Test-Retest: Results indicate there is good test-retest reliability on the HADS at 0-2 weeks (r=0.84), >2-6 weeks (r=73), and >6 weeks (r=70) for the anxiety subscale. Results from the depression subscale were 0-2 weeks (r=0.85), >2-6 weeks (r=0.76), and >6 weeks r=0.70, indicating the HADS was stable enough to withstand situation influences ((Herrmann, 1997); varying etiologies). Pearson product movement correlation was found to be 0.92 and 0.90 between the HADS total score, the HADS anxiety score, and the HADS depression score (Herrero et al., 2003; Zwick et al., 2000) (varying etiologies).

Inter-Rater Reliability: Kappa scores indicated there was no significant difference between the General Health Questionnaire-28 and HADS (total score) (kappa statistic =0.074, SE=0.089, p=0.04).

Internal Consistency: Good internal consistency was found ( .80 for the anxiety subscale scale and = 0.81 for the depression subscale) during initial testing (Zigmond & Snaith, 1983). Whelan-Goodinson et al. (2009) found that internal consistency ranges from 0.68 to 0.93, mean 0.83 (for the anxiety subscale) and 0.67 to 0.90, mean 0.82 (for the depression subscale) ((Bjelland et al., 2002); varying etiologies). In an earlier study, Lisspers et al. (1997) found Cronbach scores for the HADS total score to be 0.84, for the HADS anxiety subscale, 0.82 and for the HADS depression subscale 0.90. Scores in this study were not affected by gender or age. Herrero et al. (2003) was validating the scale with a group of Spanish patients, and found Cronbach scores to be 0.90 for the full scale, 0.84 for the depression subscale and 0.85 for the anxiety subscale. Subscales also correlated with each other r=0.68, p<0.01 and each subscale correlated with the full scale r=0.02, p<0.01.

Convergent Validity: The correlation between the HADS depression subscale and the Beck Depression Inventory Primary Care has been found to be 0.62, p<0.001 (Beck et al., 1997) (varying etiologies).

Concurrent Validity: Higher scores on the HADS-depression subscale were linked to higher scores on the SCID–IV (3.52±3.01 and 9.29±5.19, respectively; t=6.84, df=98, p<0.001). Of note 38.2% of who were diagnosed as depressed on the SCID-IV scored within the normal range on the HADS-D. Results from the SCID-IV for those diagnosed with an anxiety disorder (11.42±4.75) had a higher mean score on the HADS anxiety subscale (5.37±3.95, t=6.47, df=62.41, p=0.000). However 25% tested within the normal range of the HADS anxiety scale. Study authors suggest this was indicative of the time line in which the patient is asked to consider when completing the HADS (Whelan-Goodinson et al., 2009).

Several studies have found that the HADS total score shows a higher correlation with depression and anxiety criterion measures than the subscale does (McDowell, 2006). Lisspers and colleagues (1997) (varying etiologies) found the correlation with the Beck Depression Index (BDI) was 0.71 for the HADS-depression subscale and 0.73 for the total HADS. For hospital outpatients the HADS-depression subscale correlated 0.77 with the Montgomery-Asberg Depression Rating (MADR) scale with a group of psychiatric patients (0.70). Again with a group of elderly depressed patients the HADS and the MADR correlated 0.54 and 0.79. Overall, Mykletun and colleagues ((2001); varying etiologies) have reported the correlation between the sub-scores and the overall score as reliable.

Discriminant Validity: Correlation between the subscales of the HADS and the correlation between the HADS total score and other scales (the General Health Questionnaire –28 and the MADR scale can vary considerably. Aylard et al (1987) (varying etiologies) found the correlation of the two subscales of the HADs was r=-0.04 compared to the subscale on General Health Questionnaire –28  was r=0.54. Lewis and Wessely (1990) found the correlation between the HADS total score and the General Health Questionnaire –28 was 0.75. Schwarzbold et al. (2014) also found high discriminant validity of the HADS among participants with TBI.

Predictive validity: The HAD has depression and anxiety subscales, which were found to account for 52.6% and 60% (respectively) of variance when looking at patients who were diagnosed with a mood disorders and those with no psychiatric disorder ((Herrero et al., 2003)).