How-To Guides

Yes. In this South Australian public hospital cohort, patients with borderline personality disorder had the highest risk of hospital re-presentation compared with every other diagnostic group examined. In the mixed-effects survival models, borderline personality disorder was the referent category, and all non-BPD diagnostic groups had hazard ratios below 1, indicating lower risk of returning to an emergency department or being readmitted as an inpatient. The overall effect of diagnosis was significant in both men (c7b2b8 = 3210, P < .001) and women (c7b2b8 = 3697.1, P < .001).

Approximately 4% of mental health patients in the study period were discharged from an emergency department or inpatient admission with a primary diagnosis of borderline personality disorder. The study included 25,610 mental health patients and 86,740 emergency department and inpatient episodes across 20 South Australian public hospitals from 2014 through 2019.

No. In patients with borderline personality disorder, a longer initial inpatient stay was not associated with a longer interval before the next hospital presentation. The mean time to a second emergency department and/or inpatient presentation was 199 days for length of stay under 2 days, 179 days for 237 days, 179 days for 8314 days, and 142 days for more than 14 days.

The only statistically significant contrast was that a first admission lasting more than 14 days was associated with a second hospital presentation 58 days sooner than a stay under 2 days (P = .036). No other pairwise contrasts were statistically significant.

Patients presenting with self-harm had a significantly higher risk of hospital re-presentation than patients classified as self-harm/adjustment disorder. This difference was significant in both male and female patients (P < .001 for both). By contrast, self-harm alone did not differ significantly from non-BPD personality disorder for re-presentation risk in men (P = .457) or women (P = .738).

This was a retrospective cohort study using routinely collected, deidentified data from 20 metropolitan and regional public hospitals in South Australia between January 1, 2014, and December 31, 2019. The cohort included 25,610 mental health patients, with 13,320 men and 12,290 women.

To model recurrent presentations over time, the investigators used mixed-effects parametric survival modeling and adjusted for diagnosis, sex, age, age-squared, hospital length of stay, region, and whether the hospital journey involved emergency department care only or an inpatient admission. For the exploratory analysis of length of stay, they used inverse-probability-weighted regression adjustment to estimate potential-outcome means and average treatment effects.

The authors identified several important limitations. First, the dataset included only a limited set of patient characteristics, so potentially relevant factors such as ethnicity, education, social support, unemployment, and homelessness were not measured. Second, the retrospective cohort design means the study cannot establish causal pathways between patient characteristics and repeat hospital presentations.

The study was also limited to public hospitals in one Australian state, so the findings may not generalize to other settings and do not capture private-sector care. The authors note that, in a national study, 16% of Australians with borderline personality disorder reported using only private hospital services.

The findings suggest that, in this dataset, longer initial inpatient stays for borderline personality disorder were not associated with a longer time before the next hospital presentation. Specifically, patients whose first admission lasted more than 14 days returned to hospital sooner on average than those whose stay was under 2 days, with a 58-day shorter mean time to second presentation (P = .036).

The authors do not conclude that length of stay causes earlier return, because this was an observational study. They note that inpatient admission is not generally recommended as primary treatment for borderline personality disorder, although admission may sometimes be necessary, especially when complex comorbidities are present.