Design: Randomised, 3-arm comparative effectiveness trial, 6 US emergency departments
Population: 375 adults discharged after low-risk chest pain with at least moderate anxiety or panic
Primary outcome: Change in GAD-7 anxiety score over 12 months
Effect: Internet CBT 1.22 points greater improvement than primary care referral (95% CI 0.01–2.43)

PACER randomised 375 adults discharged from six US emergency departments after low-risk chest pain who screened positive for at least moderate anxiety or panic. All received psychoeducation, then either a recommendation to see their primary care doctor, peer-supported internet-based CBT, or therapist-delivered CBT by video. Mean age was 40.

Anxiety scores improved in all three groups over 12 months. Internet CBT improved them by 1.22 points more than primary care referral on the 21-point GAD-7 scale, a small effect, rising to 2.8 points among those with severe anxiety. Patients in either CBT group were three times as likely to report their anxiety as improved. Depression, somatic symptoms and disability did not differ between groups.

The patient who returns repeatedly with chest pain and normal tests is familiar on the medical take. This trial suggests a structured, low-intensity psychological intervention adds something, mostly for the most anxious. The difference over usual referral was small, and the lower confidence limit was close to zero.

Screen for anxiety and panic in patients with recurrent low-risk chest pain once a cardiac cause is excluded.
The GAD-7 (8 or more) and a two-item panic screen were the thresholds used here.
Consider structured CBT, including guided online programmes, for those with severe anxiety, where the benefit was largest.
Name the diagnosis clearly at discharge; psychoeducation alone produced improvement in every arm.

Why it matters

It treats anxiety as the diagnosis to act on, rather than the leftover after a negative workup.

The statistics, in plain English

The difference of 1.22 points has a 95% confidence interval of 0.01 to 2.43, so it is only just statistically significant, and an effect size of 0.25 is conventionally called small. The larger effect in severe anxiety comes from a subgroup and is less certain than the main result.