Albert L. Siu, MD, MSPH
▪ Objective: To examine, from the perspective of primary care physicians, the value of mental status findings and ancillary tests in diagnosing dementia or its causes.
▪ Data Sources: Studies identified using MEDLINE and bibliographies of pertinent articles.
▪ Study Selection: Selection based on availability of information on specific findings or tests, the criterion standard used to confirm diagnoses, the disease spectrum, and the richness of the reported data.
▪ Data Extraction: Data collected on sensitivity and specificity. In many cases, the originally reported data are re-analyzed to illustrate the usefulness of alternative cut-off values or to answer specific clinical questions. Likelihood ratios estimated to summarize test results.
▪ Data Synthesis: Probability of dementia is greatly reduced (likelihood ratio, 0.06 to 0.2) when either normal serial 7s, 7-digit span, 3-item recall, or clock drawing test results are obtained; abnormal results only moderately increase the odds of disease. Low (≤ 20), intermediate (21 to 25), or high (≥ 26) scores on the Mini-Mental State Examination (MMSE) increase (likelihood ratio ≥ 8.2), have little effect (likelihood ratio, 1.3 to 2.4), or decrease (likelihood ratio, 0.06 to 0.1) the odds of disease. The usefulness of tests to investigate the cause of dementia varies depending on the pretest probability of the specific condition.
▪ Conclusions: Several useful methods exist to screen for cognitive impairment, and clinicians need to be familiar with the strengths and limitations of their preferred screening methods. In investigating the cause of dementia, routinely obtaining a VDRL test, cerebral imaging studies, serum cobalamin level, or folate level is unwarranted.
Siu AL. Screening for Dementia and Investigating Its Causes. Ann Intern Med. 1991;115:122–132. doi: 10.7326/0003-4819-115-2-122
Download citation file:
Published: Ann Intern Med. 1991;115(2):122-132.
Results provided by:
Copyright © 2018 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use