The SCID-II Personality Questionnaire is available as a screening tool to shorten the time it takes to perform the SCID-II interview. However, very little is known about combining the questionnaire with the interview in eating disordered individuals. The aim of this study was to examine the prevalence of personality disorder (PD) among patients with eating disorders (ED) by combining initial screening self-assessment with a later interview, not performed until axis I symptoms decreased. This was performed on 154 patients with ED. According to the initial self-assessment, the rate of any PD was 90%; however, according to the later SCID-II interview, the rate decreased to 37%. In a sub-sample of 15 patients, the questionnaire was filled out twice: at intake and at the time of the interview. In the subgroup performing the questionnaire twice, there was no decline in PD symptoms over time. However, the PD rate decreased significantly when the patient was assessed by the interview. These findings indicate that the influence of measures might be greater than the influence of timing. The findings support that prevalence rates of PD in eating disordered patients are high, but not as high as it has commonly been reported.
Personality disorders (PD) are frequently diagnosed among clinical and community samples with eating disorders (ED) (for reviews, see Cassin and von Ranson, 2005 and Sansone et al., 2005). Reports on prevalence have yielded varied findings from 27% to 93%. There are methodological difficulties associated with personality assessment in general (Zimmerman, 1994) as well as some specific problems in assessing patients with ED (Vitousek & Stumpf, 2005). Among the general problems is the influence of the method of measurement, since questionnaires yield inflated estimates of PD prevalence compared to interviews (Zimmerman, 1994), particularly evident in the area of ED (Rosenvinge, Martinussen, & Ostensen, 2000). Axis I disorders also affect the measurements of personality. Symptoms of personality pathology have been shown to decline during eating disorder treatment (Ames-Frankel et al., 1992), supporting a state effect of ED on initial personality assessments. This effect is thought to be related to the impact of semi-starvation, chaotic eating, as well as denial and distortion in self-report. Consequently, the current recommendations for assessing personality disorders stress the use of semi-structured interviews, rather than questionnaires, and deferring the time of examination to a later phase of treatment in order to avoid the state-trait confusion (Vitousek & Stumpf, 2005).
The Structured Clinical Interviews for DSM, SCID-I and -II (First et al., 1996 and First et al., 1997), are the official diagnostic interviews of the DSM-IV (i.e., Diagnostic and Statistical Manual for Mental Disorders APA, 1994). The SCID-II Personality Questionnaire is available as a screening tool to shorten the time it takes for the clinician to administer the SCID-II interview (First, Gibbon, Spitzer, Williams, & Benjamin, 1997). When the SCID-II is administered, the clinician only needs to inquire about the items that screened positive on the questionnaire. This procedure has been demonstrated to be reliable with low rates of false negatives (Ekselius, Lindstrom, von Knorring, Bodlund, & Kullgren, 1994). A comprehensive search of electronic databases for studies that use the SCID-II for assessing personality disorders in patients with ED revealed that most studies used only the interview (see Table 1). The only study we found that established diagnoses by combining the questionnaire with the interview presented a lower proportion of PD (30%) (Godt, 2008), compared to other studies. In this study, efforts were made to overcome the state-trait influence by repeating the interview after three months or when it was considered appropriate. Among studies using the SCID-II interview, the weighted frequency of personality disorders at baseline was 54%. Among follow-up studies, in which the interview was performed when eating disorder symptoms had resolved, it was 32% (see Table 1).
