Abstract
Keith Bender, Inner City Mental Health Service, Royal Perth Hospital, Perth, Western Australia:
Andresen, Caputi and Oades [1] studied interrater reliability of the Camberwell Assessment of Need Short Appraisal Schedule (CANSAS). They attributed differences in staff ratings to ambiguity in the definition of need and levels of need, and/or the sources of information used by the rater making the assessment. The rating system in the CANSAS is the same as that of the full CAN [2]. Therefore, any ambiguity in the definition, and the levels, of need in CANSAS is also present in CAN. I wish to comment on this ambiguity.
A rater may score as 1 a problem that is moderate but which has not received help, or a person who has received some help but still requires more. However, all scores of 1 are counted as ‘met needs’ whereas in neither of the two examples is the need met. However, the rater does not rate the problem as 2 because it is not perceived as a ‘serious problem’.
Either the score of 1 should be a severity rating to denote a mild problem. If that is the case then the number of ‘1's should not be interpreted as number of unmet needs but number of mild problems. Alternatively, if 1 is interpreted as met need, then any unmet need should be rated as 2, even if mild. This would mean that it is no longer a severity rating but a statement that a need exists and whether or not it has been met. Then the summary scores for met needs and unmet needs accurately reflect these phenomena.
Slade [personal communication, 2000] acknowledges that the intention of CAN's authors was to have scores for met needs and unmet needs, not a severity scale. He writes ‘The use of 0, 1 and 2… implies an ordinal scale, whereas the data are actually categorical’. Unfortunately, the stated meanings of the ratings, and the examples given to guide raters, do not discourage them from treating CAN scores as severity scales.
A score of 2 should mean that action needs to be taken. If a problem is present but requires no action, then it should be scored as 0. If, in spite of receiving help, a problem continues to exist, then the question to ask is whether anything further needs to be done. If nothing needs to be done then the problem is scored 1. If further action needs to be taken, then the problem warrants a score of 2.
Similarly, with a mild problem that has received no help, the question is whether somebody needs to do something. If action is required then the score shouldbe 2, even if it is not a serious problem. If action is not required, then the problem should be rated as 0, even if a mild problem exists.
With this approach, CAN becomes useful for developing management plans. Andresen et al. [1] suggest that increasing the number of levels of need might improve interrater reliability. This is not necessary if all raters are trained to apply the same principle. If the number of categories is increased, it should only be in the research version of CAN. The clinical versions should remain simple.