Long Exercise Test in Periodic Paralysis: A Bayesian Analysis.

Area of Science:

• Neurology
• Clinical Electrophysiology

Background:

• The long exercise test (LET) is crucial for diagnosing periodic paralysis (PP).
• Current LET methodologies and normal cutoff values lack standardization, impacting diagnostic accuracy.

Purpose of the Study:

• To establish optimal LET methodology and cutoff values for diagnosing periodic paralysis.
• To compare peak-to-nadir versus baseline-to-nadir approaches and amplitude versus area decrements.

Main Methods:

• Retrospective analysis of LET data from 55 PP patients and 125 controls.
• Construction of receiver operating characteristic curves to determine area under the curve (AUC).
• Bayesian principles applied to calculate optimal decrements for 95% posttest probability of PP.

Main Results:

• Peak-to-nadir methodology yielded the highest AUC, with equal performance for amplitude and area decrements.
• Optimal peak-to-nadir cutoffs for pretest probabilities ≤50% were >40% amplitude or >50% area.
• These findings support specific cutoffs for confirming PP diagnosis.

Conclusions:

• The peak-to-nadir LET methodology is endorsed for periodic paralysis diagnosis.
• Recommended cutoffs are 40% amplitude decrement or 50% area decrement for pretest probabilities ≤50%.