Outcomes in health studies designed to compare different treatments (commonly referred to as randomized clinical trials) are often scheduled for assessment at specific times after randomization (for example, at 6 and 12 months). However, these assessments may occur before or after the scheduled time, or not at all. As a result, the number and timing of assessments can vary among study participants, generating an irregular pattern of assessment times. The degree of irregularity can be higher among participants who are poorer or have other challenging life situations. Furthermore, the assessment time may relate to a participant’s health status. For example, a participant experiencing a worsening of their illness may postpone their appointment until a time they are feeling better. When this happens—that is, when the health outcome being studied is related to the time it is assessed—the assessment time is called “informative.”
Current statistical methods used to deal with irregular assessment times rely on making somewhat arbitrary choices about which data points to include or exclude, or on making assumptions that may or may not be true. With missing data, a statistical technique called sensitivity analysis is generally recommended. However, no such technique has been developed for irregular and potentially informative assessment times.
This project has four aims. For the first aim, the project team will develop a statistical method to perform sensitivity analyses of study datasets with irregular and informative assessment times. The method depends on the selection of a range of parameters, called sensitivity parameters. Without additional information or assumptions, the range of these parameters is unknowable. For the second aim, the team will introduce a reasonable assumption and associated method that can be used to restrict the range of sensitivity parameters. For the third aim, the project will develop user-friendly software and an easy-to-use web-based application to enable other researchers to apply the project’s statistical methods. For the fourth aim, the team demonstrate the new statistical methods and software using data from four completed clinical trials conducted in populations with limited socioeconomic resources.
The project’s main product will be a sensitivity analysis software tool for randomized clinical trials with irregular and potentially informative assessment times. This tool will help researchers evaluate the reliability and strength of findings in studies with irregular assessment times as well as more accurately analyze and report their study findings.
The project will be executed by researchers with experience in developing statistical methods, creating software and online educational programs, and conducting clinical trials in populations with limited socioeconomic resources. The team includes two principal investigators of PCORI-funded randomized clinical trials conducted in low-income populations, three biostatisticians, a software developer, and a Stakeholder Advisory Board. The Stakeholder Advisory Board, which includes two principal investigators of other PCORI-funded studies, two biostatisticians, two patients, one trialist, one social scientist, and one implementation scientist, will provide guidance on all aspects of the project.
*All proposed projects, including requested budgets and project periods, are approved subject to a programmatic and budget review by PCORI staff and the negotiation of a formal award contract.