UMHS LOGOUniversity of Michigan
Department of Pediatrics

Evidence-Based Pediatrics Web Site

Historical Features Can Be Used to Differentiate Epileptic and Non-Epileptic Staring Spells


  • In children with staring spells, are there historical elements that can help differentiate epileptic and non-epileptic staring spells and how reliable are they?

Clinical Bottom Lines

  1. The age at evaluation, duration of staring spells, and frequency of staring spells were not significantly different between epileptic and non-epileptic episodes.
  2. The combination of responsiveness to touch and no interruption in playing effectively ruled in non-epileptic spells (LR+ = 3.7).  The presence of body rocking also effectively ruled in non-epileptic spells (LR+ = infinite).
  3. The presence of twitching of the arms or legs (LR+ = infinite) or the presence of urine loss (LR+ = 3.9) effectively ruled in epileptic etiology for spells.  The presence of upward eye movements greatly increased the likelihood of epileptic etiology.  Occurrence of spells when tired was somewhat predictive, but the specificity for epileptic etiology increased greatly when it was combined with upward eye movements.

Summary of Key Evidence

  1. A wriiten questionnaire was administered to parents of 40 children who presented to a neurology clinic with staring spells..
  2. A diagnosis of non-epileptic staring was made in 23 of the children.  All had normal interictal EEG and no other types of events were noted.  Ten of the children had video EEG and all were normal.  None of the children developed seizures during the follow-up period of over one year.
  3. Some selected likelihood ratios are shown below: 
Non-epileptic staring spells: Specificity Sensitivity LR+
a. No interruption of play 0.88 0.48 4
b. First seen by professional 0.88 0.43 3.6
c. Responsive to touch 0.87 0.56 4.3
d. Body rocking 1.00 0.13 Infinite
    a and b 0.94 0.22 3.7
    a and c  1.00 0.31 Infinite
    b and c 1.00 0.31 Infinite
Epileptic staring spells
a. Twitching - arms or legs 1.00 0.23 Infinite
b. Urine loss 1.00 0.13 Infinite
c. Occurrence when tired 0.74 0.58 2.2
d. Upward eye movements 0.91 0.35 3.9
    c and d 0.96 0.29 7.3

Additional Comments

  • The study size is small, leading to a large range in the likelihood ratios.  For instance, body rocking had a range of specificity from 0.8 to 1.0 and a range of sensitivity from 0.03 to 0.34.  Thus, the LR+ could range from infinity to as little as 0.2 within a 95% confidence interval.
  • The population differed from a standard clinic population.  These children were those referred for further evaluation.  The prevalence of epileptic events in this population was 43%, compared to an incidence of 9.6/100,000 in the broader population which was quoted in the article.


  1. Rosenow F, Wyllie E, Kotagal P, Mascha E, Wolgamuth B, Hamer H.  Staring spells in children: Descriptive features distinguishing epileptic and non-epileptic events.  J Peds 1998; 133:660-663.

CAT Author: Dan Elsholz, MD

CAT Appraisers: John Frohna , MD

Date appraised: February 17, 1999

Last updated June 14, 2003
Department of Pediatrics and Communicable Diseases
© 1998-2002 University of Michigan Health System