Reliability
The internal consistency reliability – measured with Cronbach’s Alpha – was good with r= 0.74 for the total scale (sample n=230 Italian women). All but one of the domains had acceptable reliability coefficients: The values were 0.73, 0.73, 0.78, 0.81, and 0.53 for the subscales “PSYCH”, “HORM”, “MENS”, “SEX”, and “ABDOM”, respectively. The alpha of the latter domain should be re-analyzed in patient groups with medically confirmed diagnosis (self-reported in this study).
The results were confirmed in the German validation study (n=108): The Cronbach’s alpha was again good with r= 0.82 for the total scale, and the for the subscales “PSYCH”, “HORM”, “MENS”, “SEX”, and “ABDOM” 0.77, 0.69, 0.92, 0.83, and 0.53, respectively. Altogether, we consider an internal consistency reliability of over 0.7 as sufficiently good.
The test-retest reliability was determined in the framework of the German validation study in 2007 (n= 108). A pretty good reliability was observed for the total score (r= 0.92), and for the five domain scores: The test-retest coefficients were r=0.83, r=0.85, r= 0.93, r=0.72, r=0.62 for the subscales “PSYCH”, “HORM”, “MENS”, “SEX”, and “ABDOM”, respectively. Altogether, the result of the test-retest reliability study can be summarized as satisfactory, i.e. confirming what has been shown in the two studies with analyses of Cronbach’s alpha.
Validity
Validity is a measure that describes to what extend a scale measures what it intends to measure. This is an ongoing process and an indepth understanding of the scale’s validity is likely to emerge over time.
Agreement with the conceptual framework
The internal structure (dimensions) of the SHE agreed with the development team’s conceptual framework and the biological plausibility of hormone-related complaints. In addition, the survey provided consistent results in Italy and Germany.
Domain and total score correlation
Idally, correlations between the total score and each of the individual domains should be high. In contrast, correlation among individual domains should be low as the sub-domains should be ‘independent’ aspects of the overall rating scale.
Initial results showed that the correlations of the five domains with the total scale ranged from r= 0.44 to r= 0.71 and the correlations among subscales oscillated around r= 0.3. In the German survey, the “domain score – total score correlations” were slightly higher. The correlation with the total score was 0.64, 0.61, 0.68, 0.67, and 0.56 for the domains PSYH, HORM, MENS, SEX, and ABDOM, respectively. The correlations among indiviudal domains were smaller, ranging well under 0.4 in both the Italian and German surveys.
Overall, the SHE scale depicts favorable psychometric characteristics taking the domain-total score correlations together with the item-domain, and item-total correlations.
Validity with other scales
The SHE scale was compared with other established (validated) scales within a small Italian survey (n=47), and a population-based German survey (n=108).
Since the SHE scale was designed also as health-related QoL scale with specific focus on short-term hormonal effects, we were particularly interested in evidence that the SHE scale really measures quality of life: The SHE total score significantly correlated with the generic QoL scale SF-12 (total, physical, and mental health score) in the German survey and similarly in the Austrian sample. However, the correlation coefficients were not high, i.e. ranging between r=0.3 and r=08)
Other significant correlations were observed between domains of the SHE-scale and the domain anxiety of the HADS as well as with the domain ‘psychosomatic QoL’ of the QSF in the German survey (correlation coefficients ranging between r=0.3 and r=0.6) .
The psychological domain of the SHE scale showed – as the total score – significant association with mental health and total score (SF-12), anxiety (HADS), and psychosomatic QoL (QSF) (range r=0.3 to r=0.6). The hormone-related domain showed correlations with SF-12 as well as QSF (range r=0.4 to r=0.5). The highest correlation of the abdominal domain of SHE was observed at least in the German sample with SF-12 (total and physical health domain) (r=0.6). The menstrual domain showed only a weak correlation with SF-12 and the self-assessment of sexual dysfunction (QSF scale) (r=0.3). The sexual domain score was correlated with anxiety (HADS) and with psychological QoL (QSF) (range r=0.3 to r=0.4).
No meaningful association of the SHE sexual score was observed with the sexual domain of the QSF scale. Data on low sexual desire, arousal, and satisfaction recorded by the FSEP were only weakly correlated with the score of the sexual domain of the SHE scale. And an inverse correlation was observed between degree of sexual arousal (excitement) and higher scores of the sexual SHE domain (the direction of the scores is inverse in the scales).
Overall, the SHE scale with its total and domain scores seem to be valid.
Ability to detect changes
Since the SHE scale was not yet applied in treatment-related observational or the randomized clinical studies, there are no data to describe responsiveness or MID.
updated