Are classification accuracy statistics underused in neuropsychological research?

Area of Science:

• Neuropsychology
• Neurology
• Medical Statistics

Background:

• Classification accuracy statistics are crucial for evaluating diagnostic tests.
• Their application in neuropsychology research may be suboptimal.
• A comparison with neurology journals is warranted.

Purpose of the Study:

• To determine the prevalence of classification accuracy statistics in neuropsychology and neurology journals.
• To assess the completeness of data reported for these statistics.
• To identify underutilized statistical measures in neuropsychology.

Main Methods:

• A systematic review of articles published in five prominent neuropsychology and five leading neurology journals in 2000-2001.
• Calculation of the prevalence of classification accuracy statistics.
• Assessment of data sufficiency for full statistical analysis.

Main Results:

• Only 29% of appropriate neuropsychological articles had sufficient data for full classification accuracy analysis.
• Classification accuracy statistics were significantly less prevalent in neuropsychology than neurology journals.
• Sensitivity/specificity were reported in 31% of neuropsychology articles, but predictive values/risk ratios in fewer than 3%.

Conclusions:

• Classification accuracy statistics, particularly predictive values and risk ratios, appear to be underused in neuropsychology.
• Neuropsychological research may benefit from increased utilization of these statistics for evaluating clinical relevance.
• Further adoption of these methods can enhance the interpretation of research findings.