QOL and Outcomes Research in Prostate Cancer Patients With Low Socioeconomic Status

QOL and Outcomes Research in Prostate Cancer Patients With Low Socioeconomic Status

Quality of life (QOL) and health status data obtained from the Veterans Administration Cancer of the Prostate Outcomes Study (VA CaPOS) have the potential to add substantially to information available from other observational databases and clinical trials with QOL outcomes. As Kim et al correctly note, most samples of prostate cancer patient in current observational databases, such as the Cancer of the Prostate Strategic Urologic Research Endeavor (CaPSURE),[1] have limited variation with respect to such variables as race and educational level. The VA system allows for the inclusion of patients with more diverse race/ethnic backgrounds, as well as more disparate educational and literacy levels. A randomized Southwest Oncology Group (SWOG) trial of stage D2 prostate cancer also had good representation (25%) of African-American patients in both treatment arms.[2]

In general, patients are required to complete QOL questionnaires in English (and, more recently, in other languages, since validated translations are available for many of the current questionnaires). We therefore assume, but rarely verify, that patients who are literate to some degree are the source of most of the QOL data reported in the literature.

Self-administered questionnaires are necessary because most studies are done in busy cancer clinics where nursing and data management personnel do not have the time to conduct in-person interviews for QOL data. (One exception to self-administration in cooperative group QOL research is the centralized telephone interview method used by the Cancer and Leukemia Group B [CALGB]).[3]

Kim et al do not clearly state whether telephone interviews are always used in the VA CaPOS, or whether certain scores on a measure, such as the Rapid Estimate of Adult Literacy (REALM),[4] are used to trigger the telephone interview. Nevertheless, the VA is to be commended for: (1) supporting more staff-intensive methods, such as telephone interviews, to broaden the population providing health outcomes data; and (2) identifying literacy levels in their population.

Disadvantages of Large Observational Databases

Although large observational databases can expand our understanding of how patients are doing, both during and after treatment for prostate cancer, it is important to remember the disadvantages of such databases. Interpretation of data provided by observational databases is problematic due to the decreased ability to control for key explanatory variables in the observational setting.

In addition, databases that rely on cross-sectional comparisons have no basis for determining change. A cross-sectional comparison is often limited to healthy survivors, thereby overestimating patient QOL. Even data from longitudinal cohorts must be examined carefully. If longitudinal analyses use only patients for whom complete data are available, such analyses can also be biased because they include survivors who are healthy enough to complete the QOL questionnaires.

Randomized trials control better for unmeasured, important covariates related to the outcome of interest and provide a basis for estimating change from pretreatment status. Patients who have chosen a particular treatment, as opposed to being randomized to a treatment approach, can differ in many ways. Since it is difficult to measure all of the potential confounding variables, the best approach to minimizing errors in estimates of QOL is to rely on randomization.

Caveats in Drawing Conclusions From Observational Databases

Observational databases may have their greatest value in generating hypotheses for randomized trials and in providing a general picture of the QOL of prostate cancer survivors. However, investigators and readers must guard against concluding that various QOL outcomes are associated with different treatments when QOL data are generated by an observational study.

Comparing one observational study with another or comparing an observational study with a randomized trial can be problematic. For example, Kim et al suggest the value of two observational databases (the CaPSURE[5] and the VA CaPOS) for assessing the long-term functioning of patients with metastatic prostate cancer; both databases include a common measure, the European Organization for Research and Treatment of Cancer (EORTC) QLQ-C30 questionnaire. Unfortunately, Albertsen et al are not clear about the timing of QOL assessments relative to the time from treatment initiation.[5]

Kim et al contrast the usefulness of the two observational databases with post-treatment follow-up to the SWOG randomized trial, which had a 6-month QOL assessment period[2] for describing patient functioning. The SWOG trial is described as finding more diarrhea during the treatment period among patients receiving an orchiectomy plus the antiandrogen flutamide (Eulexin) compared to those receiving an orchiectomy plus placebo. However, the principal finding of the SWOG QOL trial was the compromised emotional functioning of men treated with total androgen blockade (ie, orchiectomy plus flutamide) vs those patients treated with surgical castration plus placebo.

Albertsen et al[5] reported that emotional functioning scores of men, all of whom were treated with a luteinizing hormone–releasing hormone (LHRH) agonist plus flutamide, did not differ from scores of men in the general population on the same questionnaire. However, the presence of a “control” arm (orchiectomy plus placebo) in the SWOG trial[2] allowed us to gauge more accurately the impact of additional antiandrogen treatment on survival, as well as its impact on patient QOL.[6,7] This example suggests the need for caution when drawing conclusions from observational databases.

Kim et al note the limited length of QOL follow-up in the SWOG trial since it covered a 6-month treatment period. This is a valid criticism. However, the inclusion of longer-term follow-up of patients in a randomized trial can generate stronger conclusions regarding treatment comparisons than can an observational study.


Observational studies can identify important areas of QOL that require further exploration in the controlled, randomized trial setting. Consistency in the selection of QOL questionnaires for new QOL databases facilitates examination of findings from different studies. By employing a combination of both observational studies and randomized trials in prostate cancer treatment research and by remaining cognizant of the potential deficiencies based on source and design, we can more confidently inform patients about the positive and negative trade-offs associated with the treatment of prostate cancer.


1. Lubeck DP, Litwin MS, Henning JM, et al: Measurement of health-related quality of life in men with prostate cancer: The CaPSURE database. Qual Life Res 6:385-392, 1997.

2. Moinpour CM, Savage MJ, Troxel A, et al: Quality of life in advanced prostate cancer: Results of a randomized therapeutic trial. J Natl Cancer Inst 90:1537-1544, 1998.

3. Kornblith AB, Holland JC: Model for quality-of-life research from the Cancer and Leukemia Group B: The telephone interview, conceptual approach to measurement, and theoretical framework. J Natl Cancer Inst 20:55-62, 1996.

4. Davis TC, Long SW, Jackson RH, et al: Rapid estimate of adult literacy in medicine: A shortened screening instrument. Fam Med 25:391-395, 1993.

5. Albertsen PC, Aaronson NK, Muller MJ, et al: Health-related quality of life among patients with metastatic prostate cancer. Urology 49:207-217, 1997.

6. Albertsen PC: Re: Quality of life in advanced prostate cancer: Results of a randomized therapeutic trial. J Natl Cancer Inst 91:646, 1999.

7. Moinpour CM, Troxel A, Lovato LC, et al: Response. J Natl Cancer Inst, 91:646, 1999.

Loading comments...
Please Wait 20 seconds or click here to close