We would like to draw attention to the assertion by Mykletun et al (Reference Mykletun, Stordal and Dahl2001) that a two-factor structure best fits the Hospital Anxiety and Depression (HAD) scale, especially in individuals with mental problems. They stated that psychometric studies of this scale only involved small samples of non-psychiatric patients. However, we recently published the only factor analysis of the HAD scale based on a large population: 2669 ‘HAD completers’ from 3002 patients (89%) with major depression, DSM-IV criteria (Reference Friedman, Samuelian and LancrenonFriedman et al, 2001).

Contrary to Mykletun et al, we found a three-factor solution using principal-components analysis with factors defined by eigenvalues ≥1. One of Mykletun et al's reasons for rejecting the three-factor solution was that their third factor comprised heterogeneous items loading for both anxiety (items 7 and 11) and depression (item 14). Our three-factor structure discriminates the original depression factor and two separate constructs of anxiety: ‘psychic anxiety’ (items 3, 5, 9 and 13) and ‘psychomotor agitation’ (items 1, 7 and 11). This factor solution captured 48.6% of the variance and was relatively robust; it was not influenced by gender ratio and was also found in two random halves.

Two reasons may account for these discrepancies between our results. First, because of the high proportion of HAD scale non-completers (44%), Mykletunet al's sample may have been biased. Patients with depression are probably not prone to answer such surveys and may therefore be underrepresented. Second, the factor structure of the HAD scale may not be stable across different categories of subjects: those with heterogeneous mental problems and those specifically suffering from major depression.

The HAD scale is not only useful for its initial screening purpose. It also showed potential ability in assessing change in specific symptoms of anxiety (‘psychic anxiety’ and ‘psychomotor agitation’ factors of the scale) during antidepressant treatment (Reference Friedman, Samuelian and LancrenonFriedman et al, 2001). Moreover, recognition and monitoring of psychomotor agitation has several clinical implications: it is a potential side-effect of some antidepressants (Reference NuttNutt, 1999), it may predict antidepressant response (Reference Flament, Lane and ZhuFlament et al, 1999), it may predict adverse outcome and increase the risk of suicide (Reference Schatzberg and DeBattistaSchatzberg & DeBattista, 1999).