Estimands, Estimators, and Estimates With Dr Little
Aug 3, 2023
auto_awesome
Dr. Roderick J. Little, PhD, discusses estimates, estimators, and their importance in capturing the benefit or risk of a treatment. They explore estimation in clinical practice, intention to treat vs per protocol estimates, benefits and challenges of randomization, and alternative methods for estimating trial products.
Different estimates, such as the treatment policy estimate and the trial product estimate, are used to capture the benefit or risk of a treatment and inform decision-making based on the population assigned to treatment.
Valid and unbiased estimators are essential for accurately comparing outcomes, summarizing causal effects, and making informed decisions based on the findings of a clinical trial.
Deep dives
Definition and importance of estimates
An estimate is the value that we are trying to determine for a specific treatment in a target population. It captures the benefit or risk of the treatment and compares the summary effect of the treatment to a comparator. The estimate can be measured through different approaches, such as the treatment policy estimate (intention to treat estimate) that includes all individuals assigned to any treatment, regardless of compliance, and the trial product estimate (per-protocol estimate) that focuses on the outcome of individuals who actually comply with the assigned treatment. The use of these estimates depends on the specific context and the goals of the study.
The Pioneer 3 study and its estimates
The Pioneer 3 study was conducted to compare the effectiveness of different treatments for type 2 diabetes. It used two main estimates: the treatment policy estimate, which includes all individuals assigned to any treatment even if they did not comply, and the trial product estimate, which considers the outcome of individuals who actually complied with the assigned treatment. The treatment policy estimate is particularly relevant when making decisions about initial drug prescriptions as it reflects the potential benefits for the whole population assigned to treatment. The trial product estimate, on the other hand, is of interest to drug makers who want to evaluate the effectiveness of the treatment specifically in those who comply with it.
The role of estimators and characteristics of good estimates
Estimators play a crucial role in estimating treatment effects. They analyze the outcomes of a sample of participants in a study and develop an estimate based on the specific data set. Valid and unbiased estimates are key characteristics of good estimators. Good estimates should accurately compare outcomes that capture the main benefits and risks of the treatments being studied. They should summarize the causal effects of the treatments in the sample population and provide a valid and unbiased estimate of the study outcome. These characteristics are crucial for evaluating the results of a clinical trial and making informed decisions based on the findings.
Roger J. Lewis, MD, PhD, discusses Estimands, Estimators, and Estimates with Roderick J. Little, PhD, in the JAMA Guide to Statistics and Methods series. Related Content: