The study involved 99 participating individuals, with a mean age of 76 years (SD ± 6.36 years), consisting of 56 women (56.6%) and 43 men (43.4%). The age distribution was equable, since 46 persons (46.5%) were within the age range of 65 to 74 years and 53 persons (53.5%) were >75 years old (Table ). There was no statistically significant difference when we compared the health centres and sex (χ2 = 0.152, degrees of freedom (df) = 1, P = 0.697) or the health centres and the age distribution (χ2 = 0.567, df = 1, P = 0.451) (Table ).
Demographic characteristics of the sample
Comparison of the parameters of the study sample
When the total scores for SAST were examined, the test results proved negative for 58.6% (N = 58), borderline for 12.1% (N = 12), and positive for 29.3% (N = 29) (Table ).
Comparison of the Short Anxiety Screening Test (SAST) results (analysis of variance (ANOVA))
The mean score for older people with negative results was 17.6 (SD ± 2.28), whilst for those with a positive result the mean score was 28.5 (SD ± 3.24). The application of ANOVA identified a statistically significant difference between the scores (P < 0.0001, F = 188,281) (Table ). Post hoc analysis showed that the SAST score differed at the significance level P < 0.0001.
The total mean score of the SAST for the study population as a whole was 21.3 (SD ± 5.5; min 12, max 36). The mean score for women was 22.8 (SD ± 5.8) and for men 19.5 (SD ± 4.3). With the use of t test for independent samples, this difference was found to be statistically significant (t = 3.105, df = 97, P = 0.002). In contrast, there was no statistically significant difference when we compared the mean scores across age distribution (t = 0.837, df = 97, P = 0.404) or for the individual health centres (t = -0.382, df = 97, P = 0.704) (Table ).
Comparison of the Short Anxiety Screening Test (SAST) results for sex, age distribution and health centres
The translation procedures did not reveal any specific problems. The developers of the SAST made some comments on three of the backtranslated questions where minor issues were identified. These concerned the interpretation of the word 'irritable' (question 8), the differentiation of the expression 'back pain' (question 6) and the interpretation of the word 'palpitations' (question 7), emphasising the somatic complaints of older people. These comments were taken into account when finalising the Greek reconciliated version of the SAST.
During cultural adaptation, the questionnaire was found to be overall comprehensible and easy to understand, according to comments from older people. The only linguistic problem concerned question 8, where all respondents proposed to change the Greek word for 'irritable' into a less obscure word that would be more easily understood by the respondents. Their recommendation was discussed and incorporated into the final Greek translation of the questionnaire.
Feedback from the doctors demonstrated that the questionnaire was comprehensible, easy and quick (approximately 10 min) to use, and that it could be used in everyday clinical practice for primary assessment, while interviewing the patients regarding mental health issues.
Reliability and validity
The SAST questionnaire showed a very good overall internal consistency (α value: 0.763, 95% CI 0.71 to 0.82, P
< 0.001) for individual comparison. The overall Cohen κ coefficient for reproducibility (test-retest reliability) was 'very good' (0.930, 95% CI 0.918 to 0.942, P
< 0.0001) and ICC (95% CI) for reproducibility was found to be 0.763 (95% CI 0.686 to 0.827) [25
]. The Wilcoxon signed ranks test showed that there was no statistically significant difference between the total of questions (z = 0.676, P
= 0.499), as in the comparison for each question separately between the two applications of questionnaire (N = 26), with values oscillated from z = 0.0 (P
= 1.0) in question 3, to z = 1.134 (P
= 0.257) in question 9. The results are illustrated in Table .
Short Anxiety Screening Test (SAST) reproducibility (test-retest reliability)
Exploratory factor analysis indicated three factors with eigenvalues over 1.0. Those factors were responsible for 60% of variance and rotation converged in three iterations (Table ). At the same time, for the control of crosscorrelation of items among them using the interitem correlation matrix method, analysis showed that all questions correlated very well, as Cronbach α values were all greater than 0.7 (Table ).
Factor analysis for the symptoms: rotated component matrix for three factors
Short Anxiety Screening Test (SAST) interitem correlation matrix
The independent samples t test identified the SAST's ability to discriminate between older men and women, with women scoring significantly higher. Higher levels of anxiety in women have been reported in previous studies [1