PREDICTORS OF UTILITIES FOR HEALTH STATES IN EARLY STAGE PROSTATE CANCER

0022-5347/01/1663-0942/0 THE JOURNAL OF UROLOGY® Copyright © 2001 by AMERICAN UROLOGICAL ASSOCIATION, INC.®

Vol. 166, 942–946, September 2001 Printed in U.S.A.

PREDICTORS OF UTILITIES FOR HEALTH STATES IN EARLY STAGE PROSTATE CANCER CHRISTOPHER S. SAIGAL, JEFFREY GORNBEIN, ROBERT NEASE

AND

MARK S. LITWIN*

From the Departments of Urology, Health Services and Biomathematics, and Statistical-Biomathematical Consulting Clinic, University of California, Los Angeles, Los Angeles, California, and Department of Internal Medicine, Washington University School of Medicine, St. Louis, Missouri

ABSTRACT

Purpose: When faced with treatment choices for early stage prostate cancer, patients must balance the survival benefit of a treatment with its morbidity. Little is known about how patients balance these trade-offs. To further our understanding of patient decision making we assessed patient utilities for prostate cancer treatment related morbidities. We determined whether patient utilities were predicted by sociodemographic characteristics or baseline genitourinary function. Materials and Methods: We evaluated 401 men undergoing prostate needle biopsy for suspicion of prostate cancer at university, Veterans Affairs and public hospitals. Study design included a prospective cross-sectional cohort with correlation and multivariate analysis. Subjects were studied with 2 established health related quality of life instruments. Patient utilities were assessed with an interactive software application. Results: On multivariate analysis utility for current general health was a significant predictor of utilities for treatment related morbidities. Surprisingly baseline urinary, sexual and bowel function scores did not correlate well with respective utilities for potential incontinence, impotence or radiation proctitis. In other words, men with good and imperfect baseline function were equally willing to risk impairment to preserve life. Conclusions: Men who perceived that general health was better appear to place higher value on quantity of life, while those who already are suffering from poor general health place higher value on quality of life. Ethnicity appears to modify some effects of other variables on patient preference. Utility assessment provides a quantitative tool to aid physicians in counseling patients when making treatment decisions for localized prostate cancer. KEY WORDS: prostate, prostatic neoplasms, utility theory, quality of life, morbidity

The large number of early prostate cancers detected yearly creates dilemmas for many men diagnosed with early stage disease. Interventions for curing the disease impose risks of significant morbidities. Little is understood about how patients make treatment decisions when attempting to balance the dual goals of optimizing quantity and quality of life. By quantifying patient preferences for treatment side effects using a methodology known as utility assessment, we may compare the importance of various medical outcomes and how the values of patients influence their medical decision making. This method is particularly relevant for early stage prostate cancer, a disease in which biological indolence focuses attention on quality of life concerns. Patient utilities are individual valuations of various health states. A state of perfect health is assigned a utility of 1, while death is assigned a utility of 0. Suboptimal states of health correspond to values between 0 and 1. We designed this study to assess patient utilities in men at risk for prostate cancer. We assessed utilities for baseline general health, and urinary, sexual, and bowel function (“pelvic functions”). We also assessed utilities for the standard impaired health states of urinary incontinence, impotence and radiation proctitis. In addition, we evaluated baseline quality of life, sociodemographics and co-morbidity. We hypothesized that pa-

tient baseline pelvic functions would be negatively associated with their utilities for prostate cancer treatment related morbidities (higher score on quality of life scales indicates better function). For example, we predicted that an individual with poor baseline sexual function would be unwilling to trade much life span to preserve function in that domain (that is have a high utility for erectile dysfunction). We also predicted that an individual who was more bothered by baseline pelvic dysfunctions would be willing to trade more life span to avoid stress urinary incontinence, impotence and radiation proctitis (that is have low utility values for these conditions). Furthermore, we determined whether age or ethnicity would be associated with different valuations of impaired health states. We assessed empirical evidence of whether older men placed greater value on the years of life remaining (quantity) or on the life remaining in the years (quality). METHODS

Subjects. Patients scheduled for prostate biopsy between August 1997 and May 1998 were recruited to join our study at 3 clinical settings chosen to ensure a wide racial and socioeconomic mix, including a private academic medical center, a Veterans Affairs medical center and a county sponsored hospital. All English or Spanish speaking men scheduled for prostate needle biopsy due to suspected cancer at any of the institutions were eligible for enrollment. Study exclusion criteria included the inability to provide informed consent or complete the utility assessment, as evidenced by irrational patient ranking of the health states monocular and binocular blindness. Informed consent was obtained according to the

Accepted for publication April 27, 2001. Supported by American Cancer Society Research Project Grant 98-100-01-PBP. The funding agreement ensured author independence in designing the study, interpreting the data, and writing and publishing the report. * Requests for reprints: Department of Urology, University of California-Los Angeles, Box 951738, Los Angeles, California 900951738. 942

943

PREDICTORS OF UTILITIES FOR PROSTATE CANCER

guidelines of the Human Subjects Protection Committee at each institution. Instruments. General health related quality of life was measured with the RAND 36-Item Health Survey 1.0 (SF36).1 The SF-36 has been shown to be reliable and valid in various populations, including men with prostate cancer.1, 2 A validated cross-cultural Spanish translation of the SF-36 is available.3 Prostate targeted health related quality of life was measured with the University of California-Los Angeles Prostate Cancer Index (PCI), a self-administered questionnaire that quantifies prostate cancer specific health related quality of life. The PCI has been shown to be reliable and valid in populations of older men with and without prostate cancer. A validated cross-cultural Spanish translation of the PCI is available.4 – 6 Sociodemographic and co-morbidity data were collected at the time of the baseline survey with a separate instrument that includes relevant questions and a 12-item medical history checklist based on an established co-morbidity rating scale.7 Utility assessments were performed with an established, validated software application that uses the time trade-off technique. Briefly, the time trade-off technique establishes the amount of lifespan a patient would trade to avoid a health problem such as impotence. The program has been used to assess utilities in many clinical conditions, including prostate cancer. The Appendix shows a description of the time trade-off technique and scenarios used in this study. A cross-cultural Spanish language translation of the program was created for our project. Data collection. All interviews were done by 1 of 2 project staff members trained in the use of the program. The interviews were performed immediately before biopsy in the majority of cases. However, scheduling issues mandated that occasionally a patient was interviewed immediately after recovery from biopsy, usually about 30 minutes after the procedure. Thus, all baseline interviews were performed before biopsy results were available. The interviewer was present during the interview to assist with questions but otherwise remained unobtrusive. Average completion time was 25 minutes. Patients were then given the quality of life instruments to complete at home and return. At the time of the assessment serum prostate specific antigen (PSA) and digital rectal examination results were recorded from the chart. Statistical analysis. The primary unit of observation for this study was the individual patient. The primary outcomes of interest were utilities for future incontinence, impotence and radiation proctitis. Predictors included general health utility, scores on the general and disease targeted quality of life instruments, sociodemographic variables and comorbidity. The Kruskal-Wallis test was used to assess differences in utilities across stratifications of baseline quality of life scores, sociodemographic variables and PSA levels. Correlations among the utility scores, and between each utility score and the relevant PCI score and general health utility were assessed using rank based intraclass correlation coefficients. Test-retest reliability of the program interview was assessed by interviewing 21 patients again within 3 weeks of the baseline interview. The intraclass correlation for the utility domains was 1 for all current pelvic functions and future impaired states except the utility for impotence, which had an intraclass correlation of 0.56. Univariate analysis was initially performed using hypothesized predictors of patient utilities for the 3 morbidities. Candidate predictors of all morbidities were age, race and baseline general health utility. In addition, for each morbidity the relevant PCI function and bother scores were used as candidate predictors, for example urinary function and bother scores from the PCI were used for the urinary incontinence utility. Also, education and marital status were con-

sidered as covariates. All significant univariate predictors were used to construct multivariate analysis models. Multivariable models of the utilities for urinary incontinence, impotence and radiation proctitis were constructed using candidate predictors identified on univariate analysis and by clinical experience. Because the distribution of utilities was not normal, each utility score was converted into an ordinal scale of 1 to 5 with 1 as the lowest and 5 as the highest utility. Logical cutoff points in the original utility scale were used to convert it to the ordinal scale, including 1— 0 to 0.125, 2— 0.126 to 0.5, 3— 0.501 to 0.75, 4 — 0.751 to 0.92 and 5— 0.921 to 1. These logical points also roughly corresponded to the quintiles for future utility. For the 3 utility outcomes we performed ordinal logistic regression analysis using a proportional odds model and included as potential predictors all logical 2-way interactions among the main effects. Race and age were controlled for in each model. We used stepdown selection regression methods to remove nonsignificant interactions and any main effects that had no strong biological relevance and were clearly not significant (p ⬎0.25). We did not allow the stepdown procedure to determine completely the model predictors but used it primarily to determine which, if any, interactions were significant. The score test was used to test the hypothesis of proportional odds on the models.8 The c statistic,9 which provides the area under the corresponding receiver operating characteristics curve, is provided as a measure of goodness-of-fit of the final models.10 All programming and analysis in this study were performed using commercially available computer software. RESULTS

Of the 535 consecutive patients who underwent prostate needle biopsy at 1 of the 3 institutions 401 (75%) were eligible for study enrollment and agreed to participate. The most common reason for nonparticipation (55% of cases) was overlapping biopsy scheduling at 2 institutions, with only 1 interviewer available. Only 1 man did not participate because of the inability to understand the research task. Of the 401 men 246 (61%) returned the health related quality of life instrument with demographic data after biopsy. Table 1 lists the demographic characteristics of the study population. Patients who did not return the instrument did not differ significantly in terms of age or presenting serum PSA from patients who returned it. Utilities did not differ significantly between those who completed the health related quality of life survey and those who did not (table 2). Table 3 lists baseline quality of life scores from the PCI. Of the 401 respondents 308 (77%) had some degree of impairment in baseline pelvic functions. Table 3 also shows patients’ utilities for baseline function and for the 3 hypothetical impaired health states often associated with prostate cancer treatment. They were willing to trade almost no time to improve current pelvic functions, but willing to trade much more time to avoid potential future impairment. Utility for current general health, which is an indicator of self-perceived baseline health status, was associated with utilities for future stress incontinence, impotence and radiation proctitis. In other words, patients who highly valued baseline general health (had a high utility value for current general health) were less willing to trade time to avoid urinary, sexual or bowel impairment (that is had high utility values for these dysfunctions). Conversely, those who perceived baseline health status to be worse (that is had a low utility volume for current general health) were more willing to trade off time to avoid the impairments (that is had a low utility volume for these impairments). Table 4 shows the multivariate analysis for utilities of stress incontinence, impotence and radiation proctitis. Age and ethnicity were significant predictors in all 3 models except age in the impotence model. The current general

944

PREDICTORS OF UTILITIES FOR PROSTATE CANCER TABLE 1. Characteristics of the study population Hospital Overall Sample

No. subjects Mean age (range) % Study participation % Health related quality of life questionnaire response No. race (%): White Black Latino Asian No. education (%): Less than high school High school or some college College graduate or higher No. living with spouse or partner (%): Yes No Not all subjects answered all items.

University

Veterans Affairs

Public

401 63 (40–93) 75 62

219 63 (40–84) 76 70

162 64 (45–93) 87 51

20 61 (50–74) 33 70

245 101 31 15

(63) (26) (8) (4)

181 16 8 7

(85) (8) (4) (3)

60 80 13 7

(38) (50) (8) (4)

4 5 10 1

(20) (25) (50) (5)

37 105 170

(12) (34) (54)

5 45 140

(3) (24) (74)

24 55 29

(22) (51) (27)

8 5 1

(57) (36) (7)

227 166

(58) (42)

168 49

(77) (23)

49 112

(30) (70)

10 5

(67) (33)

TABLE 2. Mean differences in utilities (ordinal scale) in respondents and nonrespondents to the quality of life survey Utility

162 Nonrespondents

239 Respondents

Difference (95% CI)

p Value

4.43 4.96 4.91 4.63

4.60 4.92 4.90 4.75

⫺0.17 (⫺0.35–0.024) 0.04 (⫺0.03–0.11) 0.01 (⫺0.10–0.11) ⫺0.12 (⫺0.27–0.04)

0.09 0.71 0.94 0.14

3.51 4.17 3.75

3.57 4.15 3.89

⫺0.06 (⫺0.39–0.27) 0.02 (⫺0.25–0.28) ⫺0.14 (⫺0.43–0.18)

0.71 0.89 0.40

Current: General health Bowel function Urinary function Sexual function Future: Radiation proctitis Incontinence Impotence

TABLE 3. Age, baseline quality of life scores in the function domains of the PCI, and utilities for baseline and impaired health states No. Subjects

Mean

Median

Interquartile Range

Age 401 63 64 Baseline life quality: 246 Urinary function 90.7 100 Sexual function 57.7 62 Bowel function 86.8 94 Current health utilities: 401 Urinary function 0.98 1 Sexual function 0.95 1 Bowel function 0.99 1 General health 0.91 0.99 Future morbidity utilities: 401 Stress incontinence 0.79 0.98 Impotence 0.71 0.93 Radiation proctitis 0.60 0.75 Higher quality of life scores represent better outcomes and higher utility scores represent less

health utility was a significant predictor of utility for impotence. Older patients had significantly lower utilities for incontinence. Age was also negatively associated with the utility for radiation proctitis. White men had significantly higher utilities for incontinence and impotence than Latinos and Asian men, while black men had higher utilities for impotence than the other racial groups. Neither the SF-36 nor PCI scores were predictive of utilities for impairment after controlling for other variables. DISCUSSION

Our study has several findings. We were able to measure the time that men were willing to trade to avoid impaired pelvic functions resulting from prostate cancer treatment. This finding is important because it shows that the theoretical principles of utility assessment may be successfully applied in busy clinical situations. That the study tasks were so feasible varies from that of Stiggelbout et al, who reported that many patients with testicular or colorectal cancer required repeat explanation before they could participate in a similar utility assessment.11 Also, in our series, median utility scores for current sexual,

Range

% Greater than 0.99

58–69

40–93

Not applicable

85–100 39–78 82–100

45–100 20–100 0–100

Not applicable Not applicable Not applicable

1–1 0.99–1 1–1 0.93–0.99

0–1 0–1 0–1 0–1

91 79 95 61

0.73–0.99 0.49–0.99 0–0.99 willingness to trade off

0–1 45 0–1 40 0–1 27 time to avoid the health state.

urinary and bowel functions were 1.0, implying that respondents would not trade any life span to improve these functions. Median utility scores of the future impaired health states impotence and radiation proctitis were lower. The median utility scores for current functions were high, although the PCI scores indicated less than perfect current urinary, bowel and sexual function. An explanation for the skewed distribution of current utility scores may be the phenomenon of “response shift” described by Sprangers and Schwartz.12 Response shift represents the psychological accommodation to a health problem achieved by patients whose health status decreases. It is possible that after patients with a low utility score for radiation proctitis actually experience the morbidity, they accommodate to the problem and their rating of this state improves. However, Llewellyn-Thomas et al reported that patients accurately provided utility ratings of health states before they entered these states.13 Furthermore, respondents’ utilities for baseline general health were strongly associated with utilities for the 3 impaired health states, which represents compelling evidence that men who perceived that they had better overall health were less willing to trade life span to avoid pelvic dysfunc-

945

PREDICTORS OF UTILITIES FOR PROSTATE CANCER TABLE 4. Multivariate regression models for predictors of utility for incontinence, impotence and radiation proctitis Variable

Parameter Estimate ⫾ SE

F

Incontinence utility outcome: Intercept 5.010 ⫾ 0.572 76.78 Age ⫺0.070 ⫾ 0.012 35.76 Black 0.660 ⫾ 0.948 0.48 Latino/Asian ⫺0.486 ⫾ 0.211 5.29 Black ⫻ PCI urine bother 0.021 ⫾ 0.006 14.65 Black ⫻ baseline general health utility ⫺0.541 ⫾ 0.196 7.62 Age ⫻ baseline general health utility 0.013 ⫾ 0.002 58.76 13.62 Incontinence model (R2 ⫽ 0.285) Impotence utility outcome: Intercept 0.172 ⫾ 0.957 0.03 Age 0.018 ⫾ 0.011 2.46 Black 3.128 ⫾ 1.316 5.64 Latino/Asian ⫺1.172 ⫾ 0.498 5.54 Baseline general health utility 0.568 ⫾ 0.134 18.07 4.53 Impotence model (R2 ⫽ 0.12) Radiation proctitis utility for outcome: Intercept 3.714 ⫾ 0.855 18.86 Age ⫺0.039 ⫾ 0.019 3.92 Black ⫺0.656 ⫾ 0.311 4.46 Latino/Asian ⫺1.748 ⫾ 0.595 8.63 Age ⫻ baseline general health utility 0.009 ⫾ 0.003 11.59 Latino/Asian ⫻ education/high school or greater ⫺1.265 ⫾ 0.705 3.22 Latino/Asian ⫻ living with partner 1.445 ⫾ 0.665 4.72 5.56 Proctitis model (R2 ⫽ 0.16) Reference category is white for race, less than high school for education and not in significant relationship for marital status. Ordinal utility scale of 1— 0 to 0.125, 2— 0.126 to 0.5, 3— 0.501 to 0.75, 4 — 0.751 to 0.92 and 5— 0.921 to 1.

tion. In other words, men in better health appear to value quantity of life more highly, while those already suffering from poor health appear to value quality of life more highly. Nevertheless, the utility for baseline health may simply be a proxy for a more complicated set of issues, such as locus of control, time preference or risk aversion. Older men were no more willing than younger men to trade time to avoid impotence. This finding refutes our hypothesis that older men faced with prostate cancer treatment decisions may be more interested in preserving quality of life at the expense of survival. An exception was in the lower utilities of older men for bowel dysfunction (meaning that they would trade more life span to avoid the problem), implying that they valued this function more than younger men. This finding is notable when considering that older men with prostate cancer are often referred for radiotherapy, which may have a great impact on bowel function. The relationships of ethnicity and utilities for potential future impairment are complex and not easily characterized by the results of this study. An apparent conclusion regarding ethnicity that may be drawn from our multivariable analysis is that Latino men would trade more time than white men to avoid urinary incontinence and radiation proctitis. This finding may also have resulted from confounding differences in socioeconomic status in these 2 ethnic groups. Clearly more study is needed to explore any possible associations in this area. On multivariable models, most baseline PCI scores did not correlate highly with respective utilities for the hypothetical impaired health states. This finding was surprising, since we had speculated that men with poor baseline function would more readily accept potential impairment to maximize survival. However, in our series men with good baseline function were equally willing to risk impairment to preserve life. The multivariable models also showed that utilities for these baseline functions did not predict utilities for the impaired health states. Earlier Singer et al reported that baseline sexual function does not predict a willingness to trade time to avoid impotence.14 Despite our inclusion of all covariates that appeared relevant our regression models explained only a fraction of the total variance in utilities for the 3 impaired health states. Hence, our analysis implies that the factors driving patient preferences for treatment morbidity in early stage prostate

p Value 0.0001 0.0001 0.4870 0.0225 0.0002 0.0063 0.0001 ⬍0.0001 0.856 0.118 0.018 0.019 0.0001 ⬍0.0002 0.0001 0.0492 0.0361 0.0037 0.0008 0.0746 0.0312 ⬍0.0001

cancer are numerous, complex and highly individual. This finding reemphasizes the importance of measuring utilities in patients and not in physician proxies. Little previous study has been done of utility assessment in early stage prostate cancer but our utility findings for baseline general health, urinary function and sexual function are similar to those previously reported in patients.15 Helagson et al observed no association of age with a willingness of healthy men to trade sexual function for possibly curative treatment for prostate cancer.16 Their findings support ours, although they used a different method of preference measurement. Mazur and Merz noted that a substantial proportion of patients at a medical clinic would trade urinary or sexual function for a longer life span and older patients were more willing to trade survival time to avoid impotence.17 Their data also implied that men with baseline urinary or sexual function problems were more likely to accept the outcome of impotence, which is a finding not supported by our study. Others have observed that many patients with cancer and other chronic conditions value quantity of life so highly that they are unwilling to trade any time in utilities studies.11, 18, 19 To address this phenomenon in our data we converted the utility values from continuous into ordinal variables, which enabled us to model the characteristics, associations and utility predictors of the majority with high and the minority with lower values. The fact that many individuals were unwilling to trade much, if any, quantity of life is a strong statement of preference. Our study has several limitations. Despite high participation rates in the utilities assessment, only 61% of the men completed the health related quality of life instrument. Although we have no evidence of significant bias in nonrespondents, we did not compare them with respondents in terms of race or education. In addition, significance levels in our model may be too optimistic due to the model searching needed in an exploratory study. CONCLUSIONS

We assessed patient preferences for prostate cancer treatment related morbidity in busy clinical settings. Patients seemed to understand the utility assessment task. Those with higher utility scores for current general health tended to

946

PREDICTORS OF UTILITIES FOR PROSTATE CANCER

have high utility scores for future morbidity, indicating that men who valued current health highly placed more weight on survival time than on quality of life in the context of prostate cancer treatment. Older patients were as averse as younger ones to impotence but had lower utility scores for radiation proctitis than younger men. By guiding patients through structured interviews to assess their preferences, we hope to provide a new approach to help them evaluate the often competing priorities that they must balance.

APPENDIX

About the time trade-off technique of utility assessment. The automated interview first asks a patient to imagine that he is in a state of impaired health, such as urinary incontinence. The program presents him with a detailed clinical description of the impaired health state. He is then shown a screen and asked to choose between 2 options. One option is to live with the described morbidity for a period corresponding to his remaining life expectancy. The second option is to live with his baseline function in the relevant area but live for fewer years. The amount of time remaining in the second option is varied systematically based on subject choices. The assessment is terminated when the patient is indifferent to the 2 options. Utilities are computed by dividing the number of years the patient has chosen to live in his baseline state of functioning by the maximum possible number of years with impaired functioning offered to him. Scenarios used in utilities analysis. Current General Health: We want you to think about your current health. In a moment you’ll get a choice. You can spend your remaining life in your current health (just as it is today) or you can have ideal health but live fewer years. By ideal health we mean the best health you can imagine for someone your age. With ideal health your physical and mental health would be excellent. You would have no problem with walking, lifting or bending, thinking or talking, or hearing and seeing. You would be generally happy. Incontinence: Imagine that you have difficulty controlling your urine. Imagine that when you cough or sneeze or stand up suddenly, you leak a few drops of urine. Because of this, you need to wear a pad in your underwear. Imagine that you won’t leak a lot but you will need to wear 1 or 2 pads a day. Think about what you might trade to avoid having this problem. Impotence: Imagine that you have difficulty with erections. Imagine that you have a normal sex drive but cannot get an erection good enough to have sex (either with a partner or by yourself). Think about what you might trade to avoid having this problem. Proctitis: Imagine that you have difficulty with your bowels. Imagine that you sometimes have diarrhea, mucous from the rectum or a strong urge to move your bowels. Imagine that you have problems like this 2 or 3 times a week. Think about what you might trade to avoid having these problems.

REFERENCES

1. Lubeck, D. P., Litwin, M. S., Henning, J. M. et al: Measurement of health-related quality of life in men with prostate cancer: the CaPSURE database. Qual Life Res, 6: 385, 1997 2. Litwin, M. S., Shpall, A. I., Dorey, F. et al: Quality-of-life outcomes in long-term survivors of advanced prostate cancer. Am J Clin Oncol, 21: 327, 1998 3. Alonso, J., Prieto, L. and Anto, J. M.: La version espanola del SF-36 Health Survey (Cuestionario de Salud SF-36): un instrumento para la medida de los resultatos clinicos. Med Clin (Barc), 104: 771, 1995 4. Litwin, M. S., Hays, R. D., Fink, A. et al: The UCLA Prostate Cancer Index: development, reliability, and validity of a health-related quality of life measure. Med Care, 36: 1002, 1998 5. Litwin, M. S., Hays, R. D., Fink, A. et al: Quality-of-life outcomes in men treated for localized prostate cancer. JAMA, 273: 129, 1995 6. Krongrad, A., Perczek, R. E., Burke, M. A. et al: Reliability of Spanish translations of select urological quality of life instruments. J Urol, 158: 493, 1997 7. Greenfield, S., Apolone, G., McNeil, B. J. et al: The importance of co-existent disease in the occurrence of postoperative complications and one-year recovery in patients undergoing total hip replacement. Comorbidity and outcomes after hip replacement. Med Care, 31: 141, 1993 8. Peterson, B. and Harrell, F.: Partial proportional odds model for ordinal response variables. Appl Stat, 39: 205, 1990 9. Bamber, D.: The area above the ordinal dominance graph and the area below the ROC graph. J Math Psychol, 12: 387– 415, 1975 10. Nagelkerke, N.: A note on a general definition of the coefficient of determination. Biometrika, 78: 691– 692, 1991 11. Stiggelbout, A. M., Kiebert, G. M., Kievit, J. et al: The “utility” of the time trade-off method in cancer patients: feasibility and proportional trade-off. J Clin Epidemiol, 48: 1207–14, 1995 12. Sprangers, M. A. and Schwartz, C. E.: Integrating response shift into health-related quality of life research: a theoretical model. Soc Sci Med, 48: 1507–15, 1999 13. Llewellyn-Thomas, H. A., Sutherland, H. J. and Thiel, E. C.: Do patients’ evaluations of a future health state change when they actually enter that state? Med Care, 31: 1002–12, 1993 14. Singer, P. A., Tasch, E. S., Stocking, C. et al: Sex or survival: trade-offs between quality and quantity of life. J Clin Oncol, 9: 328 –34, 1991 15. Albertsen, P. C., Nease, R. F., Jr. and Potosky, A. L.: Assessment of patient preferences among men with prostate cancer. J Urol, 159: 158 – 63, 1998 16. Helgason, A. R., Adolfsson, J., Dickman, P. et al: Waning sexual function: the most important disease-specific distress for patients with prostate cancer. Br J Cancer, 73: 1417–21, 1996 17. Mazur, D. J. and Merz, J. F.: Older patients’ willingness to trade off urologic adverse outcomes for a better chance at five-year survival in the clinical setting of prostate cancer. J Am Geriatr Soc, 43: 979 – 84, 1995 18. Fryback, D. G., Dasbach, E. J., Klein, R. et al: The Beaver Dam Health Outcomes Study: initial catalog of health-state quality factors. Med Decis Making, 13: 89 –102, 1993 19. Tsevat, J., Dawson, N. V., Wu, A. W. et al: Health values of hospitalized patients 80 years or older. HELP Investigators. Hospitalized Elderly Longitudinal Project. JAMA, 279: 371–5, 1998