Introduction
Patient expectations are closely associated with the magnitude of the placebo response[1] and are increasingly recognized as independent predictors of treatment outcomes across therapeutic areas[2]. However, how expectations should be operationally incorporated to improve clinical trial analyses remains insufficiently defined. In this work, we evaluated the value of directly measuring patient expectations using a dedicated questionnaire (MPsQ) and incorporating expectation-related constructs as adjustment covariates to improve the precision of treatment effect estimation. In addition, we explored the socio-economic role played by country- and site-level characteristics, often associated with patient expectations[3].
Method
This post hoc analysis was conducted within CDFV890B12201, a multicenter, randomized, double-blind, placebo-controlled Phase II trial evaluating DFV890 after 12 weeks in 115 patients with knee osteoarthritis. The primary endpoint was change in KOOS Pain; secondary endpoints included Average Pain Score (APS) and the KOOS Symptoms subscore. Patient expectations were measured at baseline using the MPsQ and included as covariates in ANCOVA models adjusted for treatment and baseline values. Geographic (Western vs Eastern Europe and Europe vs USA) and site-size effects were explored using effect sizes and correlations with the national Universal Health Coverage (UHC) subindex on noncommunicable disease[4].
Results
Expectations explained additional variance of 2.3%, 4.7%, and 5.4% for KOOS-Pain, APS, and KOOS-Symptoms, corresponding to partial correlations of 0.152 (p = 0.124), 0.227 (p = 0.032), and 0.232 (p = 0.028). Expectations in small sites were higher in Eastern (N = 19) versus Western European (N = 23) countries (Cohen’s d = 0.7, p = 0.022) and were negatively correlated with UHC subindex (r = −32.8%, p < 0.001). No difference between USA and Europe was observed. Small sites presented a larger non-significant effect (N_Small = 19, N_Large = 36, d = 0.4, p = 0.204).
Conclusion
Expectations are an important driver of outcome variability in osteoarthritis trials and contribute meaningfully to between-site differences. While expectations are partly shaped by site- and country-level contextual factors, the most effective way to account for them is through direct measurement and incorporation into efficacy analyses[5,6]. Including baseline expectation-related covariates reduces unexplained variability and improves the precision of treatment effect estimates. Such use could yield gains in precision equivalent to an increase in effective sample size between 2 and 5%. Although not sufficient on its own, this approach represents an essential step toward more precise and informative clinical pain trials.
[1] Colloca L., Barsky AJ. Placebo and Nocebo Effects. N Engl J Med. 2020 Feb 6;382(6):554-561. doi: 10.1056/NEJMra1907805. PMID: 32023375.
[2] Cormier S., Lavigne GL., Choinière M., and Rainville P. Expectations predict chronic pain treatment outcomes. Pain. 2016 Feb;157(2):329-338. doi: 10.1097/j.pain.0000000000000379. PMID: 26447703.
[3] Ooghe A., Paul J., and Pereira A.Modeling the expectations for improvement in osteoarthritis trials to enhance management strategies. Osteoarthritis and Cartilage 32 (2024): S500-S501.
[4] World Health Organization 2025 data.who.int, 2021 UHC service coverage index [Subindex on noncommunicable disease]. https://data.who.int/indicators/i/3805B1E/9A706FD (Accessed on 14 July 2025)
[5] Ooghe A, Pereira A., Colloca L., et al. From theory to practice: Simple rules for improving clinical trial confidence with covariate adjustment. The Journal of Pain. 2025;34:105489.
[6] Branders S., Ooghe A., and Pereira A. Are site differences driving outcomes? the central role of expectations in oa rcts. Osteoarthritis and Cartilage 33 (2025): S434.
