In light of the ongoing debate whether agoraphobia [AG] should be viewed as a severe phobic disorder similar to specific phobia [SPE] or as a complication of panic disorder [PD] we aim to study the vulnerability structure of PD, AG and SPE.

3021 14-24 year-olds from the general population were followed-up over 10 years. DSM-IV syndromes were assessed via computerized M-CIDI interview and vulnerability factors via questionnaires. Associations were assessed with odds ratios from logistic regression. Latent class analysis (LCA) regressed on vulnerability factors was used to derive classes that underlie panic and phobic syndromes and to assess their associations with vulnerability factors.

1. 1. Vulnerability patterns were largely similar between PD, AG and SPE.

2. 2. The LCA resulted in a best fitting model with 4 classes: a healthy class, a class with moderate frequency of phobias without PD, a class characterized by PD and AG and moderate frequency of SPE (PDAG class) and one class characterized by high frequency of AG and SPE situational type and lower frequency of PD (AGSIT class).

3. 3. All classes showed different associations with multiple vulnerability measures. Subjects in the PDAG class reported less SPE in parents (OR=0.2; 95% CI=0.0-0.6) and older onset-age of any psychopathology (OR=2.0; 95% CI=1.07-3.6) than the AGSIT class.

We found indications for separate latent classes underlying PD and phobias that were characterized by different vulnerability factors. We interprete the different classes as different vulnerability clusters and evidence of multiple pathways leading to panic and phobias.