Every medical device clinical study has endpoints, although they may be called something else, such as outcomes or metrics. Regardless of the name, we are talking about the objective, measurable results of a clinical study that are used to evaluate safety, effectiveness, and benefit to patients. The overall success or failure of the study will be judged against the primary endpoint criteria, so this is a critical part of your clinical study design. In this post, we share five key qualities of well-defined primary endpoints.
Overview of primary endpoints
As the name implies, the primary endpoint is the most important endpoint in a rigorously-defined clinical study, the one on which overall success or failure will be declared. For FDA-regulated studies, two primary endpoints are often named – one for safety and one for effectiveness –although the latter is not always mandated. In addition to establishing how scientific objectives will be evaluated, primary endpoints also help determine the size of the study and the statistical methodologies needed to assess the data. A carefully defined endpoint can lower your sample size, accelerate your enrollment, and reduce overall cost, all while preserving your clinical study objectives.
So, how do you determine your primary endpoint? While most clinical studies present a broad set of choices, the primary endpoint may be obvious based on prior studies or commonly accepted metrics, such as therapy-specific consortium guidance documents or FDA precedent. For example, studies in heart failure treatment tend to rely on mortality and/or hospitalization to assess outcomes, while stroke therapies often use the widely accepted modified Rankin Scale to measure the degree of disability. If the primary endpoint is not obvious, you may need to review the literature or confer with key opinion leaders to determine how best to evaluate your product.
No matter how you envision your primary endpoints, it’s critical to ensure the criteria and definitions are carefully worded. Let’s dive in and discuss five key qualities for well-defined endpoints.
1. Clinically valid
Primary endpoints should be accepted and understood by the clinical community. While it may be tempting to pick a novel endpoint, remember your audience, and ensure you are using standardized terminology (we recommend citing references to support your study definitions unless they are broadly understood), and assessments that are meaningful to your users.
Primary endpoints should be measured free of bias by the person responsible for making the measurement. The “person responsible” can be a medical professional, such as the treating physician or site staff member, the enrolled subject, or a third party, such as a core laboratory contracted to provide consistent evaluation of imaging or lab values. For randomized studies, consider blinding the subject and/or the assessor to the treatment the subject received to minimize bias.
Primary endpoints should measure “something the subject can see and feel.” This is language often used by FDA and forms a part of regulatory thinking on good clinical study design.
Primary endpoints should be defined with specificity. When considered in its entirety, the complete set of data collected during a clinical study may give a picture of overall health or benefit, but endpoints are meant to target specifics. For example, rather than an endpoint for “quality of life,” it is desirable to use a standardized quality of life assessment at specific timepoints.
5. Statistically practical
Primary endpoints should be achievable. You “win” in a statistical sense when analysis of your data, using pre-defined statistical methods, culminates in a significant p-value. The size and length of the study needed to achieve this are related to the choice of primary endpoints.
A classic clinical study primary endpoint is all-cause mortality (i.e., death regardless of the underlying cause). This endpoint is appropriate for studies where death is a substantial risk for the patient population, such as life-saving medical devices for therapeutic areas like neurology, oncology, and cardiology. In consideration of the five key qualities above, an all-cause mortality endpoint:
- Meets criteria 1, 2, 3, and 4 (clinically valid, objective, direct, and specific)
- May or may not meet criterion 5 for a particular clinical study. Depending on the population being studied, mortality may simply not occur often enough to produce sufficient statistical evidence for p-values. A study in which mortality is expected to occur infrequently will likely have to be enormous to provide statistically valid evidence of device benefit. This suggests choosing a different endpoint since there is little value in running a study that you cannot “win.”
It is a challenging process to identify and select optimal endpoints, but it is well worth the effort considering the overall success or failure of your study will be judged based on your primary endpoint criteria and definitions. In review, the endpoint should be customized to both the field of medicine and the specific patient population expected to benefit from the treatment. For the best outcomes, ensure your study endpoints are clinically valid, objective, direct, specific, and statistically practical.