Related Experiment Videos

Interobserver agreement in the classification of stroke in the physicians' health study

K Berger1, C S Kase, J E Buring

  • 1Department of Medicine, Brigham and Women's Hospital, Boston, Mass, USA.

Stroke
|February 1, 1996
PubMed

Insights

Interobserver agreement for stroke classification is high for major types and hemorrhagic subtypes but varies for ischemic subtypes. Periodic reclassification is recommended for long-term prospective studies to ensure diagnostic consistency.

Area of Science:

  • Neurology
  • Epidemiology
  • Clinical Trials

Background:

  • Prospective studies often rely on medical records for cerebrovascular endpoint evaluation.
  • Ensuring diagnostic consistency over time and adequate medical record information is crucial.
  • The Physicians' Health Study assessed aspirin's effect on cardiovascular events, including stroke.

Purpose of the Study:

  • To evaluate interobserver agreement for stroke classification within the Physicians' Health Study.
  • To address challenges in medical record adequacy and diagnostic consistency.
  • To assess the reliability of stroke endpoint classification in a large prospective trial.

Main Methods:

  • Stroke events (1982-1988) were classified by subtype, severity, and diagnostic certainty from medical records.
  • Independent reclassification of 216 stroke events occurred in 1994.
  • Kappa statistics were used to compare initial and subsequent classifications.

Main Results:

  • Excellent agreement was found for major stroke types (kappa = 0.81) and hemorrhagic stroke subtypes (kappa = 0.95).
  • Agreement for ischemic stroke subtypes showed wide variation (kappa = 0.13 to 0.96).
  • Substantial agreement was observed for stroke severity (kappa = 0.71), but fair for diagnostic certainty (kappa = 0.33).

Conclusions:

  • Interobserver agreement is high for major stroke types and hemorrhagic subtypes based on medical records and imaging.
  • Classification of ischemic stroke subtypes demonstrates significant interobserver disagreement.
  • Periodic reclassification of endpoints in long-term studies may mitigate misclassification.
Abstract

Related Concept Videos