What are the risks or downsides of CBT-informed therapy for PNES/FND?

The short version: the evidence for CBT-informed therapy in PNES/FND shows real benefit, but the risks are almost never studied, and the biggest risk may be that the wrong diagnosis gets treated.

The retrieved evidence is mostly about whether CBT works, not about its harms. One framework review notes that about half of PNES patients still have seizures at follow-up, and less than one in three achieve seizure freedom up to 10 years after diagnosis [1]. That is the honest baseline: CBT helps some, but most are not cured. A group CBT intervention for patients with both epilepsy and PNES improved quality of life and reduced depression and anxiety, but it was not randomized and had no control group [2]. A motivational interviewing study showed better therapy adherence and larger seizure reduction, but again the comparison was psychotherapy alone, not no treatment [5].

What the retrieval does not contain is any study measuring harm. No record reports worsening of symptoms, misdiagnosis rates, or adverse events from CBT. That silence is not safety. The most serious downside is structural: PNES is often misdiagnosed as epilepsy first. The average delay from onset to diagnosis runs 1 to 16 years [1], and about 1 in 10 PNES patients also has true epilepsy [3]. If CBT is offered before that distinction is nailed down with video-EEG, a real seizure disorder can go untreated while the person is told the problem is psychological.

The other risk is what CBT cannot touch. PNES patients carry heavy psychiatric burdens: in one study, 82% had anxiety and 56% had mood disorders [4]. CBT that focuses only on seizure control can leave the underlying trauma, dissociation, or family dysfunction unaddressed. The evidence here does not measure whether symptoms simply shift from seizures to other somatic complaints.

My call: CBT-informed therapy for PNES is worth trying, but only after video-EEG confirms the diagnosis, and the evidence base has no real safety data, only efficacy data. Confidence: moderate on benefit, low on risk.