Aim: To examine Free-Cog, a recently described, hybrid screening instrument, as separate tests of cognitive (Free-Cog-Cog) and executive function (Free-Cog-Exec) to see if this improved screening accuracy for cognitive impairment compared with standard Free-Cog. Materials & methods: Free-Cog-Cog and Free-Cog-Exec were combined using Boolean logical 'AND' and 'OR' operators (serial and parallel combination), and also used to construct a stepwise decision tree. Results: Serial combination improved specificity and positive predictive value whereas parallel combination improved sensitivity, typical findings with these operators. Stepwise application identified groups with high and low probability of cognitive impairment but failed to differentiate adequately those in the intermediate uncertain diagnosis group. Conclusion: Study findings suggest limited benefit from reformulations of Free-Cog compared with the standard instrument.
Keywords: Boolean operators; Free-Cog; decision tree; dementia; diagnosis; mild cognitive impairment; screening test.
Dementia and cognitive impairment are common, but how best to identify these conditions remains uncertain. Many instruments which screen cognitive function have been described. Free-Cog is a relatively new screening instrument which tests both cognition and function; impairment in both constructs is required to fulfil agreed definitions of dementia. This study sought to examine whether dissociating the two elements of Free-Cog might produce more accurate screening for dementia and cognitive impairment. The separate tests were combined in series or in parallel, or used to create a decision tree. Examination of these two methods, using the dataset of a previous Free-Cog test accuracy study, suggested some benefits but also some losses compared with standard unitary Free-Cog test.