Abstract
Simon Stafrace, Melbourne, Australia:
I wish to comment on ‘Measuring mental health outcomes in a private psychiatric clinic: Health of the Nation Outcome Scales, Medical Outcomes Short Form SF-36’ by Page et al. [1]
Of the 1360 inpatients assessed, only 1152 (84.7%) provided questionnaires on admission, and 754 (55.4%) at discharge. The relatively poor completion of questionnaires at both points of the admission process occurred despite ‘regular HoNOS training throughout the study period’ and the fact that the measures were being collated as part of a study. This finding raises an important question about the utility of the HoNOS as a routine measure in this setting. I am reminded of the point made by Thornicroft and Slade [2] that though an outcome measure may be standardized and acceptable to clinicians, this does not necessarily mean that it is feasible for routine use.
The authors state that the data show patients in their sample were comparable in initial severity and outcome to those in other Australian hospitals and that the HoNOS and the SF-36 provide reliable and valid measures of patient function. The analyses do not allow these conclusions to be drawn. No evaluation of test-retest or interrater reliability is provided. The comparison of the HoNOS depression item and the BDI is insufficient to demonstrate construct validity. Further, no statistical analysis is used to compare the findings of the study with those from other Australian hospitals.
Finally, the authors conclude that the HoNOS was sensitive to treatment change, while the SF-36 was not. Again, this does not appear so. The statistical method used to compare HoNOS and SF-36 scores on admission and discharge was not described, though P-values were provided. These show that changes in the admission and discharge means of all 12 items of the HoNOS and six of nine items of the SF-36 were significant to a level of P < 0.001. In any case, the lack of a significant change should not be taken as evidence that the instruments are insensitive to change without further evidence of significant change using alternative instruments. Thus, a lack of change may be a true finding, and not merely an artefact of the instruments used.