Objective To analyse data from two previously published individual studies and two meta-analyses of the Ascertain Dementia 8 (AD8) cognitive screening instrument to calculate the recently described ‘likelihood to be diagnosed or misdiagnosed’ (LDM) metric, the ratio of ‘number needed to misdiagnose’ (NNM = 1/Inaccuracy) to either ‘number needed to diagnose’ (NND = 1/Youden index) or ‘number needed to predict’ (NNP = 1/predictive summary index).
Results Raw data (true positives and negatives, false positives and negatives) were extracted and LDM values for the diagnosis of dementia were calculated. For the individual studies (n=212, 67 respectively) LDM values were all <1, favouring misdiagnosis, whereas for the meta-analyses (n=3278, 3694 respectively) all LDM values were >1, favouring diagnosis. This discrepancy may be explained by the poor specificity of AD8 in the individual studies, lower than in any other study included in the meta-analyses, giving very high values for NND and NNP and lower values of NNM than in the meta-analytic data. This may relate to differences in casemix between the various studies.
Conclusion LDM is an easily calculated and potentially useful test metric. LDM values derived from meta-analytic data may differ from those of individual test accuracy studies.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.