Dataset Information

Selection bias, investment decisions and treatment effect distributions.

ABSTRACT: When making decisions regarding the investment and design for a Phase 3 programme in the development of a new drug, the results from preceding Phase 2 trials are an important source of information. However, only projects in which the Phase 2 results show promising treatment effects will typically be considered for a Phase 3 investment decision. This implies that, for those projects where Phase 3 is pursued, the underlying Phase 2 estimates are subject to selection bias. We will in this article investigate the nature of this selection bias based on a selection of distributions for the treatment effect. We illustrate some properties of Bayesian estimates, providing shrinkage of the Phase 2 estimate to counteract the selection bias. We further give some empirical guidance regarding the choice of prior distribution and comment on the consequences for decision-making in investment and planning for Phase 3 programmes.

SUBMITTER: Wiklund SJ

PROVIDER: S-EPMC9290610 | biostudies-literature | 2021 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Selection bias, investment decisions and treatment effect distributions.

Wiklund Stig Johan SJ Burman Carl-Fredrik CF

Pharmaceutical statistics 20210517 6

When making decisions regarding the investment and design for a Phase 3 programme in the development of a new drug, the results from preceding Phase 2 trials are an important source of information. However, only projects in which the Phase 2 results show promising treatment effects will typically be considered for a Phase 3 investment decision. This implies that, for those projects where Phase 3 is pursued, the underlying Phase 2 estimates are subject to selection bias. We will in this article i ...[more]

PMID: 34002467

Similar Datasets

Project description:ImportanceA goal of health care is to reduce symptoms and improve health status, whereas continuing dubious treatments can contribute to complacency, discourage the search for alternatives, and lead to shortfalls in care.ObjectiveTo test a potential bias in intuitive reasoning following a marginal improvement in symptoms after a dubious treatment (post hoc bias).Design, setting, and participantsSurveys eliciting treatment recommendations for hypothetical patients were sent to community members throughout North America recruited via an online survey platform in the early winter months of 2023 and 2024 and presented to health care professionals (pharmacists who were approached in person using a secret shopper technique) in the summer months of 2023.ExposureRespondents received randomized versions of surveys that differed according to whether vague symptoms improved or remained unchanged after a dubious treatment.Main outcomes and measuresThe primary outcome was a recommendation to continue treatment.ResultsIn total, 1497 community members (mean [SD] age, 38.1 [12.5] years; 663 female [55.3%]) and 100 health care professionals were contacted. The first scenario described a patient with a sore throat who took unprescribed antibiotics; respondents were more likely to continue antibiotics after initial treatment if there was a marginal improvement in symptoms vs when symptoms remained unchanged (67 of 150 respondents [45%] vs 25 of respondents [17%]; odds ratio [OR], 3.98 [95% CI, 2.33-6.78]; P < .001). Another scenario described a patient with wrist pain who wore a copper bracelet; respondents were more likely to continue wearing the copper bracelet after initial care was followed by a marginal improvement in symptoms vs when symptoms remained unchanged (78 of 100 respondents [78%] vs 25 of 99 respondents [25%]; OR, 16.19 [95% CI, 5.32-19.52]; P < .001). A third scenario described a patient with fatigue who took unprescribed vitamin B12; respondents were more likely to continue taking vitamin B12 when initial treatment was followed by a marginal improvement in symptoms vs when symptoms remained unchanged (80 of 100 respondents [80%] vs 33 of 100 respondents [33%]; OR, 7.91 [95% CI, 4.18-14.97]; P < .001). Four further scenarios involving dubious treatments found similar results, including when tested on health care professionals.Conclusions and relevanceIn this study of clinical scenarios, a marginal improvement in symptoms led patients to continue a dubious and sometimes costly treatment, suggesting that clinicians should caution patients against post hoc bias.

Project description:Many institutions worldwide are considering how to include uncertainty about future changes in sea-levels and storm surges into their investment decisions regarding large capital infrastructures. Here we examine how to characterize deeply uncertain climate change projections to support such decisions using Robust Decision Making analysis. We address questions regarding how to confront the potential for future changes in low probability but large impact flooding events due to changes in sea-levels and storm surges. Such extreme events can affect investments in infrastructure but have proved difficult to consider in such decisions because of the deep uncertainty surrounding them. This study utilizes Robust Decision Making methods to address two questions applied to investment decisions at the Port of Los Angeles: (1) Under what future conditions would a Port of Los Angeles decision to harden its facilities against extreme flood scenarios at the next upgrade pass a cost-benefit test, and (2) Do sea-level rise projections and other information suggest such conditions are sufficiently likely to justify such an investment? We also compare and contrast the Robust Decision Making methods with a full probabilistic analysis. These two analysis frameworks result in similar investment recommendations for different idealized future sea-level projections, but provide different information to decision makers and envision different types of engagement with stakeholders. In particular, the full probabilistic analysis begins by aggregating the best scientific information into a single set of joint probability distributions, while the Robust Decision Making analysis identifies scenarios where a decision to invest in near-term response to extreme sea-level rise passes a cost-benefit test, and then assembles scientific information of differing levels of confidence to help decision makers judge whether or not these scenarios are sufficiently likely to justify making such investments. Results highlight the highly-localized and context dependent nature of applying Robust Decision Making methods to inform investment decisions.

Dataset Information

Selection bias, investment decisions and treatment effect distributions.

Publications

Selection bias, investment decisions and treatment effect distributions.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets