Eleven Items to Flag Borderline Features in Early Adolescence
- In 112 adolescents aged 11–14 attending a community child and adolescent service, the Greek Borderline Personality Features Scale for Children-11 (BPFS-C-11) self-report separated clinician-rated cases from non-cases with an area under the ROC curve of 0.86.
- At the optimal cut-off of 26 on the summed score, sensitivity was 0.88 and specificity was 0.75 – the scale catches roughly nine of ten affected adolescents while wrongly flagging about one in four who are unaffected.
- Internal consistency was strong (omega = 0.81), but the original 11-item unidimensional model fit the data poorly, whereas a trimmed 9-item version fit well (RMSEA = 0.058, CFI = 0.986, TLI = 0.982).
- Higher scores tracked the dysregulation profile of the Youth Self-Report, supporting convergent validity: the instrument indexes the affective-behavioural dysregulation that sits at the core of emerging borderline pathology.
Clinicians still hesitate to name borderline pathology before eighteen, yet the disorder announces itself in early adolescence and responds better to treatment the earlier it is caught. What most services lack is a fast, free way to decide which 12- or 13-year-old warrants a closer look. The BPFS-C-11 is a candidate for exactly that triage role, and this Greek validation is the first to hand it a working cut-off in a real outpatient sample.
What the numbers actually show
The instrument is a self-report short form of the Borderline Personality Features Scale for Children: statements rated on a five-point scale and summed into a single total. Against the clinician's own evaluation, the summed BPFS score discriminated cases with an area under the ROC curve of 0.86 – solid for a brief screen. At a cut-off of 26, sensitivity reached 0.88, so the scale missed few genuinely affected adolescents; specificity of 0.75 means a quarter of unaffected adolescents also crossed the threshold. For an instrument whose job is to decide who gets assessed, not who gets diagnosed, that trade-off is defensible.
The factor structure carried a warning. The eleven-item unidimensional model did not fit; trimming to nine items produced a clean structure (RMSEA = 0.058, CFI = 0.986, TLI = 0.982). Reliability held at omega = 0.81, and item-response-theory analysis was applied to the items. Convergent validity was demonstrated against the Youth Self-Report dysregulation profile – reassuring, since affective and behavioural dysregulation is precisely the clinical signature the scale is meant to detect.
How to use it
Treat it as a first-stage screen in adolescents around 11–14 in outpatient or community settings, never as a diagnosis. Have the young person complete the scale, sum the items exactly as specified in the instrument's published scoring key, and read a summed score at or above 26 as a positive screen that triggers a structured clinical assessment; before clinical use, confirm against the source paper or manual which item set the cut-off of 26 is built on. A positive result is a prompt to look, not a label to apply. Read the specificity honestly: at this cut-off one in four screen-positive adolescents will not have borderline pathology, so every positive needs clinical confirmation and a single administration should never stand alone. Keep the full eleven items for now – the nine-item refinement is promising but rests on one sample of 112.
A total of 26 or more is a reason to assess, never a diagnosis to hand out.
Single-site Greek sample of 112 adolescents, cross-sectional design, and criterion validity anchored to clinical judgment rather than a structured diagnostic interview; the nine-item refinement and the cut-off both need replication before wider adoption. The published abstract does not specify whether the cut-off of 26 applies to the 11-item score or to the reduced 9-item score (full text paywalled), so verify against the full paper or the instrument's manual before using the threshold clinically.