How to avoid time-related types of bias in the analysis of clinical infectious diseases: demonstration and methods

Abstract

Background: Competing risk and immortal time bias can significantly affect the interpretation of analyses in infection disease research where patients may experience multiple outcomes. These biases can distort study results, leading to misestimation of event rates or therapy effects, and ultimately misinform clinical decisions.

Objective: This article aims to explain the concepts of competing risk and immortal time bias while providing a clear understanding of how these biases arise and offering practical methods for addressing them in clinical studies. A study of patients in the intensive care unit (ICU) is used as an illustrative example.

Sources: References were compiled through searches of MEDLINE/PubMed and Google Scholar up to December 2024.

Content: Competing risk bias occurs when a patient is at risk for more than one mutually exclusive event, such as death from different causes, which can skew survival analyses if not properly accounted for. Immortal time bias arises when periods during which the outcome cannot occur for patients with an intermediate exposure are improperly included in a survival analysis. This leads to altered effects and duration of stay estimation. In the ICU data example, ignoring competing risks results in impossible probability estimates and relative overestimation of incidence by as much as 52%. Ignoring time dependencies leads to the flawed conclusion that bloodstream infection is protective against patient death in the ICU, whereas a valid approach results in a significant harmful effect. Solutions such as multi-state models or extended Cox regression models are presented to mitigate these biases. Examples from published literature with proper handling of these biases are provided.

Implications: Properly accounting for competing risk and immortal time biases ensures more reliable and valid results, ultimately guiding better clinical decision-making and improving patient outcomes in the management of infectious diseases.

Keywords: Bias mitigation; Common mistakes; Competing risk bias; Cox regression; Immortal time bias; Methodological pitfalls; Multi-state model; Survival analysis.