(PS8-D81) Comparing Existing Self-report Measures for Borderline Personality Disorder in a College Sample

Due to the number of self-report measures of borderline personality disorder (BPD), selecting which measure to use clinically or scientifically is often difficult. Researchers and clinicians have speculated that inconsistent research findings may be due to the use of different self-report measures for the same construct (Zimmerman et al., 2020). Moreover, a recent meta-analysis found wide-ranging prevalence estimates of BPD in college samples, and the authors speculate that the use of differing BPD measures might be the culprit (Meaney et al., 2016). To address this issue, we compared BPD prevalence estimates resulting from four often-used self-report measures of BPD. A collegiate sample (N = 318) completed the BSL-23 (Bohus et al., 2009), the MSI-BPD ( Zanarini et al., 2003), the PAI-BOR (Morey, 1991), and the PID-5-BPD (Maples et al., 2015). Using cut-off scores recommended in previous research, we derived a proportion of the sample that likely met BPD criteria. Of all measures investigated, the PID-5-BPD had the highest prevalence estimate, with 25.78% of the sample meeting recommended cut-offs (i.e., x ≥ 8.85). This was followed closely by the PAI-BOR at 24.84% (x ≥ 38), and the MSI-BPD where 20.44% of the sample met the recommended cut-off (x ≥ 7). Finally, the BSL-23 was the most conservative measure, as 15.09% of the sample exceeded the recommended cutoff (i.e., x ≥ 1.5). We then looked at the agreement of each of these measures, to assess the proportion of our sample that met the recommended cut-off of each measure. We found that 6.92% of our sample met each cut-off score, this is juxtaposed to 37% of our sample that met at least one cut-off score. Taken together, these results support researchers’ concerns that heterogeneity in research methods (i.e., choice of BPD self-report measure) may be leading to heterogeneity in BPD prevalence estimates in college samples. Comparing these two groups specifically (e.g., those who exceeded all cutoffs, and those who did not meet all but met at least one), we found that the former group tended to score significantly higher in emotion dysregulation (t(137) = -5.82, p < .001) suicidality (t(137) = -3.62, p < .001), and impulsivity (t(137) = -3.46, p < .001). Taken together, these findings indicate that the cut-off scores between measures may need to be re-tuned to better suit the college population. More specifically, researchers may wish to investigate the false positive and negative rates within each of the respective measures.  In so doing, we may better understand which measures may be under or overinflating prevalence estimates.