Search for a command to run...
Batista et al.1 demonstrate the added value of inferior electrodes in a properly blinded study. At first reading, I was surprised that 12% should correspond to a number-needed-(NN)-to-test of 22. The STARD initiative2 makes recommendations on reporting both blinding and visualization. From the article, such a flowchart was constructed to illustrate that with 25 electrodes, temporal IEDs were detected in 71 or 72 EEGs; otherwise only in 63 (Figure 1). This corresponds to missing 12% of temporal IEDs with the 10–20 system, resulting in a NN-to-test of 8 for temporal epilepsy. However, from an intention-to-test perspective, all patients, EEGs, and localizations are to be counted equally. Superiority is then seen in 4.5%. The NN-to-treat is well established to quantify the effect of drugs or other interventions and to put efficacy into perspective. Stone et al. (Christmas 2002)3 were probably the first to apply the NN to diagnostic neurology in a paper on communicating the diagnosis to patients, and also on communicating statistics to physicians. More recently, Schubert et al.4 found the NN helpful to build a bridge from diagnosis to prognosis and for navigating the borderland of acute symptomatic seizures and epilepsy. To communicate the yield of the inferior chain, Batista et al.1 estimate the NN-to-test as 22. In other words, in their population six electrodes have to be added in 22 consecutive EEGs for one additional patient to be diagnosed with IEDs (and therefore with epilepsy). Reproducibility is at least as important as yield or sensitivity. Their paper reports intra-reader kappa only. It appears that the two reviewers were also blind to one another's readings, which should allow assessment of inter-rater agreement. I would hypothesize that better representation of temporal IEDs on inferior electrodes also improves concordance. Therefore, I ask the authors to provide such 2 × 2 tables to assess reliability with and without the inferior chain. Bert Kleine constructed the figure and wrote the manuscript. The author has no potential conflicts of interest to disclose. Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.