RESUMEN
BACKGROUND: Dementia is often not formally diagnosed in primary care. To what extent this is due to family physicians' (FPs) watchful waiting, reluctance to diagnose or to their unawareness of the presence of cognitive impairment is unclear. The objective of this study was to assess FPs' awareness of cognitive impairment by comparing their evaluation of the absence or presence of cognitive impairment in older patients without an established diagnosis of dementia, with a reference test of cognitive functioning. In addition, we assessed which patient characteristics were associated with con- and discordance between FPs' evaluation of cognition and results of the reference test. METHODS: The design was a nested diagnostic study. FPs (n = 29) of 15 primary care practices classified the cognitive status of all their patients ≥ 65 years of age (n = 7865) into four categories, based on recollection and medical records. All patients categorized as 'possible cognitive impairment or dementia' and a sample of patients categorized as 'no signs of cognitive impairment' randomly selected to match age and gender were offered to receive a reference test of cognitive function (the CAMCOG) to verify the FPs' label. This reference test could yield three outcomes: no cognitive impairment, amnestic mild cognitive impairment (aMCI) or dementia. Reference test results were weighted back to the original samples to provide estimates for the correct categorization of elderly as 'possible cognitive impairment or dementia' (positive predictive value [PPV]) and 'no signs of cognitive impairment' (negative predictive value [NPV]). Cognitive functioning was not assessed for patients evaluated by FPs as 'probable dementia' and 'unknown or no recent contact'. Characteristics associated with the con- or discordance of the FPs' classification and the reference test were assessed using logistic regression. RESULTS: Complete reference test results were obtained from 318 elderly. FPs labeled 8.3 % of elderly 'possible cognitive impairment or dementia'. The PPV of this label for a CAMCOG score suggestive of dementia or aMCI was 47.1 % (95 %-confidence interval: 43.5 - 62.4 %). FPs labeled 83.7 % 'no signs of cognitive impairment'. The 1-NPV of this label for a CAMCOG score suggestive of dementia or aMCI was 12.5 % (95 %-CI 8.2 - 16.8 %). FPs labeled 3.6 % as 'probable dementia' and 4.5 % as 'unknown or no recent contact'. The odds that FPs' suspicion of cognitive impairment were confirmed by the CAMCOG were higher if persons were ADL dependent (OR 2.24 [95 %-CI 1.16 - 4.35]). The odds of FPs being unaware of the presence of cognitive impairment were higher in the older elderly (OR 1.15 [95 %-CI 1.09 - 1.23] per year). CONCLUSION: Evaluation of FPs' classification of the global cognitive function of elderly without a firm diagnosis of dementia showed both over- and unawareness of the presence of cognitive impairment. FPs were more often unaware of cognitive impairment in the older elderly.