The title says it all:  Ictal eye closure is a reliable indicator for psychogenic nonepileptic seizures.

First, the bottom line: 

50/52 patients with pseudoseizures closed their eyes during their "seizure," while 152/156 of actual epileptics opened their eyes during their seizures.  That's a sensitivity of 96% and a specificity of 98%.  That's gold.

Now, the details: 

The authors took 234 consecutive "seizure" patients, hooked them up to video EEGs and stopped their medications.  There were 938 total ictal events in 221 patients.  52 (23%) had pseudoseizures, and 156 (70%) had eplieptic seizures.   There was a 3:1 female predominance in the pseudoseizures, no difference in epileptics.

In the epileptics, there was rhythmic eye blinking during tonic-clonic activity, and the eyes closed after theseizure was finished. 

An interesting corollary to this is when pseudoseizures occur in an actual epileptic: quoting the authors, "the simple question of eye opening or closure can help differentiate between the two types of events. One previous study found that patients with both types of events tended to have their eyes closed during PNES and open during ES.(6)"

Outstanding. 

Of course, this is isn't going to mean much to psychiatrists, apparently.

A questionnaire was put to neurologists (N=39) and psychiatrists (N=75) about the utility of video EEG in diagnosing pseudoseizures. 70% of the neurologists, but only 18% of the psychiatrists, thought that video-EEG was accurate "most of the time" in diagnosing pseudoseizures.  12% of the psychiatrists (no neurologists) said it is accurate "almost never."  (3% of the psychiatrists gave no clear response.  Why doesn't that surprise me?)

So here are some other differentiating symptoms:

In seizure patients, there is a crescendo-decrescendo quality to the spike-wave frequencies on EEG.  In pseudoseizure patients, however, the frequency is the same from beginning to end, and it comes on suddenly as if a switch was flicked.  The spike-wave on EEG is actually motion artifact, and typically runs around 4 Hz, while epileptics have frequencies that vary between 4-25 Hz.

In a study of 40 pseudoseizure vs. 40 matched normal controls,  the pseudoseizure group had more left handers, reduced strength and speed in both dominant and non-dominant hands, and reduction in the dominant hand advantage in strength and speed (i.e. both hands performed equally badly-- the dominant hand wasn't a little better.)  Interestingly and importantly, the authors did not think this was due to faking or psychological factors, but felt that it was due to actual neurologic impariment in bilateral pathways: 65% had had a closed head injury, 27% had had physical abuse, and 17% had had a history of substance abuse.  40% had an IQ less than 90!

A study in epileptics vs. pseudoseizure patients trying to determine how long after admission to a video EEG unit it takes for patients to have events (answer: 88% had it on day 1) also found that urinary incontinence, focal neurologic exams, and tongue biting were about the same in both groups.  But more epileptics had events less than one minute, and more pseudoseizures lasted > 5 minutes (and very few (13%) lasted less than one minute.)

Slightly different results were found in another study:  11/28 pseudoseizure patients had them  on day 1, but 9/28 needed an average of 5 days.  19/28 had an induced pseudoseizure to IV saline challenge within 3-7 minutes.  But still-- 3 days should be enough for most patients.

And alexithymia is of no value. It is found more often in epileptics and pseudoseizure patients equally, though still more than expected in the community. A larger, controlled trial had found a similar inability for alexithymia to differentiate: alexithymia was very common in epileptics (76%) and pseudoseizures (90%).  Thus, it is likely that alexithymia is a coping strategy, and not an independent trait.

Addendum 11/5/06: I did find an interesting (Greek) study finding an excess of seizures on full moons (34% vs. about 21% for the other phases.)   Importantly (and in contrast to suggestions by other studies) these were not pseudoseizures, because all patients were monitored.  The authors speculate either electromagnetic/gravitational effects (hey, it could happen) or an interaction between the intrinsic seizure threshold and the environment (i.e. you can change you rown threshold.)