mild hypertension, benign proStatic DISCUSSION hyperplasia, human immunodeficiency virus, atrial fibrillaTo construct decision analyses and cost-effectiveness anal- tion, cataract extraction, visual impairment, hearing impairyses that explore clinical outcomes associated with altema- ment, urinary incontinence, allogeneic blood transfusion and Mean: 0.968 Median: 0.950

Mean: 0.945 Median: 0.950

Mean: 0.812 Median: 0.850

-

i

1

162

ASSESSMENT OF PATIENT PREFERENCES AMONG MEN WITH PROSTATE CANCER TABLE5. Association between paper based and U-Titerbased time trade-off utilities at enrollment

Overall health, relative to ideal health All current prostate troubles, relative to no prostate troubles Current urinary troubles, relative to no urinary troubles Current bowel troubles, relative to no bowel troubles Current sexual dysfunction, relative to no sexual dvsfunction

50 50

0.33 0.40

0.02 0.0057

11

0.19

0.55

0.014

-0.010-0.038

9

0.09

0.81

-0.012

-0.066-0.042

31

0.67

0.0003

0.026

-0.014-0.066

enrollment and self-remrted health related aualitv of life

Overall health: None at alYlittle Somda lot Rostate troubles: None at all/little Somda lot Bladder problems: None at alYlittle Somela lot Bowel problems: None at all/little Somda lot Sexual dysfunction: None at alYlittle Somda lot

95% CI

pvalue

TABLE6. Association between paper based time tmde-off utilities at Health StaW Degree of Pmblem

Change (U-titer - paper) -0.004 -0.023

rho

~ 0 . h .

No. Pts.

Mean Utility

p Value

25 25

0.94

0.048

25 25

0.962 0.874

0.0021

7 6

0.843 0.95

0.14

7 2

0.979 0.975

0.87

11 20

0.95 0.867

0.0553

0.88

global health.4-9 U-Titer has also been used to investigate issues of regret in medical decision making.19 Our results did not correlate as well with HUI:3. Median HUI:3 utilities were higher for those subjects who were delighted, pleased or mostly satisfied with current health compared with those who were not (Spearman’srank correlation coefficient 0.30, p = 0.037). No differences were noted in median HuI:3 utilities between patients who described themselves as having “some” or ”a lot” of prostate troubles and those who described themselves as having “little”or “no” problems (Mann-Whitney U 0.917 versus 0.887,p = 0.66). Such results should be expected, however, based upon the different designs of the 2 instruments. The HUI:3 assigns a group average to any 1 health state while our instrument measures individual preferences. As a consequencewe would expect that estimates of central tendency should be similar between the 2 assessment techniques. This is precisely what we found. Only the self-administered paper based instrument can generate utility estimates for individual patients. Other explanations for a lack of correlation between these instruments center around the reference population. The self-administered paper based instrument relies on prostate cancer patients to estimate their own utilities. Utility scores obtained using HuI:3 are baaed upon preference weights assigned to each level that were obtained by interviewing a general population. Our most striking finding was the high relative value placed upon all 5 health states by our patients. Median values for each health state were greater than 0.97 and in some cases as high as 0.997.Equally striking was the variation in patient estimates. Patients who were delighted or pleased with prostate cancer management had high estimates of overall utility for prostate problems with minimal variation in their responses. Patients who had mixed results or who were mostly dissatisfied or unhappy with prostate cancer management, however, had much lower utility estimates of their overall prostate problems with results ranging from 0.6to 0.95.Observed differences were seen for each of the health states measured and probably reflect true differences in how patients feel about their outcomes.

-0.041-0.033 -0.057-0.010

We also tested the 3 other potential predictors of patient utility of an internal control subscale, a future satisfaction question and a mood assessment question. The internal control subscale of the locus of control questions was associated with time trade-off estimates of overall health (Spearman’s rho = 0.32, p = 0.025). Active, aggressive individuals generally have higher utilities for overall health compared with more passive people. We found little correlation between time trade-off estimates of overall health and patient estimates of future satisfaction. Estimates of patient mood, however, showed strong correlation with time trade-off estimates of overall health (Spearman’s rho = 0.46, p = 0.0012). Patients who are “down hearted” and “blue” are more likely to have lower estimates of utilities compared with men who describe their mood more positively. Our results compare favorably with efforts to assess patient utilities in other disease states. Nease et a1 recently measured patient utilities in 220 patients with chronic stable angina.4 They also found wide variations in patient response despite similarities in functional limitations. Based on these findings they suggested that guidelines for management of ischemic heart disease should be based on individual patient preference rather than on an objective assessment of disease severity. Interpreting the observed variation in this study is challenging. On the one hand we observed very good test-retest reliability with the paper based utility assessment instrument (table 3). This finding suggests that observed variation in utilities reflects true differences in preferences rather than noise. On the other hand, although the differences between utilities assessed by the paper based instrument and U-titer were small, correlations between the utilities assessed by the paper-based instrument and U-titer were modest (table 5). This finding suggests that the paper based instrument and U-titer produce similar estimates for group averages but divergent estimates for individual subjects. For any given patient the 2 instruments (paper based and U-titer) could produce consistently different estimates of utilities for the health states assessed but differences varied for each subject evaluated. This difference could possibly explain the high consistency of the paper based assessments with patients over time and some dissociation between utilities assessed by the paper based instrument and U-titer. More research is needed to interpret the observed within-patient variation. Similar estimates of mean utilities provided by the 2 instruments suggest that the paper based instrument is suitable for analyses that focus on group rather than on individual utilities. The small sample size of our pilot project limits generalization to the larger community of men diagnosed with prostate cancer. Larger studies assessing patient utilities conducted among men receiving alternative treatments for prostate cancer would provide greater detail concerning how men value health states associated with the treatments. The high test-retest reliability of the self-administered paper based instrument suggests that these studies can be accomplished successfully in large patient cohorts with modest resource expenditure. Utility estimates based upon prefer4

ASSESSMENT OF PATIENT PREFERENCES AMONG MEN WITH PROSTATE CANCER ences of a large sample of m e n with prostate cancer would greatly facilitate future cost-effectiveness analyses of prostate cancer management. For the individual patient these instruments provide new tools to help m e n understand their personal preferences a n d to improve their ability to evaluate the riskhenefit trade-off associated with alternative treatm e n t options. CONCLUSIONS

We developed a self-administered paper based instrument t o assess patient utilities for 5 key health states associated with prostate cancer management of overall quality of life, problems related to prostate cancer a n d problems related to urinary, bowel a n d sexual dysfunction. We tested this instrum e n t o n 50 m e n diagnosed with prostate cancer. Our results suggest that this instrument is a reliable and valid tool to assess patient preferences for outcomes associated with prostate cancer, a n d has distinct advantages in large survey studies because it c a n be incorporated at relatively low cost. F u r t h e r validation studies are needed to determine its usefulness in various clinical settings. Urologists practicing in the following institutions provided assistance that made this research possible: Hartford Hospital, St. Francis Hospital & Medical Center, New Britain General Hospital, Manchester Hospital, Veterans Memorial Medical Center, Bristol Hospital and the University of Connecticut Health Center/John Dempsey Hospital. REFERENCES

1. Grove, A.: Taking on prostate cancer. Fortune, 133 54, 1996. 2. DeHart, E.: Reflectionsof a prostate cancer patient. Urology, 48: 171,1996. 3. Gold, M. R., Patrick, D. L., Torrance, G. W., Fryback, D. G., Hadorn, D. C., M e t , M. S., Daniels, N. and Weinstein, M. C.: Identifying and valuing outcomes. In: CostEffectiveness in Health and Medicine. Edited by M. R. Gold, J. E. Siegel, L. B. Russell and M. C. Weinstein. New York Oxford University Press, 1996. 4. Nease, R. F., Jr., Kneeland, T., OConnor, G. T., Sumner, W., Lumpkins, C., Shaw, L., Pryor, D. and Sox,H. C.: Variations in patient utilities for outcomes of the management of chronic stable angina. J.A.M.A., 273 1185,1995. 5. Zug, K A., Littenburg, B., Baughman, R. D., Kneeland, T., Nease, R. F., Sumner, W., OConnor, G. T., Jones, R., Morrison, E. and Cimis, R.: Assessing the preferences of patients with psoriasis. Arch. Dermatol., 131: 561,1995. 6. Alexander, N. E., Ross, J., Sumner, W., Nease, R. F. and

163

Littenberg, B.: The effect of an educational intervention on the perceived risk of breast cancer. J. Gen. Intern. Med., 11: 92, 1995. 7. Gage, B., Cardinalli, A, Albers, G. and Owens, D.: Cost effectiveness of warfarin and aspirin for prophylaxis of stroke in patients with nonvalvular atrial fibrillation. J.A.M.A., 274 1839,1995. 8. Nease, R. F., Jr., Tsai, R., Hynes, L. M. and Littenberg, B.: Automated utility assessment of global health. Quality of Life Res., 6:175,1996. 9. Handler, R., Hynes, L. M. and Nease, R. F., Jr.: Effect of locus of control and consideration of future consequences on time tradeoff utilities for current health. Quality of Life Res., 8: 1997,in press. 10. Boyle, M., Furlong, W., Torrance, G. and Feeny, D.: Reliability of the Health Utilities Index Mark 111 used in the 1991 cycle 6 general social survey health questionnaire. Working Paper No. 94-7,Centre for Health Economics and Policy Analysis, McMaster University, March 1994. 11. Barr, R. D., Pai, M. K R., Weitzman, S., Feeny, D.,Furlong, W., Rosenbaum, P. and Torrance, G. W.: A multi-attribute a p proach to health status measurement and clinical management illustrated by an application of brain tumours in chddhood. Int. J. Oncol., 4 639,1994. 12. Deyo, R. A, Diehr, P. and Patrick, D. L.: Reproducibility and responsiveness of health status measures. Controlled Clin. Trials, suppl., 1 2 1425. 1991. 13. Sox, H. C., Blatt, M. A., Higgins, M. C. and Marton, K. I.: Measuring the outcome of care. In:Medical Decision Making, Boston: Butterworths, pp. 168-181,1988. 14. Litwin, M. S.,Hays, R. D., Fink, A, Ganz, P. A, Leake, B., Leach, G. E. and Brook, R. H.: Quality of life outcomes in men treated for localized prostate cancer. J.A.M.A., 273 129,1995. 15. Albertsen, P. C., Aaronson, N. K, Muller, M.J., Keller, S. D. and Ware, J. E., Jr.: Health related quality of life among patients with metastatic prostate cancer. Urology, 4 9 207,1997. 16. Fowler, F. J., Barry, M.J., Lu-Yao, G., Roman, A., Wasson, J. and Wennberg, J. E.: Patient-reported complications and follow-up treatment after radical prostatectomy. Urology, 4 2 622,1993. 17. Krahn,M. D., Mahoney, J. E., Eckman, M. H., Trachtenberg, J., Pauker, S. G. and Detsky, A. S.: Screening for prostate cancer. A decision analytic view. J.A.M.A., 272 773,1994. 18. Fleming, C., Wasson, J. H., Albertsen, P. C., Barry, M. J. and Wennberg, J. E.: A decision analysis of alternative treatment strategies for clinically localized prostate cancer. J.A.M.A., 269 2650,1993. 19. Hynes, L., Levine, A., Littenberg, B. and Nease, R.: Development and comparison of two utility-based measures of regret. Med. Decis. Makine. suod.. 1 4 433A. 1994.