ZEG Berlin

Berlin Center for Epidemiology and Health Research

Header Right

Internal structure of the SHE

A paper-based scale for patient-reported outcomes.

The format of the SHE scale was designed as paper-based scale. A Likert scale (1-5) was used to describe the personally perceived severity (or intensity) of the complaints (items). All items were phrased in a negative direction (complaints).  Two examples of how to complete the questionnaire are provided at the beginning of the scale in order to familiarize women with the questionnaire process.

The raw scale (90 items) was structured in five sections: “complaints/symptoms” (43 items), “use of hormones” (12 items), “reproductive history” (3 items), “lifestyle markers and demographic” (10 items), and “conditions/diseases” (22 items). The latter sections were deemed necessary for the interpretation of the structural (factorial) analysis  and for a meaningful item selection process.

Based on the information gathered during initial testing, the questionnaire was shortened and simplified to 43 items.  Below this point the extrinsic and intrinsic validity became compromised.

 

Five domains assessing sex hormone-related complaints

Five, theoretically independent dimensions (domains), were identified during the development and testing period; ‘sexual problems [SEX]’, ‘menstrual problems [MENS]’, ‘hormonal effects [HORM]’, ‘psychological problems [PSYCH], and ‘abdominal complaints [ABDOM]’. The factor solution for the 15-tem scale explained 66.1% of the total variance in the database, which is considered as good result.

  • Domain 1: ‘Sexual problems’ (SEX): 15% of total variance
  • Domain 2: ‘Menstrual problems’ (MENS): 14.8% of total variance
  • Domain 3: ‘Hormonal effects’ (HORM): 14% of total variance
  • Domain 4: Psychological problems’ (PSYCH):13.2% of total variance
  • Domain 5: ‘Abdominal complaints (ABDOM): 9% of total variance

The final SHE scale was validated in a German observational study of 108 women aged 26- 60 years. The German factorial analysis showed very similar results as seen in the development phase.  A slightly higher percentage of the total variance was explained by the five domains in the German sample (73.3% and 66.1% in the German and Italian sample, respectively). Again, the same 5 domains were extracted in the factorial analysis and the factor loadings were very similar if not identical with the Italian sample.