Objective To examine scoring errors in cognitive screening instruments administered to patients in primary care who were then subsequently referred to a secondary care cognitive disorders clinic.

Results Of 328 consecutive new referrals seen over 12 months (January–December 2015), 246 (75%) were from primary care, of whom 93 (37.8%) had been administered a cognitive screening instrument according to their referral letter: the most common were 6CIT (38), MMSE (30), and GPCOG (22). Using a simple ad hoc classification system (Neurodegen Dis Manag 2015;5:505–10), reported primary care test results were classified as unequivocal, ambiguous/incorrect, or incomplete.

Percentages falling into the latter two categories, which may reflect incorrect test use, were 6CIT 26%, MMSE 13%, and GPCOG 32%. 6CIT errors were mostly related to its negative scoring system; GPCOG errors were mostly related to ambiguity over the separate patient and informant components of the test.

Conclusions In this pragmatic survey, scoring errors were evident in around a quarter of patients administered cognitive screening instruments in primary care; the fraction may be higher in those administered tests (6CIT, GPCOG) specifically designed for use in primary care settings.