screening for dementia: family caregiver questionnaires reliably

Screening for Dementia: Family CaregiverQuestionnaires Reliably Predict DementiaMarilee Monnot, PhD, Meg Brosey, MS, and Elliott Ross, MD

Introduction: Because of increasing numbers of patients with diseases that cause dementia, primarycare physicians must use efficient assessment procedures in their clinics. Important advantages ofscreening for dementia include determination of the patient’s cognitive capacity to participate compe-tently in his/her own medical care and early diagnosis, which enables administration of medicationsthat preserve some cognitive functions.

Methods: A study was conducted to determine whether questionnaires completed by a family care-giver about a patient could differentiate between those with dementia and those with other neurologicaldisorders that do not cause dementia. Clinical and demographic information gleaned from more than330 consecutive multidisciplinary outpatient dementia clinic assessments were entered into an Institu-tional Review Board-approved database and analyzed post hoc to answer several research questions.

Results: Three questionnaires completed by family caregivers about patients were able to differenti-ate reliably between patients with dementia with a variety of degenerative disorders and patients with-out dementia with other neurological disorders that often are mistaken for dementia. When these ques-tionnaires are combined with a patient test (Mini-Mental State Examination), an accurate prediction ofwhich patients suffer from a true degenerative disease that causes dementia was robust (effect size ofR2 � 0.81, P < .0001 for the multiple logistic regression analysis).

Discussion: These instruments assist the primary care physician to determine which patients seem tosuffer from a disease that causes dementia and need further assessment by the physician or at a special-ized dementia clinic. The ultimate goal is to assure that patients receive appropriate medical manage-ment as early in the disease process as possible. (J Am Board Fam Pract 2005;18:240–56.)

With the aging of 75 million baby boomers,1–4

primary care physicians (PCPs) are bracing for anincrease in numbers of patients diagnosed with Alz-heimer disease (AD) and related dementias. Cur-rent estimates of AD prevalence vary from 4.5% to16.8% for patients older than 65 years; however,epidemiologic studies indicate future prevalenceand dementia-related resource usage may be higherthan current estimates.5 Some have questioned theefficacy and public health need for large-scalescreening of the elderly for dementing illnesses.6

However, the lay press is replete with warnings that“dementia often goes undiagnosed in primary care

settings,”7 and these warnings receive support fromevidence-based studies.8

There are distinct advantages to screening fordementia at the primary care level of practice. Earlydiagnosis enables the physician to administer med-ications that slow disease progression9,10 and toassist the patient and family members in planningfor diminished capacity while the patient is still ableto participate in decision-making.11 In addition,primary care physicians need to know whethertheir patients can give accurate histories or can berelied on to participate in their own medical care,including taking medications as directed. Patientswith dementia often have intact but superficial so-cial and communication skills that, if accompaniedby loss of insight, may mask their cognitive declinefrom casual observers and interfere with their abil-ity to assist the physician in their medical care. Inaddition, physicians are concerned about offendingless insightful patients regarding their current cog-nitive deficiencies and about obtaining confidentialinformation from family members. Physicians alsomust weigh how much clinic time is needed to

Submitted, revised, 22 February 2005.From the Departments of Neurology (MM, ER), Infor-

mation Technology (MB), University of Oklahoma HealthSciences Center College of Medicine, and Veterans AffairsMedical Center (ER), Oklahoma City, OK.

Conflict of interest: none declared.Corresponding author: Marilee Monnot, PhD, University

of Oklahoma Health Sciences Center, College of Medicine,Department of Neurology, 711 Stanton L. Young Boule-vard, Suite 215, Oklahoma City, OK 73104 (e-mail: [email protected]).

240 JABFP July–August 2005 Vol. 18 No. 4 http://www.jabfp.org

administer screening instruments that assess a pa-tient’s cognitive and behavioral status against thebenefits that accrue from these assessments in re-gard to patient care. From the medical consumer’spoint of view, the stress of caregiving encouragesfamily members to seek a reliable way to alert thephysician of important behavioral changes in thedemented patient.2,12 In addition, it is now gener-ally accepted that family caregivers, especially thosewho live with the patient, can provide importantinformation about recent cognitive and behavioralchanges in the patient that aid in the differentialdiagnosis of degenerative diseases that cause de-mentia.13–15

With these issues in mind, we developed a set offamily caregiver questionnaires, based on our expe-rience in diagnosing patients in a University-basedclinic for memory loss and dementia. The ques-tionnaires used in assessments consist of severalthat are recommended by the National Alzheimer’sCoordinating Center (NACC), but some weremodified for use by untrained family caregivers.

For example, the Clinical Dementia Rating(CDR)16,17 is very useful when assessing patientsfor dementia. To properly administer this instru-ment, up to 10 hours of training is needed on theWashington University School of Medicine Alzhei-mer’s Disease Research Center web site.18 It is ahighly reliable clinical staging assessment for de-mentia that uses semistructured interviews of thepatient and a reliable collateral source. It is con-ducted by a clinician who rates 6 domains of cog-nitive and functional performance: memory (recentand long-term), orientation, judgment and prob-lem solving (including insight), community affairs,home and hobbies, and personal care. Each domainis scored, and an overall CDR score is arrived at bya standard algorithm to stage the patient’s level ofimpairment: 0, no impairment; 0.5, very mild im-pairment; 1, mild dementia; 2, moderate dementia;and 3, severe dementia. After using this instrumenton approximately 400 patients/caregivers, we no-ticed a set of symptoms, including behavioral ab-normalities for each of the 6 categories, that werementioned consistently by family members of pa-tients with dementia during the interview process.We turned those observations into a caregiver-friendly questionnaire for family members to fillout about the patient during the multidisciplinaryassessment. In this modified CDR, 5 concrete ex-amples of possible patient behaviors were added to

each of the 6 major and 2 minor categories ofpatient functioning. The first 3 statements in eachcategory were clear signs of dementia, and the last2 could be observed in either mild impairment(CDR level 0.5) or clinical depression. For all 8categories, a score of “1” was assigned to each ofthe first 3 symptoms marked by the caregiver,whereas the last 2 symptoms were scored only 0.5even if both were marked. Scores for all the cate-gories were totaled and divided by 8 to arrive at amean score based on family member observationsof the patient (Appendix). This is not the scoringalgorithm used in the original clinical interview,but it served our purpose in screening new patientsfor dementia. In addition, results of administeringand scoring the CDR clinical interview in the tra-ditional manner were recorded so comparisonscould be made at a later date.

Another questionnaire, the Frontal BehavioralInventory (FBI),19 was formed from lists of symp-toms common to frontotemporal dementias inwhich behavioral and personality changes areprominent. The original FBI has now been stan-dardized and is more effective in detecting fronto-temporal dementias than traditional neuropsycho-logical testing.20–22 We adapted the list ofsymptoms for use by poor readers and those withlimited education; the family caregiver rates thepatient on each of 24 frontotemporal symptomsusing a 4-point Likert scale (0, no symptoms; to 3,severe symptoms). These scores are totaled for theentire instrument, with a maximum score of 72.

A third instrument, the caregiver burden versussatisfaction questionnaire,23 reveals positive andnegative attitudes about caregiving. It can also beused in statistical analyses if an individual indexscore is derived by subtracting the total burdenscore from the total satisfaction score. Thus, if thesatisfaction score is much larger than the burdenscore, caregiving is not stressful, and the indexscore will be large. If burden is similar to satisfac-tion, caregiving is stressful, and the index numberwill be a small positive or negative number. Theburden score alone also gives important informa-tion about how stressful it is for family members tocare for the patient.

The Yesavage Geriatric Depression Scale24 wasused to assess the elderly spouse caregiver whofilled out the forms about the patient. If a spouse issuffering from significant depressive symptoms (�9of 15 points on the shortened form), then the reli-

http://www.jabfp.org 241

ability of the patient questionnaire informationmay be suspect, because depressed persons oftensee the world as more negative than it really is.Therefore, the patient may not be as ill as thecaregiver reports. Five such cases were encoun-tered. They were not included in analyses reportedhere, and the caregivers were referred for evalua-tion and treatment of depression.

Another instrument used during the multidisci-plinary assessment was a caregiver stress symptominventory (stress warning signals) adapted from theStress Perception Scale by Herbert Benson at Har-vard University (unpublished). Family caregiverssimply checked their current symptoms of distressaccording to 6 domains: physical, emotional, spir-itual, behavioral, cognitive, and relational. Thenumber of symptoms was totaled to yield the car-egiver’s stress score. This instrument alerted themedical team about the family caregiver’s potentialhealth risks from excessive stress.

After observing the apparent efficacy of theseinstruments in the clinic, we pursued more objec-tive data by conducting an accuracy study of thesequestionnaires in screening for dementia. The hy-pothesis tested was formulated after questionnairedevelopment and use but before data collection.The explicit purpose was to formulate a screeningdevice for use in primary medical care settings, toaid in the early diagnosis of diseases that causedementia.

MethodsData in the University of Oklahoma Health Sci-ences Center (OUHSC) dementia database is com-piled continuously from medical records of consec-utive patients assessed and treated at the Universityof Oklahoma Physicians Center for Memory Lossand Dementia (CMLD). Patients seen in this clinicmust be referred by a primary care physician (PCP)or another physician who suspects that the patientmay have a disease that causes dementia. However,not all patients evaluated in this clinic have a truedegenerative disease that causes dementia; �17%of these referred patients have been found to sufferfrom other neurological diseases that share some ofthe symptoms of dementia (meningitis, stroke, de-pression, low levels of thyroid hormone, low levelsof vitamin B12, overmedication toxicity, etc). Ourde-identified, Institutional Review Board-approveddatabase contains information on more than 330

subjects who have agreed to participate (CDR level�1.0) or whose participation is approved by thepatient’s designated Power of Attorney. Using theTeleform information system (Verity Inc., Sunny-vale, CA), medical record forms are scanned toenter relevant medical information directly into thedatabase that includes demographic, neurobehav-ioral testing, neurological examination, and diag-nostic information. JMP Statistical Discovery soft-ware (SAS Institute, Cary, NC) was used to analyzethe data. In addition to specific diagnoses, patientswere classified as: (1) demented (AD, Huntingtondisease, Lewy body disease, Pick disease, fronto-temporal, corticobasal ganglionic degeneration,vascular dementia) or (2) not demented (age-asso-ciated cognitive decline, mild cognitive impair-ment, depression, stroke, seizures, encephalitis,overmedication toxicity, or other nondegenerativeneurological disorders). This was done for the pur-pose of answering the research question: “Canquestionnaires completed by a family caregiverabout the patient differentiate between patientswith dementia and patients without dementia butwith other neurological disorders?”

Traditional clinical interviews were conductedwith family members by the primary author, espe-cially focusing on clarification of inconsistencies intheir questionnaire answers. All caregiver question-naire results were compared with a cognitivescreening instrument administered to the patient;the Mini-Mental State Examination (MMSE) is awidely used and accurate predictor of cognitiveimpairment among differing racial groups.25,26 Al-though conflicting data exist regarding the diagnos-tic efficacy of combining family questionnaires withdirect patient assessment, it seems that the choiceof instruments and caregiver characteristics make adifference in the accuracy rate.27–29 Informationfrom CMLD questionnaires was subjected to sta-tistical analyses in relation to each patient’s diag-nosis and dementia status, arrived at from our mul-tidisciplinary assessments.30

Statistical Tests UsedContinuous variables, such as patient age, weresubjected to regression analyses if the variableswere parametric. Categorical (nominal) variables,such as dementia status (yes or no), were analyzedusing both analysis of variance (ANOVA) and lo-gistic regression methods. Calculations completedautomatically in the JMP statistical program for


ANOVAs, such as “CDR � dementia,” include thenumber of respondents for each level of the cate-gorical variable, the mean of the responses, thestandard error (a pooled estimate of error variance),and figures for the lower 95% and upper 95% ofscores for each level. Thus, behaviorally relevantinformation about cutoff points for each categoricalvariable is available.

When analyzing the ability of several variablesto predict a single effect, the whole model fit com-putation is similar to the ANOVA for continuousresponses; its resulting table shows tests that com-pare the whole model fit to the model that omits allthe regressor effects except the intercept parame-ters.

The computed R2 figure describes how much ofthe measured effect can be attributed to the vari-able(s) tested. The effect size is measured by severalstatistics (R2, r2, �2, g, r, and d) and the results rangefrom 0 (no association) to 1 (complete association).The squared parameters estimate what percentageof the variability is explained by the data; ie, an R2

of 0.36 indicates that 36% of the variance is ex-plained by the statistical analysis. R2 values between0.01 to 0.09 are considered small effect sizes, valuesbetween 0.09 and 0.25 are considered medium ef-fect sizes, and values �0.5 are considered largeeffect sizes.31 R2 Adjusted is a more conservativestrength of association measure than R2 because itsubtracts out the statistical variance related to theSEM. In this study, logistic regression was used tocompute the final effect; it fits nominal Y responsesto a linear model of X predictors. A nominal modelrarely has a high R2, but it is the best overallestimate of the effect size of the relationship be-tween variables analyzed in the study.

ResultsInformation about percentages of each category/diagnosis from the entire OUHSC database is sum-marized in Table 1. All variables used in this studywere parametric. Expected numbers of patientswith and without dementia, classified according tosex or ethnic group, were found, indicating that thedatabase information is similar to other publisheddata from specialized dementia clinics,32,33 and toOklahoma population figures from the latest cen-sus34 (except for Native Americans who are oftenseen in their own tribal hospitals in Oklahoma).Neither ethnic group (P � .86) nor sex (P � .64)

differentiated between patients with and withoutdementia. However, patients with dementia in ourdatabase were significantly older than patientswithout dementia (P � .007). The mean age of thepatients with dementia was 72.3 years, and themean age of patients without dementia was 66.4years, but the effect size (R2 Adjusted) of theANOVA analysis “demented � age” was negligibleat 0.04. Table 2 summarizes statistical analyses ofthese dependent and independent variables, andTable 3 summarizes analyses of the relationshipsbetween covariates in the study.

Another ANOVA analysis showed that MMSEscores were significantly different by dementia sta-tus despite a very wide range of educationalachievement among database participants (6 to 20years of formal education). Of patients diagnosed asdemented, 95% scored 20 points or less and 95% ofpatients without dementia scored more than 24points (Table 2). However, the effect size (R2 Ad-justed � 0.28) is only moderate and reflects thepoor ability of MMSE to detect diseases that causedementia that begin with predominantly behavioraland personality changes (eg, Lewy body disease,frontotemporal dementias, etc) as opposed tomemory deficits.35

Table 1. University of Oklahoma Health SciencesCenter Dementia Database Categories withDemographic Percentages and Comparisons with StateCensus Figures

CategoryUniversity of Oklahoma Database/

2000 Oklahoma Census* (%)

Female 63/50.9Male 37/49.1White 87/80.3African American 8/8.3Native American 4/11.4Asian 1/1.7Demented 73Not demented 27Alzheimer disease 27Frontotemporal/Lewy body 25Corticobasal ganglionic 11Vascular dementia 6Huntington disease 4Mild cognitive impairment 9Depression 8Other neurological diseases 10

* Does not equal 100% because of mixed racial identity reportedby citizens.


Several questionnaires were used to elicit infor-mation from family caregivers about themselvesand about behavioral changes in the patient duringCMLD multidisciplinary diagnostic assessments:the modified CDR, modified FBI, Caregiver Bur-den and Satisfaction Scale, UCLA Neuropsychiat-ric Inventory,36 Beck Depression Inventory37,38

(for use with younger caregivers), Yesavage Geri-atric Depression Scale,24 and the Stress WarningSignals. Instruments used that were not changed oradapted for use at the CMLD include the UCLANeuropsychiatric Inventory,36 Beck DepressionScale,37,38 and Yesavage Depression Scale.24

Again using ANOVA, scores derived from themodified CDR were significantly different for neu-rology patients with and without dementia. Of pa-tients with dementia, 95% scored �1.6; of patientswithout dementia, 95% scored of �0.4, as rated bytheir family members (Table 2). These figures werethe same for the traditional CDR staging algorithmscores. The scores of the modified CDR were com-pared with the traditional CDR; Pearson �2 analy-sis showed no significant difference between the 2scoring systems (P � .75) on any of the 4 staginglevels: level 0 (P � .10); level .5 (P � .17); level 1(P � .33); level 2 (P � .32); level 3 (P � .17). A

Table 2. Results of ANOVA and Linear Regression Analyses of the Dependent and Independent Variables

Statistical Analysis P R2 Adjusted F RatioPatients with

DementiaPatients

without Dementia

Demented (yes or no) � patient age .007 0.03 7.3 72.2 � 1.0(patient age)

66.7 � 1.7(patient age)

Demented � MMSE �.0001 0.28 72.4 18.7 � 0.5(MMSE)

26.1 � 0.08(MMSE)

Demented � modified CDR �.0001 0.49 191.6 1.64 � 0.05(modified CDR)

0.35 � 0.1(modified CDR)

Demented � traditional CDR �.0001 0.50 206.5 1.63 � 0.05(traditional CDR)

0.33 � 0.17(traditional CDR)

Demented � modified FBI �.0001 0.11 15.6 22.1 � 1.4(modified FBI)

11.4 � 2.3(modified FBI)

Demented � Burden .002 0.11 10.5 19.9 � 3.3(burden score)

43.3 � 6.4(burden score)

Demented � UCLA* .10 0.01 2.7Demented � Yesavage* .15 0.02 2.1Demented � Beck* .56 0.03 1.1Demented � stress* .15 0.03 2.1

* Not significant.Modified CDR, Clinical Dementia Rating scale: Y, yes; N, no; FBI, Frontal Behavioral Inventory; Burden, Caregiver Burden score;Yesavage, Yesavage Depression Rating Scale; Beck, Beck Depression Inventory; UCLA, UCLA Neuropsychiatric Inventory; MMSE,Mini-Mental State Exam.

Table 3. Linear Regression Analyses Describing Relationships Between Covariates

Statistical Analysis PR2

Adjusted F Ratio

Test Results

Demented Nondemented

Modified CDR � MMSE �.0001 0.52 95.4 CDR level 3 � mean MMSE 11 CDR level 0 � mean MMSE 27Modified CDR � Burden .0009 0.18 5.2 CDR level 3 � mean Burden

index 5CDR level 0 � mean Burden

index 28Modified CDR � modified FBI �.0001 0.25 10.8 CDR level 3 � mean FBI

score 37CDR level 0 � mean FBI

score 8MMSE � Burden �.0001 0.2 19.1 r � 0.31, P � .018 r � 0.74, P � .001MMSE � modified FBI �.0001 0.26 41.8 r � 0.44, P � .0000 r � 0.44, P � .017Modified FBI � Burden �.0001 0.37 39.0 r � 0.49, P � .0002 r � 0.8, P � .0003

CDR, Clinical Dementia Rating scale; FBI, Frontal Behavioral Inventory; Burden, Caregiver Burden score; MMSE, Mini-MentalState Exam.


linear regression of the 2 scoring systems resultedin a highly significant association [R2 Adjusted �0.83, P � .0001]. However, the statistical correla-tion between the 2 scoring systems earned by pa-tients without dementia (r � 0.51; R2 � 0.26) wasnot as robust as the correlation of the scores earnedby patients with dementia (r � 0.84; R2 � 0.71).

Analyses of 2 other questionnaires showedscores that were significantly different by dementiastatus (Table 2). For the modified FBI question-naire, 95% of patients with dementia scored 19 ormore points, whereas 95% of patients without de-mentia scored less than 7 points. For the CaregiverBurden and Satisfaction Questionnaire, 95% ofcaregivers of patients with dementia had indexscores of 26 or less, whereas 95% of caregivers ofpatients without dementia had index scores of 30points or more. However, no significant ability topredict dementia status was found in 4 of the ques-tionnaires used in our clinic assessments: theUCLA Neuropsychiatric Inventory, Beck, Yesav-age, and Stress Warning Signals (Table 2). Oneexplanation for the unexpected poor performanceof the UCLA was that many family members statedit was “too hard to read” (only 117 family caregiverscompleted that questionnaire).

Analyses of the association between one measureof caregiver observations (modified CDR) and theMMSE are also meaningful. Linear regressionanalysis of CDR � MMSE for all patients showeda significant association (Table 3). But this negativecorrelation was significant for patients with demen-tia only (r � .70; R2 � 0.49; P � .0000). Nosignificant correlation was found between CDRand MMSE scores for the patients without demen-tia (r � .23; R2 � 0.05; P � .065).

Analyses of the relationships among the instru-ments noted above indicates that the patientscreening instrument (MMSE) and the familyquestionnaires (modified CDR, modified FBI, andthe Burden score) are all significantly associated butthat each individual questionnaire assesses differentaspects of patient behavior (Table 4). They mayeach be used independently to predict dementia.However, their joint use provides much greaterpredictive power.

We subjected our data to a Nominal LogisticRegression (whole model fit) to determine howwell the combined questionnaires and the MMSEpredicted dementia. Predictors that contributedsignificantly to the clinical categorization included:

(1) MMSE, (2) modified CDR, (3) modified FBI,and (4) Caregiver Burden score; use of these 4variables in the analysis resulted in the most robusteffect size [R2 Adjusted � 0.81, df � 4, �2 � 39.9,P � .0001]. This very large effect size indicates thatthe combination of instruments reliably differenti-ates between patients with and without dementiawith a high degree of accuracy, explaining 81% ofthe variance in the data. Each individual instru-ment, or variable seems to be insufficient to predictreliably whether a patient is demented or not, buttheir combined use predicts very accurately whichpatients will later be diagnosed with a degenerativedisease that causes dementia (see Table 2 for R2

effect sizes).

DiscussionThis complete instrument (caregiver question-naires and a patient test) reliably detects patientswho are demented, even in comparison with otherneurologically ill patients. The behavioral and cog-nitive differences between patients frequently seenin primary practice settings (mildly demented andneurologically healthy) are easier to detect thanbetween patients with dementia and those withother neurological disorders.39 Therefore, this in-strument may be useful for PCPs who must decidewhich patients need referral for specialized assess-ment and disease management or must managesuch patients themselves when specialized demen-tia clinics are not available/acceptable to some pa-tients. Although general screening of all primarypractice patients is not cost-effective, this instru-ment allows concerned family members to alert thepatient’s PCP about a decline in competenceand/or disturbing changes in personality that her-ald certain diseases that cause dementia. Oftenthese symptoms are not displayed in the context ofthe primary care office; family feedback about suchchanges is the only way to detect early stages ofdiseases that cause dementia, when specializedtreatment is most effective.

At present, the packet named Dementia Screen-ing Questionnaire may be used in clinic waitingrooms when family members need to communicatewith the physician about changes in the patient’sbehavior, memory, and/or judgment (Appendix).Support staff or nurses can easily score these ques-tionnaires in approximately 5 minutes. If significantscores are noted on any of the questionnaires, then


use of time to administer the MMSE (approxi-mately 5 to 10 minutes) by a physician or nurse maybe warranted. This screening instrument can beused efficiently by a PCP to determine the patient’sneed for further detailed evaluation.

This study describes a post hoc analysis of afamily caregiver questionnaire designed for use in aspecialized dementia clinic in which patients arescreened and referred by PCPs. Although this in-strument has not been formally assessed in a pri-mary care setting, the authors felt the urgency ofproviding patients with dementia with proper treat-ment as early as possible in the disease processwarranted disclosure of these results now. It is areliable tool that complements other assessmentsused by the PCP. This screening instrument is notintended for general use with every patient seen ina primary care setting, but it can be used when thepatient’s behavioral changes warrant extra assess-ment. Including the family (spouse or adult chil-dren who know the patient very well) in the assess-ment process is not without problems, but it oftenyields very important information about subtlechanges in cognition and behavior that cannot becollected in any other way. Acceptability of thefamily caregiver questionnaire for each privatepractice setting has yet to be determined and mayvary by the ability of family members to read andunderstand the judgments they must make about

the patient. However, the educational levels repre-sented in the OUHSC dementia database rangefrom 6 to 20 years; therefore, this instrument seemsto be sufficient and tolerable for those with re-stricted reading comprehension. Thus, this instru-ment package (including the MMSE) is anotherresource available to physicians who treat those atrisk for diseases that cause dementia: middle-agedand elderly patients.

The instrument’s main limitation is that it can-not determine which disease the patient suffersfrom. That differentiation requires a detailed mul-tidisciplinary assessment by the PCP or in a spe-cialized dementia clinic. Further analyses of thisinstrument should include use in a primary caresetting and could include a blinded scoring of thequestionnaire, separate from its use within the spe-cialized dementia clinic setting where its diagnosticuse has been demonstrated. However, the objectivenature of the scoring of all questionnaire responses(either yes or no to each item) means that thequestionnaire results are not likely to change. Re-finements of the sensitivity of individual questionsand their acceptability to most levels of readingcomprehension are planned for the future. In themeantime, this instrument may be useful to PCPswho must manage the myriad health needs of largenumbers of elderly patients.


AppendixDementia Screening Questionnaire and Scoring and Results Sheet


References1. Brookmeyer R, Gray S, Kawas D. Projections of

Alzheimer’s disease in the United States and thepublic health impact of delaying disease onset. Am JPublic Health 1998;88:1337–42.

2. Ernst RL, Hay JW. The US economic and socialcosts of Alzheimer’s disease revisited. Am J PublicHealth 1994;84:1261–4.

3. Gutterman E, Markowitz J, Lewis B, Fillit H. Costof Alzheimer’s disease and related dementia in man-aged-Medicare. J Am Geriatr Soc 1999;47:1067–71.

4. Hebert LE, Beckett LA, Scherr PA, Evans DA. An-nual incidence of Alzheimer disease in the UnitedStates projected to the years 2000 through 2050.Alzheimer Dis Assoc Disord 2001;15:169–73.

5. Pressley JC, Trott C, Tang M, Durkin M, Stern Y.Dementia in community-dwelling elderly patients: acomparison of survey data, medicare claims, cogni-tive screening, reported symptoms, and activity lim-itations. J Clin Epidemiol 2003;56:896–905.

6. Boustani M, Peterson B, Hanson L, Harris R, LohrKN; U.S. Preventive Services Task Force. Screeningfor dementia in primary care: a summary of theevidence for the U.S. Preventive Services TaskForce. Ann Intern Med 2003;138:927–37 [summaryfor patients in: Ann Intern Med 2003;138:I60.].

7. Medical Study News, July 16, 20048. Boise L, Neal NB, Kaye J. Dementia assessment in

primary care: results from a study in three managedcare systems. J Gerontol A Biol Sci Med Sci 2004;59:M621–6.

9. Fillit H, Gutterman E, Lewis B. Donepezil use inmanaged Medicare: effect on health care costs andutilization. Clin Ther 1999;21:2173–85.

10. Schacter A, Davis K. Guidelines for the appropriateuse of cholinesterase inhibitors in patients with Alz-heimer’s disease. CNS Drugs 1999;11:281–8.

11. Brunton SA. Alzheimer’s disease: the importance ofearly diagnosis. Fam Pract Recertification 2001;23:33–47.

12. Max W, Webber P, Fox P. Alzheimer’s disease: theunpaid burden of caring. J Aging Health 1995;7:179–99.

13. Ready R, Ott BR, Grace J. Validity of informantreports about AD and MCI patient’s memory. Alz-heimer Dis Assoc Disord 2004;18:11–6.

14. Isella V, Villa ML, Frattola L, Appollonio I. Screen-ing cognitive decline in dementia; preliminary dataon the Italian version of the IQCODE. Neurol Sci2002;Suppl 2:S79–80.

15. Smith-Gamble V, Baiyeweu O, Perkins AJ, et al.Informant reports of changes in personality predictdementia in a population-based study of elderly Af-rican Americans and Yoruba. Am J Geriatr Psychia-try 2002;10:724–32.

16. Morris JC. The Clinical Dementia Rating (CDR):

current version and scoring rules. Neurology 1993;43:2412–4.

17. Morris JC, Ernesto C, Schafer K, et al. Clinicaldementia rating training and reliability in multi-center studies: the Alzheimer’s disease cooperativestudy experience. Neurology 1997;48:1508–10.

18. Alzheimer’s Disease Research Center [homepage onthe Internet]. St. Louis: Washington University;c2002 [cited 2005 May 27]. Available from: http://www.adrc.wustl.edu/.

19. Kertesz A, Davidson W, Fox H. Frontal behavioralinventory: diagnostic criteria for frontal lobe demen-tia. Can J Neurol Sci 1997;24:29–36.

20. Kertesz A, Davidson W, McCabe P, Munoz D. Be-havioral quantification is more sensitive than cogni-tive testing in frontotemporal dementia. AlzheimerDis Assoc Disord 2003;17:223–9.

21. Kertesz A, Nadkarni N, Davidson W, Thomas AW.The Frontal Behavioral Inventory in the differentialdiagnosis of frontotemporal dementia. J Int Neuro-psychol Soc 2000;6:460–8.

22. Kertesz A, Davidson W, Fox H. The quantificationof behavior in frontal lobe dementia. In: Kertesz A,Munoz D, editors. Pick’s disease and pick complex.New York: Wiley and Sons, Inc.; 1998. p. 47–67.

23. Lawton MP, Kleban MH, Moss M, Rovine M,Glicksman A. Measuring caregiver appraisal. J Ger-ontol 1989;44:P61–71.

24. Sheikh JI, Yesavage JA. Geriatric depression scale(GDS): recent evidence and development of ashorter version. In: Brink TL, ed. Clinical gerontol-ogy: a guide to assessment and intervention. Bing-hamton (NY): Haworth Press; 1986. p. 165–273.

25. Folstein MF, Folstein SE, McHugh PR. Mini-men-tal state: a practical method for grading the cognitivestate of patients for the clinician. J Psychiat Res1975;12:189–98.

26. Ford GR, Haley WE, Thrower SL, West CA, Har-rell LE. Utility of Mini-Mental State Exam scores inpredicting functional impairment among white andAfrican American demented patients. J Gerontol ABiol Sci Med Sci 1996;51:M185–8.

27. Zanetti O, Geroldi C, Frisoni GB, Bianchetti A,Trabucchi M. Contrasting results between caregiv-er’s report and direct assessment of activities of dailyliving in patients affected by mild and very milddementia: the contribution of the caregiver’s per-sonal characteristics. J Am Geriatr Soc 1999;47:196–202.

28. Broday H, Pond D, Kemp NM, et al. The GPCOG:a new screening test for dementia designed for gen-eral practice. J Am Geriatr Soc 2002;50:531–4.

29. Knafelc R, Lo Giudice D, Harrigan S, et al. Thecombination of cognitive testing and an informantquestionnaire in screening for dementia. Age Ageing2003;32:541–7.

30. Monnot M, Ross E. Psychosocial questionnaires ac-curately predict dementia. Poster presented at the


9th International Conference on Alzheimer’s Dis-ease and Related Disorders; 2004 Jul 17–22; Phila-delphia, Pennsylvania. Chicago: Alzheimer’s Associ-ation; 2004.

31. Rosenthal R, Rosnow RL. Essentials of behavioralresearch: methods and data analysis. 2nd ed. Boston:McGraw-Hill; 1991. p. 316–23, 350–2.

32. National Institute on Aging. Progress report on Alz-heimer’s disease. Bethesda (MD): National Instituteon Aging; 1998.

33. Evans D. Estimated prevalence of Alzheimer’s dis-ease in the United States. Milbank Q 1990;68:267–89.

34. 2000 Census Results. Washington DC: UnitedStates Census Bureau; 2000. Available from http://www.census.gov/main/www/cen2000.html

35. Gregory CA, Orrell M, Sahakian B, et al. Can fron-totemporal dementia and Alzheimer’s disease be dif-ferentiated using a brief battery of tests?. Int J Geri-atr Psychiatry 1997;12:375–83.

36. Cummings J. The Neuropsychiatric Inventory: as-sessing psychopathology in demented patients. Neu-rology 1997;48(5 Suppl 6):S10–6.

37. Depression in primary care: detection and diagnosis.Volume 1. Rockville (MD): Agency for Health CarePolicy and Research, US Department of Health andHuman Services; 1993.

38. Beck AT. Beck Depression Inventory (BDI). SanAntonio: Harcourt Brace & Company; 1996.

39. Spreen O, Strauss E. A compendium of neuropsy-chological tests: administration, norms, and com-mentary, 2nd ed. New York: Oxford UniversityPress; 1998.


screening for dementia: family caregiver questionnaires reliably

Documents