Assessing patient-reported outcomes in prevention trials.

CS16-02

The public health and personal implications of a diagnosis of cancer are being increasingly recognized. For this reason, research programs now encompass prevention as well as treatment clinical trials.

Quality of life (QOL) is an end point of interest in both treatment and prevention clinical trials. Quality of life is an important outcome measure to monitor when healthy participants are asked to take a chemopreventive agent for a protracted period of time. Chemopreventive agents can have broad impacts on a participant's functioning as well as specific side effects associated with the agent. Systematic monitoring of participant HRQL also communicates to the participant that the research team is concerned about his well-being, a factor that can affect willingness to enroll in a long-term trial (1). In the primary prevention setting, study participants are healthy individuals and it is important to monitor effects on QOL and functioning, particularly when chemopreventive agents are administered. From a research perspective, participants' expectations regarding the impact of the chemopreventive regimen will affect their willingness to enroll and participant experience with the prevention intervention will affect adherence to trial requirements (1). Collection of QOL data reinforces that effects on functioning are being monitored by the researchers. Recent trials such as the Postmenopausal Estrogen/Progestion Intervention (PEPI) (2-4), the Breast Cancer Prevention Trial (BCPT) (5), and the Women's Health Initiative (WHI) (6) have included comprehensive QOL assessments. A comprehensive QOL assessment should also include lifestyle behaviors; these data can serve as important covariates in analyses. For example, physical activity has been found to have a protective effect on the incidence of prostate cancer (7).

The prevention context has implications for assessment, design, and statistical considerations for a trial. This presentation will describe HRQL measures that work well for healthy participants (e.g., the SF-36 [8]). An important step in establishing a QOL instrument's feasibility, reliability, and validity is evaluating its psychometric properties in a broad range of clinical populations (9). Most prevention trials have a goal of enrolling sufficient numbers of minority group participants to be able to evaluate whether or not there are racial/ethnic differences in the efficacy of the preventive agent. The investigators could be faced with low literacy levels (= sixth grade reading level). To compensate for low literacy levels, investigators might need to use a centralized phone bank, interviewer-assessment strategy as an alternative to self-administration of HRQL questionnaire. The feasibility, validity, and reliability of telephone-administered questionnaires have been documented for a group of VA patients (10). In general, little is known about the HRQL of persons who are in lower socioeconomic strata and who have low literacy skills.

Potential designs for limited HRQL samples should be discussed because most prevention trials require very large samples, which are not necessary to answer the HRQL research questions. In most cases, based on knowledge about the preventive agent or intervention, the HRQL hypothesis is one of equivalence between the prevention treatment arms; that is, equivalence is a reasonable hypothesis if there are no previous data suggesting that HRQL would be either negatively or positively affected by the study agents. Equivalence can be defined in a number of ways. One possibility is to define it as an effect that is not clinically significant and a method for gauging a meaningful effect is to use the categories proposed by Cohen (11) where a small effect is between .2 and .5 standard deviations. Therefore, we define HRQL equivalence for the treatment arms as any effect less than .2 standard deviations on the primary HRQL continuous measure; on a 0-100 scale, this would be less than a two-point difference. Most prevention trial sample sizes provide more than adequate power to detect these small differences.

Depending on resources available for the trial, it may be necessary and scientifically justified to limit the HRLQ assessments to a smaller sample of trial participants. There are several ways to do this. It is better to select institutions and not patients within institutions to reduce complexity. For example, the trial could limit the collection of HRLQ data to Community Clinical Oncology Programs (CCOPs); these institutions could earn cancer control credits for obtaining baseline and follow-up HRQL assessments. Since these institutions are required to obtain cancer control credits in order to retain their CCOP status, the HRLQ study can serve as an incentive. The HRQL study can be conducted with only the first XX number of participants registered to the prevention trial (i.e. the HRLQ study is closed once a specified number of participants have been enrolled).

There are potential problems with such methods and these will be discussed in the presentation.

Another issue is the length of follow-up period for the HRQL outcome. It should be as long as participants are followed for primary efficacy outcomes but a number of variables will affect the frequency with which the HRLQ are collected during the study period.

References

1. Nayfield SG. Roles of government agencies in support and application of quality of life research. Spilker B, editor. Quality of life and pharmacoeconomics in clinical trials. 2nd ed. Philadelphia: Lippincott-Raven; pp. 779-783, 1996.

2. The Writing Group for the PEPI Trial. Effects of estrogen or estrogen/progestin regimens on heart disease risk factors in postmenopausal women. JAMA 1995;273(3):199-208.

3. Espeland MA, Bush TL, Mebane-Sims I (Writing Group for the PEPI Trial Investigators). Rationale, design, and conduct of the PEPI trial. Controlled Clin Trials 1995;16(4 (Suppl)):1S-19S.

4. Miller VT, Byington RL, Espeland MA, et al. Baseline characteristics of the PEPI participants. Controlled Clin Trials 1995;16(4 (Suppl)):54S-72S.

5. Ganz Patricia A, Day Richard, Ware JrJE, Redmond Carol, Fisher Bernard. Base-line quality-of-life assessment in the National Surgical Adjuvant Breast and Bowel Project Breast Cancer Prevention Trial. J Natl Cancer Inst 205;87(18):1372-82.

6. Matthews Karen A, Shumaker Sally A, Bowen Deborah J, et al. Women's health initiative: Why now? What is it? What's new? Am Psychol 1997:101-116.

7. Hartman TJ, Albanes D, Rautalahti M, et al. Physical activity and prostate cancer in the Alpha-Tocopherol, Beta-Carotene (ATBC) Cancer Prevention Trial (Finland). Cancer Causes Control 1998;9(1):11-18.

8. Ware JE Jr., Sherbourne CD. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med Care 1992;30:473-483.

9. Nunnally JC. Psychometric theory. New York, NY: McGraw-Hill; 1978.

10. Knight SJ, Chmiel JS, Kuzel T, Sharp L, Albers M, Fine R, Moran EM, Nadler RB, Sharifi R, Bennett CL. Quality of life in metastatic prostate cancer among men of lower socioeconomic status: Feasibility and criterion related validity of 3 measures. J Urol 1998;160:1765-1769.

11. Cohen J. Statistical power analysis for the behavioral sciences. (Edition 2). Hillsdale, NJ: Lawrence Erlbaum & Associates, Publishers, 1988.