Diagnostic accuracy of the Gerontological Personality Disorder Scale (GPS) in Dutch general practice

Objective: Personality disorders (PDs) often remain unrecognized in older adults by doctors in general practice. Therefore, this study evaluated the diagnostic accuracy of a screening instrument, the Gerontological Personality Disorder Scale (GPS), in a Dutch general-practice population of older adults.Method: The psychometric properties of the GPS patient (GPS-pv) and informant (GPS-iv) versions were assessed in a sample of 302 (144 male) patients (average age: 69.9 years) and 302 (124 male) informants (average age: 64.7 years), respectively, using an informant-based personality questionnaire (the Hetero-Anamnestische Persoonlijkheidsvragenlijst ) as a reference criterion.Results: The internal consistency (average item correlation) of the subscale and total scores of the GPS-pv and GPS-iv were .12 (HAB), .16 (BIO), and .10 (total); and .16 (HAB), .15 (BIO), and .12 (total), respectively. The test--retest reliability was strong for both the GPS-pv (r(s) = .56 [HAB], r(s) = .67 [BIO], r(s) = .66 [total]) and the GPS-iv (r(s) = .52 [HAB], r(s) = .65 [BIO], r(s) = .68 [total]) versions. The sensitivity and specificity of the GPS-pv were .83 and .27, respectively, with a cutoff score of 1. Raising the cutoff score to 2, the sensitivity dropped to .59, whereas the specificity rose to .57. For the GPS-iv, a cutoff score of 3 maximized the sensitivity (.78) and specificity (.65).Conclusion: The diagnostic accuracy of the GPS-iv was preferable to that of the GPS-pv. This is the first psychometric study to use the GPS as an age-specific screening instrument for PDs.