The conclusion reached by Kelleher & DeVylder ^{Reference Kelleher and DeVylder1} that hallucinations are no more prevalent in borderline personality disorder (BPD) than in depression and anxiety was unexpected and raises some interesting points. The prevalence of BPD in the community population they studied is low (0.4%) compared with previously published prevalence rates of between 0.7 and 2.7%. ^{Reference Trull, Jahng, Tomko, Wood and Sher2} This suggests that the Adult Psychiatric Morbidity Survey may lack sensitivity for detecting BPD in a community sample (57% response rate; possible self-selection bias). Similarly, the correlation between hallucinations and BPD in this study is considerably lower than clinical experience and research estimates would suggest. Schroeder et al reported rates between 20 and 50% in a review of findings on the prevalence and clinical management of psychotic symptoms, particularly hallucinations, in BPD. The disparity between these findings may be explained by the use of community versus clinical sampling, since community sampling may underreport severe presentations of BPD, which in turn are more likely to be accompanied by hallucinations. ^{Reference Slotema, Niemantsverdriet, Blom, van der Gaag, Hoek and Sommer4}

Well-conducted studies have demonstrated that hallucinations in BPD are highly correlated with the experience of childhood trauma, including childhood sexual abuse. ^{Reference Schroeder, Fisher and Schafer3} Childhood trauma is not as prevalent in patients with depression and anxiety. Although a similar prevalence of hallucinations in BPD, depression and anxiety, as reported by Kelleher & DeVylder, could suggest that previous prevalence estimates in BPD may be compounded by these co-occurring disorders, ^{Reference Zanarini, Frankenburg, Hennen, Reich and Silk5} we regard this as unlikely.

The probable association between hallucinations and severity of BPD presentation ^{Reference Slotema, Niemantsverdriet, Blom, van der Gaag, Hoek and Sommer4} suggests that hallucinations might respond well to effective treatment. This is our clinical experience, supported by a small pilot study we conducted involving 38 women diagnosed with BPD (aged 18–56 years at intake). The presence and nature of hallucinations was recorded using the Psychotic Symptom Rating Scale (PSYRATS). At intake, 34% of participants reported hallucinations (PSYRATS score: 15.30 s.d. = 17.22); 50% also reported a history of childhood sexual abuse (Childhood Trauma Questionnaire). After 12 months of individual psychotherapy using a common factors approach, the PSYRATS score was 7.00 (s.d. = 13.93; P = 0.04). This correlated with a reduction in the number of BPD symptoms assessed using the Structured Clinical Interview for DSM-IV Axis II Personality Disorders (SCID-II: intake: 7.00 (s.d. = 1.63); 12 months: 4.60 (s.d. = 1.84); P = 0.01). These preliminary data suggest, in agreement with previous findings, ^{Reference Slotema, Niemantsverdriet, Blom, van der Gaag, Hoek and Sommer4} that hallucinations may positively correlate with BPD severity. Our findings also suggest that treating BPD using appropriate psychotherapy reduces the experience of hallucinations.