Screening tests for cognitive impairment not very accurate
How reliable are screening tests for cognitive impairment when used in primary care settings?
Although none of the existing screening tests are great, the modified Mini-Mental State Examination and Mini-Cog are modestly accurate in identifying patients with cognitive impairment or dementia. (LOE = 1b)
Holsinger T, Plassman BL, Stechuchak KM, Burke JR, Coffman CJ, Williams JW Jr. Screening for cognitive impairment: comparing the performance of four instruments in primary care. J Am Geriatr Soc 2012;60(6):1027-1036.
[PubMed ® abstract]
Study design: Cross-sectional
Funding source: Government
Setting: Outpatient (primary care)
These researchers evaluated more than 600 veterans older than 65 years without a prior diagnosis of dementia or cognitive impairment. The men were randomly selected from the appointment schedules of 3 Veterans Administration clinics. Because of the paucity of female veterans, all of them were invited to participate. A research assistant administered 4 cognitive impairment screening tests in random order and a trained nurse performed the gold standard evaluation that consisted of several neuropsychiatric tests adjudicated by an expert consensus panel. None of the evaluators knew the results of the gold standard or screening tests, respectively. The screening tests were: Mini-Cog, Memory Impairment Screen (MIS), a 2-item functional memory screen (MF2), and a modified Mini-Mental State Examination (3MS). The Mini-Cog is made up of a 3-item recall test and the clock-drawing test. It takes approximately 3 minutes to administer, and scores range from 0 to 5 (0 to 2 represent impairment). The MIS, which takes approximately 4 minutes to administer, consists of a 4-word recall test with an interference task (eg, counting forward and backward, saying the alphabet forward and backward, and counting by 3s for 2 to 3 minutes). Two points are earned for each word recalled without a cue and 1 point for each word recalled after a cue. Scores range from 0 to 8 (less than 5 indicates impairment). The MF2, taking approximately 90 seconds to administer, consists of 2 questions given to the patient and the patients’ caregiver: (1) “During the past 12 months, have you noticed a decline in your memory such that you have trouble remembering where you put things, remembering to take your medications, or remembering to pay bills?” and (2) “During the past 12 months, have you had problems with your memory or thinking that interfere with your ability to do things that you regularly do, such as taking care of your home, managing your checkbook, or keeping up with TV programs?” The 3MS is a 100-point instrument that includes the Mini-Mental State Examination and adds items to assess memory, verbal fluency, similarities, and delayed recall. It takes approximately 17 minutes to administer and a score of less than 83 indicates impairment. Based on the gold standard assessment, approximately 3% of participants had dementia and 39% showed cognitive impairment. The diagnostic accuracy for screening for dementia alone was: Mini-Cog = 76% sensitive, 73% specific, positive likelihood ratio (LR+) 2.8, negative likelihood ratio (LR-) 0.3; MIS = 43% sensitive, 93% specific, LR+ 5.9, LR- 0.6; MF2 = 38% sensitive, 87% specific, LR+ 2.9, LR- 0.7; and 3MS = 86% sensitive, 79% specific, LR+ 4.1, LR- 0.2. The diagnostic accuracy for screening for either dementia or cognitive impairment was: Mini-Cog = 39% sensitive, 78% specific, LR+ 1.8, LR- 0.8; MIS = 17% sensitive, 98% specific, LR+ 7.6, LR- 0.8; MF2 = 24% sensitive, 92% specific, LR+ 3.6, LR- 0.8; and 3MS = 39% sensitive, 89% specific, LR+ 3.4, LR- 0.7. Keep in mind that a LR+ higher than 10 and LR- lower than 0.1 provide the most useful information. None of these tests are great, but the 3MS and the Mini-Cog are moderately accurate. Since the 3MS takes approximately 17 minutes to administer, however, its utility in the real world is limited. When any of these tests are used for screening, patients will need additional evaluation to validate the results. Since we have no effective treatment for cognitive impairment, one can question whether we really need to be screening elders for cognitive impairment at all.
Henry C. Barry, MD, MS
Michigan State University
East Lansing, MI