Canadian Journal of Cardiology



      Vasovagal syncope (VVS) is a common clinical condition that lacks effective medical therapies despite being associated with significant morbidity. The norepinephrine transport inhibitors reboxetine, sibutramine, and atomoxetine (Strattera) all prevent the induction of vasovagal syncope on tilt table testing. We hypothesized that atomoxetine would be effective in suppressing syncope in patients with recurrent VVS.

      Methods and Results

      This was a retrospective, open-label, observational case series of 12 patients taking atomoxetine for compassionate-use suppression of recurrent vasovagal syncope. We compared syncope frequency in the periods 1 year before atomoxetine and while subjects were taking atomoxetine. We used novel applications of the Poisson distribution to describe the results as a collection of n=1 studies. The Poisson distribution is ideal for assessing the significance of distributions with few events per subject. There were 12 subjects, 8 female, with mean age 47+/-22 years and a mean Calgary Syncope Score of 2. The dose of atomoxetine that the patients received was 66+/-16 mg (1.06+/-0.21 mg/kg). In the preceding year they had had a mean 9.5+/-11.1 syncopes and a median 5.5 (IQR 4, 6.75) syncopes. While taking atomoxetine all patients appeared to improve and 8/12 had no syncope in followup (p = 0.0013). The mean syncopes per year decreased from 9.5+/-11.1 to 0.51+/-0.92 (p=0.019, T test). Syncope frequency decreased from a median 5.5 (IQR 4, 6.75) syncope per year to 0 (IQR 0, 0.88) syncope per year (p=0.0015, Wilcoxon test). All 4 patients who fainted in follow-up improved from a previous year count of 5.75+/-1.26 syncopes to 1.52+/-1.03 syncopes per year on atomoxetine (p=0.0006). According to the Poisson distribution 7/12 subjects were each significantly improved with p values of < 0.0001 to 0.023, and an eighth subject had borderline significant improvement (p=0.082). Of the 5 subjects who did not improve significantly one fainted once and 4 did not faint but lacked a significantly long follow-up duration. Therefore, in total 8/12 subjects were significantly or nearly significantly improved (p=0.005, T test), and 4/12 subjects had insignificantly long follow-up times to test whether they responded significantly to atomoxetine.


      In this case series atomoxetine was effective in preventing vasovagal syncope. The novel use of the Poisson distribution permits per patient assessment of improvement and detects insufficient followup despite apparent improvement. Atomoxetine merits a formal randomized trial. The Poisson distribution merits further study for n=1 clinical trials.