Triage-weighted kappa: a more appropriate triage reliability measure
Article Outline
To the Editor:
van der Wulp et al. [1] make a compelling argument regarding the assessment of emergency department triage systems. We agree that neither the linear nor quadratically weighted kappa statistics adequately takes into account the severity of mistriage in ordinal triage scales (despite the fact that these appear to be the most commonly quoted statistics) and that an elaborated weighting scheme is long overdue.
Although we congratulate the authors on the first step in this direction, we believe that when triage scales stipulate different care pathways for different triage categories, algorithms for determining over- or undertriage should be constructed so as to account for these differences. It is not clear whether this is the case with this new algorithm. For instance, a patient undertriaged using the Australasian triage scale (ATS) will have a different path through the emergency care system to one undertriaged using the Manchester triage system (MTS). If a patient was triaged ATS-3 instead of ATS-2, she would be seen in 30 minutes rather than 10 minutes—if she was triaged MTS yellow (third) instead of orange (second), she will be seen in 60 minutes rather than 10 minutes. Undertriage would be more severe using the MTS. This difference needs to be accounted for in some way—either quantitatively, semi-quantitatively, or qualitatively. The effect of overtriage is also different in different settings. For instance, the consequences of overtriage would be more severe in systems with fewer resources and increased overcrowding (e.g., South Africa as opposed to Canada). From a qualitative perspective, setting is critical. A new tool, therefore, should ideally account for both qualitative factors, such as setting, and quantitative factors, such as different weightings among levels of mistriage. As with the ideal triage system, such an analytical tool may not be completely realizable, but we echo the authors' call for further research in this direction.
Reference
PII: S0895-4356(09)00358-8
doi:10.1016/j.jclinepi.2009.11.005
© 2010 Elsevier Inc. All rights reserved.
