Patient Preferences for Treatment of Intermediate-risk Prostate Cancer: A Discrete Choice Experiment

In 88 men with intermediate-risk prostate cancer, bowel symptoms ranked as the #1 factor to avoid — above erectile dysfunction — while urinary symptoms didn't significantly sway choices at all.

Journal: Urology | Published: 2026-05-04 | Type: Journal Article | PMID: 42086402 Authors: Madhavaram Avanish et al. (University of Pittsburgh School of Medicine; Allegheny Health Network; University of Iowa) Funding/COI: Funding not disclosed. Authors declare no competing financial interests.

Summary

A discrete choice experiment asked 88 men with intermediate-risk prostate cancer to make hypothetical trade-offs between treatment attributes — bowel symptoms, erectile dysfunction, catheter use, travel burden, and more. Bowel side effects came out on top as the most-weighted concern, edging out erectile dysfunction, while urinary symptoms had no statistically significant influence on choices. Latent class analysis split the sample into two camps: one prioritizing avoidance of surgery and catheters, the other prioritizing avoidance of erectile and bowel symptoms — with older patients and racial/ethnic minorities skewing toward the first group.

Claims

• Bowel symptoms were the single most important attribute patients weighted against in treatment decisions
• Erectile dysfunction was a significant driver of avoidance, but ranked below bowel symptoms
• Urinary symptoms did not significantly influence patient choices — the opposite of what many clinicians might assume
• Patients placed low weight on travel time and undergoing major surgery at the aggregate level
• Latent class analysis identified two distinct preference subgroups among the 88 respondents
• Older age and racial/ethnic minority status were associated with membership in the surgery/catheter-averse class (multivariable linear probability model)

Study Quality

This is a discrete choice experiment (DCE), a well-established preference-elicitation methodology from health economics. The survey design included 8 attributes with 3–4 levels each, which is within normal range for DCE complexity. The conditional logit model is the standard analytic choice for DCEs. Adding latent class analysis to uncover preference heterogeneity is methodologically appropriate.

The sample size, however, is the binding constraint. With only 88 completers from a single urology clinic at one institution, the conditional logit estimates are credible but the latent class analysis — which is effectively carving that 88 into subgroups — is statistically fragile. Latent class models require larger samples to reliably identify distinct classes; two-class solutions with small n tend to be sensitive to model initialization and may not replicate. The demographic predictors of class membership (older age, minority status) are interesting but underpowered as multivariable findings.

Red Flags

• n=88 is small for any DCE; it's notably underpowered for latent class subgroup analysis
• Single center (Pittsburgh-area urology clinic) — patient preferences may reflect local clinical culture and demographics, not a generalizable population
• Funding source not disclosed — unusual absence for an academic paper; unclear if industry had any role
• DCE measures stated preferences under hypothetical conditions, not revealed preferences from actual treatment decisions; the gap between what patients say they'd prioritize and what drives real choices can be substantial
• "Racial and ethnic minorities" as a category is too coarse — lumping heterogeneous groups together limits interpretive precision
• Urinary symptom non-significance is surprising and should prompt skepticism about whether the attribute levels were constructed to be clinically meaningful

Strengths

• DCE is the appropriate methodology for this question — more rigorous than simple Likert surveys
• Latent class analysis at least attempts to capture that patient preferences aren't monolithic
• Demographic analysis of class membership is clinically useful for shared decision-making, even if underpowered
• No declared conflicts of interest from manufacturer-affiliated authors
• Results directly contradict the assumption that urinary function is patients' primary concern — a finding worth taking seriously if replicated

Verdict

The headline finding — that bowel side effects outweigh erectile dysfunction as a decision driver, and that urinary symptoms barely register — is clinically counter-intuitive enough to be worth noting. But 88 patients from one Pittsburgh clinic is a thin empirical base for that conclusion, and the latent class subgroup analysis is doing more work than the sample size can support. This paper is useful as hypothesis generation: if a larger, multi-site DCE replicates these findings, it would have real implications for how clinicians frame treatment trade-offs with patients. As a standalone study, it's suggestive rather than definitive.