Of Geriatric Psychiatry and Neurology VOLUME 16, NUMBER 1, MARCH 2003

Journal of Geriatric Psychiatry and Neurology VOLUME 16, NUMBER 1, MARCH 2003

Editor-in-Chief Ned Cassem, MD Benjamin Liptzin, MD Andrew Satlin, MD Massachusetts General Hospital Bay State Medical Center McLean Hospital Alan M. Mellow, MD, PhD Boston, Massachusetts Springfield, Massachusetts Belmont, Massachusetts Ann Arbor VA Medical Center Ann Arbor, Michigan Bruce M. Cohen, MD, PhD Linda A. Miner, PhD Cary Savage, PhD McLean Hospital Southern Nazarene University Massachusetts General Hospital Editorial Assistant Belmont, Massachusetts Tulsa, Oklahoma Boston, Massachusetts Sherry P. Becker, MPH Kenneth L. Davis, MD George B. Murray, MD Lon Schneider, MD Mount Sinai School of Medicine Massachusetts General Hospital USC School of Medicine Associate Editors New York, New York Boston, Massachusetts Los Angeles, California Jeffrey L. Cummings, MD Associate Editor of Behavioral Neurology David A. Drachman, MD James A. Nathanson, MD, PhD Charles A. Shamoian, MD West Los Angeles VA Medical Center University of Massacuhusetts Massachusetts General Hospital Cornell Medical Center Los Angeles, California Medical Center Boston, Massachusetts White Plains, New York Worcester, Massachusetts Charles A. Marotta, MD, PhD Michael Otto, PhD Gary W. Small, MD The Miriam Hospital Carl Eisodorfer, MD, PhD Massachusetts General Hospital UCLA Neuropsychiatric Institute Providence, Rhode Island University of Miami School of Med. Boston, Massachusetts Los Angeles, California Miami, Florida Gary D. Miner, PhD Eric Pfeiffer, MD Theodore Stern, MD Associate Editor for Alzheimer’s and Marshall Folstein, MD University of South Florida Massachusetts General Hospital Other Dementia Tufts University School of Medicine College of Medicine Boston, Massachusetts Alzheimer’s Foundation Boston, Massachusetts Tampa, Florida Tulsa, Oklahoma Paul Summergrad, MD Abstract Editor Alan J. Gelenberg, MD Chester M. Pierce, MD Massachusetts General Hospital University of Arizona Cambridge, Massachusetts Boston, Massachusetts Cornelia Cremens, MD Tuscon, Arizona Massachusetts General Hospital Boston, Massachusetts Bruce G. Pollock, MD, PhD Trey Sunderland, MD Donald Goff, MD Western Psychiatric Institute and Clinic National Institute of Mental Health Freedom Trail Clinic Pittsburgh, Pennsylvania Bethesda, Maryland Editorial Board Boston, Massachusetts George S. Alexopoulos, MD Peter V. Rabins, MD Owen Surman, MD Cornell University Medical College C.G. Gottfries Johns Hopkins University Massachusetts General Hospital White Plains, New York University of Goteborg School of Medicine Boston, Massachusetts Hisings Backa, Sweden Baltimore, Maryland

William H. Anderson, MD Virginia E. Tay, RN, MSN Augusta Mental Health Institute Paula Grammas, PhD Murray A. Raskind, MD Cambridge, Massachusetts Augusta, Maine University of Oklahoma VA Geriatric Research Health Sciences Center Education and Clinic Center Oklahoma City, Oklahoma Seattle, Washington Leon J. Thal, MD Lee Baer, PhD University of California San Diego Massachusetts General Hospital La Jolla, California Boston, Massachusetts Cyril I. Gryfe, MD, FRCPC Scott L. Rauch, MD Toronto, Canada Massachusetts General Hospital Boston, Massachusetts William Yamanishi, PhD Dan G. Blazer, MD, PhD University of Oklahoma Health Duke University Medical Center Stephen E. Hyman, MD Sciences Center, Tulsa, Oklahoma Durham, North Carolina Massachusetts General Hospital Barry Reisberg, MD Boston, Massachusetts New York University Medical Center New York, New York Jerome A. Yesavage, MD Andrew Brotman, MD VA Medical Center Deaconness Hospital Lissy Jarvik, MD, PhD Palo Alto, California Boston, Massachusetts UCLA Neuropsychiatric Institute Paavo J. Riekkinen, MD Los Angeles, California University of Kupio Kupio, Finland George S. Zubenko, MD, PhD Roger A. Brumback, MD University of Pittsburgh Creighton University Medical Center Hideyo Katsunama, MD School of Medicine Saint Joseph Hospital Pathology Tokyo Medical College Hospital Carl Salzman, MD Pittsburgh, Pennsylvania Omaha, Nebraska Tokyo, Japan Massachusetts Mental Health Center Boston, Massachusetts

For Sage Publications: Eric Moran, Kendra Kimball, and Kelli Palma Journal of Geriatric Psychiatry and Neurology VOLUME 16, NUMBER 1, MARCH 2003

3 Editor’s Note

ORIGINAL ARTICLES

4 Association of Depression With Agitation in Elderly Nursing Home Residents Oscar Heeren, MD, Luda Borin, MD, Allen Raskin, PhD, Ann L. Gruber-Baldini, PhD, A. Srikumar Menon, MD, PhD, Bruce Kaup, MD, David Loreck, MD, Paul E. Ruskin, MD, Sheryl Zimmerman, PhD, and Jay Magaziner, PhD

8 Correlates of Behavioral Disturbances and Pattern of Pyschotropic Medication Use in Five Skilled Nursing Facilities Fadi H. Ramadan, MD, Bruce J. Naughton, MD, and Roger Prior, ScD

15 Health Service Utilization by Alzheimer’s Disease Patients: A 2-Year Follow-up of Primary Versus Subspecialty Care Peter M. Aupperle, MD, MPH, Edward R. MacPhee, BA, Andrew C. Coyne, PhD, Jonathan Blume, BS, and Betty Sanchez, MS

18 Smoking and Cognitive Performance in the Community Elderly: A Longitudinal Study Wei-Ta Chen, MD, Pei-Ning Wang, MD, Shuu-Jiun Wang, MD, Jong-Ling Fuh, MD, Ker-Neng Lin, PhD, and Hsiu-Chih Liu, MD

23 Construct Validity of the 15-Item Geriatric Depression Scale in Older Medical Inpatients R. Antonelli Incalzi, MD, M. Cesari, C. Pedone, and P. U. Carbonin

29 Treatment of Delirium in Older Adults With Quetiapine Kye Y. Kim, MD, Geoffrey M. Bader, MD, Victor Kotlyar, MD, and Debra Gropper, PharmD

32 Recorded Delirium in a National Sample of Elderly Inpatients: Potential Implications for Recognition Helen C. Kales, MD, Barbara A. Kamholz, MD, Stephanie G. Visnic, BA, and Frederic C. Blow, PhD

39 Long-Term Effects of Donepezil on P300 Auditory Event-Related Potentials in Patients With Alzheimer’s Disease Eiichi Katada, MD, PhD, Koichi Sato, MD, PhD, Akira Sawaki, MD, Yasuaki Dohi, MD, PhD, Ryuzo Ueda, MD, PhD, and Kosei Ojika, MD, PhD

44 Age-Related Impairments in Comprehending Affective Prosody With Comparison to Brain-Damaged Subjects Diana M. Orbelo, PhD, Julie A. Testa, PhD, and Elliott D. Ross, MD

53 Effects of Cognitive Challenge on Gait Variability in Patients With Parkinson’s Disease Jeffrey M. Hausdorff, PhD, Jackov Balash, MD, PhD, and Nir Giladi, MD

59 Familial Dementia With Lewy Bodies With an Atypical Clinical Presentation Lauren T. Bonner, MD, Debby W. Tsuang, MD, MSc, Monique M. Cherrier, PhD, Charisma J. Eugenio, BS, Jennifer Q. Du, RN, Ellen J. Steinbart, RN, Pornprot Limprasert, MD, PhD, Albert R. La Spada, MD, PhD, Benjamin Seltzer, MD, Thomas D. Bird, MD, and James B. Leverenz, MD JOURNAL OF GERIATRIC PYSCHIATRY AND NEUROLOGY (JGPN) presents the results of clinical and research studies considering all aspects of the psychiatric and neuralgic care of aging patients, including age-related biologic, neuralgic, and psychiatric illness; psychosocial problems; forensic issues; and family care. It pursues advances in allied sciences as diverse as molecular biology and genetics, brain imaging, neuropathology, neuropsychology, pharmacology, epidemiology, and health sciences research, which have fueled the burgeoning body of knowledge in geriatric psychology and neurology.

JOURNAL OF GERIATRIC PYSCHIATRY AND NEUROLOGY (JGPN) (ISSN 0891-9887) is published quarterly in March, June, September, and December by Sage Publications, 2455 Teller Road, Thousand Oaks, CA 91320; telephone (800) 818-SAGE (7243) and (805) 499-9774; fax/order line (805) 499-0871; e-mail [email protected]; http://www.sagepub.com. Copyright © 2003 by Sage Publications. All rights reserved. No portion of the contents may be reproduced in any form without written permission of the publisher.

Periodicals postage paid at Thousand Oaks, CA, and additional mailing offices.

Disclaimer: The authors, editors, and publisher will not accept any legal responsibility for any errors or omissions that may be made in this publication. The publisher makes no warranty, express or implied, with respect to the material contained herein. Advertisements in JGPN for products, services, courses, and symposia are published with a “caveat emptor–buyer beware” understanding. The authors, editors, and publisher do not imply endorsement of products, nor quality, validity, or approval of the educational material offered by such advertisements.

Subscriptions: Annual subscription rates for institutions and individuals are based on the current frequency and are valid through August 31, 2003. Prices quoted are in U.S. dollars and are subject to change without notice. Canadian subscribers add 7% GST (and HST as appropriate). Outside U.S. subscription rates include shipping via air-speeded delivery. Institutions: $198 (within the U.S.) / $218 (outside the U.S.) / single issue: $54 (worldwide). Individuals: $125 (within the U.S.) / $145 (outside the U.S.) / single issue: $35 (worldwide).

This journal is abstracted or indexed in Biological Abstracts, Current Contents: Clinical Medicine, EMBASE/ Excerpta Medica, Index Medicus, MEDLINE, Neuroscience Citation Index, PsycINFO, Psychological Abstracts, Science Citation Index, Science Citation Index Expanded, and e-psyche.

Back issues: Information about availability and current prices of back issues may be obtained from the publisher’s order department (address above). Single-issue orders for 5 or more copies will receive a special discount. Contact the order department for details.

Inquiries: Address all correspondence and permissions requests to Sage Publications, 2455 Teller Road, Thousand Oaks, California 91320; e-mail: [email protected]; website: http://www.sagepub.com.

Authorization to photocopy material for internal or personal use under circumstances not falling within the fair use provi- sions of the Copyright Act is granted by Sage Publications to libraries and other users registered with the Copyright Clearance Center Transactional Reporting Service, provided that the fee of $.50 per copy plus $.10 per copy page is paid directly to the CCC, 21 Congress Street, Salem, MA 01970. The identification code for Journal of Geriatric Psychiatry and Neurology is 0891-9887/2003/$.50 + .10.

Advertising: Current rates and specifications may be obtained by contacting Bob Vrooman or Pat Chojnicki at (805) 499- 0721 ext. 7160 (phone), [email protected] (e-mail), Sage Publications, 2455 Teller Road, Thousand Oaks, California 91320.

Claims: Claims for undelivered copies must be made no later than 6 months following month of publication. The publisher will supply missing copies when losses have been sustained in transit and when the reserve stock will permit.

Changes of address: Please inform the publisher at least 6 weeks prior to move. Enclose present mailing label with change of address.

Postmaster: Send address changes to Journal of Geriatric Psychiatry and Neurology, 2455 Teller Road, Thousand Oaks, CA 91320. Editor’s Note

This issue of the Journal marks another transition in cess. I am truly excited to welcome our new publisher–Sage its journey.We bid adieu to B.C. Decker, Inc., which has pub- is a distinguished publishing house which has already lished this journal for nearly a decade, and welcome our shown its commitment to helping further the mission of new publisher, Sage Publications, which has recently the Journal. I look forward to working with them to pro- acquired JGPN. I want to personally express my gratitude vide the very best to our readership. to the staff at Decker for their support during the past 5 years I have worked with them. It has been gratifying to Alan M. Mellow, MD, PhD see the quality and impact of the Journal increase over this Editor-in-Chief time, and their stewardship has been critical to this suc-

Oscar Heeren, MD, Luda Borin, MD, Allen Raskin, PhD, Ann L. Gruber-Baldini, PhD, A. Srikumar Menon, MD, PhD, Bruce Kaup, MD, David Loreck, MD, Paul E. Ruskin, MD, Sheryl Zimmerman, PhD, and Jay Magaziner, PhD

ABSTRACT

Agitation is a serious problem for elderly individuals with dementia. It is often the major reason for admission to a restrictive environment such as a nursing home or hospital. The objectives of the current study were to (1) identify the components of agitation embedded in the Psychogeriatric Dependency Rating Scale (PGDRS) and (2) find race, gender, depression, and cognitive deficits associated with the factors extracted from the PGDRS in demographic variables and clinical variables. The study sample comprised 2285 subjects who were admitted to 59 nursing homes across Maryland. The factor analysis of the PGDRS confirmed that agitation is made up of a number of different elements ranging from physical and/or verbal aggression to wandering. Correlates of these elements varied, as did possible treatments. For example, physical and/or verbal aggression often accompanied severe depression, suggesting that treating the depression may alleviate this problem. However, wandering and psychotic behavior may be less amenable to existing treatments as these behaviors were associated with severe cognitive impairment. (J Geriatr Psychiatry Neurol 2003; 16:4–7).

Agitation is a serious problem for elderly demented indi- and a more general category, which includes pacing, wan- viduals. It is often the primary reason for admission to a dering, and inappropriate behavior. restrictive environment such as a nursing home or hospital. A major interest of several of the studies has been to Efforts to assess agitation have included behavior rating identify background and clinical variables that are either scales,1–3 direct observation,4,5 and direct measurement of precursors or correlates of agitation in elderly demented motor behavior.6 individuals. Some of the variables that have been identi- A number of these scales have been rationally or fied include cognitive deficits, failure to participate in empirically factor analyzed.6–11 The number of factors social activities, and depression.3,4,10,12,13 extracted from these scales has varied from three6 to The objectives of the current study were to (1) iden- eight,9 depending on how broadly the term agitation was tify the components of agitation embedded in the Psy- defined by the scale items. The major components on all chogeriatric Dependency Rating Scale (PGDRS),14 and (2) of the scales have been physical aggression, verbal abuse, find correlates of the factors extracted from the PGDRS in the following variables: race, gender, depression, and cognitive deficits. The significance of this study is that it sep- From the Mental Health Service Line (Drs. Menon, Kaup, and Loreck), Veterans Affairs Maryland Health Care Center, Baltimore, Maryland; arates the various components embedded in the construct Veterans Affairs Capitol Network Research Education and Clinical Center, labeled agitation. Further, it seeks to identify the contri- Baltimore, Maryland (Dr. Ruskin); Department of Epidemiology and bution that race, gender, depression, and cognitive deficits Preventive Medicine (Drs. Gruber-Baldini and Magaziner), University of Maryland School of Medicine, Baltimore, Maryland; Department of play in initiating these behaviors. This new information Psychiatry (Dr. Raskin), University of Maryland School of Medicine; School may assist nursing home staff in preventing the onset of of Social Work (Dr. Zimmerman), University of North Carolina, Chapel various forms of agitation by providing targeted treat- Hill, North Carolina. ment options for each component of agitation. This research was supported by a grant from the National Institute on Aging (RO1 AG 8211), by the Veterans Affairs Maryland Healthcare System, and by the Veterans Affairs Capitol Network Mental Illness METHOD Research Education and Clinical Center (MIRECC). Reprint requests: Dr. Paul E. Ruskin, Mental Health Service Line (116/MH), Subjects Baltimore VA, 10 N. Greene Street, Baltimore, MD 21201. Subjects were recruited from a stratified random sample DOI: 10.1177/0891988703252157 of 59 nursing homes across Maryland. All new admissions

4 © 2003 Sage Publications Association of Depression With Agitation in Elderly Nursing Home Residents / Heeren et al. 5 from September 1992 through March 1995, aged 65 years and 3 = severe depression (scores 21+). These categoriza- and older, who had not been a resident in a nursing home tions follow the recommendations of Alexopoulos and asso- in the previous year were eligible for the study. Informed ciates.17 This scale has been validated as a depression consent was obtained from either the resident or a sig- assessment instrument in both demented and nonde- nificant other.The Institutional Review Board of the Uni- mented persons.17 versity of Maryland approved the consent procedure. There were a total of 3283 eligible subjects, of whom 2285 were Agitation enrolled in the study. There were 998 who did not partic- Behavioral functioning was assessed using the behavior ipate. The major reason for not participating was because subscale of the PGDRS, which assessed the presence of the patients or their significant others refused consent. A aggression, passive hostility,attention seeking, wandering, more detailed account of subject selection and data col- restlessness, and fearfulness. The PGDRS behavior sub- lection procedures may be found elsewhere.15 scale included 17 items rated on a frequency of occurrence scale as 0 = none, 1 = one to two times a week, and Procedure 2 = more often. The original reference to this scale cited Data were collected from multiple sources, including inter- evidence of high inter-rater reliability.14 views with residents, nursing staff, and significant others; medical records; and hospital discharge summaries. The Data Analysis current article reports on a secondary, post hoc analysis, Prior factor-analytic studies of scales assessing agitation using data from a National Institute on Aging–sponsored in elderly nursing home residents have identified at least study conducted by the Department of Epidemiology and three independent factors.6,7,9 Based on these results, a fac- Preventive Medicine at the University of Maryland School tor analysis of the PGDRS behavior subscale was under- of Medicine. The latter, larger study was an epidemiologic taken. To obtain relatively independent factors, a normal study to determine the prevalence of dementia in persons varimax rotation, which provides an orthogonal solution, admitted to nursing homes in Maryland.15 was performed on all factors with eigenvalues of 1 or greater. Five factors met this criterion and were extracted. Measures We also looked at the four-, three-, and two-factor solutions The choice of instruments selected for this study was but decided that the five-factor solution best described the based on two factors. First, as noted above, this was a sec- components imbedded in the PGDRS behavior subscale. ondary analysis, and we were limited to the assessment This was a judgment on our part, but subsequent findings measures included in the larger study. Second, however, seemed to justify this decision. An exact factor scoring we did have a wide range of instruments and demographic method was used that includes all items in a factor, but variables from which to choose. The instruments and vari- the items are assigned weights based on their loading on ables selected were those that had shown promise in prior the factor. This approach maintains the orthogonality studies as predictors of agitated behavior. among factors. Separate analyses of variance were computed using each of the five factors as the dependent vari- Cognitive Deficit able and race, gender, CSDD score, and MMSE score as the Cognitive deficits were assessed using the Mini-Mental independent variables. Range tests (Scheffé) were also State Examination (MMSE).16 The MMSE was adminis- performed to assess where the significant differences were tered to every patient who did not refuse, who could com- when the independent variable contained three or more municate, and for whom testing could be done within 2 groups. An analysis of variance was also performed using months of admission (n = 1446; 63% of the sample). Res- the total score of the PGDRS behavior subscale as the idents were classified by severity of cognitive impairment dependent variable and the same variables noted above based on MMSE scores: 1 = severe (< 10), 2 = moderate as the independent variables. (10–19), 3 = mild (20–26), and 4 = absent (27–30). This is a widely used screen for dementia that has shown high RESULTS inter-rater reliability and validity.16 The mean age of the study population was 81.45 years, with Depression a median age of 81. Female residents accounted for 70.9% Depressive symptoms were ascertained with a modified ver- of the sample. In terms of race, 80.4% were white and 19.4% sion of the Cornell Scale for Depression in Dementia were nonwhite. The majority were widowed (62.2%); 23.9% (CSDD),17 administered to a nurse informant. The CSDD were still married. Fifty-one percent had at least high is an observer-rated scale designed to rate depression in school education. Based on the 4-point scale developed residents with dementia. It contains 19 items, and total for the MMSE, 19.8% of the sample had severe cognitive scores range from 0 to 38. Based on their scale scores, res- impairment, 34.3% had moderate cognitive impairment, idents were categorized as follows: 0 = no depression 27.3% had mild cognitive impairment, and 18.6% had no (scores 1–7), 1 = mild to moderate depression (scores 8–20), cognitive impairment. Based on the CSDD, 21.7% of the 6 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003 study population was classified with either mild or mod- Table 1. Item Loading on the Five-Factor Solution of the Behavior erate depression. Only 0.09% had severe depression. Subscale of the Psychogeriatric Dependency Rating Scale As noted, five factors were extracted from the PGDRS Factor Loading* behavior subscale. The factor loadings of the scale items Item Content 1 2 3 4 5 are listed in Table 1. Based on items with the highest loading for each factor (0.40 or higher), factors were Interferes with and disturbs .51 .42 .24 .13 .17 assigned the following names: 1 = physical aggression/ver- activities of other residents or staff bal abuse (5 items), 2 = wandering (3 items), 3 = intru- Gets you to do things for .13 .05 .75 .03 .04 siveness (2 items), 4 = psychotic behavior (2 items), and him/her that he/she can do 5 = manic/destructiveness (2 items). Factor 1, physical Wanders around the floor/ .13 .81 .01 .01 .08 nursing home aggression/verbal abuse, describes hitting, biting and Shows behavior that are .32 .48 .01 .33 .09 scratching, that is, physical abuse against others. Factor objectionable (spit, undress, 5, manic/destructiveness, describes behavior inflicted on urinate) Demands conversation .17 .16 .74 .07 .11 one’s belongings, that is, destroying and damaging clothes Has difficulty understanding .12 .36 .36 .31 .07 and belongings. Factor 5 differs from factor 4, psychotic what you want behavior, in that the latter includes visual and auditory Makes noises that are .48 .08 .33 .29 .13 objectionable hallucinations and incoherent speech, behaviors that are Hits, bites, scratches .75 .19 .14 .04 .08 frankly psychotic. Deliberately refuses to obey .66 .27 .16 .08 .06 The F-ratios for the PGDRS behavior subscale total commands Verbal abuse and threats .80 .02 .09 .06 .05 score and for factors 1 to 4 were significant at the .01 Restlessly paces .12 .80 .08 .02 .05 level (Table 2). Cognitive impairment significantly predicted Self-destructive acts .29 .26 .12 .24 .38 outcome for the PGDRS total score and for factors scores Destroys and damages .16 .06 .06 .22 .68 clothes and belongings 1, 2, and 4 (see Table 2). Depression measured by the Inappropriate elation .14 .09 .16 .12 .76 CSDD significantly predicted the PGDRS total score and Thinks people are trying to .61 .15 .10 .32 .09 factor scores 1, 3, and 4 (see Table 2). Gender significantly harm him/her Sees and hears things that .11 .04 .08 .63 .19 predicted factors scores 1 and 2 (see Table 2). Race sig- are not there nificantly predicted outcome only for factors score 1 (see Speech incoherent .13 .31 .06 .66 .08 Table 2). *Factor names: 1 = physical aggression/verbal abuse ; 2 = wandering; 3 = intrusiveness; There were also eight significant interactions among 4 = psychotic behavior; 5 = manic/destructiveness. the independent variables (see Table 2). For the PGDRS Numbers in bold indicate items that were used in that factor total score, the only significant interaction was between cognitive deficit and depression. For factor 1, the significant interactions were between cognitive deficit and depres- depressed white females and males with moderate cogni- sion, between cognitive deficit and gender, between tive impairment. In this context, bear in mind that whites depression and gender, between depression and race, and comprised 80% of the sample and women 70%. Wandering between gender and race. For factor 3, the only significant was a problem in the severely cognitively impaired and interaction was between depression and race. There were more so in males than females. Attention seeking occurred no significant interactions for factor 2, 4,or 5. most often in residents with severe depression, especially white residents with moderate cognitive deficits. Finally, DISCUSSION as one might expect, psychotic behaviors were seen in residents with severe cognitive impairment. These findings The factor analysis of the PGDRS confirmed that agita- are summarized in Table 3. tion is a construct made up of a number of different ele- The results indicate that physical and/or verbal aggres- ments ranging from physical and/or verbal aggression to sion often accompanies severe depression in nursing home wandering. This finding is consistent with the results of residents, suggesting that treating depression may alle- other factor analyses, including the work of Bogner and viate this problem. Similarly,treating depression in atten- associates,1,7) who view agitation as a general or unitary tion-seeking or intrusive patients may also alleviate this construct with three underlying factors. This is an impor- problem. However, wandering and psychotic behavior may tant point to keep in mind because the predisposing fac- be less amenable to treatment as these behaviors were asso- tors and treatment of these elements or underlying factors ciated with severe cognitive impairment, for which there can be quite different. For example, we found that different is currently no effective treatment. As noted above, the sam- variables predicted physical and/or verbal aggression (fac- ple studied was primarily white and female. This may tor 1), wandering (factor 2), and intrusiveness (factor 3). explain, in part, why many of the results seem targeted Moderate cognitive deficits and severe depression to this cohort of nursing home residents. were associated with physical aggression. In particular, high These results are consistent with prior findings that levels of physical aggression were present in severely identified cognitive deficits and depression as precursors Association of Depression With Agitation in Elderly Nursing Home Residents / Heeren et al. 7

Table 2. Significant F-Ratios for Main Effects and Interactions Table 3. Summary of Results Based on Significant Main Effects (M) and Interactions (I) Factors Factors Predictors Variable PGDRS 1 2 3 4 5 Physical aggression/ Mild to moderate cognitive MMSE 13.994** 3.85** 19.07** 0.913 15.12** 0.82 verbal abuse impairment (M) Severe depression (M) Cornell 13.38** 13.69** 1.35 10.35** 7.52** 1.74 Severe depression and mild cognitive impairment (I) Gender 1.73 6.45* 4.09* 2.06 1.74 2.09 Severe depression and white (I) Females more than males (M) Race 0.77 5.62* 0.007 3.19 1.07 0.375 Whites more than nonwhites (M)

MMSE 4.64** 8.79** 1.34 3.18* 1.25 0.68 Wandering More cognitive impairment X Cornell more wandering (M) Males more than females (M) MMSE 0.634 2.79* 0.30 0.59 0.83 0.19 Intrusiveness High depression (M) X Gender High depression and mild to moderately cognitively Cornell 1.61 3.11* 1.84 1.52 2.17 0.81 impaired (I) X Gender Highest in severely depressed whites (I) MMSE 0.206 0.365 0.69 0.56 0.03 0.52 X Race Psychotic behavior High cognitive impairment (M) Mild depression (M) Cornell 1.57 3.55* 0.39 3.89* 1.69 0.42 X Race

Gender 0.340 4.50* 2.26 0.19 2.03 1.75 X Race 5. Bliwise DL, Lee KA. Development of an agitated behavior rat- Factor names: 1 = physical aggression/verbal abuse; 2 = wandering; 3 = intrusiveness; 4 = ing scale for discrete temporal observations. J Nurs Meas 1993; psychotic behavior; 5 = manic/destructiveness. 1:115–124. *Significant at .05 level (two-tailed); **significant at .01 level (two-tailed). 6. Cohen-Mansfield J. Assessment of agitation. Int Psychogeriatr PGDRS = Psychogeriatric Dependency Rating Scale; MMSE = Mini-Mental State Examination. 1996; 8:233–245. 7. Bogner JA, Corrigan JD, Stange M, et al. Reliability of The Agi- tated Behavior Scale. J Head Trauma Rehabil 1999; 14:91–96. 3,4,10,12,13 8. Logsdon RG, Teri L, Weinar MF, et al. Assessment of agitation and correlates of agitation. The present study in Alzheimer’s disease: The Agitated Behavior in Dementia extended these findings by pinpointing the levels of cog- Scale. J Am Geriatr Soc 1999; 47:1354–1358. nitive deficits and depression associated with various com- 9. Tariot DM, Mack JL, Patterson MB, et al. The Behavior Rating ponents of agitation, such as physical aggression. This Scale for Dementia of the Consortium to Establish a Registry for information is valuable for identifying specific underly- Alzheimer’s Disease. Am J Psychiatry 1995; 152: 1349–1357. ing psychiatric problems that may be amenable to treat- 10. Beck C, Frank L, Chumbler MR, et al. Correlates of disruptive behavior in severely cognitively impaired nursing home residents. ment as a way of managing specific aspects of agitated Gerontologist 1998; 38:189–198. behavior. 11. Corrigan JD, Bogner JA. Factor structure of the Agitated Behav- ior Scale. J Clin Exp Neuropsychol 1994; 16:386–392. Acknowledgments 12. Cohen-Mansfield J, Marx MS. Do past experiences predict agi- We would like to acknowledge the cooperation of the facilities, residents, and tation in nursing home residents? Int J Aging Hum Dev 1989; families participating in the University of Maryland Long-Term Care Project. 28:285–294. 13. Cohen-Mansfield J, Marx MS. Relationship between depression and agitation in nursing home residents. Compr Gerontol 1988; References 2:141–146. 1. Bogner JA, Corrigan JD, Bode RK, et al. Rating scale analysis of The Agitated Behavior Scale. J Head Trauma Rehabil 2000; 14. Wilkinson IM, Grahm-White. J Psychogeriatric Dependency 15:656–669. Scale (PGDRS): a method of assessment for use by nurses. Br J 2. Yudofsky SC, Kopecky HJ, Kunik M, et al. Assessment of agi- Psychiatry 1980; 137:558–565. tation in Alzheimer's disease: The Agitated Behavior in Demen- 15. Magaziner J, German PS, Zimmerman SI. The prevalence of tia Scale. J Neuropsychiatry Clin Neurosci 1997; 9:541–548. dementia in a statewide sample of new nursing home admissions 3. Cohen-Mansfield J, Marx MS, Werner P.Agitation in elderly per- age 65 and older: diagnosis by expert panel. Gerontologist 2000; sons. An integrative report of findings in a nursing home. Int 40:663–672. Psychogeriatr 1992; 4:221–240. 16. Folstein MF,Folstein SE, McHugh PR. Mini Mental State: a prac- 4. Cohen-Mansfield J, Marx MS, Werner P. Observational data on tical method for grading cognitive state of patients for the cli- time use and behavior problems in the nursing home. J Appl nician. J Psychiatr Res 1975; 12:189–198. Gerontal 1992; 11:111–121. 17. Alexopoulos GS, Abrams RC, Young RC, et al. Cornell Scale for Depression in dementia. Biol Psychiatry 1988; 23:271–284. Correlates of Behavioral Disturbances and Pattern of Psychotropic Medication Use in Five Skilled Nursing Facilities

Fadi H. Ramadan, MD, Bruce J. Naughton, MD, and Roger Prior, ScD

ABSTRACT

There are several treatment options for behavioral disturbances (BDs) in dementia. However, the choice of a specific psychotropic agent is directed by personal preferences and local community practice patterns. We examined the relationship between common clusters of BDs and the use of different classes of psychotropic agents in our community. A cross-sectional study of 430 long-term care residents from 5 nursing homes was undertaken. The Behavior Measurement Scale (BMS) was used to measure the frequency of BDs grouped in 4 categories. Residents with > 4 BD episodes in at least one category during a 2-week observation period were the behavior group and were considered to have clinically significant BDs. A sample of patients who had < 4 BDs in all BMS categories during the same observation period defined the nonbehavior group. A BD cluster was defined as > 4 BDs occurring in one or more BMS categories during the 2- week observation. Data on functional status, comorbidity, use of benzodiazepines, antidepressants, and neuroleptic agents were collected with chart review. The chi-square test was used to examine the correlation between variables. Clinically significant BDs were identified in 27.2% (117/430) of the residents in the sample. Five of 15 behavior clusters accounted for 73% of all clinically significant BDs. The 5 clusters were verbally nonaggressive behaviors (cluster 1, 20.5%), behaviors from all 4 categories (cluster 2, 17.9%), verbally and physically nonaggressive behaviors (cluster 3, 14.5%), physically nonaggressive behaviors (cluster 4, 12.8%), and verbally aggressive and nonaggressive behaviors (cluster 5, 7.7%). Cluster 5 had a negative correlation with functional impairment (P = .009). There was a significant correlation between cluster 2 and benzodiazepine use (P = .014). No other significant correlation was found between any of the 5 clusters and demographic variables, comorbidity status, and use of antidepressant or neuroleptic medications. Residents in the behavior group had higher impairment in self-feeding (P = .036) and bathing (P < .001) and were more likely to be treated with benzodiazepines (P = .004) and neuroleptic agents (P = .009) than residents in the nonbehavior group (n = 116). The higher use of neuroleptics and benzodiazepines in the behavior group compared with the nonbehavior group indicates that BDs are being identified for treatment, but the medications used may not be efficacious. The lack of association between specific classes of psychotropic medications and distinct behavior clusters indicates that clinicians are not using a standardized approach to target the neurochemical abnormalities that may underlie certain behavior clusters. Some behavior clusters correlate with impairment in specific activities of daily living categories such as bathing and feeding, making room for nonpharmacologic interventions. (J Geriatr Psychiatry Neurol 2003; 16:8–14).

These facilities are the Deaconess Center,Waterfront Health Care Center, Department of Medicine (Drs. Ramadan, Naughton, and Prior), School of Degraff Hospital, Millard-Fillmore Gates Hospital, and Greenfield Nursing Medicine and Biomedical Sciences, State University of New York at Buffalo, Home. Buffalo, New York; Department of Medicine (Drs. Ramadan and Naughton), The abstract of this article was presented at the Chicago 2001 annual Kaleida Health System, Division of Geriatric Medicine, State University American Geriatrics Society Conference. of New York at Buffalo, Buffalo, New York. Reprint requests: Dr. Fadi Ramadan, 1910 Sheridan Drive, #9, Kenmore, This study was supported by a grant from the Oishei Foundation in Buffalo, NY 14223. New York. It was conducted in five skilled nursing facilities, two of which are affiliated with the major teaching hospital of the University at Buffalo. DOI: 10.1177/0891988703252177

The necessity for managing patients with significant ple would be to determine the extent to which SSRIs are behavioral problems in nursing homes is increasing, owing used for verbal agitation, a behavior that is often associ- in part to the growth of the “very old” or “frail elderly”.1 ated with depressive symptoms and reduced central sero- With advanced age, physical and mental health problems tonin levels. increase significantly.1 Both pharmacologic2–6 and nonpharmacologic interventions6–8 have been effective for cer- METHODS tain behavioral disturbances (BDs); however, consensus on specific treatment strategies remains elusive.9 Even for Participants instances in which there is consensus, treatment often Four hundred and thirty-one residents from five skilled varies with local community practices10 and clinician nursing care units in five SNFs were included in the study. biases.11 The physician staffing, medical directorship, and hospital Research on agitated behaviors in dementia is char- affiliation differed among these SNFs. In addition, the acterized by inconsistent and imprecise definition of BD,12 study SNFs represented different populations that included unavailability of reliable and uniform clinically applica- inner city and suburban communities. Because of the high ble measurement tools,12 and absence of consensus guide- prevalence of dementia in SNFs35 and the frequent asso- lines for treatment.13 The 7th International Psychogeriatric ciation of psychiatric symptoms and BDs with cognitive Association Congress (1995) issued a number of recom- impairment,36, 37 only residents who had the diagnosis of mendations for research priorities.14,15 These recommen- dementia and resided on one of the long-term care units dations emphasized the need for determining the incidence during the data collection period were eligible for the and characteristics of BDs14 and developing treatment study. We excluded residents who were comatose or in a guidelines based on etiology.15 Longitudinal evaluation of vegetative state and residents from subacute rehabilita- therapeutic responses of BDs was identified as another crit- tion units. ical area for research.15 Reliable and accurate measurement Residents with > 4 BD episodes in at least one Behav- of agitated behaviors is crucial for tracking illness pro- ior Measurement Scale (BMS) category38 during a 2-week gression and monitoring the effects of treatment.16 observation period were defined as the behavior group Despite improved understanding of BDs and their and considered to have clinically significant BDs. A ran- underlying etiologies,10 traditional neuroleptic medica- dom sample (n = 116) of residents who did not meet the tions are often used to treat BDs.10,17 However, the efficacy definition for clinically significant BDs represented the of neuroleptic medications may be limited,18–22 and adverse comparison or nonbehavior group. effects are common.18,23 Although other agents, such as selective serotonin reuptake inhibitors (SSRIs),24–26 valproic Measures acid,27,28 and benzodiazepines,19,20,29,30 are also being used Demographic data were collected on the study sample. The for BDs, it is not clear how clinicians select a specific BMS was used for behavior measurement. The BMS meas- agent for a particular BD.10,11 ures the frequency of 33 behaviors38 grouped in four cat- It has been previously shown that psychiatric symp- egories: (1) verbally aggressive, (2) verbally nonaggressive, toms and BD may co-occur in dementia.31–33 However, (3) physically aggressive, and (4) physically nonaggressive. treatment of these psychiatric symptoms may have either The BMS, derived from the Cohen-Mansfield Agitation no effect on BD10 or a beneficial effect that is independent Inventory (CMAI),39 allows the rating of the same behav- of the medication’s effect on psychosis.34 A recent study by ior items included in the CMAI, daily,and per nursing shift. Sultzer et al found that treatment of delusions in concert The BMS represents a prospective version of the CMAI.38 with BDs did not predict greater behavioral improvement Although there are other scales available for behav- when compared with treatment of subjects without signs ior measurement, such as the Neuropsychiatric Inven- of psychosis.10 On the other hand, treatment of depressive tory (NPI)39 and the Manchester and Oxford Universities symptoms with trazodone resulted in greater behavioral Scale for the Psychopathological Assessment of Dementia improvement.10 These findings support the distinction (MOUSEPAD),40 most of them include psychiatric com- between psychoses and BDs and can be used to improve ponents, such as delusions and paranoia. In addition, most the prescribing patterns of psychotropic medications,10 of these scales are time consuming,41 and some require the specifically a more targeted approach to the use of neu- skills of research assistants for completion.42 On the other roleptic agents.11 hand, the BMS is a simple scale that rates BDs only, and The objective of our study was to examine the pattern it can be completed by any member of the nursing team.38 of psychotropic medication use in five skilled nursing facil- The 33 CMAI behavior items on the BMS are shown in the ities (SNFs) in relation to the five most frequently encoun- Appendix. tered clusters of BD during a 2-week sampling period. In each category, behaviors were rated either present Our goal was to determine whether pharmacologic inter- (occurred at least once during a nursing shift) or absent ventions were consistent with current recommendations (never occurred) at the end of nursing shifts. Behavior was and to identify potential areas for improvement. An exam- recorded when the corresponding category (from 1 to 4) was 10 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003 circled on the BMS chart.38 The inter-rater reliability38 and daily living44 listed on the Katz index, and number of validity38,43 of the BMS have been previously reported. comorbid illnesses. Data on medication use were reported The Katz Index of Activities of Daily Living44 was using the number 1 (medication class used) or zero (med- used to measure functional status. The research coordi- ication class not used). The same data were collected for nators completed this measurement in collaboration with the nonbehavior group. the head nurse on each nursing unit. A score of > 10 on this index defined significant functional impairment. The level Analyses of comorbidity was determined based on chart review.Sig- The whole study sample was used to calculate the fre- nificant comorbidity was defined as having two or more quencies of the 4 behavior categories and examine their medical illnesses, such as congestive heart failure, chronic distribution in relation to morning and evening nursing obstructive lung disease, and cancer. Data on the use of shifts. We used the cutoff score of > 4 episodes/2-week antidepressants, neuroleptic agents, and benzodiazepines period/category/resident, to identify behavior clusters were collected. defined as BDs occurring in one or more categories. Cor- The diagnosis of dementia was obtained from the SNF relation between the 5 most commonly encountered behav- medical records. We did not collect data on neuroimaging ior clusters and demographic variables, functional status, or mental state examination results since studies have comorbidity status, and medication use was examined. found only modest differences in the prevalence of psy- Data on demographic variables, clinical characteristics, fre- chiatric symptoms and BDs in different types of demen- quency of behaviors, and medication use were entered tia (Alzheimer's disease, vascular, etc.) and at different into a desktop computer and analyzed using the SPSS 9.0 stages of the illness.37 program. Functional status and comorbidity scores were dichotomized to > 10 or < 10, and > 2 or < 2, respectively. Procedures The chi-square test (Fisher exact test) was used to study Institutional Review Board approval for the use of the BMS correlation between variables in the group with clinically was obtained. Prior to data collection, in-service training significant BDs. All possible combinations of clusters were sessions for all members of the nursing staff were conducted tested for any significant differences in demographic vari- in each of the five nursing homes. The training sessions ables. The chi-square test was also used to examine the dif- consisted of a 1-hour group presentation in each nursing ferences in clinical characteristics or medication use facility on the purpose of the study,the value of consistent between the behavior and nonbehavior groups. and complete data, and appropriate use of the BMS. It was emphasized that certified nurse assistants (CNAs) should RESULTS perform the ratings. Fifteen residents from each unit were rated during each Clinically significant BDs were observed for 27% (117/431) 2-week observation period. The sequence was repeated of the study sample. Of those with significant BDs, 58% until all residents satisfying the inclusion criteria from all (68/117) had BDs in two or more categories. A comparison units were rated. The research coordinators reviewed the between the behavior and the nonbehavior groups on data collection sheets from each unit twice a week and demographic data, Katz scores, and comorbidity level is reported any problems to the head nurse. Residents with shown Table 1. The only statistically significant differ- missing ratings (no items circled on the BMS on a nurs- ences found were higher impairment in self-feeding (P = ing shift) were observed for an extended period (beyond 2 weeks) to complete the 2-week data. .036) and bathing (P < .001) for the behavior group. The The BMS sheets were collected at the end of the study differences between behavior frequencies on the morning period. All participants had 14 observations completed. A versus the evening shifts were not statistically signifi- 2-week score in each behavior category was obtained for cant. There were 15 behavior clusters identified in the 117 each nursing home resident by adding the daily positive patients who had clinically significant BDs (Table 2). Five ratings. A score of 4 or more per the 2-week study period clusters comprised 73% of the behaviors. These were ver- per category (corresponding to 2 behavior episodes or more bally nonaggressive behaviors (cluster 1, 20.5%), behaviors per category per week) was considered to be clinically sig- from all 4 categories (cluster 2, 17.9%), verbally and phys- nificant. This cutoff score is based on the Manchester and ically nonaggressive behaviors (cluster 3, 14.5%), physically Oxford Universities Scale for the Psychopathological nonaggressive behaviors (cluster 4, 12.8%), and verbally Assessment of Dementia (MOUSEPAD),40 which sets the aggressive and nonaggressive behaviors (cluster 5, 7.7%). frequency of moderate symptoms at one or more episodes No statistically significant correlations were found per week. between the 5 behavior clusters and demographic variables A data collection sheet was created for the demo- or comorbidity level. There was a negative correlation graphic variables, functional status, comorbidity status, and between cluster 5 (verbally aggressive and nonaggressive medication use. The sheet included information on age, sex, behaviors) and functional impairment (P = .009). total Katz index scores, scores on individual activities of Correlates of Behavioral Disturbances and Pattern of Psychotropic Medication Use in Five Skilled Nursing Facilities / Ramadan et al. 11

Table 1. Demographic and Diagnostic Variables of Patients with and Table 2. Distribution of the 117 Patients in Relation without Clinically Significant Behavioral Disturbances to the 15 Behavior Clusters Group with Behavior Cluster* Number of Patients % Clinically Significant BDs Control Group C2 (cluster 1) 24 20.5 Characteristics (n = 117) (n = 116) C1, C2, C3, C4 (cluster 2) 21 17.9 C2, C4 (cluster 3) 17 14.5 Age 76.1 (SD 14.0) 8.2 (SD 12.0) C4 (cluster 4) 15 12.8 Female 68% (79/117) 73% (85/116) C1, C2 (cluster 5) 9 7.7 High comorbidity 29% (34/117) 39% (45/116) C1 6 5.1 (> 2 chronic illnesses) C1, C2, C3 5 4.3 Katz index score > 10 85% (99/117) 88% (102/116) C3 4 3.4 Requires assistance 75% (88/117) 59% (68/116) C3, C4 3 2.6 with feeding* (P = .036) C2, C3, C4 3 2.6 Requires assistance 90% (105/117) 48% (56/116) C1, C2, C4 3 2.6 with bathing* (P < .001) C1, C3 2 1.7 Receiving an 46% (54/117) 43% (50/116) C1, C4 2 1.7 antidepressant C2, C3 2 1.7 Receiving an SSRI 17% (20/117) 22% (26/116) C1, C3, C4 1 0.9 Receiving a neuroleptic* 27% (32/117) 14% (16/116) Total 117 100 (P = .009) Receiving a 15% (17/117) 3% (4/116) C1 = verbally aggressive; C2 = verbally nonaggressive; C3 = physically aggressive; C4 = benzodiazepine* (P = .004) physically nonaggressive.

*Difference is statistically significant. 45,50 BDs = behavioral disturbances; SSRI = selective serotonin reuptake inhibitor. pendent and that BDs from different categories co- occur with varying frequencies.45 Patients with clinically significant BDs had higher Analysis of medication use showed that 27% (32/117) impairment in self-feeding and bathing compared with of patients with clinically significant BDs received a neu- patients without significant BDs. Our results are consis- roleptic medication, and 15% (17/117) received a benzo- tent with previous reports that examined correlates of diazepine (see Table 1). Individuals in the behavior group agitated behaviors.38,51–53 In a sample of 24 cognitively were treated more frequently with a benzodiazepine impaired nursing home residents, Cohen-Mansfield et al (17/117) (P = .004) or a neuroleptic (32/117) (P = .009) com- found that eating, bathing, and toileting were associated pared with those in the nonbehavior group (4/116 and with more agitation when nursing staff members initiated 16/116, respectively). The use of neuroleptic medications these activities.53 Other studies have found significant and antidepressants did not differ between the 5 clusters correlation between verbal agitation (screaming, cursing) of BD. Patients in cluster 2 (behaviors from all categories) and routine care activities such as bathing and feed- received benzodiazepines more frequently (P = .014) than ing.38,48,52 These findings have implications for the use of patients in other clusters. nonpharmacologic interventions to reduce triggers for BDs.54 DISCUSSION The finding of increased benzodiazepine use among patients with the widest range of BDs indicates that BDs This study reports on the prevalence and treatment pat- are being identified for treatment; however, the treatment tern of BDs in five nursing homes in one community. may not be efficacious. The results also support the evi- Patients with clinically significant BDs were more depend- dence that benzodiazepines may induce paradoxical agi- ent in bathing and eating compared with those without clin- tation in the elderly.55,56 Recent reports do not recommend ically significant BDs. Higher use of neuroleptics and the use of benzodiazepines for treatment of agitation in benzodiazepines was found among patients with clini- older adults.57,58 In addition to paradoxical agitation,56 cally significant BDs compared with patients without clin- these medications have been associated with falls,20 ically significant BDs. Among patients with BDs, those with impaired cognition,19 and daytime sedation.20 They may also BDs in all four categories (cluster 2) were more often induce delirium,59 leading to a variety of agitated behav- treated with benzodiazepines. All remaining clusters did iors.60 not correlate with specific classes of psychotropic med- The significantly higher use of neuroleptic medications ications. for patients with clinically significant BDs suggests that The finding that verbal nonaggression is the most these medications are of limited efficacy for certain BDs.19,22 prevalent behavior in nursing homes is consistent with pre- A meta-analysis of neuroleptic agents for the treatment 45–49 vious reports. A majority of patients (72%) with clin- of agitated behaviors found only a modest response to ically significant BDs had verbal nonaggression either these medications.34 Placebo response rate ranged between alone or in combination with other BDs. Physically aggres- 40% and 60%.18 In contrast to the studies of psychoses in sive behaviors were the least prevalent. It has previously dementia,23,61 a recent study of BDs in dementia found no been noted that the four behavior categories are not inde- difference in outcome between treatment with haloperidol, 12 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003 trazodone, behavior management, or placebo.62 A study on 9. Sultzer D, Gray K, Gunay I. A double blind comparison of trazodone and haloperidol for treatment of agitation in patients with withdrawal of neuroleptics in SNFs found short-term dementia. Am J Geriatr Psychiatry 1997; 5:60–69. improvement in 55% of patients and no change in another 10. Sultzer DL, Gray KF, Gunay I, et al. Does behavioral improve- 22%.63 Overall, there is a paucity of placebo-controlled ment with haloperidol or trazodone treatment depend on psy- trials to show the effectiveness of neuroleptic medications chosis or mood symptoms in patients with dementia? J Am Geriatr Soc 2001; 49:1294–1300. in the treatment of BDs.20 11. Rojas-Fernandez CH, Lanctot KL, Allen DD, et al. Pharma- There are several limitations to this study. First, the cotherapy of behavioral and psychological symptoms of demen- BMS simplifies the rating procedure and does not permit tia: time for a different paradigm? Pharmacotherapy 2001; rating of individual BDs within each category. Second, 21:74–102. the results were based only on frequency of behaviors. There 12. Khachaturian Z. Future research directions. Int Psychogeriatr 1996; 8(Suppl 3):493–495. were no data on the severity of specific BDs, such as phys- 13. Lebowitz B. Consensus conference goals and objectives. Int Psy- ically aggressive behaviors, and their correlation with any chogeriatr 1996; 8(Suppl 3):217–218. of the variables tested. Third, the study did not address 14. Neistein S, Siegal A. Agitation, wandering, pacing, restlessness, medication doses or differentiate between various med- and repetitive mannerisms. Int Psychogeriatr 1996; 8(Suppl 3):399–402. ications within the same class. Fourth, the large number 15. Finkel S, Costa e Silva J, Cohen G, et al. Behavioral and psy- of correlations examined may increase the likelihood of sig- chological signs and symptoms of dementia: a consensus state- nificant findings occurring by chance. Last, the pattern of ment on current knowledge and implications for research and medication use found in this study may differ from that treatment. Int Psychogeriatr 1996; 8(Suppl 3):497–500. in other communities or regions in the United States. 16. Davis L, Buckwalter K, Burgio L. Measuring problem behaviors in dementia: developing a methodological agenda. Adv Nurs In conclusion, this study found no pattern of medica- Sci 1997; 20:40–55. tion use to suggest targeted treatment of specific clusters 17. Maixner SM, Mellow AA, Tandon R. The efficacy,safety and tol- of BDs. There does not appear to be a specific treatment erability of antipsychotics in the elderly.J Clin Psychiatry 1999; pattern of BDs in our community in which psychotropic 60(Suppl 8):29–41. 18. Schneider LS, Pollock VE, Lyness SA. A metaanalysis of controlled medications are chosen based on distinct characteristics trials of neuroleptic treatment in dementia. J Am Geriatr Soc of a particular behavior cluster. We anticipate that the 1990; 38:553–563. results of the present study can be used in the development 19. Lantz MS, Marin D. Pharmacologic treatment of agitation in of treatment guidelines for BD in dementia. dementia: a comprehensive review. J Geriatr Psychiatry Neu- rol 1996; 9:107–119. 20. Stoppe G, Brandt CA, Staedt JH. Behavioral problems associ- Acknowledgments ated with dementia. Drugs Aging 1999; 14:41–54. The authors would like to acknowledge the Departments of Nursing at the Dea- 21. Narayan M, Nelson JC. Treatment of dementia with behavioral coness, Waterfront, Degraff, Millard-Fillmore, and Greenfield skilled nursing disturbance using divalproex or a combination of divalproex facilities for their participation in this research project. We acknowledge the and a neuroleptic. J Clin Psychiatry 1997; 58:351–354. efforts of Angela Vacanti, Maureen Fassl, and Eileen Green in ensuring proper 22. Cohen-Mansfield J, Lipson S, Werner P, et al. Withdrawal of data collection and providing expert data entry. The authors are grateful to Dr. haloperidol, thioridazine, and lorazepam in the nursing home. Joseph Mylotte, who reviewed the final version of the manuscript and provided Arch Intern Med 1999; 159:1733–1740. expert opinion. 23. Devanand DP, Marder K, Michaels KS, et al. A randomized, placebo-controlled dose-comparison trial of haloperidol for psychosis and disruptive behaviors in Alzheimer’s disease. Am J Psy- chiatry 1998; 155:1512–1520. References 24. Burke WJ, Dewan V,Wengel SP, et al. The use of selective sero- 1. National Center for Health Statistics. The National Nursing tonin reuptake inhibitors for depression and psychosis com- Home Survey: 1985 summary for the United States. DDHS plicating dementia. Int J Geriatr Psychiatry 1997; Publication No. PHS 89-1758. Washington, DC: U.S. Government 12:519–525. Printing Office, 1989. 25. Pollock BG, Mulsant BH, Sweet R, et al. An open pilot study of 2. Zubenko GS, Rosen J, Sweet RA, et al. Impact of psychiatric hos- citalopram for behavioral disturbances of dementia. Am J Geri- pitalization on behavioral complications of Alzheimer’s disease. atr Psychiatry 1997; 5:70–78. Am J Psychiatry 1992; 149:1484–1491. 26. Pollock BG, Rosen J, Mulsant BH, et al. Placebo-controlled com- 3. Kunik ME, Yudofsky SE, Silver JM, et al. Pharmacologic approach parison of citalopram versus perphenazine for acute treatment to management of agitation associated with dementia. J Clin Psy- of severe behavioral disturbances associated with dementia. chiatry 1994; 55: (Suppl)13–17. Presented at the Annual Meeting of the American Association 4. Tariot PN. Treatment of agitation in dementia. J Clin Psychia- for Geriatric Psychiatry, 2001; Sanfrancisco, California. try 1999; 60:11–20. 27. Kunik ME, Puryear L, Orengo CA, et al. The efficacy and tol- 5. Lantz MS, Marin D. Pharmacologic treatment of agitation in erability of divalproex sodium in the elderly demented patients dementia: a comprehensive review. J Geriatr Psychiatry Neu- with behavioral disturbances. Int J Geriatr Psychiatry 1998; rol 1996; 9:107–119. 13:29–34. 6. Teri L, Logsdon RG. Assessment and management of behavioral 28. Porsteinsson AP,Tariot PN, Erb R, et al. An open trial of valproate disturbances in Alzheimer disease. Compr Ther 2000; 26:169–175. for agitation in geriatric neuropsychiatric disorders. Am J Geri- 7. Allen-Burge R, Stevens AB, Burgio LD. Effective behavioral atr Psychiatry 1997; 5:344–351. interventions for decreasing dementia-related challenging behav- 29. Mintzer JE, Hoernig KS, Mirski DF. Treatment of agitation in ior in nursing homes. Int J Geriatr Psychiatry 1999; 14:213–232. patients with dementia. Clin Geriatr Med 1998; 14:147–175. 8. Carlson DL, Fleming KC, Smith GE, et al. Management of 30. Christensen DB, Benfield WR. Alprazolam as an alternative to dementia-related behavioral disturbances: a non-pharmacologic low-dose haloperidol in older, cognitively impaired nursing facil- approach. Mayo Clin Proc 1995; 70:1108–1155. ity patients. J Am Geriatr Soc 1998; 46:620–625. Correlates of Behavioral Disturbances and Pattern of Psychotropic Medication Use in Five Skilled Nursing Facilities / Ramadan et al. 13

31. Rockwell E, Jackson E, Vilke G, et al. A study of delusions in a 47. Ryan DP, Tainsh SM, Kolodny V, et al. Noise-making amongst large cohort of Alzheimer’s disease patients. Am J Geriatr Psy- the elderly in long-term care. Gerontologist 1988; 28:369–371. chiatry 1994; 2:157–164. 48. Cariaga J, Burgio L, Flynn W, Martin D. A controlled study of 32. Aarsland D, Cummings JL, Yenner G, et al. Relationship of aggres- disruptive vocalizations among geriatric residents in nursing sive behavior to other neuropsychiatric symptoms in patients with homes. J Am Geriatr Soc 1991; 39:501–507. Alzheimer’s disease. Am J Psychiatry 1998; 6:104–121. 49. Cohen-Mansfield J, Werner P, Marx MS. Screaming in nursing 33. Gilley DW,Wilson RS, Beckett LA, et al. Psychotic symptoms and home residents. J Am Geriatr Soc 1990; 38:785–792. physically aggressive behavior in Alzheimer’s disease. J Am 50. Cohen-Mansfield J. Measurement of inappropriate behavior Geriatr Soc 1997; 45:1074–1079. associated with dementia. J Gerontol Nurs 1999; 25:42–51. 34. Katz I, Jetse DV, Mintzer JE, et al. Comparison of risperidone 51. Beck C, Rossby L, Baldwin B. Correlates of disruptive behavior and placebo for psychosis and behavioral disturbances associ- in cognitively impaired elderly nursing home residents. Arch Psy- ated with dementia: a randomized, double-blind trial. J Clin Psy- chiatr Nurs 1991; 5:281–291. chiatry 1999; 60:107–115. 52. Beck C, Frank L, Chumbler NR, et al. Correlates of disruptive 35. Chandler JD, Chandler JE. The prevalence of neuropsychiatric behavior in severely cognitively impaired nursing home residents. disorders in a nursing home population. J Geriatr Psychiatry Gerontologist 1998; 38:189–198. Neurol 1988; 1:71–76. 53. Cohen-Mansfield J, Marx MS, Werner P. Observational data on 36. Mega MS, Cummings JL, Fiorello T, et al. The spectrum of time use and behavioral problems in nursing home. J Appl behavioral changes in Alzheimer's disease. Neurology 1996; Gerontol 1992; 11:111–121. 46:130–135. 54. Allen-Burge R, Stevens AB, Burgio LD. Effective behavioral 37. Lyketsos CG, Steinberg M, Tschanz JT,et al. Mental and behav- interventions for decreasing dementia-related challenging behavioral disturbances in dementia: findings from the Cache County ior in nursing homes. Int J Geriatr Psychiatry 1999; 14:213–232. study on memory in aging. Am J Psychiatry 2000; 157:708–714. 55. Dietch JT, Jennings RK. Aggressive dyscontrol in patients with 38. Ramadan FH, Naughton BJ. Prospective measurement of behav- benzodiazepines. J Clin Psychiatry 1988; 49:184–188. ioral disturbances in nursing homes. J Am Med Dir Assoc 2000; 56. Lechin F, Van der Dijs B, Benaim M. Benzodiazepines: tolera- 1:149–153. bility in elderly patients. Psychother Psychosom 1996; 39. Cohen-Mansfield J. Conceptualization of agitation: results based 65:171–182. on the Cohen-Mansfield agitation inventory and the agitation 57. Verma S, Daiello LA, Hurley DJ. Managing behavioral distur- behavior mapping instrument. Int Psychogeriatr 1996; 8(Suppl bances in geriatric population. Ann Long Term Care 2001; 3):309–315. 9(Suppl):1–4. 40. Gordon S, Hope T, Burns A. Manchester and Oxford Universi- 58. Alexopoulos GS, Silver JM, Kahn DA, et al. Treatment of agi- ties Scale for the Psychopathological Assessment of Dementia tation in older persons with dementia. A special report. [Special (MOUSEPAD). Br J Psychiatry 1996; 169:293–307. Issue] Postgrad Med 1998; April: 1–88. 41. Finkel S, Lyons JS, Anderson RL. A brief agitation rating scale (BARS) for nursing home elderly.J Am Geriatr Soc 1993; 41:50–52. 59. Gray SL, Lai KV, Larson EB. Drug-induced cognition disorders in the elderly: incidence, prevention and management. Drug 42. Shah A, Heather E, Parkash N. Evaluation of three aggression/agitation behavior rating scales for use on an acute admis- Saf 1999; 21:101–122. sion and assessment psychogeriatric ward. Int J Geriatr 60. Jacobson S, Schreibman B. Behavioral and pharmacologic treat- Psychiatry 1998; 13:415–420. ment of delirium. Am Fam Physician 1997; 56:2005–2012. 43. Ramadan FH, Naughton BJ. Paroxetine for verbal agitation. Ann 61. Street JS, Clark S, Gannon KS, et al. Olanzapine treatment of Long Term Care 1999; 7:282–286. psychotic and behavioral symptoms in patients with Alzheimer 44. Katz S, Downs TD, Cash HR, et al. Progress in development of disease in nursing facilities. Arch Gen Psychiatry 2000; the index of ADL. Gerontologist 1970; 10:20–30. 52:968–976. 45. Cohen-Mansfield J. Agitated behaviors in the elderly.Part II. Pre- 62. Teri L, Logsdon RG, Peskind E, et al. Treatment of agitation in liminary results in the cognitively deteriorated. J Am Geriatr AD. A randomized, placebo-controlled clinical trial. Neurology Soc 1986; 34:722–727. 2000; 55:1271–1278. 46. Cohen-Mansfield J, Werner P. Longitudinal predictors of non- 63. Risse SC, Cubberly L, Lampe TH, et al. Acute effects of neuroleptic aggressive agitated behaviors in the elderly. Int J Geriatr Psy- withdrawal in elderly demented patients. J Geriatr Drug Ther chiatry 1999; 14:831–844. 1989; 2:65–77. 14 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003 Health Service Utilization by Alzheimer’s Disease Patients: A 2-Year Follow-up of Primary Versus Subspecialty Care

Peter M. Aupperle, MD, MPH, Edward R. MacPhee, BA, Andrew C. Coyne, PhD, Jonathan Blume, BS, and Betty Sanchez, MS

ABSTRACT

All dementia patients and their caregivers who had received a university-based comprehensive evaluation and a diagnosis of Alzheimer’s disease during 1997 (N = 80) were identified. Of the original cohort, 48.8% (n = 39) were able to be contacted approximately 2 years after their initial assessment, and the caregivers were the informants for this follow- up. Two subgroups were defined: 22 patients were being seen only by their primary care physicians (MED), while 17 patients were also being treated by a geriatric psychiatry faculty member (GERO). There were statistically significant differences between the 2 groups (MED versus GERO, respectively) at follow-up in terms of (1) institutionalization (30.0% versus 4.6%, P < .05), (2) CDR (2.3 versus 1.5, P < .005), and (3) prescription of donepezil at follow-up (45.5% versus 76.5%, P = .05). These differences are being assessed in a larger scale prospective study. (J Geriatr Psychiatry Neurol 2003; 16:15–17)

Keywords: Alzheimer’s disease; primary care; health service utilization

The cost of caring for patients with dementia is twice as However, many patients rely on primary care physi- high as that of the average Medicare patient. With approx- cians to treat dementia, and it is often difficult for them imately 4 million Americans currently diagnosed with to provide the 3 areas of treatment suggested by the Alzheimer’s disease (AD) accounting for 10% to 30% of guidelines. They are under time limitations and are some- nursing home admissions, this amounts to nearly 70 bil- times not fully informed about diagnosing patients with lion dollars annually.Therefore, considering the cost and dementia. As a result, they may fail to appropriately the growing number of AD patients, it is imperative to assess the severity of the patient’s cognitive deficits and address psychosocial issues and delay institutionaliza- often lack a well-developed plan for managing the long- tion.1,2 Current practice guidelines stress 3 treatments to term psychosocial problems of the disease. The constraints accomplish this goal: cognitive enhancers, specifically cho- on primary care physicians may also lead to differences linesterase inhibitors, to provide symptomatic treatment in care for their patients.6 This was previously demon- of cognitive deficits; psychotropics for behavioral compli- strated in a pilot 1-year follow-up study that looked at cations; and psychoeducation for caregivers.3-5 patients with AD who were treated by their primary care physician only or in combination with a geriatric psychiatrist. It noted significant differences in treatments and Received January 8, 2001. Received revised February 28, 2002. Accepted outcomes between the 2 groups. There existed a greater for publication March 25, 2002. rate of hospitalization and cognitive decline in the cohort From the Division of Geriatric Psychiatry, Department of Psychiatry, treated only by a primary care physician. In addition, the University Behavioral HealthCare, University of Medicine and Dentistry percentage of patients that maintained donepezil treat- of New Jersey, Robert Wood Johnson Medical School, Piscataway, New Jersey. ment, the only cognitive enhancer available at the time, at the 1-year follow-up was significantly lower in this This research was performed at Robert Wood Johnson Medical School and 7 presented as a paper presentation at the 2000 annual meeting of the cohort. American Association for Geriatric Psychiatry. Given the results of the above study, a 2-year follow- Reprint requests: Dr. Peter M. Aupperle, University of Medicine and up was conducted to reevaluate the aforementioned Dentistry of New Jersey, 667 Hoes Lane, P.O. Box 1392, Piscataway, NJ patients and to expand our understanding of the poten- 08855-1392. tial differential outcomes. Specifically, the study was DOI: 10.1177/0891988702250507 broadened to include rates of institutionalization and psy-

© 2003 Sage Publications 15 16 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003 chotropic prescriptions. In addition, donepezil usage, hos- graphic characteristics of the MED (n = 22) and GERO pitalization, and cognitive status were reassessed at a 2- (n = 17) groups at year 2 were compared, and there were year interval. no statistical differences between the 2 cohorts. Analysis of variance also revealed no significant differences in the METHODS CIRS scores of the groups. Chi-square analyses revealed a significantly greater All dementia patients and their caregivers who had rate of institutionalization among MED patients, specif- received a comprehensive neuropsychiatric evaluation ically with respect to placement in a nursing home or the (including neuropsychological testing) and a diagnosis of combination of admission to an assisted living facility or AD at a university-based diagnostic clinic during 1997 nursing home (P < .05). Hospitalization rates, usage of (N = 80) were identified. From the original cohort, 57 home health aides, emergency room visits, primary care were contacted after 1 year, and 39 of those remaining could doctor visits, and the usage of dementia day programs were be contacted at the 2-year follow-up. Of these 39 patients, statistically similar in both groups. Chi-square analyses 22 had opted to be seen only by their primary care physi- also revealed a smaller percentage of MED patients receiv- cian (MED) for their AD, while 17 continued to be treated ing donepezil (P = .05). Behavioral complications were at the university-based clinic by a geriatric psychiatry fac- also taken into account by determining if the patient was ulty member (GERO) in collaboration with a case manager on a psychotropic medication (an antidepressant, antipsy- (eg, geriatric social worker or geriatric nurse). The com- chotic, or antianxiety prescription), and no significant dif- prehensive case management provided to the GERO cohort ferences were found between the MED and GERO cohorts included AD education, review of caregiver coping skills, (Table 1). The baseline CDRs for each cohort were statis- behavioral management, community resources, long-term tically equivalent (P = .53). Analyses of variance showed care planning, and legal/financial planning. a significantly greater CDR in the MED cohort relative to All baseline data were collected from the initial com- the GERO cohort at year 2 (2.3 vs 1.5, F = 11.71, P < .005). prehensive evaluation, and demographic data and assess- This difference was evidenced by a divergence in the CDR ment of physical impairment with the Cumulative Illness scores between the 2 cohorts. Rating Scale (CIRS)8 were collected via standardized chart reviews. Data collected at follow-up were obtained by one DISCUSSION of the authors via telephone contact with the patient’s primary caregiver.This information included assessments of In this 2-year follow-up study,3 important themes emerged use of health services by the patient (institutionalization, in comparing the treatment of AD by primary care physi- hospitalizations for nonpsychiatric problems, usage of cians versus geriatric psychiatrists: (1) a significantly home health aides, emergency room visits, primary care greater rate of institutionalization for those being treated doctor visits, and the usage of dementia day programs), only by primary care physicians; (2) less usage of donepezil prescriptions, and cognition (Clinical Dementia Rating (45.5% vs 76.5%), consistent with the results of the 1- Scale [CDR]).9 Nonparametric and correlational analyses year follow-up; and (3) greater cognitive decline via CDR of the baseline and follow-up data were performed using scores, also consistent with the previous study. the SAS Language for Personal Computers. The protocol Data on the usage of psychotropic medications were was approved by the University of Medicine and Den- evaluated to assess for behavioral complications between tistry of New Jersey Institutional Review Board, and the 2 cohorts, and no significant differences existed. Hos- informed consent was obtained from the caregiver after pitalization rates (38.7% vs 14.8%) and usage of home a complete description of the study. health aides (45.2% vs 18.5%) were significantly different at the 1-year follow-up but not at the 2-year follow-up. RESULTS There exist several areas of concern regarding the results of this study. Since only 39 of the original cohort The cohort at the 2-year follow-up (n = 39) was a subset of 80 could be contacted at the 2-year follow-up, there was of the original cohort of patients diagnosed with AD at base- a selection bias. However, no significant baseline differences line (n = 80). To assess whether this subset was not demo- were noted between these cohorts, leading to the conclu- graphically or clinically disparate from the original cohort, sion that comparisons between the remaining subjects the 2-year follow-up cohort was compared with the 41 are useful and that any bias due to attrition was minimal. patients that could no longer be assessed at year 2. The It should also be noted that in this retrospective study,the cohorts were not statistically different in terms of age cohorts were not randomized, the interviewers were not (78.8 vs 78.4 years), gender, marital status, living arrange- blinded, and the informant was the caregiver. ment, and relationship of caregiver. In terms of cognition In summary, significant differences exist at the 2- and medical comorbidity,there were also no statistical dif- year follow-up in this pilot study of 2 different models of ferences: CDR = 1.7 versus 1.5 (t = 0.03, df = 50, P = .97), care. Primary care intervention trials would be useful in CIRS = 2.3 versus 1.8 (t = 0.57, df = 54, P = .57). The demo- assessing differences in outcomes after an educational Health Service Utilization by Alzheimer’s Disease Patients / Aupperle et al. 17

Table 1. Two-Year Outcome: Primary Care Physicians (MED) Versus Geriatric Psychiatrist Care (GERO)

MED (n = 22) GERO (n = 17)

n%n% χ2 P OR 1/OR

Service usage Hospitalization 5 22.7 2 11.8 0.78 .38 0.45 2.21 Assisted living 4 18.2 1 5.9 1.30 .26 0.28 3.56 Nursing home 5 22.7 0 0.0 4.06 .04 0.00 Undefined Assisted living/nursing home 9 40.9 1 5.9 5.30 .01 0.09 11.08 Medication Donepezil 10 45.5 13 76.5 3.81 .05 3.90 0.26 Antidepressant 8 36.4 8 47.1 0.15 .70 1.56 0.64 Antianxiety 1 4.5 1 5.9 0.04 .85 1.31 0.76 Antipsychotic 3 13.6 3 17.6 0.12 .73 1.36 0.74

OR = odds ratio. intervention to determine if a collaborative care model is disease: a randomized, controlled trial. JAMA 1996; 276:1725- efficacious. Specifically, one would randomize equivalent 1731. clinical sites and provide specialized intensive education 2. Brummel-Smith V.Alzheimer’s disease and managed care: how much will it cost? J Am Geriatr Soc 1998; 46:780-781. and consultation for the physicians at the “intervention” 3. Rogers SL, Freidhof LT,and the Donepezil Study Group. The effi- sites, while providing only minimal general education at cacy and safety of donepezil in patients with Alzheimer’s dis- the “control” sites. One would then assess the clinical out- ease: results of a U.S. multicentre, randomized, double-blind, comes and health service utilization. In addition, the placebo-controlled trial. Dementia 1996; 7:293-303. assessment of the cost/benefit implications (both direct and 4. Mittelman MS, Ferris SH, Shulman E, et al. A comprehensive support program: effect on depression in spouse-caregivers of indirect) of such an intervention would be necessary. A AD patients. Gerontologist 1995; 35:792-802. large-scale prospective study with this format is currently 5. Knopman D, Schneider LS, Davis K, et al. Long-term tacrine in progress. (cognex) treatment: effects on nursing home placement and Such research is akin to the large body of data demon- mortality. Neurology 1996; 47:166-177. strating success in enhancing the treatment of depression 6. Edwards RM, Plant MA, Novak DS, et al. Knowledge about in general adults in the primary care setting.10 Given Alzheimer’s disease among primary care physicians, psychiatrists, nurses and social workers. J of Nursing Edu 1992; 3:127- that the majority of older adults receive treatment for cog- 135. nitive or behavioral changes from their primary care 7. Aupperle PM, Coyne AC. Primary vs. subspecialty care: a struc- physicians, the goal of the above investigations is to tured follow-up of dementia patients and their caregivers. Am enhance the outcome in this setting. It is hoped that in the J Geriatr Psychiatry 2000; 8:167-170. future, all AD patients will have access to state-of-the-art 8. Linn BS, Linn MW, Gurel L. Cumulative illness rating scale. J Am Geriatr Soc 1968; 16:622-626. care, with a resultant improvement in quality of life for 9. Hughes CP, Berg L, Danziger WL, et al. A new clinical scale for themselves and their caregivers. the staging of dementia. Br J Psychiatry 1982; 140:566-572. 10. Katon W, Korff MV, Lin E, et al. Stepped collaborative care for References primary care patients with persistent symptoms of depression. 1. Mittelman MS, Ferris SH, Shulman E, et al. A family intervention Arch Gen Psychiatry 1999; 56:1109-1115. to delay nursing home placement of patients with Alzheimer’s Smoking and Cognitive Performance in the Community Elderly: A Longitudinal Study

Wei-Ta Chen, MD, Pei-Ning Wang, MD, Shuu-Jiun Wang, MD, Jong-Ling Fuh, MD, Ker-Neng Lin, PhD, and Hsiu-Chih Liu, MD

ABSTRACT

This prospective study investigated the association between smoking and cognitive performance in a community of nondemented elderly subjects aged 65 or older. All subjects were categorized as current smokers, former smokers, or never smokers. The lifetime cigarette exposure was computed. At baseline, we found the abstainers from smoking had better cognitive performances; however, the differences were not significant after adjusting for age, education, hypertension, diabetes, and vascular events. The lifetime cigarette exposure was not predictive of the cognitive status. At a 3-year follow-up, neither the smoking status nor the lifetime cigarette exposure predicted the declination of cognition. (J Geriatr Psychiatry Neurol 2003; 16:18–22)

Keywords: smoking; elderly; cognitive performance

Previous researchers have shown that nicotine administra- article is to use cohort data from the Kinmen Neurologi- tion had cognitive-enhancing effects in normal adults.1 Nico- cal Disorder Survey (KINDS)13,14 to examine whether cig- tine administration would activate brain areas pertaining arette smoking is predictive of cognitive status (cross- to attention, working memory,motivation, mood, and emo- sectional relationship) or cognitive decline (longitudinal tion.2,3 Similar effects also exist in patients with Alzheimer’s relationship). disease (AD); nicotine injection or skin patches significantly improve patients’ attention, learning, and memory.4 METHODS Even though these biological evidences suggest a pro- tective effect of smoking against AD or cognitive decline, Study Setting and Design the results from epidemiological studies did not show con- The general methodology of the KINDS has been detailed sistent benefits of smoking to cognition. Notably,smoking elsewhere.13-14 The KINDS was a community-based, 2- may adversely increase the risk of cognitive impairment phase epidemiological study of several neurological dis- through atherosclerotic and hemodynamic processes.5 orders including dementia, Parkinson’s disease, essential Recently, several large-scale longitudinal studies have tremor, stroke, and headache. The target population was suggested that smoking might be a risk factor for AD or registered residents aged 65 years and older living in cognitive decline.6-9 In addition, some longitudinal stud- Kinmen, a rural islet located west of Taiwan and near the ies did not indicate any relationship between smoking coast of mainland China. The population in Kinmen was and cognitive function.10-12 quite stable; the rate of immigration and emigration was To date, the impact of long-term cigarette smoking very low. Most of the population lived on farms and were on cognition has yet to be clarified. The purpose of this illiterate since compulsory education was not introduced until the 1950s. A total of 798 nondemented subjects (424 women, 374 men) received a test of cognitive functioning and com- Received July 27, 2001. Received revised August 2, 2002. Accepted for pleted the first phase of the KINDS in 1993. In the follow- publication August 13, 2002. up phase of the KINDS in 1996, 671 subjects (84.1%, 357 From the Neurological Institute, Taipei Veterans General Hospital, and women and 314 men) successfully repeated the cognitive Department of Neurology, National Yang-Ming University School of Medicine, Taipei, Taiwan. testing. Data including demographics as well as medical histories were collected, and the smoking histories of each Reprint requests: Hsiu-Chih Liu, MD, the Neurological Institute, Taipei Veterans General Hospital, No. 201 Shih-Pai Road, Sec. 2, Taipei 112, subject at baseline were assessed. Subjects with demen- Taiwan; e-mail: [email protected]. tia identified during the first phase were excluded from DOI: 10.1177/0891988702250510 data analysis because their recall of medical and smok-

18 © 2003 Sage Publications Smoking and Cognitive Performance in the Community Elderly / Chen et al. 19 ing histories was unreliable and their cognitive function they are associated with either smoking or cognitive func- declined beyond that associated with normal aging. tion and possibly mediate the relationship between them both.19,20 Smoking History Assessment The assessment of smoking history started with a trigger Statistical Analyses question: “Did you ever smoke during your lifetime?” When In this study,the CASI score was the measure of cognitive the answer was “yes,” further detailed information would performance; the CASI score change between both phases be collected. Those who answered “yes” were asked at (follow-up CASI score minus baseline CASI score) was what age they began smoking, whether they were still computed as a measure of cognitive decline. Data analy- smoking, at what age they quit smoking (or the number ses included mainly 2 parts. First, all subjects were clas- of years since they quit smoking), and how many cigarettes sified by the smoking status at baseline as current smokers, on average they had smoked or still smoked per day. former smokers, or never smokers. We examined the cross- sectional relationship between cognitive performance and Outcome Variables: Cognitive Measures smoking status. Subsequently, we tested the longitudinal The cognitive performances were assessed by the Chi- relationship between smoking status and cognitive decline. nese version of the Cognitive Abilities Screening Instru- The differences were analyzed by the χ2 test for categori- ment (CASI).15,16 The CASI is a comprehensive cognitive cal variables and one-way analysis of variance (ANOVA) test validated against clinically diagnosed dementia17 and for continuous variables. Crude correlations of the smok- has a higher sensitivity/specificity than the Mini-Mental ing status to baseline cognitive performance or cognitive State Examination (MMSE).17 The CASI was originally decline were further examined by a multiple linear regres- designed for cross-cultural studies and is globally recog- sion model, adjusting for potential confounders. In this nized nowadays. Pilot studies conducted in Japan and in model, age and education were entered as continuous the United States have demonstrated its cross-cultural variables; hypertension, diabetes, and vascular events applicability and usefulness in screening for dementia, in were considered as categorical variables. monitoring disease progression, and in providing profiles For the second part of data analyses, the number of of cognitive impairment.15 The CASI has a score range of pack years of each subject was computed as a measure of 0 (worst) to 100 (best).15 It takes about 20 minutes to lifetime cigarette exposure by the formula (daily num- administer and includes 9 domain subscales: attention, con- bers of cigarettes smoked/20 cigarettes per pack) × years centration, orientation, long-term memory, short-term smoked. We tested the associations between lifetime cig- memory, abstraction and judgment, language abilities, arette exposure amount and baseline cognitive perform- constructional praxis, and category fluency. The MMSE ance, or the cognitive decline, by Pearson’s correlation. score could be derived from CASI items according to the Crude associations were adjusted for potential confounders CASI manual.15 by linear regression analysis. The CASI has shown good reliability and validity in Additional analyses focused on the influence of quit- the Chinese illiterate population of Kinmen.16 Educational ting smoking on cognitive performance. Using the same sta- disparity influences the performance in the CASI test; tistical strategy as the second part, we checked the people with varied educational achievements score dif- relationship between quitting duration, which was recorded ferently on the CASI test.18 Our recent study18 obtained as the number of years since smoking cessation, and the cutoff scores of the Chinese version of the CASI in demen- cognitive status or decline. tia diagnosis for different educational groups: schooling All statistical analyses were carried out using SPSS year = 0: 49/50 (sensitivity = 0.83; specificity = 0.85); for Windows (SPSS Inc., Chicago, release 10.0, 1999), and schooling year = 1~5: 67/68 (sensitivity = 0.83; specificity = the statistical levels quoted (P values) were two-tailed. 0.91); schooling year ≥ 6: 79/80 (sensitivity = 0.89; specificity = 0.90). RESULTS

Potential Confounders Subjects Selection The potential confounders considered in the present study The preliminary results of baseline KINDS revealed that included age, educational attainment, hypertension, dia- most men (78%) had smoked; however, few women (only betes, and vascular events. Data of these confounders 10%) were current or former smokers. The enormous gen- were based on demographics and medical histories collected der disparity also existed in educational achievement: at baseline. Notably, educational attainment was 65% of men had received formal education, although only recorded as schooling years. Vascular events was an arti- a few years of schooling, but only 11% women were edu- ficially composed variable indicating the presence of one cated. Because the women smokers were few, and consid- of the following self-reported histories: stroke, transient ering the education effect on their cognitive performance, ischemic attack, angina pectoris, and myocardial infarc- we decided to use data from only male subjects to elimi- tion. These variables were considered confounders because nate the confounding effect of gender differences. The data 20 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003

Table 1. Demographics, Hypertension, Diabetes, Vascular Events, and Cognitive Measures in All of the Subjects and Subgroups Categorized by Smoking Status

Smoking Status

Total Subjects (n = 314) Never Smokers (n = 68) Former Smokers (n = 51) Current Smokers (n = 195)

Age in years (SD)a 72.0 (6.0) 72.3 (6.2) 70.0 (4.3) 72.5 (6.3) Education in years (SD)a 3.4 (3.6) 4.4 (3.8) 3.9 (4.1) 2.9 (3.4) Hypertension (%) 26 25 35 23 Diabetes mellitus (%) 8 15 4 7 Vascular events (%) 9 10 6 10 Baseline CASI score (SD)a 79.0 (11.6) 80.9 (9.7) 82.0 (10.3) 77.5 (12.3) Follow-up CASI score (SD) 78.1 (14.5) 78.9 (13.2) 81.3 (14.3) 76.9 (15.0) CASI score change (SD) –0.9 (10.7) –2.0 (9.0) –0.7 (10.4) –0.6 (11.4) Baseline MMSE-CE score (SD)a 24.8 (4.0) 25.3 (3.7) 25.8 (3.1) 24.3 (4.2) Follow-up MMSE-CE score (SD) 24.5 (4.6) 24.6 (4.4) 25.3 (4.4) 24.2 (4.7) MMSE-CE score change (SD) –0.3 (4.0) –0.7 (3.4) –0.5 (3.4) –0.1 (4.2)

CASI = Cognitive Abilities Screening Instrument; MMSE-CE = CASI-estimated Mini-Mental State Examination. a. Variables that are significantly different across smoking status categories (P < .05; one-way ANOVA for continuous variables, chi-square test for proportions).

Table 2. Results of Linear Regression Analyses to Test Smoking Status (Model 1) and Quitting Duration (Model 2) as a Predictor of Baseline CASI Score Controlling for Potential Confounders

Model 1 Model 2

Independent Variable β Coefficient SD P Value β Coefficient SD P Value

Smoking status Current smokers –1.12 1.40 .424 — — Former smokers 0.69 1.85 .707 — — Never smokers Reference Reference Reference — — Quitting duration — — — 0.01 0.02 .753 Age –0.33 0.09 < .001 –0.35 0.09 < .001 Education 1.55 0.16 < .001 1.58 0.15 < .001 Hypertension 0.95 1.34 .476 1.09 1.33 .411 Diabetes –0.04 2.13 .984 –0.04 2.11 .986 Vascular event –3.59 1.95 .067 –3.70 1.95 .059

CASI = Cognitive Abilities Screening Instrument. Adjusted R 2 = 0.30 for both model 1 and model 2.

analyses thus finally focused on 314 nondemented male CASI scores than other smokers did (P = .013); however, subjects who completed the CASI tests in both phases the difference disappeared after adjustment for age, edu- and had no missing data in all variables of interest. A com- cation, hypertension, diabetes, and vascular events (see parison of the included with the excluded men disclosed Table 2, model 1). The CASI score changes between both similar age, educational attainment, and prevalence of phases were comparable regardless of smoking status. hypertension, diabetes, and vascular events. After grouping the former and the current smokers (ie, the “ever smokers”) for further comparison with those Smoking Status and Cognitive Performance that never smoked, these 2 groups had similar mean age Of the 314 included subjects, 68 (21%) were never smok- (72.3 ± 6.2 vs 72.0 ± 6.0 years, P = .665), but the ever smokers, 51 (16%) were former smokers, and 195 (63%) were cur- ers had fewer schooling years (3.1 vs 4.4 years, P = .012). rent smokers. The duration of smoking was 24.0 ± 17.4 The baseline CASI score (78.4 ± 12.0 vs 80.9 ± 9.7, P = .116) years for former smokers and 40.8 ± 19.0 years for current and the CASI score change (–0.6 ± 11.1 vs –2.0 ± 9.0, P = smokers. The distributions of the demographics, hyper- .323) of ever smokers and never smokers were comparable. tension, diabetes, vascular events, and the CASI scores Since the former and current smokers had signifi- (baseline, follow-up, and decline) are summarized in cantly fewer schooling years than the never smokers did, Table 1. Those who continued to smoke were older (P = .025) this may act to minimize any differences related to smok- and had fewer schooling years (P = .008). The proportion ing status. Thus, we divided all the subjects into 2 groups of the subjects with hypertension, diabetes, and vascular by their education (illiterate vs literate) for subanalysis. events did not differ across smoking status categories. At In the illiterate subgroup, never and ever smokers had a baseline, the former smokers had significantly higher comparable mean CASI score (74.9 ± 10.0 vs 69.9 ± 11.7, Smoking and Cognitive Performance in the Community Elderly / Chen et al. 21

P = .132). In the literate subgroup, both smokers did not dose effect was unavailable. A Dutch cohort6 study disclosed score differently (82.5 ± 9.0 vs 83.8 ± 8.7, P = .348). harmful effects of smoking on cognition, but the participants with dementia were not excluded at the baseline. The Influence of Lifetime Exposure Moreover, it might be of concern that the data collection and Duration of Quitting for cognitive outcomes was administered in different sit- The lifetime cigarette exposure ranged from 0 to 140 pack uations: research center at baseline versus home at follow- years, with a mean of 27.9 ± 28.5 pack years. Age and edu- up. Another cohort study conducted in London7 indicated cation had no association with the lifetime exposure. Like- that smoking might be a prospective risk factor for inci- wise, the baseline CASI score or the CASI score change did dental cognitive impairment. However, this finding was still not correlate with lifetime exposure. The lack of associa- limited by a short follow-up interval (1 year), less com- tion remained unchanged in analyses confined to the ever prehensive cognitive measures, and absence of dose effect smokers or the current smokers alone. analyses. The Honolulu-Asia Aging Study (HAAS)8 used The quitting duration among the former smokers the CASI to measure cognition and disclosed positive asso- averaged 7.7 ± 10.5 years. It was slightly associated with ciation between smoking during middle age and later risk education (= 0.13, P = .02) but not with age. A significant of cognitive impairment. Nevertheless, congruent find- correlation existed between the quitting duration and the ings in dose effect are lacking. In conclusion, many possi- CASI scores at both phases (= 0.44, P = .002 for baseline; ble unmeasured confounders (eg, genetic susceptibility) of = 0.12, P = .038 for follow-up). Nonetheless, the crude asso- the relation between smoking and cognition remain to be ciations disappeared after adjusting for all the potential considered before these “positive” studies can be validly confounders (Table 2, model 2). Likewise, the CASI score interpreted. change was not associated with the quitting duration. A Our study results reflect the possibility that harmful similar lack of correlation occurred in further analyses pool- effects after long-term cigarette exposure in atherosclerosis ing data from the current and the former smokers, in and hemodynamic change might “counterbalance” the which the quitting duration of the latter was regarded as beneficial effects mediated through nicotine receptors. zero (year). However, current smokers did not have a higher prevalence of vascular risks in our study (Table 1). This result might DISCUSSION be due to confounding from the “healthy smokers effect.”6 Our results of regression analyses indicated education The present study did not demonstrate any consistent was the most predictive among all potential confounders effects of smoking histories (including the current smok- (Table 2). When the effects of all the potential confounders ing status, lifetime cigarette exposure, and number of of interest were considered individually, we found that years since smoking cessation) on cognitive status and cog- all the crude associations disappeared after adjusting for nitive decline. The crude data showed a better cognitive education but not for age or any other confounder. performance by those who once smoked but quit and a pos- The major limitation of the present study occurs in gen- itive association between the years after quitting and the eralization of the results to women and populations with cognitive status. However, these findings were not signif- high educational attainment. To date, the role of education icant after adjustment for age, education, hypertension, dia- in predicting cognition is still a matter of discussion.20,24 betes, and vascular events. The rate of cognition decline in nondemented aging may Our results were in line with those obtained from a change with different education attainment.25 Furthermore, 10 11 Cleveland cohort study and a Boston cohort study. the age of our study subjects were on average 72 years at These 2 studies did not disclose any substantial impact of baseline and 75 years at the end of the 3-year cohort smoking on cognition, although they focused on some spe- study. Age also may play a role in the speed of cognitive cific cognitive fields rather than the global cognitive func- decline.26 Further studies with a longer follow-up interval 12 tion. One study from the Medical Research Council (MRC) in subjects with older age are necessary. Finally, some also revealed negative association between smoking and potential confounders that may distort the relationship cognition, but interpretation of the study results was lim- between smoking and cognition should be controlled, such ited to individuals with hypertension. Similarly,recent lab- as head injury, medication use, and so forth. oratory studies of brain function via quantitative We examined the predictive value of smoking on electroencephalography (EEG) or event-related potentials global cognitive function and found little evidence of any (ERP) did not evidence any marked differences between significant association. Further studies with longer follow- 21,22 elderly lifelong nonsmokers and smokers. up interval and using more cognitive measures are manda- Contrary to these negative findings, some longitudi- tory to confirm our results since the impact of smoking nal studies suggested smoking as an isolated predictor of history on late-life cognition is enormously complex, involv- 23 cognition. A French cohort study showed some benefits ing at least lifelong trends in behavior, morbidity, and of smoking on attentional and visuospatial functioning, but mortality. 22 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003

Acknowledgments 14. Wang PN, Wang SJ, Fuh JL, et al. Subjective memory complaint This study was supported by grants from the National Science Council, NSC86- in relation to cognitive performance and depression: a longitu- 2314-B075-070. dinal study of a rural Chinese population. J Am Geriatr Soc 2000; 48:295-299. References 15. Teng EL, Hasegawa K, Homma A, et al. The Cognitive Abilities Screening Instrument (CASI): a practical test for cross-cultural 1. Heishman S, Taylor R, Henningfield J. Nicotine and smoking: a epidemiologic studies of dementia. Int Psychogeriatr 1994; 6:45- review of effects on human performance. Exp Clin Psychophar- 58. macol 1994; 2:345-395. 16. Liu HC, Chou P, Lin KN, et al. Assessing cognitive abilities and 2. Ghatan PH, Ingvar M, Eriksson L, et al. Cerebral effects of dementia in a predominantly illiterate population of older indi- nicotine during cognition in smokers and non-smokers. Psy- viduals in Kinmen. Psychol Med 1994; 24:763-770. chopharmacology 1998; 136:179-189. 17. Graves AB, Larson EB, Kukull WA, et al. Screening for demen- 3. Stein EA, Pankiewicz J, Harsch HH, et al. Nicotine-induced tia in the community in cross-national studies: comparison limbic cortical activation in the human brain: a functional MRI between the cognitive abilities screening instrument and the study. Am J Psychiatry 1998; 155:1009-1015. Mini-Mental State Examination. In: Corain B, Iqbal K, Nicolini 4. Rezvani AH, Levin ED. Cognitive effects of nicotine. Biol Psy- M, et al, eds. Alzheimer’s disease: advances in clinical and basic chiatry 2001; 49:258-267. research. New York: Wiley, 1993:113-119. 5. Shinton R, Beevers G. Meta-analysis of relation between ciga- 18. Lin KN, Wang PN, Liu CY, et al. Cutoff scores of Cognitive Abil- rette smoking and stroke. Br Med J 1989; 298:789-794. ities Screening Instrument, Chinese version (CASI C-2.0) in 6. Launer LJ, Feskens EJM, Kalmijn S, Kromhout D. Smoking, screening of dementia. Dement Geriatr Cogn Disord. In press. drinking, and thinking: the Zutphen Elderly Study. Am J Epi- 19. Breteler MMB, Claus JJ, Grobbee DE, Hofman A. Cardiovascular demiol 1996; 143:219-227. disease and distribution of cognitive function in elderly people: 7. Cervilla JA, Prince M, Mann A. Smoking, drinking, and incident the Rotterdam study. Br Med J 1994; 308:1604-1608. cognitive impairment: a cohort community based study included 20. Zhang M, Katzman R, Salmon D, et al. The prevalence of demen- in the Gospel Oak project. J Neurol Neurosurg Psychiatry 2000; tia and Alzheimer’s disease in Shanghai, China: impact of age, 68:622-626. gender, and education. Ann Neurol 1990; 27:428-437. 8. Galanis DJ, Petrovitch H, Launer LJ, et al. Smoking history in 21. Knott VJ, Harr A. Assessing the topographic EEG changes asso- middle age and subsequent cognitive performance in elderly ciated with aging and acute/long-term effects of smoking. Neu- Japanese-American men: the Honolulu-Asia Aging Study. Am J ropsychobiology 1996; 33:210-227. Epidemiol 1997; 145:507-515. 22. Knott VJ, Harr A, Mahoney C. Smoking history and aging-asso- 9. Launer LJ, Andersen K, Dewey ME, et al. Rates and risk fac- ciated cognitive decline: an event-related brain potential study. tors for dementia and Alzheimer’s disease: results from EURO- Neuropsychobiology 1999; 40:95-106. DEM pooled analyses. Neurology 1999; 52:78-84. 23. Leibovici D, Ritchie K, Ledésert B, Touchon J. The effects of 10. Ford AB, Mefrouche Z, Friedland RP, et al. Smoking and cogni- wine and tobacco consumption on cognitive performance in the tive impairment: a population-based study. J Am Geriatr Soc elderly: a longitudinal study of relative risk. Int J Epidemiol 1999; 1996; 44:905-909. 28:77-81. 11. Hebert LE, Scherr PA, Beckett LA, Debanne SM. Relation of 24. Cobb JL, Wolf PA, Au R, et al. The effect of education on the inci- smoking and low-to-moderate alcohol consumption to change in dence of dementia and Alzheimer’s disease in the Framingham cognitive function: a longitudinal study in a defined community study. Neurology 1995; 45:1707-1712. of older persons. Am J Epidemiol 1993; 137:881-891. 25. Evans DA, Beckett LA, Albert MS, et al. Level of education and 12. Cervilla JA, Prince M, Joels S, et al. Long-term predictors of cog- change in cognitive function in a community population of older nitive outcome in a cohort of older people with hypertension. Br persons. Ann Epidemiol 1993; 3:71-77. J Psychiatry 2000; 177:66-71. 26. Verhaeghen P, Salthouse TA. Meta-analyses of age-cognition 13. Liu HC, Wang SJ, Fuh JL, et al. The Kinmen neurological dis- relations in adulthood: estimates of linear and nonlinear age orders survey (KINDS): a study of a Chinese population. Neu- effects and structural models. Psychol Bull 1997; 122:231-249. roepidemiology 1997; 16:60-68. Construct Validity of the 15-Item Geriatric Depression Scale in Older Medical Inpatients

R. Antonelli Incalzi, MD, M. Cesari, C. Pedone, P. U. Carbonin

ABSTRACT

The construct validity of the 15-item Geriatric Depression Scale (sfGDS) has been assessed in selected populations. The aim of this study was to assess the appropriateness of applying the sfGDS to unselected older inpatients. The main component analysis of sfGDS was performed in 2032 medical inpatients (mean age = 76.3 ± 8.4). sfGDS did not qual- ify as a unidimensional test. Three factors explained 47.7% of variance and explored the following dimensions: positive attitude toward life, distressing thoughts/negative judgment about the own condition, and inactivity/reduced self- esteem. The internal homogeneity was poor (Cronbach’s α = .46). A higher fraction of variance was explained in patients independent in all or dependent in ≥ 1 activity of daily living (ADL). In older medical inpatients, sfGDS is not a single construct, which prevents the univocal interpretation of the final score. The higher fraction of explained variance in patients with comparable ADL performance probably reflects the dependency of affective from physical status. (J Geriatr Psychiatr Neurol 2003; 16:23–28)

Keywords: screening of depression; elderly; medical inpatients; short form GDS

The short form of the Geriatric Depression Scale (sfGDS) people and not in the broad hospitalized geriatric popu- is currently used for clinical and research purposes.1 Its lation, which is more commonly screened for depression.6 appropriateness as a screening tool for home-dwelling The present study aimed at verifying whether the elderly people has been questioned by Alden et al.2 Burke final score of sfGDS can be considered really representa- et al3 reported that the sfGDS is an effective screening tive of the outcome depression and to what extent impair- instrument in cognitively intact but not in mildly demented ment of selected psychological dimensions can be patients. The sfGDS was very effective in identifying recognized by analyzing the performance of older medical depressed subjects in the Greek but not in the Israeli eld- inpatients. erly population.4,5 The only study that validated the sfGDS among inpatients was performed in a selected population METHOD that included depressed, demented, and thought-disordered The present study uses data from a large collaborative observational study group, the Italian Group of Pharma- coepidemiology in the Elderly (Gruppo Italiano di Far- Received July 24, 2001. Received revised April 30, 2002. Accepted for pub- macoepidemiologia nell’Anziano [GIFA]). The GIFA is a lication April 30, 2002. multicenter study involving wards of geriatrics and inter- From Centro di Medicina dell’Invecchiamento, Università Cattolica del nal medicine in community or university hospitals scat- Sacro Cuore, Roma, Italy (Drs. Incalzi, Cesari, Pedone, and Carbonin) and tered over the whole Italian territory.The main objective the Department of Community Health, Brown University, Providence, of the GIFA study is to survey drug consumption, incidence RI, USA (Dr. Pedone), for the Gruppo Italiano di Farmacovigilanza nell’Anziano (GIFA). and type of adverse drug reactions, and quality of hospital care. For the present study, we used data collected by This work was supported by grants from Ministero della Sanità No. 900.3 CONV. 1987/535, from Ministero della Pubblica Istruzione fondi 40% 24 centers in the last survey performed (May-June and Sep- (1987), and from Regione Molise (Progetto Active Ageing). tember-October 1998) because the sfGDS was not used in Reprint requests: R. Antonelli Incalzi, MD, Centro di Medicina previous surveys. dell’Invecchiamento, Università Cattolica del Sacro Cuore, Largo A. Procedures were approved by the Catholic University Gemelli, 8, 00168 Roma, Italy; e-mail: [email protected]. Ethical Committee as well as by the Steering Committee DOI: 10.1177/0891988702250532 of CNR–Aging Project. A detailed description of proce-

© 2003 Sage Publications 23 24 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003 dures is available elsewhere.7,8 Briefly,patients underwent closely related variables and then being more meaning- a multidimensional assessment covering several domains: ful and easily interpretable than those obtained in the sociodemographic characteristics, smoking and alcohol extraction phase. The varimax method pursues this objec- intake history,medical problems, pharmacological therapy tive by minimizing the number of items having loadings and adverse drug reactions, complete blood count, cogni- on a factor.The strength of the relationship of items to indi- tive and affective status, and functional capabilities. Each vidual factors is directly proportional to the magnitude of domain was assessed according to a standardized and the corresponding correlation coefficient in the rotated fac- validated method, for example, functional capabilities by tor matrix. This allows the identification of items having activities of daily living (ADL) and instrumental activities loadings on a given factor. The optimal number of factors of daily living (IADL) scales, comorbidity by the Charlson’s to be selected in the final model was chosen by the scree 14 index, and cognitive status by the Abbreviated Mental Test pattern method. (AMT).9-11 Diseases and drugs were codified by the Inter- The Cronbach’s α was measured to assess the degree 13 national Classification of Diseases, 9th revision (ICD-9), of internal homogeneity of the sfGDS. Factor analysis was and the Anatomical and Therapeutical Classification, repeated on patients matched for cognitive performance respectively.8,12 A detailed description of the GIFA proto- (AMT score less than 7, n = 435) or physical impairment col is available elsewhere.7,8 (dependency in at least 1 ADL, n = 780). Sensitivity,speci- Affective status was assessed using the sfGDS. Five ficity,and diagnostic accuracy of the sfGDS versus a diag- items explore a positive attitude toward life (happy most nosis of major depression made according to the Diagnostic of the time, satisfied with life, wonderful to be alive, full and Statistical Manual of Mental Disorders (3rd ed., 15 of energy, in good spirit) and thus can be defined as posi- revised) criteria (n = 128) were computed. tive items. The 10 negative items generically assess dis- satisfaction with life (life is empty, feel pretty worthless, RESULTS feel helpless, often get bored, worry about the future, situation is hopeless, others are better off) or point at selected Table 1 shows the general characteristics of patients. personal problems (problems with memory,prefer to stay Most of them had been admitted to wards of geriatrics. home, dropped activities/interests). Every item contributes Their mean age was 76 years, and 37.2% were aged 80 or 1 point to the final score, which ranges from 0 to 15. A score older.The overall educational level was low: only 21.6% had greater than 5 is commonly considered as indicative of a an 8-year formal education or more. Cerebrovascular, car- depressive trait.1 diovascular, respiratory, and metabolic diseases were the The present analysis refers to patients older than 60 most common diagnoses. Dependency,as reflected by need years of age consecutively admitted to the participating of assistance in at least 1 ADL, affected 32.1% of patients. centers. By protocol, a study physician administered the Impairment in 1 or more IADL was recorded in 69.7% of sfGDF on the day before the planned discharge because patients. Twenty percent of patients scored greater than a great proportion of patients were too seriously ill to be 7 on the AMT; this figure rose to 28.9% in patients older interviewed on admission. Patients who died during hos- than 80 years. pital stay (n = 138) were excluded from the study.A total Table 2 shows results from the main component analy- of 363 patients were excluded because of 1 or more of the sis on the whole population. The Kaiser-Meyer-Olkin following reasons: aphasia, illiteracy, deafness, uncon- measure of sampling adequacy was 0.865; this value is consciousness, unwillingness to cooperate, or severe demen- sistent with the correlations between individual pairs of tia. Due to these exclusion criteria, the analysis was variables being explained by the other variables. The conducted on 2032 patients. value of the Bartlett’s test of sphericity excludes that the correlation matrix for the 14 items is an identity matrix. Statistical Analysis Results from these 2 tests show that the main component Statistical analyses were performed using SPSS software. analysis is appropriate to data. A 3-factor model was Two complementary methods, the Kaiser-Meyer-Olkin developed and could explain 47.7% of variance in GDS per- Measure of Sampling Adequacy and Bartlett’s test of formance. The ratio between variances explained by the sphericity, were used to evaluate the appropriateness of first and the second factor was 2.59, far from the cutoff of factor analysis.13 The aim of this preliminary phase was 3.5, which is considered to characterize a scale as unidi- 16 to evaluate whether items are reciprocally related so that mensional. Factor 1 had large loadings on 4 out of the 5 a factor model can be developed. Weakly correlated vari- positive items, factor 2 on 5 negative items, and factor 3 ables cannot share common factors, thus making factor on 1 positive and 4 negative items. Distressing analysis inappropriate. Factors were extracted by the thoughts and negative judgment about the own condition principal component analysis. From the basic component were the main areas related to factor 2, whereas inactiv- matrix, a rotated component matrix was obtained through ity and reduced self-esteem were related to factor 3. The a varimax with Kaiser normalization. The rotation phase item “worry about the future” remained unrelated to any allows the identification of factors summarizing sets of factor. Construct Validity of the 15-Item Geriatric Depression Scale in Older Medical Inpatients / Incalzi et al. 25

Table 1. General Characteristics of the Patients Table 2. Construct Validity of the Geriatric Depression Scale–Short Form in Older Medical Inpatients n % M ± SE Component Demographics and functional status Age 76.3 ± 0.18 123 Gender (female) 976 48 Education (years) 5.26 ± 0.09 Are you basically satisfied with your life? 0.686 –0.286 –0.11 Living alone 348 17.3 Have you dropped many of your Abbreviated Mental Test 8.07 ± 0.04 activities and interests? –0.223 0.069 0.61 Geriatric Depression Scale–short form 5.43 ± 0.08 Do you feel that your life is empty? –0.434 0.578 0.13 Length of stay (days) 16.9 ± 3.7 Do you often get bored? –0.355 0.489 0.239 Activities of daily living (impaired in 1 Are you in good spirits most of the time? 0.729 –0.147 –0.138 or more) 780 38.3 Are you afraid that something bad is Instrumental activities of daily living going to happen to you? –0.014 0.358 0.341 (impaired in 1 or more) 1485 73.1 Do you feel happy most of the time? 0.775 –0.087 –0.199 Comorbid conditions Do you often feel helpless? –0.25 0.724 0.113 Charlson’s index 1.8 ± 0.04 Do you prefer to stay at home rather Chronic obstructive pulmonary disease 350 17.2 than going out and doing new things? –0.02 0.197 0.58 Cerebrovascular disease 351 17.3 Do you feel you have more problems Coronary artery disease 595 29.3 with memory than most? 0.153 0.509 0.199 Congestive heart failure 568 28 Do you think it is wonderful to be alive Dementia 86 4.2 now? 0.731 –0.016 –0.051 Depression 113 5.6 Do you feel pretty worthless the way Diabetes 419 20.6 you are now? –0.266 0.282 0.417 Hypertension 862 42.4 Do you feel full of energy? 0.443 0.309 –0.589 Acute myocardial infarction 142 7 Do you feel that your situation is Renal disease 162 8 hopeless? –0.222 0.72 0.054 Cancer 163 8 Do you think that most people are better Stroke 130 6.4 off than you are? 0.087 0.365 0.499 Medication Antidepressant agents 191 9.4 Variance explained by individual Digoxin 486 23.9 components (%) 28.903 11.154 7.681 Angiotensin-converting enzyme Total variance explained inhibitors 199 9.8 (cumulative variance, %) 47.738 Diuretics 499 24.6 Calcium channel blockers 364 17.9 Loadings of identified components on individual items and explained variance are β-blockers 54 2.7 reported. Corticosteroids 202 9.9 Admission ward Geriatrics 1549 76.2 α Internal medicine 483 23.8 The Cronbach’s for the general model was .46, which indicates a low level of homogeneity among items of the sfGDS. Values for partial models ranged between .40 and Comparable results were obtained in the 2 strata .48. Sensitivity and specificity of sfGDS > 5 versus a first- with AMT < 7 or AMT > 6. The cumulative variance explained listed or secondary diagnosis of major depression were in these samples was 47% and 47.4%, and the best model 78.9% and 61.7%, respectively. was a 3-factor one with loadings of individual items fairly comparable to those observed in the general model. DISCUSSION In subjects dependent in at least 1 ADL, a more complex model was developed and is summarized in Table 3. Our findings show that in broad hospitalized geriatric It includes 5 main components explaining 28.9%, 11.7%, populations, the sfGDS explores 3 different psychological 7.9%, 7.3%, and 6.7% of the variance, respectively. Thus, areas, has good sensitivity but suboptimal specificity ver- the cumulative variance explained was 62.5%, but the sus a diagnosis of depression, can explain less than 50% relationship between factors and items was quite complex: of total variance, and lacks unidimensionality.The latter while positive and most negative items had loadings with 2 findings limit the interpretation of the sfGDS because factors 1 and 2, respectively, factor 3 was related to loss the final score may reflect in variable proportions factors of usual activity (prefer to stay home, dropped activi- unrelated to the state of mood. After stratification by ties/interests); factors 3 and 4 were weakly related to presence of physical impairment, the sfGDS could explain items assessing self-esteem or memory and overall con- a larger proportion of variance, probably because of the ditions in comparison with those of the general population. strong relationship between physical and affective status Table 4 shows the model for subjects independent in in the elderly.17 Indeed, physical disability was found to all ADLs. It included four factors and could explain 53.6% predict the onset of depression in subjects older than 64 of the variance. The ratio of variance explained by the first years of age.18 factor to that explained by the second factor was 2.6. The None of the previous studies assessed the construct loadings of factors on the items did not reflect an easily validity of sfGDS on an unselected hospitalized geriatric interpretable pattern of psychological dimensions. population. Furthermore, the size of previously studied 26 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003

Table 3. Construct Validity of the Geriatric Depression Table 4. Construct Validity of the Geriatric Depression Scale–Short Form in Older Medical Inpatients Scale–Short Form in Older Medical Inpatients Independent in All Activities of Daily Living Dependent in at Least One Activity of Daily Living

Component Component

12 3 4 12 3 4

Are you basically satisfied with Are you basically satisfied with your life? 0.648 –0.325 –0.097 –0.036 your life? 0.745 –0.046 –0.129 –0.021 Have you dropped many of Have you dropped many of your activities and interests? –0.17 –0.075 0.634 0.187 your activities and interests? –0.209 0.098 0.031 0.666 Do you feel that your life Do you feel that your life is is empty? –0.315 0.574 0.337 –0.026 empty? –0.538 0.257 0.457 0.172 Do you often get bored? –0.287 0.421 0.447 0.035 Do you often get bored? –0.411 0.261 0.417 0.248 Are you in good spirits most Are you in good spirits most of the time? 0.725 –0.146 –0.144 –0.105 of the time? 0.76 –0.047 –0.031 –0.128 Are you afraid that something Are you afraid that something bad is going to happen to bad is going to happen to you? 0.094 0.237 0.608 –0.059 you? –0.016 0.661 0.013 –0.028 Do you feel happy most of the Do you feel happy most of the time? 0.781 –0.094 –0.203 –0.102 time? 0.768 –0.108 0.123 –0.156 Do you often feel helpless? –0.144 0.71 0.174 0.178 Do you often feel helpless? –0.363 0.577 0.369 0.03 Do you prefer to stay at home Do you prefer to stay at home rather than going out and rather than going out and doing new things? –0.157 0.054 –0.001 0.775 doing new things? –0.047 0.067 0.172 0.781 Do you feel you have more Do you feel you have more problems with memory problems with memory than most? 0.03 0.413 –0.033 0.475 than most? 0.117 0.099 0.603 0.243 Do you think it is wonderful to Do you think it is wonderful to be alive now? 0.758 –0.063 0.02 0.057 be alive now? 0.69 –0.088 0.206 –0.053 Do you feel pretty worthless Do you feel pretty worthless the way you are now? –0.18 0.259 0.582 –0.047 the way you are now? –0.183 0.629 –0.239 0.204 Do you feel full of energy? 0.426 0.391 –0.488 –0.198 Do you feel full of energy? 0.389 –0.134 0.54 –0.406 Do you feel that your situation Do you feel that your situation is hopeless? –0.167 0.692 0.05 0.196 is hopeless? –0.304 0.575 0.419 –0.057 Do you think that most people Do you think that most people are better off than you are? 0.022 0.124 0.288 0.627 are better off than you are? 0.063 0.535 0.177 0.198

Variance explained by Variance explained by individual components (%) 27.325 10.758 8.215 6.939 individual components (%) 27.890 12.077 7.654 6.740 Total variance explained Total variance explained (Cumulative variance, %) 53.236 (cumulative variance, %) 54.361

Loadings of identified components on individual items and explained variance are Loadings of identified components on individual items and explained variance are reported. reported. samples ranged between 72 and 285 subjects.1-6 The pres- likely limits the possibility of a reliable screening of depres- ent findings add to the knowledge on sfGDS by present- sion.21 Although we studied our patients immediately ing results generalizable to the older medical inpatients. before discharge, it is unlikely that this limitation was com- The main limit of sfGDS lies in its doubtful meaning: its pletely overcome. We ignore whether the same limitation internal structure does not guarantee that the final score affects the other scales commonly used for screening depres- of sfGDS measures what it is assumed to measure, that sion in older medical inpatients because their construct is, the state of mood. Examples of construct validity for the validity has been assessed in samples that are small and/or Mini-Mental State Examination (MMSE), the test most poorly representative of this broad population.22-26 commonly used for screening dementia, help to clarify Even if its overall reliability was poor, the sfGDS had this issue: explaining a cumulative variance of 68% and good sensitivity versus a diagnosis of depression and a log- being unidimensional, MMSE really expresses what it ical structure of relationships between individual items and measures in community-dwelling older adults.19 The 56.1% factors. The latter finding suggests that sfGDS can explore explained variance and the clustering of individual items selected dimensions of mood such as self-esteem, dis- into well-defined dimensions of neurocognitive ability tressing thoughts, positive attitude toward life, and judg- indicate that MMSE can effectively screen for cognitive ment about own condition. Thus, an analytical interpretation impairment in older nursing home residents, but the final of sfGDS performance might provide potentially important score should be interpreted with some caution because of information. It is likely that impairment of some of the the lack of unidimensionality.20 explored psychological dimensions has specific practical The complex psychological status of older patients implications. The demonstration of a well-defined rela- experiencing both an acute illness and the hospital stay tionship between memory complaint and incident demen- Construct Validity of the 15-Item Geriatric Depression Scale in Older Medical Inpatients / Incalzi et al. 27 tia in the general geriatric population and between sui- per se from those of disability on the internal cidal ideas and nonsuicidal death in medical older inpa- structure of sfGDS. tients are examples of such a possibility.27,28 • The data are taken from Italian hospitals, and the Multidimensional assessment is the cornerstone of the conclusions cannot be generalized to other settings. current approach to the elderly patient.29 Assessing social, • Men outnumbered women in our population. medical, neuropsychological, and functional status allows SfGDS might have a different internal structure in for complete information on the health status of the patient the presence of a female to male ratio similar to and the planning of interventions aimed at improving that that of the older general population. status or preventing further worsening. Multidimensional assessment is based on diagnostic instruments that should References require little time and be easy to use, reproducible, sensi- 1. Sheikh VI, Yesavage VA. Geriatric Depression Scale (GDS): ble to small changes, and related to major outcomes such recent evidence and development of a shorter version. In: Brink TL, ed. Clinical gerontology: a guide to assessment and inter- as disability and mortality.Several instruments exploring vention. New York: Haworth Press, 1986:165-174. functional and mental status meet these requirements, 2. Alden D, Austin C, Sturgeon R. A correlation between the Geri- while the same is not true of tests screening older medical atric Depression Scale long and short form. J Gerontol 1989; inpatients for depression. The prominent role and the poor 44:124-125. specificity of somatic symptoms in geriatric depression 3. Burke WJ, Roccaforte WH, Wengel SP.The short form of the Geri- 30 atric Depression Scale: a comparison with the 30-item form. J could partly explain this finding. Furthermore, visual Geriatr Psychiatry Neurol 1991; 4:173-178. problems, illiteracy, and, mainly, lack of motivation fre- 4. Fountoulakis KN, Tsolaki M, Iacovides A, et al. The validation quently prevent older medical patients from completing of the short form of the Geriatric Depression Scale (GDS) in the self-rated depression scale.31 Theoretically, lack of Greece. Aging 1999; 11:367-372. motivation could also affect the quality of self-rating. The 5. Cwikel J, Ritchie K. Screening for depression among the elderly in Israel: an assessment of the Short Geriatric Depression Scale important association between depression and cognitive (S-GDS). Isr J Med 1989; 25:131-137. impairment in older inpatients might further affect the 6. Lesher EL, Berryhill JS. Validation of the Geriatric Depression 32 quality of self-rating. Accordingly, the observed limita- Scale–Short Form among inpatients. J Clin Psychol 1994; tions of sfGDS might be common to other instruments 50:256-260. based on self-rating. A comparison between self-rated and 7. Carbonin PU, Pahor M, Bernabei R, et al. Is age an independ- observer-rated depressive status in older medical patients ent risk factor of adverse drug reactions in hospitalized medical patients? J Am Geriatr Soc 1991; 39:1093-1099. could help clarify this issue. 8. Pahor M, Chrischilles EA, Guralnik JM, et al. Drug data cod- In conclusion, analysis of construct validity of sfGDS ing and analysis in epidemiologic studies. Eur J Epidemiol disclosed a well-defined and logical set of component fac- 1994; 10:405-411. tors but failed to demonstrate a structure that could guar- 9. Katz S, Ford AB, Moskowitz RW, et al. Studies of illness in the antee reliability of results. It is likely that selecting out aged. The index of ADL: a standardized measure of biological and psychosocial function. JAMA 1963; 185:84-99. the 15 items from the parent GDS according to findings 10. Charlson ME, Pompei P, Ales KL, et al. A new method of clas- obtained from a home-dwelling older population resulted sifying prognostic comorbidity in longitudinal studies: devel- in the development of a scale only partially fitting the needs opment and validation. J Chronic Dis 1987; 40:373-383. 1 of older medical inpatients. Thus, future efforts should be 11. Jitapunkul S, Pillay I, Ebrahim S. The abbreviated mental test: finalized to develop an instrument specifically designed its use and validity. Age Ageing 1991; 20:332-336. for this very large population to assess both prevalence and 12. PHS-HCF. International classification of diseases, 9th revision. Clinical modifications. Washington, DC: Public Health Service– implications of depression as well as its responsiveness to Health Care Financing Administration, 1980. therapeutic interventions. 13. Gruijter DNM, Van der Kamp LJT. Statistical models in psychological and educational testing. Lisse, the Netherlands: CLINICAL IMPLICATIONS Swets & Zeitlinger, 1984. 14. Zwick W,Velicer W.Comparison of five rules for determining the • SfGDS cannot be considered an optimal instru- number of components to retain. Psychol Bull 1986; 99:432- 442. ment for screening depression in unselected older 15. American Psychiatric Association. Diagnostic and statistical medical inpatients. manual of mental disorders. 4th ed. Washington, DC: American • Cognitive performance, as assessed by the AMT, Psychiatric Association, 1994. does not affect the internal structure of sfGDS. 16. Teresi J, Golden R. Latent structure methods for estimating item • SfGDS performs better in patients matched for bias, item validity and prevalence using cognitive and other geriatric screening measures. Alzheimer Dis Assoc Disord 1994; level of physical capabilities. 8:S291-S298. 17. Phifer JF, Murrell SA. Etiologic factors in the onset of depres- LIMITATIONS sive symptoms in older adults. J Abnorm Clin Psychol 1986; 95:282-291. • The lack of a symptom-rating instrument prevent- 18. Turner RI, Noh S. Physical disability and depression: a longi- ed us from distinguishing the effects of symptoms tudinal analysis. J of Health and Social Behavior 1988; 29:23- 37. 28 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003

19. Jones RN, Gallo JJ. Dimensions of the Mini-Mental State Exam- Depression Rating Scale in depressed geriatric inpatients. Am ination among community dwelling older adults. Psychol Med J Geriatr Psychiatr 2001; 9:35-40. 2000; 30:605-618. 27. Jonker C, Geerling MI, Schmand B. Are memory complaints pre- 20. Abraham IL, Manning CA, Snustad DG, et al. Cognitive screen- dictive for dementia? A review of clinical and population-based ing of nursing home residents: factor structures of the Mini- studies. Int J Geriatr Psychiatry 2000; 15:983-991. Mental State Examination. J Am Geriatr Soc 1994; 42:750-756. 28. Shah A, Hoxey K, Mayadunne V. Some predictors of mortality 21. Antonelli Incalzi R, Gemma A, Caparella O, et al. The effects of in acutely medically ill elderly inpatients. Int J Geriatr Psychiatry the hospitalization on affective status of elderly patients. Inter- 2000; 15:493-499. national Psychogeriatrics 1991; 3:67-74. 29. Stuck AE, Siu AL, Wieland GD, et al. Comprehensive geriatric 22. Kitchell MA, Barnes RF, Veith RC, et al. Screening for depres- assessment: a meta-analysis of controlled trials. Lancet 1993; sion in hospitalized geriatric medical patients. J Am Geriatr Soc 342:1032-1036. 1982; 30:174-77. 30. Brink TL, Yesavage JA, Lum O, et al. Screening tests for geri- 23. Rapp SR, Parisi SA, Walsh DA. Psychological dysfunction and atric depression. Clin Gerontol 1982; 1:37-43. physical health among elderly medical inpatients. J Consult Clin 31. Toner J, Gurland B, Teresi J. Comparison of self-administered Psychol 1980; 56:509-513. and rater-administered methods of assessing levels of severity 24. Koenig HG, Meador KG, Cohen HJ, et al. Screening for depres- of depression in eldelry patients. J Gerontol: Psychological Sci- sion in hospitalized elderly medical patients: taking a closer look. ences 1988; 43:136-140. J Am Geriatr Soc 1982; 40:1013-1017. 32. Geerlings MI, Schoevers RA, Beekman AT,et al. Depression and 25. Hammond MF. Rating depression severity in the elderly phys- risk of cognitive decline and Alzheimer’s disease. Results of two ically ill patient: reliability and factor structure of the Hamil- prospective community-based studies in the Netherlands. Br J ton and the Montgomery-Asberg depression rating scales. Int Psychiatry 2000; 176:568-575. J Geriatr Psychiatry 1998; 13:257-261. 26. Moberg PJ, Lazarus LW,Mesholam RI, et al. Comparison of the standard and structured interview guide for the Hamilton Treatment of Delirium in Older Adults With Quetiapine

Kye Y. Kim, MD, Geoffrey M. Bader, MD, Victor Kotlyar, MD, and Debra Gropper, PharmD

ABSTRACT

Delirium is a neuropsychiatric syndrome characterized by impairment of consciousness, changes in cognition, or perceptual disturbances. In addition, delirium is often accompanied by delusions, hallucinations, and agitation. In this study, 12 older patients with delirium were treated for neuropsychiatric symptoms with quetiapine. The mean duration for stabilization was 5.91 ± 2.22 days, and the mean dose was 93.75 ± 23.31 mg/day. None of the 12 patients developed extrapyramidal symptoms. There were significant improvements on all measures used in this study.Interestingly, the Delirium Rating Scale scores along with scores of the Mini-Mental State Examination and Clock Drawing Test continued to improve throughout the 3-month study period. In our study,we found that quetiapine was a safe and effective treatment in hospitalized older patients with delirium. (J Geriatr Psychiatry Neurol 2003; 16:29–31)

Keywords: delirium; older adults; quetiapine; behavioral symptoms

Delirium is a syndrome of disturbed consciousness, cog- still had 1 or more symptoms as long as 6 months after hos- nition, and perception that develops over a short period of pital discharge. Older adults with delirium can present with time and tends to fluctuate during the course of the day, a wide variety of neuropsychiatric symptoms.5 The prin- and it is caused by 1 or more physical conditions.1 Although cipal treatment of delirium is still the diagnosis and treat- delirium is significantly common among people of all ages, ment of the underlying physical conditions contributing to older adults are more prone to develop this syndrome delirium. However, psychopharmacological intervention is because they are at higher risk for underlying brain dis- a major component of all interventions for delirium. Typ- ease.2 The prevalence of delirium has been estimated from ically,high-potency neuroleptic agents such as haloperidol 10% to 40% among hospitalized elderly patients on med- have been used as first-line treatment for neuropsychiatric ical and surgical wards.3 Also, hospital mortality esti- symptoms of patients with delirium.6 However, they are fre- mates range from 10% to 65%.4 quently associated with extrapyramidal symptoms (EPS), Delirium is traditionally considered a transient syn- particularly in older adults. Recently, there have been drome that ends in recovery after several days to weeks several reports of use of the atypical antipsychotic agents, in most cases.3 This notion tends to lead most clinicians’ risperidone and olanzapine.7-10 psychopharmacological interventions to a brief one as Quetiapine is an atypical neuroleptic agent and diben- long as the underlying causes of delirium are identified and zothiazepine derivative structurally related to clozapine resolved. Interestingly, Levkoff and her colleagues2 in and olanzapine.11 Quetiapine is well tolerated and asso- their prospective study demonstrated that many patients ciated with improvement in psychotic symptoms, although it demonstrates lack of EPS and minimal seda- tive, hypotensive, and anticholinergic side effects in the dose range used in older adults.12,13 This study was designed to Received March 25, 2002. Received revised April 29, 2002. Accepted for publication April 30, 2002. determine the efficacy and safety of open-label quetiapine treatment in patients with delirium over a period of 3 From Salem Veterans Affairs Medical Center, Department of Psychiatric Medicine, School of Medicine, University of Virginia (Drs. Kim and Bader), months. Institute of Living, Hartford, CT (Dr. Kotlyar), and Perry Point Veterans Affairs Medical Center, Perry Point, MD (Dr. Gropper). METHODS This study was conducted at Veterans Affairs Medical Center, Salem, Virginia, and supported by a grant from AstraZeneca. All patients were enrolled from the acute medical units of Reprint requests: Kye Y. Kim, MD, Psychiatry Service, Veterans Affairs Salem Veterans Affairs Medical Center, a 271-bed, general Medical Center, Salem, VA 24153; e-mail: [email protected]. and teaching hospital for the University of Virginia, School DOI: 10.1177/0891988702250533 of Medicine. The hospital’s Human Studies Subcommittee

Table 1. Clinical Characteristics of Patients

Time to Patient Age Stabilization Adverse

Number (years) Gender Medical Diagnosis Dose (mg) at T1 (days) Reactions

1 80 Male DM, CRIa, hypertension, hypothyroidism, dementia 50 mg bid 6 None 2 64 Male UTI, delirium tremensa 75 mg bid 8 None 3 68 Male S/P laryngectomy, hypertension, hypothyroidism, UTIa 50 mg bid 5 None 4 70 Male DM, S/P hip hemiarthroplastya, BPH 50 mg bid 10 None 5 82 Male CRIa, CHF, hypertension, DM, BPH 25 mg qam 75 mg ghs 4 None 6 79 Male GI bleedinga, DM, chronic anemia, dementia 50 mg ghs 6 None 7 75 Male CHFa, hypertension, BPH, sleep apnea 50 mg bid 8 Vivid dreams, drowsiness 8 67 Male S/P coronary bypassa, hypertension, DM, chronic anemia 50 mg bid 8 None 9 71 Male MIa, cardiomyopathy, hypertension, DM 25 mg bid 4 None 10 77 Male UTI, dehydrationa 25 mg bid 3 None 11 74 Male Pneumoniaa, PE, duodenal ulcer, esophageal stricture 25 mg qam, 50 mg ghs 4 None 12 88 Male GI bleedinga, glaucoma, BPH 50 mg qam, 100 mg ghs 5 Drowsiness

DM = diabetes mellitus; CRI = chronic renal insufficiency; UTI = urinary tract infection; BPH = benign prostatic hypertrophy; GI = gastrointestinal; MI = myocardial infarction; PE = pulmonary embolism. a. Main cause of delirium. and Research and Development Committee approved same measurements. The efficacy of quetiapine was eval- the study protocol. Informed consent was required for uated using responder analyses and paired t tests. Side participation in this study. All patients were referred to effects were assessed with clinically oriented, open-ended Consultation-Liaison Psychiatry Service for evaluation of questions. mental status changes. Patients who had known histories of psychotic disorders and were treated with neuroleptic RESULTS agents within the previous 4 weeks prior to the enrollment were excluded from this study. They were required to be Eleven of the 12 patients completed the study to T3. One at least 60 years old. Patients provided written informed patient was started on quetiapine and followed at T1, but consent before admission to the study.When they did not he died of acute myocardial infarction a few days later.This have a capacity to consent due to significant changes in patient had preexisting cardiac conditions, and his death mental status, the consent was sought from their next of was unrelated to the study drug. All 12 patients were con- kin. All patients enrolled in this study met Diagnostic sidered evaluable (Table 1). There were no dropouts due and Statistical Manual of Mental Disorders (4th ed.) cri- to side effects. The age (mean ± SD) of patients was 74 ± teria for delirium.1 They were administered the Mini- 7 years, and all of them were male. None of the 12 patients Mental State Examination (MMSE),14 Delirium Rating had prior history of psychiatric treatment. Two patients Scale (DRS),15 Clock Drawing Test (CDT),16 and Clinical had a diagnosis of early dementia, but they had no behav- Global Impression (CGI).17 At their admission to the study ioral symptoms prior to the development of delirium. Only

(T0), they were administered the MMSE, the DRS, and the 1 patient required a one-time dose of lorazepam (1 mg) CGI. When an attending psychiatrist determined that administered intramuscularly. The mean ± SD dose of patients were in need of antipsychotic treatment based on quetiapine at T1 was 93.75 ± 23.31 mg/day. The mean clinical grounds, they were given a starting dose of queti- duration for stabilization at T1 was 5.91 ± 2.22 days. None apine 25 mg twice a day.The dosages were then increased of the 12 patients developed EPS, and rates of other side by 25 mg every 2 days until patients were maximally sta- effects were considered minimal (sedation in 2 patients and bilized (T1). When patients required an adjunctive psy- vivid dreams in 1). Average scores on the efficacy meas- chotropic therapy for acute symptoms, they were given ures at T0 through T3 are reported in Table 2. All t tests oxazepam or lorazepam by mouth or by intramuscular were highly significant, indicating that there were sig- injection as needed. At stabilization, patients were again nificant improvements on all measures used in this study. given the above-mentioned measures and discharged on Interestingly, the DRS scores along with scores of MMSE the stabilizing dose of quetiapine. All patients had follow- and CDT continued to improve during the study period from up visits at the first month of therapy (T2) and third month T0 to T3. Although a monitoring of the patient’s blood pres- of therapy (T3). At the first-month visit, the quetiapine was sure was not one of the measurements in this study,there tapered off by 25 mg every 3 days when they were con- was no indication of any significant changes in blood pres- sidered stable. At each follow-up visit, they were given the sure secondary to quetiapine use. Treatment of Delirium in Older Adults With Quetiapine / Kim et al. 31

Table 2. Paired t Tests Comparing Scores on the MMSE, DRS, implication for the duration of intervention, especially CDT, and CGI Across Time Periods pharmacotherapeutic treatment. In our study, we found that quetiapine was a safe and effective treatment in hos- Measure T T T T 0 1 2 3 pitalized older patients with delirium. However, it will be MMSE difficult to absolutely determine whether the neuropsy- Mean 14.50 21.17 24.27 25.18 chiatric symptoms improved because of quetiapine, reso- SD 5.90 4.55 4.86 4.45 lution of the underlying medical conditions, or a N12121111 ta 4.07 combination of the two. Larger controlled studies are tb 3.96 needed to further explore these preliminary findings and c t 1.99 conclusions. P .002 .003 ns DRS Mean 18.25 8.00 2.27 0.63 References SD 6.05 2.34 2.28 1.21 1. American Psychiatric Association. Diagnostic and statistical n12121111 a manual of mental disorders. 4th ed. Washington, DC: American t 5.60 Psychiatric Association, 1994. tb 7.88 tc 2.57 2. Levkoff SE, Evans DA, Lipzin B, et al. Delirium: the occurrence P .0002 .0001 .03 and persistence of symptoms among elderly hospitalized patients. CDT Arch Inter Med 1992; 152:334-340. Mean 3.25 6.25 7.91 8.09 3. Lipowski ZJ. Delirium (acute confusional state). JAMA 1987; SD 2.77 4.07 3.45 3.21 258:1789-1792. n12121111 a 4. Inouye SK. Delirium in hospitalized elderly patients: recogni- t 3.35 tion, evaluation, and management. CT Med 1993; 57:309-315. tb 3.46 tc 1.0 0 5. Sandberg O, Gustafson Y, Braennstroem B, Bucht G. Clinical pro- P .007 .006 ns file of delirium in older patients. JAGS 1999; 47:1300-1306. CGI-S 6. Shapiro BA, Warren J, Egol AB, et al. Practice parameters for Mean 3.00 1.91 1.36 intravenous analgesia and sedation for adult patients in the SD 0.43 0.83 0.67 intensive care unit: an executive summary. Crit Care Med 1995; n12111123:1596-1600. tb –5.16 7. Sipahimalani A, Masand PS. Use of risperidone in delirium: case tc –2.63 reports. Ann Clin Psychiatry 1997; 9:105-107. P .0004 .03 8. Ravona-Springer R, Dolberg OT, Hirschmann S, Grunhaus L. MMSE = Mini-Mental State Examination; DRS = Delirium Rating Scale; CDT = Clock Delirium in elderly patients treated with risperidone: a report Drawing Test; CGI-S = Clinical Global Impressions–Severity. of three cases. J Clin Psychopharmacol 1998; 18:171-172. 9. Sipahimalani A, Sime RM, Masand PS. Treatment of delirium a. Paired t test between T0 and T1. b. Paired t test between T and T with risperidone. Int J Geriatric Psychopharmacology 1997; 1 2. 1:24-26. c. Paired t test between T and T 2 3. 10. Sipahimalani A, Masand PS. Olanzapine in the treatment of delirium. Psychosomatics 1998; 39:422-430. DISCUSSION 11. Goren JL, Levin GM. Quetiapine, an atypical antipsychotic. Pharmacotherapy 1998; 18:1183-1194. 12. McManus DQ, Arvanitis LA, Kowalcyk BB. Quetiapine, a novel Since this was a small, open-label study with no control antipsychotic: experience in elderly patients with psychotic dis- group, our findings are certainly limited for generalization. orders. J Clin Psychiatry 1999; 60:292-298. This study was also limited in that all subjects were male. 13. Madhusoodanam S, Brenner R, Alcantra A. Clinical experience All patients had multiple physical disorders. They were also with quetiapine in elderly patients with psychotic disorders. J Geriatr Psychiatry Neurol 2000; 13:28-32. taking various drugs for their medical conditions. However, 14. Folstein MF, Folstein SE, McHugh PR. “Mini-Mental State”: a all patients were free of previous psychiatric diagnosis and practical method for grading the cognitive state of patients for were not on any type of psychotropic agents for at least 4 the clinician. J Psychiatr Res 1975; 12:189-198. weeks prior to admission to this study; thus, it was an 15. Trzepacz PT, Baker RW, Greenhouse J. A rating scale for delir- appropriate group of patients for a psychotropic drug ium. Psychiatry Res 1988; 23:89-97. study. 16. Wolf-Klein GP, Silverstone FA, Levy AP, et al. Screening for Quetiapine was well tolerated by all patients in this Alzheimer’s disease by clock drawing. J Am Geriatr Soc 1989; 37:730-734. study. Vivid dreams and sedation were reported, but no 17. Guy W. ECDEU assessment manual for psychopharmacology–revised patients developed EPS. Peak response times were com- (DHEW Publ No ADM 76-338). Rockville, MD: U.S. Department parable to what has been reported for olanzapine and of Health, Education, and Welfare, Public Health Service, Alco- haloperidol.10 In addition, it was of clinical interest that hol, Drug Abuse, and Mental Health Administration, NIMH Psychopharmacolgy Research Branch, Division of Extramural the main measure in this study, the DRS, continued to Research Program, 1976:218-222. improve across time periods. This finding might have some Recorded Delirium in a National Sample of Elderly Inpatients: Potential Implications for Recognition

Helen C. Kales, MD, Barbara A. Kamholz, MD, Stephanie G. Visnic, BA, and Frederic C. Blow, PhD

ABSTRACT

This retrospective study examined delirium and related confusional diagnoses recorded in patients older than age 60 discharged from Veterans Affairs (VA) acute inpatient units nationally in 1996 (n = 267,947). Only 4% of patients had delirium or related confusional diagnoses recorded. Patients with recorded delirium had significantly higher mortality than did those without recorded delirium or those with other confusional diagnoses (“organic psychoses”); the most common delirium types were dementia with delirium and alcohol intoxication/withdrawal delirium. Organic psychoses patients had the longest lengths of stay and significantly more admissions to nonmedical/surgical units and discharges to nursing homes; almost 20% were African American. The recorded rate of delirium in the VA health system likely underestimates true prevalence and possibly reflects nonrecognition of delirium in many older veterans. Certain motoric and etiologic types of delirium may be more commonly diagnosed and recorded. Future research should prospectively examine recognition of motoric and etiologic delirium subtypes and racial differences in delirium diagnoses. (J Geriatr Psychiatry Neurol 2003; 16:32–38)

Keywords: delirium; elderly; veterans

Delirium has been estimated to occur in 14% to 56% of hos- ence of delirium. Other studies have consistently noted a pitalized elderly patients.1 Prior studies have shown that different problem associated with delirium in clinical set- delirium may be associated with prolonged hospital lengths tings: that of its underrecognition and misidentification. of stay,increased mortality,functional and cognitive decline, It has been estimated that clinicians fail to recognize and increased rates of nursing home placement.2-10 Inouye delirium one third to two thirds of the time.12-14 Various rea- et al have noted that delirium may serve as a “window” to sons have been suggested for the underrecognition of delir- the hospital care of elderly patients as it is frequently ium: the use of different diagnostic terms; varying clinical iatrogenic, closely linked to the processes of care, and, as presentations of delirium due to fluctuation or hypoactive such, serves as a useful outcome measure.11 The above- forms of the syndrome; the attribution of cognitive changes mentioned rates of delirium and negative associated out- to dementia, depression, or senescence; the lack of expe- comes generally have been found in prospective studies in rience of health practitioners in testing for attentional which investigators were actively assessing for the pres- disturbances; the absence of baseline cognitive assessments; and the failure to appreciate the significance of delirium as a marker for severe illness and mortality.11,15 Received March 23, 2001. Received revised April 1, 2002. Accepted for Studies involving record review at single institutions publication May 13, 2002. have found low rates of provider recording of delirium. Also, From the Serious Mental Illness Research and Evaluation Center although “delirium” is the gold standard term in clinical (SMITREC), Health Services Research and Development, Ann Arbor VA and research use, providers may still use less precise, Medical Center, Ann Arbor, Michigan (Drs. Kales, Visnic, and Blow); Psychiatry Service, Ann Arbor VA Medical Center, Ann Arbor, Michigan vague descriptions of symptoms such as “confusion” or (Drs. Kales and Kamholz); the Division of Geriatric Psychiatry,University “agitation” or alternative diagnostic terms to describe of Michigan, Ann Arbor, Michigan (Drs. Kales and Blow). delirium such as “acute brain syndrome,” “toxic psychosis,” Presented in part at the annual meeting of the American Association for and “metabolic encephalopathy.”16 A study comparing the Geriatric Psychiatry, Miami Beach, Florida, March 2000. prospective diagnosis of delirium by a psychiatrist with evi- Reprint requests: Dr. Helen Kales, Psychiatry Service (116A), Ann Arbor dence for delirium in the medical record found that only VA Medical Center, 2215 Fuller Road, Ann Arbor, MI 48105; e-mail: 17% of the delirious patients could be recognized as deliri- [email protected]. ous by record review (International Classification of Dis- DOI: 10.1177/0891988702250535

32 © 2003 Sage Publications Recorded Delirium in a National Sample of Elderly Inpatients / Kales et al. 33 eases, 9th revision [ICD-9] codes, admission notes, and examination of the rates and characteristics of recorded progress notes).17 Identifying cases through chart review delirium and related diagnoses in the VA system is highly at the Ann Arbor Veterans Medical Center, Kamholz et al18 relevant to the study of delirium; further data regarding found delirium in approximately 20% of hospitalized eld- the demographics of recognized delirium in a large health erly patients using the Confusion Assessment Method.19 care system may give clues as to the types of delirium that The records of these patients with delirium were then may go undetected as well as suggesting the possible searched for any provider notations indicative of the recog- extent of underrecognition. nition of delirium. Only 45% of the patients with delirium Given our awareness of past prospective prevalence on medical units had any such notations made by physi- estimates, as well as prior studies demonstrating poor cians in the chart. Far fewer (18%) surgical patients with physician recognition and recording of delirium in the delirium had any such notations recorded. medical record, we hypothesized that the rate of recorded The recognition of delirium may be particularly affected delirium in the VA system would be lower than the preva- by both the fluctuation of delirium symptoms and the lence noted in prospective studies. We also suspected that appearance of hypoactive or “quiet” delirium. Physicians patients recorded as having delirium would be older and may see inpatients during morning rounds when behav- more likely to have dementia than elderly patients with- ior may be less disturbing than in the afternoon or night.20 out recorded delirium. Nursing documentation of delirium is more frequent, likely due to nurses’ closer contact with patients through- METHODS out the day, enabling them to spot delirium despite fluctuation.20 In terms of motoric subtypes, O’Keefe and Data were obtained retrospectively from 2 sources, both Lavan21 have noted that it is the patients with agitated national computerized abstracts of VA administrative data. delirium who are most likely to attract medical and nurs- The Patient Treatment File contains abstracts of discharge ing attention, while patients with “quiet” delirium may records for each episode of care in all VA medical centers appear to be compliant “model” patients.22 and extended care facilities across the United States by Delirium is a syndrome associated with a multitude fiscal year. The abstracts include patient demographic of possible risk factors including advanced age, chronic cog- characteristics such as age, race, and gender and episode nitive impairment, severe medical comorbidity, and func- of care summary information such as admission date, dis- tional impairment.5,10,23-25 It may be these underlying risk charge date, and discharge diagnoses. The latter include factors (as opposed to the delirium itself) that may account primary diagnoses responsible for length of stay and as for the poor outcomes associated with delirium. These many as 9 accompanying secondary diagnoses. The Annual risk factors or comorbidities may also affect detection. Patient Census files contain analogous abstracts of records Elie et al26 found that elderly emergency room patients with to date for all patients residing in VA hospitals and detected delirium were more likely to have a primary neu- extended care facilities at the end of each fiscal year. rologic diagnosis than those with undetected delirium. The sample for our study was defined as all veterans The authors speculated that neurologic disorders may discharged from a VA medical center or extended care have led emergency room physicians to pay more atten- facility in FY96 (October 1, 1995, to September 30, 1996) tion to mental status. Notably, however, another study who were 60 years of age or older at discharge and who noted that delirium was less likely to be recognized by had any one of the following (International Classification nurses and physicians in hospitalized elderly patients of Diseases, 9th revision [ICD-9CM])29 delirium diagnoses with dementia.27 recorded for at least 1 episode of care: acute delirium Therefore, there are factors including providers’ choice (293.0), subacute delirium (293.1), drug-induced delirium of diagnostic terms, behavioral manifestations of delir- (292.0, 292.81), alcohol use or withdrawal delirium (291.0), ium, and associated comorbidities that may affect the senile dementia with delirium (290.3), arteriosclerotic detection and recording of delirium. No studies to our dementia with delirium (290.41), and presenile dementia knowledge have examined the rates of recorded delirium with delirium (290.11). As alternative terms for delirium and associated characteristics within a large health care might also be used by physicians such as “confusion,” system. As Gustafson et al have noted, if delirium is “not “encephalopathy,” “organic brain syndrome,” or “organic psy- properly diagnosed and documented, subsequent lack of chosis,” ICD-9CM was also searched for these and any appropriate treatment and care is expected.”20 The Veter- like terms. The FY96 entire sample was then examined for ans Affairs (VA) health care system, the largest single any of these related diagnoses recorded, which included provider of health care in the United States, serves a large acute encephalopathy (348.3); toxic encephalopathy (349.82); clinical population of elderly veterans. Prior research has hypertensive encephalopathy (437.2); hepatic encephalopa- indicated that the VA patient population has poorer health thy (572.2); drug-induced paranoid/hallucinatory state status and more medical conditions than non-VA patient (292.1); pathological drug intoxication resulting in brief populations, related in part to older age of the veteran psychotic states (292.2); unspecified drug-induced mental patients.28 Thus, the veteran patient population might be disorder, subcoded as organic psychosis not otherwise considered at higher risk for developing delirium. A detailed specified (292.9); organic delusional syndrome, subcoded 34 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003 as transient organic psychotic condition, paranoid type recorded delirium or encephalopathy diagnoses that con- (293.81); organic hallucinosis syndrome, subcoded as tran- stituted the recorded delirium group. An additional 6680, sient organic psychotic condition, hallucinatory type or 2.5%, had organic psychosis diagnoses that made up the (293.82); organic affective syndrome, subcoded as transient other recorded confusional diagnoses group. Therefore, organic psychotic condition, depressive type (293.83); other 4% (10,658 of the 267,947) of elderly patients admitted in specified transient organic mental disorder (293.89); 1996 had a delirium or a related confusional diagnosis unspecified transient organic mental disorder, subcoded as recorded. organic psychosis (293.9); other specified organic brain Table 1 lists demographics, utilization rates, and mor- syndromes (chronic) (294.8, includes mixed paranoid and tality for the 3 groups: (1) no recorded confusion, (2) affective organic psychotic states); and unspecified organic recorded delirium, and (3) other recorded confusional diag- brain syndrome (chronic), subcoded as organic psychosis noses. Patients with no recorded confusion were signifi- (294.9). cantly younger than those with either recorded delirium The overall rate of delirium diagnoses within the VA or other recorded confusional diagnoses. Patients with inpatient system for FY96 was calculated, as well as the other recorded confusional diagnoses were also signifi- rates of various delirium types and related confusional diag- cantly older than the recorded delirium group. There were noses recorded. Demographic variables included age at dis- no gender differences found among the 3 groups. Signifi- charge from index hospitalization, gender, and race. Because cant racial differences were found among the 3 study of small numbers, Hispanic white, Hispanic black, Amer- groups. The other recorded confusional diagnoses group had ican Indian, and Asian patients were grouped together as a significantly higher percentage of African Americans “other.” Length of stay at index hospitalization was com- (19%) than did the recorded delirium or no recorded con- puted as the number of days from admission to discharge, fusion groups. less days on leave. Other variables examined included In terms of health care utilization patterns, the groups index location, discharge location, index mortality, and with recorded delirium and other recorded confusional medical comorbidity. The percentage of patients with diagnoses were significantly more likely to be admitted to dementia in each group was calculated by searching for the psychiatry and nursing home units (and less likely to be following dementia codes: 331.0 (Alzheimer’s disease), admitted to medical or surgical units) than were patients 331.1 (Pick’s disease), 290.0 (senile dementia, uncompli- with no recorded confusion. Patients with other recorded cated), 290.10-3 (presenile dementia), 290.20-1 (senile confusional diagnoses had the highest rates of admission dementia), 290.3 (senile dementia with delirium), 290.40- to psychiatry and nursing home units. The groups with 3 (arteriosclerotic dementia), 291.2 (alcoholic dementia), recorded delirium and other recorded confusional diagnoses 292.82 (drug-induced dementia), and 294.1 (dementia in were also significantly more likely to be discharged to conditions classified elsewhere). nursing homes as compared to those with no recorded Patients without recorded delirium and without related confusion. The group with other recorded confusional diag- confusional diagnoses constituted the “no recorded con- noses had the highest rate of nursing home discharges fusion” group. As the results for recorded cases of delirium (23.2%) as compared to patients with no recorded confu- and encephalopathy cases were similar in their charac- sion (5.6%) or recorded delirium (17.1%). teristics, these cases were added together to form the Patients with recorded delirium had slightly shorter “recorded delirium” group. The organic psychoses and lengths of stay than did those with no recorded confu- organic brain syndrome cases were significantly different sion, but this difference was not significant. Patients with enough in their characteristics from the “recorded delir- other recorded confusional diagnoses had significantly ium” cases that they were grouped into a third group, longer lengths of stay on index admission than either the “other confusional diagnoses.” There were 3 patients with no recorded confusion or recorded delirium groups. Con- both recorded delirium and other confusional diagnoses; versely,those with recorded delirium had the highest mor- these patients were grouped under recorded delirium. The tality rates during index hospitalization (13.3%); these 3 groups, no recorded confusion, recorded delirium, and mortality rates were significantly higher than the no other recorded confusional diagnoses, were then compared recorded confusion (5.2%) or other recorded confusional for the demographic and utilization variables; categorical diagnoses (8.7%) groups. differences were assessed using chi-square tests of independence, and continuous variables were compared by 3- Types of Delirium: group 1-way univariate analyses of variance (ANOVA). Related Confusional Diagnoses Recorded Table 2 lists the types of delirium noted among patients RESULTS recorded as having delirium or encephalopathy diagnoses. Among the recorded delirium group, the highest rates There were a total of 267,947 patients older than the age were found for dementia with delirium (30.1%) and alco- of 60 admitted to VA inpatient units nationally in 1996. hol intoxication/withdrawal delirium (18.7%). Among Of these, a total of 3978, or 1.5%, of elderly patients had patients having dementia with delirium recorded, demen- Recorded Delirium in a National Sample of Elderly Inpatients / Kales et al. 35

Table 1. Demographic and Utilization Variables for the Study Groups

Group 3: Group 1: Group 2: Other Recorded No Recorded Confusion Recorded Delirium Confusional Diagnoses (n = 257,289) (n = 3978) (n = 6680)

n % n % n % Significance Test

Demographics a a,b * Age (years, mean [SD]) 71.4 (6.8) 72.0 (7.4) 75.5 (7.3) F[df] = 1186.22[2,267,944] χ2 Race [df] = 108.25[4] Caucasian 198,520 77.2 3,086 77.6 5,088 76.2 African American 40,458 15.7 605 15.2 1,276 19.1 All other 18,311 7.1 235 5.9 316 4.7 Missing 0 0.0 52 1.3 0 0.0 Gender

Female 5,356 2.1 70 1.8 137 2.1 0.35[2] Utilization * Index location 4706.47[4] Medical/surgery 233,474 90.7 3,238 81.4 4,444 66.5 Psychiatry 11,644 4.5 464 11.7 1,028 15.4 Nursing home 12,171 4.7 276 6.9 1,208 18.1 * Discharge location 5562.79[6] Community 214,717 83.4 2,474 62.2 3,893 58.3 Nursing home 14,344 5.6 681 17.1 1,549 23.2 Death (index) 13,347 5.2 530 13.3 579 8.7 Other 14,881 5.8 293 7.4 659 9.9 Index length of stay a,b (days, mean [SD]) 44.4 (288.5) 42.2 (171.9) 150.5 (485.7) F[df] = 426.02[2,267,944]*

Significant post hoc (Tukey-Kramer’s) comparisons (P < .05): asignificantly greater than group 1; bsignificantly greater than group 2. *P < .0001.

Table 2. Types of Delirium Recorded (n = 3978) higher than the other 2 groups. Among hospitalized elderly patients with no recorded confusion (n = 257,289), 5.4% Diagnosis (ICD-9CM Code) n % (n = 13,828) had a recorded dementia diagnosis. The low- est percentage of recorded dementia (4.9%, n= 326) was Dementia with delirium (290.3, 290.41, 290.11, or dementia code plus delirium or found among patients with other recorded confusional encephalopathy code) 1197 30.1 diagnoses. Alcohol intoxication/withdrawal delirium (291.0) 743 18.7 Other medical comorbidities recorded were analyzed Acute/subacute delirium (293.0, 293.1) 701 17.6 Hepatic encephalopathy (572.2) 560 14.1 for all patient groups. The top 3 comorbidities recorded did Drug-induced delirium (292.0, 292.81) 492 12.4 not differ between groups and were (1) diseases of the cir- Acute encephalopathy (348.3) 172 4.3 culatory system (390-459.9), (2) infectious diseases (001- Hypertensive encephalopathy (437.2) 89 2.2 Toxic encephalopathy (349.82) 17 0.4 139.8), and (3) endocrine, nutritional and metabolic, and Combination diagnoses (348.3/293.83, immunity disorders (240-279.9). 290.2/293.81, 572.2/437.2, 572.2/292.0, 572.2/348.3, 348.3/292.0) 7 0.2 DISCUSSION

The current study was undertaken to obtain a snapshot tia of the Alzheimer’s type (n = 652, 54.5%) was the most of the face of recorded delirium and related diagnoses common dementia. within a large national health care system, the VA health As seen in Table 3, among other recorded confusional care system, to see what characteristics might be associ- diagnoses, the most common recorded diagnosis was chronic ated with recorded delirium. The overall rate of the recorded organic brain syndrome (294.8, 84.1%). Taking all patients diagnosis of delirium and very closely related diagnoses with either recorded delirium or other recorded confu- among all hospitalized elderly veterans was found to be sional diagnoses together, the most common confusional only 1.5%. More common were other confusional diag- diagnosis recorded was chronic organic brain syndrome, noses, recorded in another 2.5% of all elderly inpatients. which was found in almost 53% of all study patients with In total, 4% of hospitalized elderly veterans in FY96 had delirium or related diagnoses. delirium or a related confusional diagnosis. Among the delirium diagnoses recorded, the highest rates were found Dementia or Other Medical Comorbidities for dementia with delirium and alcohol intoxication/ Thirty percent (n = 1197) of those with recorded delirium withdrawal delirium. Patients with recorded delirium also had a recorded dementia diagnosis, significantly diagnoses were significantly older, more likely to be admit- 36 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003

Table 3. Other Recorded Confusional Diagnoses (n = 6680) the group of patients who received other recorded confusional diagnoses is likely heterogeneous, containing patients Diagnosis (ICD-9CM Code) n % with chronic psychiatric diagnoses as well as undiagnosed Chronic organic brain syndrome (294.8) 5616 84.1 dementia. Another indication of this heterogeneity is pos- Chronic organic psychosis (294.9) 367 5.5 sibly found in the large standard deviations in length of Specified transient organic psychosis (293.8) 327 4.9 stay found for all groups (but largest for the group with Unspecified transient organic psychosis (293.9) 189 2.8 Drug-induced paranoid/hallucinatory state (292.1) 88 1.3 other recorded confusional diagnoses); we suspect this Combination diagnoses (292.1/292.9, 293.9/292.9, reflects the tremendous variability in hospital and nurs- 293.9/293.83, 294.8/293.83, 294.8/293.89, ing home stays of this elderly patient population. 294.8/293.81, 294.8/293.82, 294.8/293.9, 294.8/292.9, 294.8/294.9, 294.8/292.1, The racial differences found between groups were 294.9/293.82, 294.9/293.9) 48 0.7 particularly striking: almost 20% of patients receiving Organic psychosis not otherwise specified related other recorded confusional diagnoses or organic psychoses to drugs (292.9) 39 0.6 Pathological drug intoxication (brief psychotic state) were African American. While race was missing in 52 (292.2) 6 0.1 cases in the recorded delirium group, even had all 52 patients been African American, the percentage of African Americans would still have been only 16.5% in this group. One explanation for the racial differences found may per- ted to nonmedical/surgical units, and more likely to be dis- tain to medical comorbidities with high prevalence rates charged to nursing homes than were those without delirium/ in African Americans such as chronic renal failure,33 as confusional diagnoses recorded. Patients with recorded encephalopathy associated with renal dialysis is one diag- delirium also had significantly higher mortality rates nosis coded as chronic organic brain syndrome. While than either those with no delirium/confusional diagnoses African Americans account for about 13% of the popula- recorded or those with other confusional diagnoses tion, they account for 29% of new cases of end-stage renal recorded. The profile of those with other recorded confu- disease and 37% of those undergoing dialysis.34 However, sional diagnoses appeared distinct; patients who received based on the data available to us, we were unable to show these diagnoses were significantly older, significantly more different patterns of medical comorbidity between groups. likely to be African American, significantly more likely to Another explanation might be found in the findings of be admitted to nonmedical/surgical units and discharged prior studies including our own, that is, that elderly African to nursing homes, and had significantly longer lengths of American patients are significantly more likely to receive stay on index admission than did both other patient groups. clinical diagnoses of psychotic disorders than are Cau- The overall rate of recorded delirium/related confu- casian patients.35-37 In a previous national study of elderly sional diagnoses found retrospectively among all elderly veterans, we found that elderly African Americans were hospitalized veterans, 4%, is most certainly an underes- significantly more likely to receive psychotic, cognitive, and timate of true prevalence. In previous prospective studies, substance abuse disorder diagnoses and significantly less in which investigators actively assessed for the presence likely to receive mood disorders diagnoses.35 In that study, of delirium, delirium has been estimated to occur in 14% we hypothesized that differential rates of recognition/ to 56% of hospitalized elderly patients.1 Our results are diagnosis of depression by race might pertain to both likely both a reflection of the consistent underrecognition patient (differential symptom presentation) and provider and underreporting (and thus underrecording) of delir- (conscious or unconscious bias) factors. This could be the ium by physicians and nurses found in previous stud- case with delirium as well. All of the diagnoses contained ies14,30-32 and a possible warning that the problem of poor in the other recorded confusional diagnoses category recognition may be widespread as indicated by the low rates include organic psychosis in either the coding or subcod- of recorded delirium and related diagnoses found in the VA ing in ICD-9CM; it is possible that in some cases, psychotic national health care system. symptoms in the context of a delirium syndrome are coded The finding of 53% of study patients with preferentially as organic psychosis in African Americans delirium/other related confusional diagnoses being assigned as opposed to actual delirium diagnoses. a diagnosis of chronic organic brain syndrome is puzzling. In terms of patients given actual delirium or Interestingly, while patients with other recorded confu- encephalopathy diagnoses, the most common type of delir- sional diagnoses were older and had longer lengths of ium recorded was delirium with dementia; 30% of the stay than did the patients with delirium diagnoses, their delirium group also had dementia recorded. This finding mortality rates were significantly lower. Based on their could reflect both the increased risk for delirium in demen- demographic and utilization profile, we suspected that tia, as dementia itself is the most common risk factor for patients with other recorded confusional diagnoses would delirium in the elderly with the possibility of almost any have higher rates of recorded dementia than the other 2 physical illness precipitating delirium in the elderly when groups, but surprisingly,only 5% of these patients had such there exists an underlying cerebral disease such as diagnoses recorded. These findings lead us to suspect that Alzheimer’s or vascular dementia.38,39 The finding may Recorded Delirium in a National Sample of Elderly Inpatients / Kales et al. 37 also point toward the greater detection of delirium in care system. Some patients in the group with no recorded dementia. Elie et al26 found that patients in their study with delirium very likely had actual episodes of delirium that detected delirium were more likely to have a primary neu- were either unrecognized, unrecorded, or both. rologic diagnosis than were those with undetected delir- Despite the caveats noted, the current study provides ium. The authors hypothesized that neurologic disorders an important window to recorded delirium in elderly inpa- may lead physicians to pay more attention to mental sta- tients within a large health care system. Taking into tus; however, Fick and Foreman27 found that delirium was account prior prospective prevalence estimates, the over- less likely to be recognized by nurses and physicians in hos- all rate of delirium/related confusional diagnoses recorded pitalized elderly patients with dementia. in the VA health system is very likely an underestimate The second highest recorded delirium rate among of true prevalence and possibly reflects poor recognition patients with delirium/encephalopathy was for alcohol and recording of delirium syndromes in older veterans. We intoxication/withdrawal delirium. Age is a potential pre- suspect that dementia-related delirium, perhaps in part dictor of alcohol withdrawal severity and complications; due to greater attention to mental status in a preexisit- Kraemer et al40 found that 14% of elderly patients admit- ing neurological disorder, and organic psychoses and alcohol- ted to an alcohol detoxification unit had delirium during related delirium, perhaps in part due to hyperactive or alcohol withdrawal. Prior studies have found that elderly disruptive behavioral symptoms, are more commonly diag- patients had significantly more alcohol withdrawal symp- nosed and recorded. Other types of delirium (other etiologies toms for a longer duration than did younger patients.41,42 or motoric subtypes) are likely underrecognized and under- Delirium has been phenomenologically divided into 3 treated. Racial differences in delirium or other confusional different motoric subtypes according to the presence of psy- diagnoses received may reflect both patient (differential chomotor and behavioral symptoms: hyperactive, hypoac- symptom presentation) and provider (conscious or uncon- tive, and mixed.38 Prior studies have suggested that the scious bias) factors. Future research should prospectively hyperactive delirium subtype may be associated with bet- examine the relationship between motoric and etiologic ter outcomes43 including lower mortality rates44 and eti- delirium subtypes and recognition, as well as racial dif- ologically related to drug intoxication/withdrawal.45 ferences in the diagnosis of delirium. Hypoactive delirium has been associated both with poorer outcomes21 including higher mortality rates43,44 as well as References greater underrecognition.46,47 Delirium in the elderly is 1. Inouye SK. The dilemma of delirium: clinical and research con- often quiet or hypoactive.47 Those with agitated, hyperactive troversies regarding diagnosis and evaluation of delirium in hospitalized elderly medical patients. Am J Med 1994; 97:278- delirium may be most likely to attract provider attention, 288. while those who are quietly delirious may appear to be 2. Murray AM, Levkoff SE, Wetle TT,et al. Acute delirium and func- “model patients.”21 Thus, delirium presenting quietly and tional decline in the hospitalized elderly patient. J Gerontol: Med without disruptive behaviors in elderly patients often may Sci 1993; 48:M181-M185. be missed by physicians.46 Notably,among the 3 most com- 3. Cole MG, Primeau FJ. Prognosis of delirium in elderly hospitalized patients. Can Med Assoc J 1993; 149:41-46. mon confusional diagnoses assigned in our study (chronic 4. van Hemert AM, van der Mast RC, Hengeveld MW,et al. Excess organic brain syndrome, dementia with delirium, alcohol mortality in general hospital patients with delirium: a 5-year intoxication/withdrawal delirium), 1 is related to organic follow-up of 519 patients seen in psychiatric consultation. J Psy- psychosis and the other to alcohol withdrawal. While we chosom Res 1994; 38:339-346. cannot say for certain what the phenomenological subtypes 5. Pompei P, Foreman M, Rudberg MA, et al. Delirium in hospitalized older persons: outcomes and predictors. J Am Geriatr Soc were in our sample with recorded delirium/confusional diag- 1994; 42:809-815. noses, we speculate that many of these patients may have 6. Inouye SK, Rushing JT, Foreman MD, et al. Does delirium con- had hyperactive delirium with behavioral/physical man- tribute to poor hospital outcomes? J Gen Intern Med 1998; ifestations (hallucinations, withdrawal symptoms) increas- 13:234-242. ing their visibility to clinicians. 7. Francis JF,Kapoor WN. Prognosis after hospital discharge of older There are several limitations to our study. These medical patients with delirium. J Am Geriatr Soc 1992; 40:601-606. include the fact that all patients were veterans and most 8. O’Keefe S, Lavan J. The prognostic significance of delirium in older hospitalized patients. J Am Geriatr Soc 1997; 45:174-178. patients were men. Thus, our results cannot be general- 9. Stevens LE, de Moore GM, Simpson JM. Delirium in hospital: ized to nonveteran populations containing equal numbers does it increase length of stay? Aust N Z J Psychiatry 1998; of men and women. In addition, our study was retrospec- 32:805-808. tively performed, using information from an administra- 10. Levkoff SE, Evans DA, Liptzin B, et al. Delirium: the occurrence tive database. Consequently, we could not prospectively and persistence of symptoms among elderly hospitalized patients. Arch Intern Med 1992; 152:334-340. assess actual delirium recognition rates. It is notable that 11. Inouye SK, Schlesinger MJ, Lydon TJ. Delirium: a symptom of the rate of delirium/related confusional diagnoses in FY96 how hospital care is failing older persons and a window to was only 4% of all elderly patients hospitalized on VA improve quality of hospital care. Am J Med 1999; 106:565-573. inpatient units. We suspect this to be a gross underesti- 12. Levkoff SE, Besdine RW,Wetle T.Acute confusional states in the mate of the true prevalence of delirium within the health hospitalized elderly. Ann Rev Gerontol Geriatr 1986; 6:1-26. 38 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003

13. Francis J, Strong S, Martin D, et al. Delirium in elderly general Ann Arbor, MI: Commission on Professional and Hospital Activ- medical patients: common but often unrecognized. Clin Res ities, 1978. 1988; 36:711A. 30. Knights EB, Folstein MF.Unsuspected emotional and cognitive 14. Williams MA, Holloway JR, Winn MC, et al. Nursing activities disturbance in medical patients. Ann Intern Med 1977; 87:723-724. and acute confusional states in elderly hip-fractured patients. 31. Palmateer LM, McCartney JR. Do nurses know when patients Nurs Res 1979; 28:25-35. have cognitive deficits? J Gerontol Nurs 1985; 11:6-16. 15. Johnson J. Identifying and recognizing delirium. Dement Geri- 32. Levkoff SE, Safran C, Cleary PD, et al. Identification of factors atr Cogn Disord 1999; 10:353-358. associated with the diagnosis of delirium in elderly hospitalized 16. Francis J, Kapoor WN. Delirium in hospitalized elderly. J Gen patients. J Am Geriatr Soc 1988; 36:1099-1104. Int Med 1990; 5:65-79. 33. Young CJ, Gaston RS. Renal transplantation in black Americans. 17. Johnson J, Kerse M, Gottlieb G, et al. Prospective versus retro- New Engl J Med 2000; 343:1545-1552. spective methods of identifying patients with delirium. J Am Geri- 34. United States Renal Data System 1999 Annual Data Report. Am atr Soc 1992; 40:316-319. J Kidney Dis 1999; 34(Suppl 1):S1-S176. 18. Kamholz BK, Kudelko K, Tyrell V. Prevalence and recognition 35. Kales HC, Blow FC, Bingham CR, et al. Race and inpatient psy- of delirium at a VA teaching hospital. Scientific proceedings. chiatric diagnoses among elderly veterans. Psychiatr Serv 2000; Bethesda, MD: American Association for Geriatric Psychiatry, 51:795-800. 2000. 36. Fabrega H, Mulsant BH, Rifai AH, et al. Ethnicity and psy- 19. Inouye SK, van Dyck CH, Alessi CA, et al. Clarifying confusion: chopathology in an again hospital-based population. J Nerv the Confusion Assessment Method: a new method for detection Ment Dis 1994; 182:136-144. of delirium. Ann Intern Med 1990; 113:941-948. 37. Leo RJ, Narayan DA, Sherry C, et al. Geropsychiatric consultation 20. Gustafson Y, Brannstrom B, Norberg A, et al. Underdiagnosis for African-American and Caucasian patients. Gen Hosp Psy- and poor documentation of acute confusional states in elderly chiatry 1997; 19:216-222. hip fracture patients. J Am Geriatr Soc 1991; 39:760-765. 38. Lipowski ZJ. Delirium: acute confusional states. Oxford, UK: 21. O’Keefe ST, Lavan JN. Clinical significance of delirium sub- Oxford University Press, 1990. types in older people. Age Ageing 1999; 28:115-119. 39. Lindesay J, Mac Donald A, Starke I. Delirium in the elderly. 22. Inouye SK, Foreman MD, Mion LC, et al. Nurses’ recognition of Oxford, UK: Oxford University Press, 1990. delirium and its symptoms. Arch Intern Med 2001; 161:2467-2473. 40. Kraemer KL, Mayo-Smith MF, Calkins DR. Impact of age on 23. Francis J, Martin D, Kapoor WN. A prospective study of delir- severity, course and complications of alcohol withdrawal. Arch ium in hospitalized elderly. JAMA 1990; 263:1097-1101. Intern Med 1997; 157:2234-2241. 24. Rockwood K. Acute confusion in elderly medical patients. J Am 41. Brower KJ, Mudd S, Blow FC, et al. Severity and treatment of Geriatr Soc 1989; 37:150-154. alcohol withdrawal in elderly vs younger patients. Alcohol Clin 25. Marcantonio ER, Goldman L, Mangione CM, et al. A clinical pre- Exp Res 1994; 18:196-201. diction rule for delirium after elective noncardiac surgery. JAMA 42. Mudd SA, Boyd CJ, Brower KJ, et al. Alcohol withdrawal and 1994; 271:134-139. related nursing care in older adults. J Gerontol Nurs 1994; 26. Elie M, Rousseau F, Cole M, et al. Prevalence and detection of 20:17-26. delirium in elderly emergency department patients. CMAJ 43. Kobayashi L, Takeuchi O, Suzuki M, et al. A retrospective study 2000; 163:977-981. on delirium type. Jpn J Psychiatry Neurol 1992; 46:911-917. 27. Fick D, Foreman MD. Consequences of not recognizing delirium 44. Liptzin B, Levkoff SE. An empirical study of delirium subtypes. superimposed on dementia in hospitalized elderly individuals. Br J Psychiatry 1992; 161:843-845. J Gerontol Nursing 2000; 26:30-40. 45. Ross CA, Peyser CE, Shapiro I, et al. Delirium: phenomenologic 28. Agha Z, Lofgren RP, VanRuiswyk JV, et al. Are patients at the and etiologic subtypes. Int Psychogeriatr 1991; 3:135-147. Veterans Affairs medical centers sicker? Arch Intern Med 2000; 46. Francis J, Martin D, Kapoor WN. A prospective study of delir- 160:3252-3257. ium in hospitalized elderly. JAMA 1990; 263:1097-1101. 29. Commission on Professional and Hospital Activities. International 47. Inouye SK. Delirium in hospitalized older patients: recognition classification of diseases: 9th revision, clinical modification. and risk factors. J Geriatr Psychiatry Neurol 1998; 11:118-125. Long-Term Effects of Donepezil on P300 Auditory Event-Related Potentials in Patients With Alzheimer’s Disease

Eiichi Katada, MD, PhD, Koichi Sato, MD, PhD, Akira Sawaki, MD, Yasuaki Dohi, MD, PhD, Ryuzo Ueda, MD, PhD, and Kosei Ojika, MD, PhD

ABSTRACT

The P300, one of the cognitive event-related potentials (ERPs) of the cerebral cortex, reflects the functioning of the neurochemical system involved in cognitive processes. We investigated clinical significance of the components of auditory P300 ERPs, in comparison with neuropsychologic tests including the Mini-Mental State Examination and the Japanese version of the Alzheimer’s Disease Assessment Scale–cognitive subscale (ADAS–J cog), for evaluating of the effect of donepezil (DPZ) (5 mg daily for 6 months), an acetylcholinesterase inhibitor, in patients with Alzheimer’s disease (AD). Reduction of P300 latency associated with a parallel improvement of ADAS–J cog scores was observed after administration of 5 mg/day of DPZ in patients with AD. P300 latency gives very useful information on the progression of AD, especially in the longitudinal follow-up of patients with AD during treatment with DPZ acting on cholinergic pathways. (J Geriatr Psychiatry Neurol 2003;16:39–43)

Keywords: Alzheimer’s disease; event-related potentials; auditory P300; donepezil

Alzheimer’s disease (AD), one of the most common causes these methods listed above, P300, one of the cognitive of mental deterioration in the elderly, is a progressive ERPs of the cerebral cortex, reflects the functioning of neurodegenerative disorder characterized by cognitive the neurochemical systems involved in cognitive processes.6 and behavioral dysfunction. The etiology of AD is not yet The anticholinergic scopolamine delays P300 latency and elucidated, but the cholinergic hypothesis of AD is stated decreases P300 amplitude, and physostigmine on the ground of the presynaptic deficits found in the (cholinesterase inhibitor) reduces P300 latency in the brains of patients with AD and studies of the role of acetyl- short term, indicating that cholinergic neurons are impor- choline in animal and human behavior.1 To assess the pro- tant in the neuronal networks generating the P300 poten- gression of AD, neuropsychological (Alzheimer’s Disease tial.6,7 P300 latency is very useful in determining the Assessment Scale–cognitive subscale [ADAS-cog],2 Mini- progression of AD while the ADAS-cog2 and the MMSE3 Mental State Examination [MMSE]3), neuroimaging (com- show the degree of cognition. Furthermore, P300 latency puted tomography, magnetic resonance imaging, is known to change even at the early stage of AD with lit- single-photon emission computed tomography, positron tle cognition deterioration that cannot be detected by the emission tomography),4 and neurophysiological (elec- MMSE.5 Therefore, changes in P300 latency may reflect the troencephalography,auditory P300 event-related potentials effects of treatment in patients with AD. In this study, we [ERPs])5 investigations have been widely used. Among investigated clinical significance of P300 latency in comparison with MMSE and the Japanese version of the ADAS-cog (ADAS–J cog), for evaluating the effect of 8 Received February 12, 2002. Received revised June 3, 2002. Accepted for donepezil (DPZ), an acetylcholinesterase inhibitor, in publication June 7, 2002. patients with AD. From the Department of Internal Medicine, Nagoya City Johoku Hospital, Japan (Drs. Katada, Sato, and Sawaki); Second Department of Internal PATIENTS AND METHODS Medicine, Nagoya City University Medical School, Japan (Drs. Dohi and Ueda); and Department of Neurology, Nagoya City University Medical School, Japan (Dr. Ojika). Subjects Thirteen subjects, aged 70 to 88 (5 men, 8 women; aver- Reprint requests: Eiichi Katada, Department of Internal Medicine, Nagoya City Johoku Hospital, 2-15 Kaneda-cho, Kita-ku, Nagoya 462-0033, Japan; age age [mean ± SD] = 78.0 ± 6.1) were enrolled in this e-mail: [email protected]. study.The subjects were outpatients diagnosed as having DOI: 10.1177/0891988702250561 probable AD by the diagnostic criteria of the National

Institute of Neurological and Communicative Disorders and ton with their right index finger.We assessed the push but- Stroke and the Alzheimer’s Disease and Related Disorders ton response capability and accuracy by observing the Association9 and had undergone a neurological examina- Neuropack Sigma display. The responses were amplified tion, neuroimaging, and laboratory workup to rule out with filter band passes of 0.1 and 50 Hz and were aver- other treatable causes of dementia. None had other sig- aged until the individual artifact-free trial was completed nificant psychiatric or medical illness or had taken med- after 30 target tones. The P300 wave was identified accord- ications that might affect cognitive ERPs. Patients in ing to the operating definition based on its probability sen- whom no reliable ERPs to “odd” stimuli could be recorded sitivity and the sequence of preceding components. The at baseline (since undetectable P300 is a common minor- latency of P300 was measured at the peak of the poten- ity finding7,10) were excluded because we considered record- tial at Pz, and the amplitude of the potential at Pz was ing of detectable P300 to be the basic prerequisite in this measured in comparison with prestimulus baseline. study. The patients orally received DPZ (3 mg/day) for the first Statistical Evaluation week to avoid gastrointestinal symptoms such as nausea All 13 patients had a detectable P300 component during or vomiting that might appear during the initial admin- all recording sessions. Amplitude and latency measure- istration period and 5 mg/day thereafter over 6 months. ments of P300, which were recorded during this study,were Neuropsychologic findings and auditory P300 ERPs were fed into a computer data management and analysis sys- obtained at baseline (ie, before initiation of DPZ treatment) tem program for statistical evaluation of data recorded at and after 1, 3, and 6 months of the treatment with DPZ. different times of the study. Continuous variables were The neuropsychologic tests were performed on the days the reported as mean ± SD. Systematic differences of ampli- ERPs were recorded. tude and latency of P300 between baseline and each trial period were evaluated using repeated-measure one-way Neuropsychologic Assessment analysis of variance. If the difference was significant, Neuropsychologic tests included the MMSE3 and the Fisher’s protected least significant difference was applied. ADAS–J cog11 in this study.The MMSE was used for grad- Systematic differences of the scores of the MMSE and ing the cognitive state of patients with AD. To assess the ADAS–J cog between baseline and each trial period were efficacy of DPZ, ADAS–J cog was used. The ADAS–J cog, evaluated using the Friedman test, and if the difference the Japanese version of ADAS-cog that is a sensitive and was significant, the Wilcoxon signed-rank test with Bon- reliable psychometric scale,2,12,13 is used as the selection cri- ferroni correction was applied. The Spearman rank cor- teria of a clinical trial.11,14 ADAS–J cog consists of 11 items relation was applied to verify the correlation between that evaluate selected aspects of memory, orientation, MMSE scores and P300 and ADAS–J cog scores and P300. attention, language, reasoning, and praxis, and the range of its total score is from 0 to 70.11 At baseline in this study, RESULTS the selection criteria were as follows: MMSE scores ranging from 12 to 26 (average [mean ± SD] = 18.5 ± 4.3) and All the patients completed the whole study without any ADAS–J cog scores ranging from 9.4 to 58.0 (average side effects. The premedication and postmedication effects [mean ± SD] = 23.6 ± 12.6). of treatment at DPZ were assessed in all 13 subjects.

Event-Related Potential Recordings MMSE Baseline Versus 1, 3, and 6 Months ERPs were recorded with Ag/AgCl disk scalp electrodes Values of the MMSE demonstrated an improvement in 10 placed on the Fz, Cz, and Pz positions of the international patients (76.9%) treated with DPZ at 1-month evaluation 10-20 system. Reference electrodes corresponded to the in comparison with those at premedication baseline (Fig- linked earlobes. Ground was placed at Fpz. Interelectrode ure 1A). An increase in mean MMSE scores was signifi- impedance was reduced to below 5 Kohm. The binaurally cant (vs premedication baseline) at 1 month and 3 months presented auditory stimuli were produced by a system (P < .05, P < .01, respectively) (Table 1; Figure 2A). MMSE applying averaging and a P300 stimulator: Neuropack score decrements were significant (vs 3 months) at 6 Sigma (Nihon Kohden, Tokyo, Japan). We used a simple months (P < .05) (Table 1; Figure 2A). auditory oddball paradigm. Tones of 2 different frequencies, 1 kHz (frequent nontarget tone) and 2 kHz (rare tar- ADAS–J cog Baseline Versus 1, 3, and 6 Months get tone), were presented at a rate of 80% and 20%, Decrements of the ADAS–J cog score were observed in 11 respectively, in a random order at an interstimulus inter- patients (84.6%) treated with DPZ at 1 month relative to val of 1.4 seconds. The rise-fall and plateau times were 10 those at premedication baseline (Figure 1B). ADAS–J cog and 100 msec, respectively. The stimuli were delivered score decrements were significant (vs premedication base- binaurally through light headphones at an intensity of 70 line) at 1, 3, and 6 months (P < .01) (Table 1; Figure 2B). dB nHL. The subjects were required to respond to the The ADAS–J cog increase was significant (vs 3 months) at stimulus of the rare target tone by pressing a push but- 6 months (P < .05) (Table 1; Figure 2B). Long-Term Effects of Donepezil on P300 Auditory Event-Related Potentials in Patients With Alzheimer’s Disease / Katada et al. 41 (ms)

Figure 1. Individual values of Mini-Mental State Exam (MMSE)

MMSE scores (A), Alzheimer’s Disease ADAS-J cog P300 latency Assessment Scale–cognitive subscale (Japanese version) (ADAS–J cog) scores (B), and P300 latencies (C) showing the difference between baseline and 1 month after the first adminis- Baseline 1M Baseline 1M Baseline 1M tration of donepezil in 13 patients (A) (B) (C) with Alzheimer’s disease.

Table 1. MMSE Mean Scores, ADAS–J Cog Mean Scores, ** and P300 Components in Patients With Alzheimer’s 28 ** Disease Treated With Donepezil From Baseline 26 to Completion of the Study (6 Months) 24 Baseline 1 Month 3 Months 6 Months 22 MMSE 18.5 ± 4.3 21.8 ± 4.8 22.1 ± 4.6 20.5 ± 4.9 (A) MMSE ADAS–J cog 23.6 ± 12.6 18.4 ± 9.6 18.3 ± 10.7 19.9 ± 12.2 20 P300 18 latency (ms) 404.3 ± 49.9 381.5 ± 42.2 394.8 ± 39.2 405.4 ± 51.7 P300 16 amplitude (µV) 12.7 ± 8.1 13.0 ± 9.4 10.1 ± 7.2 11.3 ± 6.2 14 MMSE = Mini-Mental State Examination; ADAS-J cog = Alzheimer’s Disease Assessment Baseline 1 month 3 months 6 months Scale–cognitive subscale (Japanese version). Values are expressed as mean ± standard ** 40 ** deviation. ** * 35

P300 Baseline Versus 1, 3, and 6 Months 30 Typical recordings of P300 obtained from a subject in this study from baseline to 6 months are shown in Figure 3. 25

(B) ADAS-J cog Decrements of P300 latency were observed in 10 patients 20 (76.9%) treated with DPZ at 1 month relative to those at 15 premedication baseline (Figure 1C). P300 latency reduction was significant (vs premedication baseline) at the 1- 10 month follow-up after the first administration (P < .05) 5 (Table 1; Figure 2C). P300 latency was delayed in 6 patients Baseline 1 month 3 months 6 months * (46.2%) treated with DPZ at the 6-month evaluation rel- * ative to those at the 1-month evaluation. P300 latency delay 460 was significant (vs 1 month) at the 6-month follow-up 440 (P < .05) (Table 1; Figure 2C). 420 Correlation Between Neuropsychological (C) 400 Tests and P300 380

Figure 4 shows the scatter plot of P300 latencies and P300 latency (ms) ADAS–J cog scores recorded at baseline (A) and on com- 360 pletion of the study (6 months) (B) in patients with AD who 340 were treated with DPZ. Because of the reduction of P300 latency and the modest improvement of ADAS–J cog 320 Baseline 1 month 3 months 6 months scores, the correlation values changed on completion of the study.The ADAS–J cog score was significantly correlated Figure 2. Effect of donepezil (5 mg/day) on Mini-Mental State Exam with P300 latency but not with P300 amplitudes (data not (MMSE) scores (A), Alzheimer’s Disease Assessment Scale–cognitive subscale (Japanese version) (ADAS–J cog) scores (B), and shown) at baseline (r = 0.676, P < .01) (Figure 4A) and on P300 latencies at Pz (C) in 13 patients with Alzheimer’s disease. The completion of the study (6 months) (r = 0.701, P < .01) (Fig- mean ± SD values are expressed as changes from baseline to 6 ure 4B). However, the MMSE score was not correlated months. *P < .05. **P < .01. 42 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003

reduced the P300 latency (mean pretreatment latency = a 382.7 msec; mean posttreatment latency = 368.8 msec at b 1 month, 366.9 msec at 3 months, and 367.4 msec at 6 months). Our data concerning the changes of P300 latency c were similar to the results of these previous reports. How- ever, there were a few differences. Our study indicated that d the P300 latency reduction was significant (versus base- 5 µV line) at 1 month but the P300 latency delay was significant (versus 1 month) at 6 months, although Thomas et al reported that the P300 latency reduction persisted for 6 months after the initial treatment with DPZ. The characteristics of patients with AD were quite different between our study and that of Thomas et al; the mean age was 200ms 400ms 0 66.5 ± 9.1 years, the mean MMSE score was 16 ± 0.5, and Figure 3. Typical recordings of P300 obtained from a subject in this the mean ADAS-cog score was 33.34 ± 2.7 in the study by study from baseline to 6 months. The markers of peak P300 are positioned at 426 ms (a; baseline), 404 ms (b; 1 month), 428 ms (c; 3 Thomas et al. Thus, the difference in the effect of DPZ on months), and 468 ms (d; 6 months). P300 latency between the studies is probably based on the selection of responders and the dosage of DPZ. with P300 latency in patients with AD throughout the The effect of DPZ on P300 latency was rapid and con- study. sistent, indicating that P300 ERP recordings might represent a rapid and convenient measurement that can be DISCUSSION performed repeatedly in the evaluation of patients with AD during treatment with DPZ acting on cholinergic pathways. Our study showed that P300 latency was reduced by the However, there are several problems related to P300 ERP 6 months’ treatment with 5 mg/day of DPZ with a paral- recordings in the evaluation of patients with AD. For lel improvement of ADAS–J cog scores in patients with AD. example, methodologically, no reliable P300 components Two previous reports were published concerning the effect to odd stimuli at baseline could be recorded in some sub- of DPZ on auditory P300 ERPs.13,15 Reeves et al15 reported jects: undetectable P300 is a common finding in a minority that auditory P300 latency was not normalized in 11 (5%-10%) of normal subjects.7 Moreover, neurophysiolog- patients with AD after 1 month of treatment with a dosage ical change does not necessarily imply clinical or behav- of 5 mg/day of DPZ, but the treatment was associated ioral change. Results of neuropsychological tests sometimes with a significant reduction in P300 latency (mean pre- remained in the abnormal range, and clinical improvement treatment latency = 401.5 msec; mean posttreatment was not significant despite of the improvement of P300 latency = 392.7 msec). They also indicated that DPZ had latencies. The discrepancy between P300 latencies and no discernible effect on the auditory P300 peak-to-peak clinical features may be explained by the fact that P300 amplitude or baseline-to-P300 peak amplitude. Thomas et latency reflects cognitive dysfunction more sensitively al13 reported that in 20 patients with AD, treatment with than clinical or behavioral features in patients with AD. the DPZ (5 mg/day for 15 days and 10 mg/day thereafter) Indeed, P300 latency changes in patients with AD at an

Figure 4. Correlations between the Alzheimer’s Disease Assessment Scale–cognitive subscale (Japanese version) (ADAS–J cog) scor e and P300 latency before (A) and after 6 months of treatment with donepezil (5 mg/day) (B) in 13 patients with Alzheimer’s disease. Linear regression analysis represents (A) r = 0.676, P < .01, and (B) r = 0.701, P < .01. Long-Term Effects of Donepezil on P300 Auditory Event-Related Potentials in Patients With Alzheimer’s Disease / Katada et al. 43 early stage with very little cognitive deterioration.5 Thus, diagnosis and follow-up of patients with Alzheimer’s disease. Rev Neurol 2001; 32:525-528. P300 is very useful in monitoring the efficacy of treatment 6. Polich J, Kok A. Cognitive and biologic determinants of P300: an with DPZ in AD. integrative review. Bio Psychol 1995; 41:103-146. We also found that the ADAS–J cog score was signif- 7. Meador KG, Loring DW, Sethi KD, et al. Cholinergic and sero- icantly correlated with P300 latency at baseline and on com- tonergic effects on the P3 potential and recent memory. J Clin pletion of the study (6 months). Correlation between Exp Neuropsychol 1989; 11:252-260. latency of P300 ERPs and cognitive dysfunction appears 8. Bryson HM, Benfield P.Donepezil. Drugs & Aging 1997; 10:234- 239. to be well established: P300 latency is correlated negatively 9. McKhann G, Drachman D, Folstein M, et al. Clinical diagnosis with mental function in normal subjects, and severe delays of Alzheimer’s disease: reports of the NINCDS-ADRDA Work are observed in patients with AD in comparison with age- Group under the auspices of Department of Health and Human matched control subjects.13,16 These results suggest that Services Task Force on Alzheimer’s Disease. Neurology 1984; P300 latency may serve as a marker for measuring the sub- 34:939-944. clinical and clinical course of AD. 10. Clair DS, Blackburn I, Blackwood D, Tyrer G. Measuring the course of Alzheimer’s disease: a longitudinal study of neu- We should note that there were no elderly normal- ropsychological function and changes in P3 event-related poten- control subjects treated with DPZ, and the specificity of tial. Br J Psychiatr 1988; 152:48-54. these observations for AD is unknown. Furthermore, this 11. Homma A, Fukuzawa K, Tsukada Y, et al. Development of a was not a double-blind, placebo-controlled study. How- Japanese version of Alzheimer’s disease assessment scale (ADAS). Jpn J Geriatr Psychiatry 1992; 3:647-655. ever, we can conclude that P300 latency gives very useful 12. Rogers SL, Doody RS, Mohs RC, Friedhoff LT,and the Donepezil information on the progression of AD, especially in the lon- Study Group. Donepezil improves cognition and global func- gitudinal follow-up of patients with AD during treatment tion in Alzheimer disease. Arch Intern Med 1998; 158:1021- with DPZ acting on cholinergic pathways. 1031. 13. Thomas A, Iacono D, Bonanni L, et al. Donepezil, rivastigmine, References and vitamin E in Alzheimer disease: a combined P300 event- related potential/neuropsychologic evaluation over 6 months. Clin 1. Francis PT, Palmer AM, Snape M, Wilcock GK. The cholinergic Neuropharmacol 2001; 24:31-42. hypothesis of Alzheimer’s disease: a review of progress. J Neu- rol Neurosurg Psychiatry 1999; 66:137-147. 14. Homma A, Takeda M, Imai Y, et al. Clinical efficacy and safety of donepezil on cognitive and global function in patients with 2. Mohs RC, Rosen WG, Davis KL. The Alzheimer’s Disease Assess- Alzheimer’s disease: a 24-week, multicenter, double-blind, ment Scale: an instrument for assessing treatment efficacy. Psy- placebo-controlled study in Japan. Dement Geriatr Cogn Disord chopharmacol Bull 1983; 19:448-450. 2000; 11:299-313. 3. Folstein MF, Folstein SE, McHugh PR. “Mini-Mental State”: a 15. Reeves RR, Struve FA, Patricks G, et al. The effects of donepezil practical method for grading the cognitive state of patients for on the P300 auditory and visual cognitive evoked potentials of the clinician. J Psychiat Res 1975; 12:189-198. patients with Alzheimer’s disease. Am J Geriatr Psychiatry 4. Jagust WJ. Neuroimaging in dementia. Neurol Clin 2000; 18:885- 1999; 7:349-352. 902. 16. Polich J, Ladish C, Bloom FE. P300 assessment of early 5. Fernandez LA, Penzol DF, Morales RM. Neurophysiological Alzheimer’s disease. Electroenceph Clin Neurophysiol 1990; study and use of P300 evoked potentials for investigation in the 77:179-189. Age-Related Impairments in Comprehending Affective Prosody With Comparison to Brain-Damaged Subjects

Diana M. Orbelo, PhD, Julie A. Testa, PhD, and Elliott D. Ross, MD

ABSTRACT

This study evaluated the ability to produce and comprehend affective prosody across age groups and compared patterns of impaired performance to deficits observed after focal brain damage. Sixty-nine healthy subjects, ages 22 to 83 years, were given the Aprosodia Battery, a test that distinguishes between affective prosodic processing deficits following right- versus left-brain damage through the use of stimuli with progressively reduced verbal articulatory content. Production of affective prosody, measured by variation in fundamental frequency, was unimpaired in older subjects, whereas comprehension of affective prosody was impaired, particularly for tasks with reduced verbal articulatory content. The pattern of performance across affective comprehension tasks in the older subjects resembled the pattern found after right-brain damage. The results demonstrate age-related loss in comprehension of affective prosody that is most likely due to a processing deficit involving the right hemisphere. (J Geriatr Psychiatry Neurol 2003; 16:44–52)

Keywords: affective prosody; aging; auditory comprehension; stroke

Affective prosody is the “melody of speech” that provides duration), quality (measured as spectral information), emotional and attitudinal information during discourse. and variations in syllable structure8 also provide acoustic The acoustic features of affective prosody include aspects cues to affective prosodic intent.9,10 of frequency, duration, and intensity.1 Varying funda- Although affective prosody appears to be a dominant 7,11,12 mental frequency (Fo) over time is perceived by listeners function of the right hemisphere, production and as pitch or intonation variation and has been shown to be comprehension deficits have been reported following both one of the most salient cues to determining the affective right-brain damage (RBD) and left-brain damage (LBD).13-15 intent of an utterance.2-6 The absence of normal pitch Such deficits have been explained in various ways. For variation in speech causes individuals to sound “flat,” as example, brain injury may cause individuals to rely pref- if they have no feeling even though they may experience erentially on one set of acoustic cues opposed to another.5 normal emotion.3,7 Although pitch variation is important, Van Lancker and Sidtis5 found that RBD patients relied loudness (measured as intensity), timing (measured as mostly on durational cues whereas LBD patients relied

mostly on Fo variability,but these findings were not repli- cated by Pell and Baum.6 Also, in LBD patients with dense Received September 5, 2001. Received revised May 17, 2002. Accepted for aphasias, the severity of aphasic comprehension deficits publication May 22, 2002. has been correlated with the severity of affective prosodic From the Departments of Neurology, Communication Sciences and comprehension deficits,16,17 suggesting that the verbal and Disorders (Drs. Orbelo and Ross), and Psychiatry and Behavioral Sciences articulatory impairments associated with aphasia might (Drs. Testa and Ross), University of Oklahoma Health Sciences Center; 15 and the VA Medical Center (Drs. Orbelo and Ross), Oklahoma City,Oklahoma. explain impaired affective prosodic performance. In an attempt to avoid the confounding contribution This work was supported by a Merit Review Grant from the Medical Research Service of the Department of Veterans Affairs, Washington, DC, of aphasic deficits, an Aprosodia Battery was developed and in part by an Oklahoma Center for the Advancement of Science and using stimuli that are reduced progressively in their ver- Technology grant, Oklahoma City,Oklahoma, to Dr. Ross. We thank Sarah bal articulatory content.15 It was assumed that reducing Christman, PhD, William Beatty,PhD, and William Lovallo, PhD, for their comments. the verbal articulatory content causes the tasks to shift from bihemispheric to primarily right hemispheric pro- Reprint requests: Diana Orbelo or Elliot Ross, VA Medical Center (11AZ), 921 NE 13th Street, Oklahoma City, OK 73104; e-mail: diana- cessing. This assumption is consistent with findings from [email protected]. psychoacoustical studies that suggest a specialized role for DOI: 10.1177/0891988702250565 the right auditory cortical areas in processing pitch direc-

44 © 2003 Sage Publications Age-Related Impairments in Comprehending Affective Prosody With Comparison to Brain-Damaged Subjects / Orbelo et al. 45 tion18 and a specialized role for the left hemisphere in control subjects seemed to have difficulty comprehending processing temporal information.19 affective prosody compared to younger subjects. A litera- In the Aprosodia Battery, acoustic cues to affective ture review showed that other investigators have reported prosody are reduced along with the reduction of verbal age-related impairments in the recognition of auditory 11,27-30 28 articulatory content. The word stimuli have a full com- affect. Brosgole and Weisman compared 6 groups plement of prosodic cues, including variations in pitch, with ages spanning from 3 to 83 years on tasks of audi- rhythm, loudness, and voice quality along with normal syl- tory and facial affect recognition and discrimination, as well labic structure, thus recruiting both left and right hemi- as recognition of nonemotional voice inflections. They spheres for processing. The monosyllabic stimuli also found that the ability to identify facial affect, nonemotional include a full complement of acoustic cues to affective voice inflections, and emotional voice intonations began to prosody,although the syllabic information is no longer lin- decline in their 45- to 64-year-old group with further wors- guistically meaningful. Finally, the asyllabic stimuli con- ening in their 65- to 83-year-old group. The authors pos- 31 sist mainly of suprasegmental variations in pitch, loudness, tulated, based on their findings and others, that plaque and voice quality that are carried over the entire utterance deposits may be the underlying cause of age-related with limited rhythmic and no segmental cues. The asyl- changes in mood recognition and that the deficits might 28 labic condition therefore taxes predominantly right hemi- represent an early sign of impending dementia. sphere processing. The goals of the present study were (1) to explore the Using the Aprosodia Battery, 2 distinct patterns of relationship between Aprosodia Battery scores and age and affective prosodic performance were observed in brain- (2) to compare impaired patterns of performance found in damaged subjects.19 LBD performance, for both compre- apparently healthy subjects to those observed after RBD hension and production, improved relative to control and LBD. subjects as the verbal-articulatory content of the test stimuli was reduced. In contrast to previous findings in METHODS densely aphasic patients,16,17 affective prosodic deficits in the LBD group did not correlate with aphasic deficits. Subjects This difference may have been due to subject selection The primary study group included 69 control subjects (29 because some of the LBD patients had mild or no apha- men, 40 women) from our previous clinical studies of 15,24,26 sic deficits.15 In this less impaired group of LBD patients, affective prosody. Demographic information is pro- the presence of deep white matter lesions adjacent to the vided in Table 1. All subjects were free from medical con- corpus callosum best predicted affective prosodic deficits. ditions known to cause cognitive deficits, such as history Thus, affective-prosodic deficits after LBD were attributed of major psychiatric (eg, schizophrenia, bipolar illness), to a disruption of callosal integration between affective medical (eg, serious cardiovascular disease), or neurologic prosody centers of the right hemisphere and propositional (eg, head trauma, vascular dementia) conditions. Sub- language centers of the left hemisphere. Conversely,RBD jects were also free from obvious hearing loss. All control performance for both production and comprehension subjects scored 27 or better on the Mini-Mental State 32 remained impaired or worsened as the verbal-articulatory Exam. Individuals were excluded if taking medications content of test stimuli was reduced with the cortical rather that could interfere with testing, such as neuroleptics, than the deep distribution of the lesions best predicting antidepressants, high-dose β-blockers, or benzodiazepines. affective-prosodic deficits.20 Thus, the affective prosodic All subjects were native speakers of English and right deficits after RBD were attributed to a disruption of the handed based on self-report and the Edinburgh Handed- right hemisphere’s dominant processing of affective ness Scale.33 prosody, a conclusion also reached by other investigators Because the second goal of this study was to compare using different test methods.11,12,21-23 patterns of performance in apparently healthy subjects to While studying processing of affective prosody in var- those found after brain damage, data from 37 brain- ious clinical populations,20,24-26 we observed that our older damaged patients, 19 of whom have been reported else-

Table 1. Demographic Information for Experimental Subjects

Normal Subjects Brain-Injured Subjects

Demographics Entire Group Young Middle Elderly Left-Brain Right-Brain (N = 69) (n = 38) (n = 11) (n = 20) Damaged (n = 17) Damaged (n = 20)

Age in years (Mean ± SD) 58 ± 20 30 ± 6 55 ± 7 74 ± 5 62 ± 12 64 ± 10 Years of educationa (Mean ± SD) 14 ± 3 15 ± 2 13 ± 2 13 ± 3 12 ± 3 12 ± 2 Gender (male/female) 29/40 9/11 3/8 12/21 11/6 14/6 a. Young subjects had significantly more years of education than the left-brain-damaged and right-brain-damaged subjects did, F (4, 101) = 4.01, P = .005. 46 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003 where,34 were included. The brain-damaged subjects met been found that in speakers of English, a nontone language, the above criteria with the exception that each had expe- it correlates with clinical loss of affective prosody after brain 15 rienced a unilateral ischemic infarction documented by injury. If individuals lose general affective expression, we magnetic resonance imaging involving either the right would expect to find lower mean Fo CV% compared to the (n = 20) or left (n = 17) hemisphere (see Table 1 for demo- younger subjects. There is, however, no evidence in the lit- graphic information). Testing of the brain-damaged patients erature to suggest a change in affective production with age. was completed within 4 to 8 weeks postictus to exclude acute deficits due to diaschisis and improvement from Comprehension spontaneous recovery due to neural reorganization. Comprehension of affective prosody was assessed using 3 identification tasks—word, monosyllabic, and asyllabic— Aprosodia Battery and an affective discrimination task. The stimuli for the 3 identification tasks were the same as those described above Production for repetition. The 12 stimuli were randomized and pre- Production of affective prosody was assessed by tape sented twice for 24 trials per task. Subjects were asked to recording subjects on 3 tasks of affective repetition and on identify the emotional intonation of each stimulus by choos- a spontaneous production task. The repetition stimuli ing the appropriate affect from a vertical array of 6 line consisted of 3 sets of randomized utterances reduced pro- drawings of faces expressing different affects. Next to gressively in verbal articulatory content. Each set consisted each face was the corresponding written label of “neutral,” of 12 utterances, 2 renditions each of 6 emotions (happy, “happy,”“sad,”“disinterested,”“surprised,” and “angry.” Before sad, disinterested, neutral, surprised, and angry), with 1 testing, each subject demonstrated the ability to identify rendition having emphatic stress early in the utterance the facial expressions and to read the written labels. and the other rendition having emphatic stress late in the The affect discrimination stimuli were the same as utterance. The carriers were “I am going to the other those used for word repetition after band-pass filtering movies” for word repetition, “ba ba ba ba ba ba” for mono- between 70 and 300 Hz (using a Krohn-Hite Variable fil- syllabic repetition, and “aaaaahhhhh” for asyllabic repe- ter [Model 3550]), a process that distorted the phonetic tition. Subjects were instructed to listen to each stimulus information while leaving overall prosodic information 35 and then repeat what they heard using the same tone of intact. Twenty-four pairs of stimuli were recorded; 12 voice. They were allowed 4 practice utterances before each pairs had the same affective intonation with different set of stimuli. All stimuli were played in a free sound field stress patterns, and 12 had different intonation with the at a comfortable loudness level using a prerecorded audio same stress pattern. Subjects were asked to indicate compact disc and a Sony compact disc player (Sony Elec- whether the represented emotions were the same or dif- tronics Inc., Park Ridge, NJ). ferent for each pair. Subjects were expected to perform Spontaneous production of affective prosody was poorly when basing their answers on stress rather than assessed by recording each subject during a 5- to 10- affective prosodic information. minute interview, during which they were encouraged to The scores for each of the comprehension tasks were discuss various emotionally provocative personal life the total number of correct responses out of 24. events that made them feel happy,frightened, angry,or sad. Subject responses were recorded on analogue tape Data Analysis using a Marantz PMD 340 tape recorder and a Shure To address the first goal of this study, regression analy- SM12A microphone mounted on an adjustable boom ses were used to explore the relationship between age attached to a headset. The microphone was positioned and scores on the Aprosodia Battery tasks. To address the just to the side of the subject’s air stream to avoid record- second goal of this study, the primary study group was ing artifacts from head movement and to ensure a high divided into 3 age groups—young (ages 22 to 44 years), mid- signal-to-noise ratio. dle (ages 45 to 64 years), and elderly (ages 65 to 83)—cor- Utterances were evaluated using a PM Pitch Analyzer responding to the adult groups used by Brosgole and 28 (Voice Identification Inc., Somerville, NJ), which extracted Weisman. Then, patterns of impaired affective prosodic processing were explored across relevant tasks using fundamental frequency (F0) in Hertz (Hz). Next, the mean, repeated-measure ANOVAs. Finally, any impaired per- standard deviation, and coefficient of variation (CV) of Fo formance was compared to patterns found after LBD and were calculated for each utterance. Finally,a mean Fo CV RBD using paired repeated-measure ANOVAs. Alpha was percentage (Fo CV%) was calculated for all the utterances within each repetition task. The same analyses were con- set at .05 for all analyses. Statistical analyses were car- ducted on a total of 10 seconds of each subject’s sponta- ried out using SPSS 8.0 for Windows. neous production, using phrases 2.5 to 4 seconds long, taken In the original study using the Aprosodia Battery, from sections of discourse involving recall of emotional sit- subject raw scores were converted to z scores by sub- uations. Fundamental frequency variation was chosen as tracting the control group mean score from the individual an objective measure of the affect in speech because it has subject score and then dividing by the control group stan- Age-Related Impairments in Comprehending Affective Prosody With Comparison to Brain-Damaged Subjects / Orbelo et al. 47 dard deviation.15 This removed the variability in per- 45 formance across the repetition and comprehension tasks Word Rep. 40 attributable to unknown factors in the control group while y = 0.0455x + 19.839 R2 = 0.062 leaving the variability attributable to brain damage intact. 35 Because the focus of the present study is on healthy subjects’ performance, all results are presented as raw rather 30 than z scores. 25 Mean Fo CV % RESULTS 20

Preliminary Analysis 15 To screen for the presence of any unexpected predictive 20 30 40 50 60 70 80 value for gender or education, a preliminary regression Age in Years analysis was conducted that included these factors along with age in the model. Education did not predict performance on any of the Aprosodia Battery tasks; thus, it 45 was dropped from further analyses. Gender was predictive Asyllabic Rep. 2 of asyllabic repetition (R = 0.116; df = 1, 67; P = .004); there- 40 y = 0.07x + 22.9 fore, regression models predicting asyllabic repetition R2 = 0.09 scores with age were run for each gender group sepa- 35 rately.Within each gender, the predictive value of age for 30 asyllabic repetition was virtually identical for women 2 2 (R = 0.111, r = +0.333) and men (R = 0.117, r = +0.342). 25 Mean Fo CV % Although the relationship between gender and asyllabic repetition is an interesting one that may deserve further 20 attention, because the focus in the present study was on age and the relationship between age and asyllabic rep- 15 etition was similar for both genders, gender was not con- 20 30 40 50 60 70 80 sidered in further analyses. Age in Years Figure 1. Affective prosodic repetition as a function of age. Aprosodia Battery Scores and Age Age was found to be a significant predictor of Aprosodia Table 2. Regression Analysis of Battery scores on 6 of the 8 tasks (Table 2). For produc- Aprosodia Battery Scores Using Age tion, age was related to word and asyllabic repetition 2 (Figure 1). Because these relationships were positive and r Value R F Value P Value 2 statistically small (R < 0.10), there was no indication a Word repetition +0.249 0.062 4.430 .039 that age impairs the ability to produce affective prosody Monosyllabic repetition +0.004 0.000 0.001 .972 in speech. Asyllabic repetitiona +0.296 0.088 6.440 .013 Spontaneous production +0.074 0.005 0.367 .547 For comprehension, age was negatively related to all a Word comprehension –0.274 0.075 5.446 .023 of the comprehension tasks (Table 2). No sharp changes Monosyllabic comprehensionb –0.458 0.210 17.758 < .001 occurred with age (Figure 2). Instead, there was a grad- Asyllabic comprehensionb –0.650 0.423 49.056 < .001 b ual fall of test scores with greater variability in test per- Affect discrimination –0.401 0.160 12.803 .001 formance as age increased. This finding was most apparent a. Age was predictive at significance levels less than .05. for asyllabic and monosyllabic identification and least b. Age was predictive at significance levels less than .01. apparent for word identification and affective discrimination. The best-fitting trend lines were linear for all tasks except asyllabic identification, which was curvilin- Patterns of Performance ear.Trend line equations and R2 values are noted on each To further explore patterns of impairment on affective com- graph. Ceiling effects, observed for all the tasks, may have prehension, the performance across identification tasks for influenced the slope of the trend lines and may account the 3 age groups (Figure 3) were compared using a for the curvilinear best fit for asyllabic identification. repeated-measure ANOVA. The ANOVA yielded a signif- However, it should be noted that these ceiling effects icant main effect for task, F(2, 132) = 29.30, P < .001, and make it more rather than less difficult to observe differ- group, F(2, 66) = 20.29, P < .001, and a significant group ences related to age. Based on regression analyses, there by task interaction, F(4, 132) = 7.50, P < .001. Post hoc tests appears to be impairment of affective prosodic compre- comparing groups within each identification task showed hension associated with age. no significant differences among the age groups on word 48 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003

24 24 22 Young 20 22 18

16 20 Middle 14 Word ID Number Correct 12 y = -0.03x + 23.1 18 2 10 R = 0.0733 Elde rly 8 16 20 30 40 50 60 70 80 Age in Years 14 Word ID Monosyllabic ID Asyllabic ID 24

22 Figure 3. Mean affective prosodic identification scores for the 3 age groups. 20

18 interactions on paired repeated-measure ANOVAs (Table 3), 16 the two older groups were found to have similar patterns 14 Monosyllabic ID of performance across identification tasks (no significant Number Correct 12 y = -0.06x + 23.8 interaction). However, there was a main effect for group 2 10 R = 0.21 indicating that the elderly group was more impaired than 8 the middle group. The young group’s performance across 20 30 40 50 60 70 80 tasks was different from both the middle and elderly groups Age in Years based on the presence of significant interactions. Finally, the patterns of performance from the age 24 groups were compared to those found following brain 22 damage (Figure 4). Based on presence or absence of sig- 20 nificant interactions on the paired repeated-measure

18 ANOVAs (Table 4), similar patterns of performance were

16 observed for the elderly, middle, and RBD groups, and similar patterns of performance were observed for the 14 Asyllabic ID

Number Correct young and LBD groups. 12 2 y = -0.003x + 0.17x + 18.6 2 10 R = 0.46 DISCUSSION 8 20 30 40 50 60 70 80 The primary goal of the present study was to explore the Age in Years relationship between age and the ability to produce and comprehend affective prosody.The results suggest that pro- 24 duction of affective prosody remains intact with age, while 22 comprehension of affective prosody is impaired in the 20 older relative to younger subjects. The data in the pres- 18 ent study showed a small positive correlation with the 16 amount of fundamental frequency variation produced by subjects on word and asyllabic repetition, while no rela- 14 Discrimination

Number Correct y = -0.04x + 24.2 tionship between age was found for monosyllabic repeti- 12 2 R = 0.16 tion or spontaneous production. Although the positive 10 correlations with repetition and age are interesting, the 8 findings do not indicate any impairment in the ability to 20 30 40 50 60 70 80 produce affect as a function of age. It may be that speech Age in Years is such an overlearned and well-developed skill that it Figure 2. Affective prosodic comprehension as a function of age. remains relatively impervious to the aging process. The ability to comprehend affective prosody,however, identification, but they did show a significant difference showed a consistent negative relationship with age. Per- between the young and elderly groups on monosyllabic formance on all 4 of the comprehension tasks was nega- identification and among all 3 groups on asyllabic iden- tively correlated to age, with the effect being most tification. Based on the presence or absence of significant impressive for asyllabic identification. The asyllabic task Age-Related Impairments in Comprehending Affective Prosody With Comparison to Brain-Damaged Subjects / Orbelo et al. 49

Table 3. Repeated-Measure ANOVAs for Age-Group Pairs on Affective Identification

Age-Group Pair Interaction for Group by Task Main Effect for Group Main Effect for Task

Young–middle F (2, 58) = 3.62, P = .033 F (1, 29) = 7.25, P = .012 F (2, 58) = 11.94, P < .001 Young–elderly F (2, 112) = 14.14, P < .001 F (1, 56) = 38.78, P < .001 F (2, 112) = 27.95, P < .001 Middle–elderly F (2, 94) = 2.11, P = .127 F (1, 47) = 6.48, P = .014 F (2, 94) = 27.30, P < .001

Table 4. Repeated-Measure ANOVAs for Age- and Brain Damage–Group Pairs on Affective Identification

Age- and Brain Damage–Group Pair Interaction for Group by Task Main Effect for Group Main Effect for Task

Young–LBD F(2, 70) = 1.28, P = .285 F(1, 35) = 64.80, P < .001 F(2, 70) = 3.295, P = .043 Middle–LBD F(2, 52) = 3.19, P = .049 F(1, 26) = 17.43, P < .001 F(2, 52) = 6.15, P = .004 Elderly–LBD F(2, 106) = 11.57, P < .001 F(1, 53) = 8.46, P = .005 F(2, 106) = 20.68, P < .001 Young–RBD F(2, 76) = 11.44, P < .001 F(1, 38) = 94.10, P < .001 F(2, 76) = 25.66, P < .001 Middle–RBD F(2, 58) = 1.05, P = .357 F(1, 29) = 34.10, P < .001 F(2, 58) = 26.28, P < .001 Elderly–RBD F(2, 112) = 0.41, P = .663 F(1, 56) = 44.24, P < .001 F(2, 112) = 59.51, P < .001

LBD = left-brain damaged; RBD = right-brain damaged.

24 The second goal of this study was to compare patterns Young of performance across tasks that are impaired with age, 22 specifically affective identification. To do this, the healthy study subjects were split into 3 groups after Brosgole and 20 Middle Weisman,28 and their performance across tasks was compared to brain-damaged subjects. First, we noted that the 18 Elderly middle (ages 45-64) and elderly (ages 65–83) groups were LBD 16 impaired relative to the young group on tasks of affective prosodic identification, with the middle group being inter- 14 mediate and the elderly group performing the worst, a find- 28 RBD ing consistent with those of Brosgole and Weisman. Second, 12 we noted that the pattern of performance exhibited by the 2 older groups was similar to that of the RBD group, whereas 10 the young group was similar to the LBD group. The 2 Word ID Monosyllabic ID Asyllabic ID older groups’ and the RBD groups’ performance declined Figure 4. Mean affective prosodic identification scores for the 3 age across tasks as the verbal articulatory content of test stim- groups and left-brain-damaged and right-brain-damaged groups. uli was reduced, whereas the performance of the young and LBD = left-brain damaged; RBD = right-brain damaged. LBD groups remained consistent across tasks. Thus, the impaired pattern of performance across identification uses stimuli with the least amount of verbal cues to affect tasks found in the older groups appears consistent with and no cues associated with segmentation. Subjects must a right rather than a left hemisphere processing disorder. make affective distinctions based primarily on changes in Although our subjects had no overt hearing difficulty, pitch, a task modulated predominately by the right hemi- hearing loss due to changes in both the peripheral and cen- sphere.36 Although intonation or pitch variation over time tral auditory system are prevalent in the elderly.37 Thus, has been shown to be one of the most salient acoustic it is possible that the older subjects were not able to hear cues for signaling affective prosody in spoken English, a the pitch variations well enough to make affective iden- nontone language,3,7 older individuals may be relying tifications. In past research, it has been assumed that more heavily than younger individuals on cues such as hearing deficits are not a contributing factor to age-related rhythm, loudness, and voice quality or may need greater deficits in affect recognition because subjects had no obvi- redundancy of cues to make affective determinations. ous difficulty with verbal communication in test situations Either way, the current results suggest that when into- and they passed screening tests involving verbal instruc- nation is the primary cue to affective prosody, the elderly tions.30 Likewise, in the present study,intact performance exhibit more difficulty than do the young. This may reflect on the affective repetition tasks suggests that test stim- a right hemisphere processing disorder based on the idea uli were presented at audible levels sufficient for nascent that the Aprosodia Battery changes from a bihemispheric perceptual analysis. task to primarily a right hemispheric task as the verbal If hearing is not an issue, than what could be the articulatory content of test stimuli is reduced. cause of the relatively poor performance on comprehen- 50 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003 sion tasks exhibited by the older individuals? One possi- tive processes.47 Because older subjects performed worse bility is that the tasks simply become more difficult as the on asyllabic identification compared to word identification, stimuli become more unnatural due to reduction in ver- it is unlikely that deficits in working memory can entirely bal articulatory content. As mentioned above, the reduced explain our findings. However, frontal circuits in the right verbal articulatory content probably forces subjects to hemisphere, important for the retention and comparison rely solely on intonational cues in the absence of sup- of pitch information,36 may be compromised in older adults. porting segmental and rhythmic indices of affective prosody. Although the elderly subjects in the present study scored The relatively poor performance on affective prosodic 27 or better on the Mini-Mental State Exam, it is possi- tasks by older subjects compared to younger subjects may ble that deficits in cognitive and executive function could imply a difference in cognitive strategy employed by the be found using more sensitive neuropsychological assess- older subjects. ment tools that might partially explain the results of the Differences in cognitive strategy have been inferred present study. from neuroimaging studies showing activation of differ- The idea that the right hemisphere ages faster than ent brain areas when young and elderly perform identi- the left is controversial. Proponents of the right hemi- 38 cal tasks. More specifically, imaging studies suggest sphere aging hypothesis cite evidence from neuropsycho- 39 greater use of frontal cognitive systems with age. This logical studies reporting deterioration in nonverbal recruitment of different and, by extension, possibly less effi- compared to verbal skills48 and increased lateralization of cient brain areas may result in either equivalent or language functions to the left hemisphere with age.49 In impaired levels of task proficiency in elderly compared to contrast, studies involving visual perception of emotional 40,41 young subjects. Such recruitment has been postulated words50 and the production and perception of right and left to function as a compensatory mechanism for reduced hemifacial emotions51,52 have suggested a continued and efficiency in other brain areas that occurs with increas- competent role for the right hemisphere in modulating 42 ing age. If in the present study older individuals are affective behavior across the life span. However, recent relying mostly on the temporal cues associated with seg- studies probing the ability to process facial blends of emo- mentation to infer affective intent, and the processing of tion have reported age-related deficits in the ability to temporal cues are modulated primarily by the left hemi- attend and perceive emotion conveyed on the upper half 19 sphere, then the older subjects may be “compensating” of the face,53,54 a finding that was consistent with right for right hemisphere deficits by using a more left hemi- hemisphere aging. Additional hypotheses posit that both sphere strategy. Our findings of affective prosodic com- hemispheres age equally55,56 or that the frontal lobes are prehension deficits in the current study suggest that if the focus of age effects.57 elderly subjects are using a different cognitive strategy,it Lastly, an important question to ask is, How do age- is not sufficient to bring performance levels to those of the related findings in the comprehension of affective prosody young subjects. affect elderly people in everyday communication situations? If a difference or a deficit in cognitive strategy is the The aging effect observed on regression analysis for word underlying cause for the relatively poor performance identification was relatively small compared to those exhibited by the elderly subjects in the present study, found for monosyllabic and asyllabic identification. Using then what brain areas may be affected in the elderly? tests similar to word identification, others have found Assuming right hemisphere dominance for affective that only a subgroup of healthy elderly was impaired on 7,15,22,43 prosody, it is possible that aging adversely alters affect recognition.27,30 Thus, many elderly people may be right hemisphere processing. Based on previous work in able to interpret basic emotions such as happiness, sad- 15,20,44,45 brain-damaged individuals, our finding of pre- ness, and anger in typical communication situations that served production of affective prosody across ages suggests involve a synthesis of affective prosodic cues with context that the perisylvian and related frontal language areas of and facial expression. This argument, however, should be the right hemisphere remain intact. However, the finding made cautiously because age-related impairments have of impaired comprehension of affective prosody in the eld- also been reported for comprehension of facial53,58 and lex- erly is similar to findings obtained from subjects described ical59 emotions. In addition, the ability of elderly subjects as having transcortical sensory aprosodia, which has been to interpret the subtler cues underlying attitudinal prosody associated with lesions involving the posterior parasylvian that convey important psychosocial information in every- cortices of the right hemisphere. day situations, such as sarcasm, tacit approval, and dis- Another possibility is that the impairment exhibited approval, needs exploration. by the older subjects on comprehension tasks may be due, Regardless of our speculation about underlying mech- at least in part, to age-related deficits in working mem- anisms, the results have empirical validity.Age-associated ory or executive function. Working memory has been asso- loss of affective prosodic comprehension, coupled with a 46 ciated with frontal and parietal structures and includes possible reduction in the ability to detect facial and lexi- 3 levels of processing: one that acts as a buffer for storage, cal emotion, carries certain implications. An elderly per- one for rehearsal, and an executive one for higher cogni- son’s perception of social and emotional support may be Age-Related Impairments in Comprehending Affective Prosody With Comparison to Brain-Damaged Subjects / Orbelo et al. 51 attenuated if the affective aspects of communication involv- 22. Bowers D, Bauer R, Heilman K. The nonverbal affect lexicon: ing social interaction are missed or misinterpreted. Con- theoretical perspectives from neuropsychological studies of affect perception. Neuropsychology 1993; 7:433-444. sequently, they may be less adept at interacting with 23. Schmitt JJ, Hartje W,Willmes K. Hemispheric asymmetry in the family, friends, and professionals and may be at a disad- recognition of emotional attitude conveyed by facial expres- vantage during deceitful situations. sion, prosody and propositional speech. Cortex 1997; 33:65-81. 24. Testa J, Beatty W, Gleason A, et al. Impaired affective prosody References in AD: relationship to aphasic deficits and emotional behaviors. 1. Lehiste I. Suprasegmentals. Cambridge, MA: MIT Press, 1970. Neurology 2001; 57:1474-1481. 2. Kent RD, Rosenbeck JC. Prosodic disturbances and neurologic 25. Monnot M, Nixon S, Lovallo W, Ross E. Altered emotional per- lesion. Brain Lang 1982; 15:259-291. ception in alcoholics: deficits in affective prosody comprehension. Alcohol Clin Exp Res 2001; 25:362-369. 3. Ross ED, Edmondson JA, Seibert GB, Homan RW. Acoustic analysis of affective prosody during right-sided wada test: a 26. Ross E, Burgard M, Hansel S, Swenson R. Leukoaraiosis is not within-subjects verification of the right hemisphere’s role in lan- associated with loss of interhemispheric functions or dementia. guage. Brain Lang 1988; 33:128-145. Neurology 1997; 48(suppl 2):A168. 4. Shapiro B, Danly M. The role of the right hemisphere in the con- 27. Cohen ES, Brosgole L. Visual and auditory affect recognition in trol of speech prosody in propositional and affective contexts. senile and normal elderly persons. Int J Neurosci 1988; 43:89-101. Brain Lang 1985; 25:19-36. 28. Brosgole L, Weisman. Mood recognition across the ages. Int J 5. Van Lancker D, Sidtis JJ. The identification of affective-prosodic Neurosci 1995; 82:169-189. stimuli by left- but not right-hemisphere-damaged patients. 29. Oscar-Berman M, Hancock M, Mildworf B, et al. Emotional per- Brain Lang 1992; 32:256-277. ception and memory in alcoholism and aging. Alcohol Clin Exp 6. Pell M, Baum S. Unilateral brain damage, prosodic compre- Res 1990; 14:383-393. hension deficits, and the acoustic cues to prosody. Brain Lang 30. Allen R, Brosgole L. Facial and auditory affect recognition in 1997; 57:195-214. senile geriatrics, the normal elderly and young adults. Int J Neu- 7. Ross ED. Affective prosody and the aprosodias. In: Mesulam M, rosci 1993; 68:33-43. ed. Principles of behavioral and cognitive neurology. New York: 31. Kurucz J, Charbonneau R, Kurucz A, Ramsey P. Quantitative Oxford University Press, 2000:316-331. clinicopathologic study of senile dementia. J Am Geriatr Soc 1981; 8. Lakshminarayanan K, Ben Shalom D, van Wassenhove V,et al. 29:158-163. The effect of spectral manipulations on identification of active 32. Folstein M, Folstein S, McHugh P.“Mini-Mental State”: a prac- and linguistic prosody. Brain Lang. In press. tical method for grading the cognitive state for the clinician. J 9. Scherer KR. Vocal affect expression: a review and a model for Psychiatr Res 1975; 12:189-198. future research. Psychol Bull 1986; 99:143-165. 33. Oldfield RC. The assessment and analysis of handedness: the 10. Williams CE, Stevens KN. Emotions and speech: some acousti- Edinburgh inventory. Neuropsychologia 1971; 9:97-113. cal correlates. J Acoust Soc Am 1972; 52:1238-1250. 34. Ross E, Orbelo D, Cartwright J, et al. Affective-prosodic deficits 11. Tompkins CA, Flowers CR. Perception of emotional intonation in schizophrenia: profiles of patients with brain damage and com- by brain-damaged adults: the influence of task processing lev- parison with relation to schizophrenic symptoms. J Neurol Neu- els. JSHR 1985; 28:527-538. rosurg Psychiatry 2001; 70:597-604. 12. Borod JC, Cicero BA, Obler LK, et al. Right hemisphere emo- 35. Lenhardt ML. Factors in affective discrimination of speech. tional perception: evidence across multiple channels. Neu- Neurology 1978; 28:309-310. ropsychology 1998; 12:445-458. 36. Zatorre R, Evans A, Meyer E, Gjedde A. Lateralization of pho- 13. de Bleser R, Poeck K. Analysis of prosody in spontaneous speech netic and pitch discrimination in speech processing. Science of patients with CV-recurring utterances. Cortex 1985; 21:405- 1992; 256:846-849. 416. 37. Willott JF. Aging and the auditory system anatomy, physiology, 14. Cancelliere AEB, Kertesz A. Lesion localization in acquired and psychophysics. San Diego, CA: Singular, 1991. deficits of emotional express and comprehension. Brain Cogn 38. Grady C, Craik F. Changes in memory processing with age. 1990; 13:133-147. Curr Opin Neurobiol 2000; 10:224-231. 15. Ross ED, Thompson RD, Yenkosky J. Lateralization of affective 39. Grady C. Brain imaging and age-related changes in cognition. prosody in brain and the callosal integration of hemispheric lan- Exp Gerontol 1998; 33:661-673. guage functions. Brain Lang 1997; 56:27-54. 40. Horwitz B, McIntosh A, Haxby J, et al. Network analysis of 16. Schlanger BB, Schlanger P,Gerstmann LJ. The perception of emo- PET-mapped visual pathways in Alzheimer type dementia. Neu- tionally toned sentences by right hemisphere-damaged and roreport 1995; 6:2287-2292. aphasic subjects. Brain Lang 1976; 3:396-403. 41. Madden D, Turkington T, Provenzale J, et al. Adult age differ- 17. Seron X, Van der Kaa MA, Vanderlinden M, et al. Decoding ences in the functional neuroanatomy of verbal recognition paralinguistic signals: effect of semantic and prosodic cues on memory. Hum Brain Mapp 1999; 7:115-135. aphasic comprehension. J Commun Disord 1982; 15:223-231. 42. Grady C. Functional brain imaging and age-related changes in 18. Johnsrude IS, Penhune VB, Zatorre RR. Functional specificity cognition. Biol Psychol 2000; 54:259-281. in the right human auditory cortex for perceiving pitch direc- 43. Heilman K, Scholes R, Watson R. Auditory affective agnosia: dis- tion. Brain 2000; 123:155-163. turbed comprehension of affective speech. J Neurol Neurosurg 19. Zatorre RR, Berlin P. Spectral and temporal processing in Psychiatry 1975; 38:59-72. human auditory cortex. Cereb Cortex 2001; 11:946-953. 44. Ross E. The aprosodias: functional-anatomic organization of 20. Ross E, Orbelo D, Burgard M, Hansel S. Functional-anatomic the affective components of language in the right hemisphere. correlates of aprosodic deficits in patients with right brain dam- Arch Neurol 1981; 38:561-569. age. Neurology 1998; 50(suppl 4):A363. 45. Gorelick PB, Ross ED. The aprosodias: further functional- 21. Blonder L, Bowers D, Heilman K. The role of the right hemisphere anatomic evidence for the organization of affective language in in emotional communication. Brain 1991; 114:1115-1127. the right hemisphere. J Neurol Neurosurg Psychiatry 1987; 50:553-560. 52 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003

46. Jonides J, Schumacher EH, Smith EE, et al. The role of parietal 54. Prodan C, Orbelo D, Testa J, Ross E. Hemispheric differences cortex in verbal working memory. J Neurosci 1998; 18:5026-5034. in recognizing upper and lower facial displays of emotion. Neu- 47. Baddeley A. Working memory. Science 1992; 255:556-559. ropsychiatry Neuropsychol Behav Neurol 2001; 14:206-212. 48. Reuter-Lorenz PA. Cognitive neuropsychology of the aging 55. Mittenberg W,Seidenberg M, O’Leary DS, DiGiulio DV.Changes brain. In: Park D, Schwarz N, eds. Cognitive aging, a primer. in cerebral functioning associated with normal aging. J Clin Exp Philadelphia: Psychology Press, 2000:93-114. Neuropsychol 1989; 11:918-932. 49. Brown J, Jaffe J. Hypothesis on cerebral dominance. Neu- 56. Nocentini U, Goulet P, Drolet M, Joanette Y. Age-related evolu- ropsychologia 1975; 13:107-110. tion of the contribution of the right hemisphere to language: 50. Hutner N, Oscar-Berman M. Visual laterality patterns for the absence of evidence. Int J Neurosci 1999; 99:59-67. perception of emotional words in alcoholic and aging individu- 57. Greenwood P. The frontal aging hypothesis evaluated. J Int als. J Stud Alcohol 1996; 57:144-154. Neuropsychol Soc 2000; 6:705-726. 51. Levine S, Levy J. Perceptual asymmetry for chimeric faces 58. McDowell CL, Harrison DW,Demaree HA. Is right hemisphere across the life span. Brain Cogn 1986; 5:291-306. decline in the perception of emotion a function of aging? Int J 52. Moreno C, Borod JC, Welkowitz J, Alpert M. Lateralization for Neurosci 1994; 79:1-11. the expression and perception of facial emotion as a function of 59. Grunwald I, Borod J, Obler L, et al. The effects of age and gen- age. Neuropsychologia 1990; 28:199-209. der on the perception of lexical emotion. Appl Neuropsychol 53. Prodan C, Orbelo D, Ross E. Age-related differences in recognizing 1999; 6:226-238. upper and lower facial displays of emotion. Neurology 2001; 56(suppl 3):A188. Effects of Cognitive Challenge on Gait Variability in Patients With Parkinson’s Disease

Jeffrey M. Hausdorff, PhD, Jackov Balash, MD, PhD, Nir Giladi, MD

ABSTRACT

Attention plays an important role in the gait disturbances of patients with Parkinson’s disease (PD); however, the effects of “dual tasking” on fall risk and gait instability have not been well studied. The authors tested the hypothesis that gait variability increases when subjects with PD walk while performing a cognitively challenging task (CCT). Subjects with idiopathic PD walked under normal conditions and while performing a CCT. During normal walking, gait variability was significantly (.0001 < P < .05) correlated with fall risk, disease duration, disease severity, motor function, mentation, behavior and mood, and cognitive function. When walking while cognitively challenged, gait variability increased (from 47 ± 29 msec to 223 ± 281 msec, P < .002). During dual-task walking, only disease duration remained significantly associated with gait variability.These results highlight the profound effects of attention on gait and indicate that walking while cognitively challenged impairs the ability of patients with PD to maintain a stable walk. (J Geriatr Psychiatry Neurol 2003; 16:53–58)

Keywords: walking; falls; dynamics; dual tasking; attention

Gait disturbances and falls are common in Parkinson’s dis- talking” apparently does not predict falls in patients with ease (PD), often leading to institutionalization and a loss PD. Thus, while it is clear that enhancing attention to gait of independence in the advanced stages of the disease.1-3 via “cueing” improves walking in PD,6 the effects of reduced Gait disturbances typically include slowed walking and attention on gait and fall risk in PD are not yet fully increased gait instability,often manifest by increased stride- understood. to-stride variability in walking.4-6 Despite the incapacitating The present study is motivated by 4 factors: (1) the effects of falls,7 the specific factors that contribute to falls strong relationship between increased gait variability and fall prediction in PD remain elusive.8 Impaired atten- and fall risk observed in other populations,13,14 (2) the tion and a reduced ability to carry out dual tasks may exac- observation that gait variability is increased in PD,4 (3) the erbate gait disturbances and contribute to fall risk in PD,1,9 evidence highlighting the important influences of atten- as is the case in other populations.10,11 Recent work, how- tion on balance and gait,1,15 and (4) the poor understand- ever, suggests that the relationship between fall risk and ing of the factors that contribute to falls in PD.7 Here, we dual-task performance may not be so simple in PD.12 For sought to gain insight into the effects of dual-task per- example, Bloem et al12 found that “stops walking while formance and attention on fall risk and gait instability in PD. More specifically, in this pilot study, we tested the hypothesis that walking while performing a cognitive Received November 16, 2001. Received revised May 2, 2002. Accepted for publication May 13, 2002. task would alter walking rhythmicity, bringing about an increased gait variability in patients with PD. In addition, From the Movement Disorders Unit, Department of Neurology, Tel Aviv Sourasky Medical Center and Sackler School of Medicine, Tel Aviv,Israel we sought to begin to determine which effects of PD (Drs. Hausdorff, Balash, and Giladi), and the Gerontology Division, Beth pathology are associated with gait variability during dual- Israel Deaconess Medical Center and Division on Aging, Harvard Medical task performance. School, Boston (Dr. Hausdorff).

This work was supported in part by NIH grants AG-14100, RR-13622, METHODS AG-08812, and HD-39838.

Reprint requests: Jeffrey M. Hausdorff, PhD, C/O Movement Disorders Subjects Unit, Tel Aviv Sourasky Medical Center, 6 Weizman Street, Tel Aviv 64239, Israel; e-mail: [email protected]. This pilot study is based on the analysis of data from 10 subjects with idiopathic PD. The subjects ranged in age DOI: 10.1177/0891988702250580

© 2003 Sage Publications 53 54 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003 from 52 to 82 years old. Hoehn and Yahr16 stage during test- applied to remove data points that were 3 standard devi- 5,20 ing ranged between 1 and 4 (mean = 3.1). Patients were ations greater than or less than the median value. Sub- receiving routine care for advanced PD in a movement dis- sequently, the average stride time was determined for orders unit. Disease duration ranged between 6 and 26 each segment. Stride time variability,the magnitude of the years (mean = 13.4 years). Medication usage was not stride-to-stride fluctuations in the gait cycle duration, altered. All subjects were tested while in an “on” state (ie, was calculated by determining the standard deviation on medications). All subjects provided informed consent. and the coefficient of variation (CV) of each subject’s stride Cognitive status and subject characteristics were time. The CV assesses the magnitude of the deviations of assessed using the Mini-Mental State Exam (MMSE)17 and the stride time with respect to each subject’s mean value. the Unified Parkinson’s Disease Rating Scale (UPDRS).18 These 2 measures of the magnitude of the stride-to-stride Average total score on the UPDRS during “on” was 33.9 variability,the standard deviation and the CV,were highly (range = 7-66). Subtotals for items grouped by mental correlated with one another (r > 0.95, P < .001), and all function (UPDRS part I), activities of daily living (UPDRS results were similar with both measures. In addition to the part II), and motor function (UPDRS part III) were 2.5, above measures, we also determined if a subject stopped 11.8, and 14.0, respectively. Mean score on the MMSE walking during each trial (ie, the occurrence of “stop walk- 11 was 27.1 (range = 23-30). No patient met the DSM-IV cri- ing while talking” or SWWT ). Finally,a neurologist with teria for dementia. Fall risk (retrospective) and postural expertise in PD and freezing of gait documented the occur- instability were assessed using questions 13 and 30 of the rence of any freezing episodes (ie, the subject tries to walk UPDRS, respectively. but is stuck in place). To a large degree, SWWT and freezing can be distinguished visually. In contrast to freezing, Assessment of Gait Dynamics during SWWT,there is no effort made to continue walking. The protocol and methods used to assess the dynamics of walking, for example, gait variability, were similar to Dual Task those used previously.5,19,20 Briefly,subjects were instructed After completing the normal walk, subjects walked along to walk at their normal pace on level ground from one end the same pathway and under the same conditions while of a hallway to the other end (approximately 20 meters being asked to serially subtract 7s from a 3-digit number away) and then back again. (During normal walking and (eg, 200, 193, 186). Specifically, subjects were instructed dual-task walking, a clinician walked near the subject to to walk at their usual pace, without stopping, while per- ensure safety.) To measure the gait rhythm and the tim- forming serial 7 subtractions out loud. Serial 7 performing of the gait cycle, force-sensitive insoles were placed in ance was not evaluated. Serial 7 subtraction has been the subject’s shoe. These inserts produce a measure of widely used as means of providing a distraction and a cog- 21,22 the force applied to the ground during ambulation. A nitive challenge. small, lightweight (5.5 × 2 × 9 cm; 0.1 Kg) recorder was worn on the ankle and held in place using an ankle wallet. An Statistical Analysis on-board A/D converter (12 bit) sampled the output of the Results are reported as mean ± standard deviation. The footswitches at 300 Hz and stored the data. Subsequently, Wilcoxon signed rank test, a nonparametric analog of the the digitized data were transferred to a Linux worksta- paired t test, was used to compare the results from nor- tion for analysis using software that extracts the initial mal walking with those from walking while performing the contact time of each stride (for 1 foot). With this infor- cognitive challenge. Spearman’s correlation coefficient mation, the stride time or duration of the gait cycle (time was used to evaluate the association between measures. from initial contact of one foot to subsequent contact of the A P value less than .05 (2-tailed) was considered statisti- same foot) was determined for each stride during the cally significant. Statistical analysis was performed using walk. SAS version 7.0 software. To study the intrinsic dynamics of the gait rhythm, the time series of the stride time was divided into 3 time seg- RESULTS ments for both normal walking and walking while cognitively challenged. We analyzed the first walk down the hall, An example of the effect of dual tasking on gait dynam- before any turns (FirstLap), the complete walk (Com- ics in a subject with PD is shown in Figure 1. During nor- pleteWalk), and the initial 5 strides (FirstStrides). Three mal walking, relatively small stride-to-stride fluctuations subjects experienced freezing of gait, but all freezing are seen in the stride time of this subject. In contrast, when episodes were excluded from the analysis. For analysis of walking under the dual-task condition, the subject’s gait the complete walk, turns and any starting and stopping becomes hesitant and the stride-to-stride fluctuations episodes were automatically excluded by filtering out become very large. In addition, the average stride time is those stride times that were greater than 2.5 seconds or longer compared to walking without dual tasking. less than 0.5 seconds in duration. In addition, to focus on In general, a similar pattern was observed for all of the dynamics of continuous walking, a median filter was the subjects with PD, no matter which time segment Effects of Cognitive Challenge on Gait Variability in Patients With Parkinson’s Disease / Hausdorff et al. 55

Figure 1. Example of the effects of walking while performing a Figure 2. Example of the effects of walking while performing a cognitively challenging task in a subject with Parkinson’s disease. cognitively challenging task in a subject without Parkinson’s disease The average stride time increases as does the magnitude of the but with “primary freezing of gait.” Contrast this to the response stride-to-stride fluctuations during this dual-task condition. seen in Figure 1. was analyzed (see Table 1). For example, during First- youngest subjects (the oldest of these subjects was 61 Lap, stride time variability increased in each of the 10 years of age), we find results that are consistent with PD subjects, and the average stride time increased in 9 those of the entire group. For example, stride time vari- of the 10 PD subjects. Note that for all 3 analysis seg- ability increased in all 5 of these subjects during FirstLap ments, the percentage increase in stride time variabil- (∆ variability = 153 msec, P = .043), during Complete- ity tended to be much larger than the percentage increase Walk (∆ variability = 79 msec, P = .043), and during First- in average stride time (in 26 of 30 pairwise compar- Strides (∆ variability = 223 msec, P = .043). isons), and the 2 measures of change were not correlated Dual tasking consistently affected gait variability in (P > .05). the subjects with PD; however, SWWT was observed in only The observed, marked effect of the cognitive chal- 2 subjects. These 2 had fairly advanced PD (Hoehn and lenge on the dynamics of gait in PD is likely abnormal. Con- Yahr’s scores of 3 and 4). Nonetheless, other subjects with sistent with this, we note anecdotally a very different advanced PD did not exhibit SWWT. For these 2 subjects, response to dual tasking in one subject with “primary relatively high and extreme values of gait variability were freezing of gait.”23 In this older adult, gait dynamics was obtained when walking while cognitively challenged (eg, similar during normal walking and during dual-task walk- during FirstLap, they had the highest gait variability ing (see Figure 2). For example, in the FirstStrides segment, from among all subjects). stride time variability was 53 msec during normal walk- In the 10 subjects with PD, gait variability during nor- ing and 45 msec during dual-task waking. In contrast to mal walking was associated with disease severity, dis- the subjects with PD, in this 77-year-old elderly woman ease duration, fall risk, and motor and cognitive function who did not have PD, the cognitive challenge essentially (Table 2). There was a wide range in age, but as noted had no effect on gait dynamics. above, variability was not associated with age. Although This initial result suggests that age alone may not the response to dual tasking was more severe in some sub- explain the effects of dual tasking on gait. A closer look jects with more advanced disease, the associations between at the PD subjects supports this idea. In all 3 analysis peri- clinical characteristics (eg, motor function) and gait vari- ods (ie, FirstLap, FirstSrides, and CompleteWalk), age ability became weaker during dual-task walking. The was not associated with the mean stride time (P > .35) or average stride time was not significantly associated with either measure of stride time variability (P > .56) during age, disease duration, fall risk, or any other clinical meas- dual-task walking. Moreover, when studying only the 5 ure, under both walking conditions. The percentage

Table 1. Effects of Cognitive Challenge on Stride Time Dynamics in Parkinson’s Disease

Time Segment Measure of Stride Time Normal Walking Cognitively Challenged Percentage Increase P Value

FirstLap Average (msec) 940 ± 67 1137 ± 228 21 ± 21 .006 Variability (msec) 47 ± 29 223 ± 281 384 ± 595 .002 CompleteWalk Average (msec) 946 ± 74 1046 ± 126 10 ± 9 .006 Variability (msec) 64 ± 37 142 ± 109 154 ± 164 .014 FirstStrides Average (msec) 950 ± 45 1169 ± 283 23 ± 28 .019 Variability (msec) 33 ± 28 243 ± 361 915 ± 1912 .008

Values are mean ± SD. 56 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003

Table 2. Association Between Gait Variability nitive function. Perhaps SWWT is a mark of only very and Parkinson’s Disease Characteristics severe impairment of the ability to walk while performing another task. On the other hand, gait variability may Walking While Cognitively provide a more sensitive, continuous measure of this abil- Normal Walking Challenged ity and hence of fall risk in PD subjects with intact cognitive function, as is the case in relatively healthy idiopathic Disease duration 0.81 (.005) 0.76 (.010) 14,20 Hoehn and Yahr stage 0.95 (< .0001) 0.57 (.088) elderly fallers. Indeed, in the present study, gait vari- Disease severity (UPDRS total) 0.84 (.002) 0.63 (.053) ability was retrospectively associated with fall risk (Table Mental function (UPDRS part I) 0.72 (.019) 0.17 (.64) 2). A larger scale, prospective study should help address Activities of daily living (UPDRS part II) 0.56 (.089) 0.59 (.074) this issue. Motor function (UPDRS part III) 0.74 (.015) 0.63 (.052) Further study is required to more clearly elucidate the Fall risk 0.68 (.042) 0.55 (.098) pathophysiologic mechanisms that impinge on dual-task Postural instability 0.76 (.023) 0.52 (.122) performance, in general, and gait variability,more specif-

Entries are the Spearman correlation coefficent (P value). From a potential list of variables ically, in PD. In this regard, the association between gait that includes age, the 4 subcategories of the Unified Parkinson’s Disease Rating Scale variability and mental function (Table 2) is worth noting. (UPDRS), and the listed headings, only those associations in which P < .10 during at least 1 walking condition are included. Values are calculated from the first lap using the Adequate dopaminergic transmission is apparently nec- coefficient of variation of gait; similar results were obtained in the other time segments essary for concurrent processing of cognitive information and using the Pearson correlation coefficient instead. Fall risk and postural instability were determined by questions 13 and 30 of the UPDRS, respectively. and striatal integration of sensorimotor information required to program cognitive acts.25 However, stride-to- increases in the average and variability of the stride time stride variability of gait in PD may be dopa independent.24 were not related to baseline measures. Some have suggested that depleted central processing resources may impair dual-task performance or the pos- DISCUSSION sibility that this represents a deficit in switching processing resources between 2 tasks as the combined Previous investigations demonstrated that gait and bal- demands outweigh available resources.26 Alternatively, ance are altered during dual-task performance in PD.1,6,9,15 there is a close connection between the basal ganglia and These studies of walking largely focused on the effects on frontal lobe function, and the frontal lobe–like deficits gait speed or the number of steps taken during the walk. that have been reported in PD may affect dual-task per- However, gait variability, on one hand, and gait speed formance.27 (and the closely related measure average stride time), on Further investigations are required to examine if the the other, are often affected by different mechanisms.5,14,20,24 observed effect of dual tasking on gait variability was age For example, among older adults, gait variability is prospec- related, if there is a disproportional effect in subjects with tively associated with fall risk, while gait speed is not.14,20 PD, over and above a healthy population, and/or how spe- In subjects with PD, gait speed was reported to be dopa cific the observed response is to PD. It seems likely, how- responsive, while gait variability was not.24 The present ever, that aging alone does not explain the sensitivity of findings extend the results of earlier studies and demon- gait variability to dual-task performance in PD. On a pre- strate, for the first time, that not only does dual-task per- liminary basis, we note that dual-task walking did not affect formance exacerbate the bradykinetic manifestations of the gait of a 77-year-old woman who did not have PD. More- PD, but it also apparently markedly heightens gait vari- over, we note that age was not related to dual performance ability and instability. In fact, the magnitude of the effect among the subjects with PD, and even relatively young sub- on variability is apparently much larger than the effect jects consistently showed increased stride time and on bradykinetic manifestations (compare the percentage increased variability during dual-task walking. changes in average stride time and stride time variabil- Other studies suggest that PD pathology contributes ity in Table 1). to the effect of dual tasking on gait. Camicioli et al10 In frail older adults, the inability to walk while talk- observed that dual-task performance did not affect gait ing has been shown to be a predictor of fall risk.11 Since speed in healthy older adults. They also found that a ver- subjects with PD also have difficulty performing multiple bal fluency task had a much greater effect on gait in sub- tasks, Bloem et al12 hypothesized that SWWT would also jects with PD compared to aged-matched controls.9 predict falls in PD. In contrast to their hypothesis, they Similarly, Bond and Morris15 reported that PD subjects found that in PD patients with intact cognition, SWWT was experienced considerable difficulty when they were a poor predictor of falls. In fact, while many of their sub- required to walk while attending to a complex visuomo- jects reported falls, only a small number of subjects could tor task, while aged-matched controls did not. Recent not walk and talk. We also found that most of the PD sub- findings by Stolze et al28 are consistent with the sugges- jects studied could walk while performing serial 7 sub- tion that the effects of dual tasking and attention on gait tractions. This is consistent with the idea that SWWT is may be specific to PD. Thus, previous studies and our pre- not a good predictor of falls in PD subjects with intact cog- liminary evidence suggest that performance of a second- Effects of Cognitive Challenge on Gait Variability in Patients With Parkinson’s Disease / Hausdorff et al. 57 ary task has little or no effect on the gait of healthy older by distinct pathways outside the dopaminergic system adults. Future studies of the effects of dual-task per- that affects general motor function. In this regard, it may formance on gait variability will be helpful for identifying be helpful to investigate the relationship between sensi- any role of age. It seems likely, however, that age-related tive measures of neuropsychological/cognitive function effects are much more subtle than the dramatic response and dual-task walking ability. observed and that the more than 5-fold increase in vari- Much further work is needed to help reduce falls in ability is a consequence of PD. PD.7 Future investigations should confirm the present The present study has a number of limitations, per- findings and attempt to more fully identify the factors that haps most notably the small number of subjects, the lack modulate the ability of a person to maintain a stable walk of a control group (see above), and the nature of the dual while performing a secondary task. The present study task. Future, larger scale studies should explore the effects provides new insights into PD and gait variability and sets of other dual tasks, compare the results to age-matched the stage for additional investigations that may help controls, and study other aspects of gait and motor func- reduce the ubiquity and incapacitating effects of falls in tion (eg, left-right asymmetry, kinematics, dyskinesias; PD. Walking while performing a dual task markedly exac- we did not observe any marked increase in dyskinesias dur- erbates gait variability and impairs the ability of patients ing dual-task walking, however, dyskinesias were not with PD to maintain a stable walk. Behavioral or phar- quantified). It is possible that there was an ordering effect. macological interventions designed to minimize the dis- All subjects performed dual-task walking after walking tractions of dual-task walking may be helpful in under usual conditions. Direct confrontation with the maintaining gait stability in subjects with PD. most difficult task could have induced even more gait abnormalities than presently observed.29 Although we did References not quantify serial 7 performance, it might also be help- 1. Ashburn A, Stack E, Pickering RM, Ward CD. A community- ful in the future to compare performance on the dual task dwelling sample of people with Parkinson’s disease: characteristics of fallers and non-fallers. Age Ageing 2001; 30:47-52. with motor performance. In contrast to subjects with PD, 29 2. Gray P, Hildebrand K. Fall risk factors in Parkinson’s disease. Bloem et al noticed that healthy controls tend to give pri- J Neurosci Nurs 2000; 32:222-228. ority to cognitive tasks over the motor task when atten- 3. Koller WC, Glatt S, Vetere-Overfield B, Hassanein R. Falls and tional resources are divided during dual-task performance. Parkinson’s disease. Clin Neuropharmacol 1989; 12:98-105. Despite these limitations, the highly consistent results 4. Blin O, Ferrandez AM, Serratrice G. Quantitative analysis of gait among the PD subjects studied support the idea that in in Parkinson patients: increased variability of stride length. J Neurol Sci 1990; 98:91-97. subjects with PD and intact cognitive function, perform- 5. Hausdorff JM, Cudkowicz ME, Firtion R, et al. Gait variability ance of a secondary task increases gait variability. and basal ganglia disorders: stride-to-stride variations of gait Further study is also needed to understand why dual- cycle timing in Parkinson’s disease and Huntington’s disease. task performance causes a relatively small increase in gait Mov Disord 1998; 13:428-437. variability in some patients while it almost literally knocks 6. Morris ME, Iansek R, Matyas TA, Summers JJ. Stride length regulation in Parkinson’s disease. Normalization strategies and some people off their feet. All subjects were tested while underlying mechanisms. Brain 1996; 119(Pt 2):551-568. their medications were active (ie, in the “on” state). 7. Playfer JR. Falls and Parkinson’s disease. Age Ageing 2001; Subject-to-subject variability in the “on” state may be 30:3-4. related to differences in the response to dopamine. Either 8. Bloem BR, van Vugt JP,Beckley DJ. Postural instability and falls way, it is important to note that the present results sug- in Parkinson’s disease. Adv Neurol 2001; 87:209-223. gest that the clinical features measured in the UPDRS may 9. Camicioli R, Oken BS, Sexton G, et al. Verbal fluency task affects gait in Parkinson’s disease with motor freezing. J Geri- not explain the degree to which some subjects respond. Per- atr Psychiatry Neurol 1998; 11:181-185. haps this may also be related to patient-specific differences 10. Camicioli R, Howieson D, Lehman S, Kaye J. Talking while in the perception of the stimulus provided by serial 7 sub- walking: the effect of a dual task in aging and Alzheimer’s distraction. Studies of dual-task effects on balance have ease. Neurology 1997; 48:955-958. shown that both arousal and attention influence balance 11. Lundin-Olsson L, Nyberg L, Gustafson Y.“Stops walking when but differentiate between the specific effects.30 For some talking” as a predictor of falls in elderly people. Lancet 1997; 349:617. patients, serial 7 subtraction may simply divide attention 12. Bloem BR, Grimbergen YA, Cramer M, Valkenburg VV. “Stops and provide a distraction. For other patients, an element walking when talking” does not predict falls in Parkinson’s dis- of arousal/stress may also play a role. Differences in cog- ease. Ann Neurol 2000; 48:268. nitive function not captured by the MMSE, a crude meas- 13. Nakamura T, Meguro K, Sasaki H. Relationship between falls ure of mental function, and variable engagement in the and stride length variability in senile dementia of the Alzheimer type. Gerontology 1996; 42:108-113. secondary task may also explain subject-to-subject vari- 14. Maki BE. Gait changes in older adults: predictors of falls or indi- ability. The fact that clinical features evaluated in the cators of fear. J Am Geriatr Soc 1997; 45:313-320. UPDRS did not fully explain the degree of the response 15. Bond JM, Morris M. Goal-directed secondary motor tasks: their may suggest that the ability to walk while performing a effects on gait in subjects with Parkinson disease. Arch Phys Med cognitive task follows an individual course and is influenced Rehabil 2000; 81:110-116. 58 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003

16. Hoehn MM, Yahr MD. Parkinsonism: onset, progression and mor- 24. Blin O, Ferrandez AM, Pailhous J, Serratrice G. Dopa-sensitive tality. Neurology 1967; 17:427-442. and dopa-resistant gait parameters in Parkinson’s disease. J Neu- 17. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state.” A rol Sci 1991; 103:51-54. practical method for grading the cognitive state of patients for 25. Malapani C, Rakitin B, Levy R, et al. Coupled temporal mem- the clinician. J Psychiatr Res 1975; 12:189-198. ories in Parkinson’s disease: a dopamine-related dysfunction. J 18. Martinez-Martin P, Gil-Nagel A, Gracia LM, et al. Unified Cogn Neurosci 1998; 10:316-331. Parkinson’s disease rating scale characteristics and structure. 26. Brown RG, Marsden CD. Dual task performance and process- The Cooperative Multicentric Group. Mov Disord 1994; 9:76-83. ing resources in normal subjects and patients with Parkinson’s 19. Hausdorff JM, Edelberg HK, Mitchell SL, et al. Increased gait disease. Brain 1991; 114( Pt 1A):215-231. unsteadiness in community-dwelling elderly fallers. Arch Phys 27. Dubois B, Malapani C, Verin M, et al. Cognitive functions and Med Rehabil 1997; 78:278-283. the basal ganglia: the model of Parkinson disease. Rev Neurol 20. Hausdorff JM, Rios D, Edelberg HK. Gait variability and fall risk (Paris) 1994; 150:763-770. in community-living older adults: a 1-year prospective study. Arch 28. Stolze H, Kuhtz-Buschbeck JP, Drucke H, et al. Comparative Phys Med Rehabil 2001; 82:1050-1056. analysis of the gait disorder of normal pressure hydrocephalus 21. Ganguli M, Ratcliff G, Huff FJ, et al. Serial sevens versus world and Parkinson’s disease. J Neurol Neurosurg Psychiatry 2001; backwards: a comparison of the two measures of attention from 70:289-297. the MMSE. J Geriatr Psychiatry Neurol 1990; 3:203-207. 29. Bloem BR, Valkenburg VV, Slabbekoorn M, Gert vD. The mul- 22. Karzmark P. Validity of the serial seven procedure. Int J Geri- tiple tasks test. Strategies in Parkinson’s disease. Exp Brain Res atr Psychiatry 2000; 15:677-679. 2001; 137:478-486. 23. Achiron A, Ziv I, Goren M, et al. Primary progressive freezing 30. Maki BE, McIlroy WE. Influence of arousal and attention on the gait. Mov Disord 1993; 8:293-297. control of postural sway. J Vestib Res 1996; 6:53-59. Familial Dementia With Lewy Bodies With an Atypical Clinical Presentation

Lauren T. Bonner, MD, Debby W. Tsuang, MD, MSc, Monique M. Cherrier, PhD, Charisma J. Eugenio, BS, Jennifer Q. Du, RN, Ellen J. Steinbart, RN, Pornprot Limprasert, MD, PhD, Albert R. La Spada, MD, PhD, Benjamin Seltzer, MD, Thomas D. Bird, MD, and James B. Leverenz, MD

ABSTRACT

The authors report a case of a 64-year-old male with Alzheimer’s disease (AD) and dementia with Lewy bodies (DLB) pathology at autopsy who did not manifest the core symptoms of DLB until very late in his clinical course. His initial presentation of early executive and language dysfunction suggested a cortical dementia similar to frontotemporal lobar degeneration (FTLD). Core symptoms of DLB including dementia, hallucination, and parkinsonian symptoms were not apparent until late in the course of his illness. Autopsy revealed both brainstem and cortical Lewy bodies and AD pathology. Family history revealed 7 relatives with a history of dementia including 4 with possible or probable DLB. This case is unique because of the FTLD-like presentation, positive family history of dementia, and autopsy confirmation of DLB. (J Geriatr Psychiatry Neurol 2003; 16:59–64)

Keywords: Lewy bodies; familial dementia; Alzheimer’s disease

Dementia with Lewy bodies (DLB) may be the second Clinically, DLB is frequently characterized by dementia, leading cause of dementia after Alzheimer’s disease (AD).1 fluctuation in cognition and attention, visual hallucinations, Autopsy studies have demonstrated Lewy bodies (LBs) in and parkinsonian symptoms.1 “Cortical” symptoms, such the brains of at least 15% to 25% of patients with demen- as executive dysfunction and aphasia, occur in more tia.2-4 Approximately 50% to 70% of DLB cases also have advanced stages of DLB but have not been reported as pre- sufficient Alzheimer’s pathology to meet criteria for AD.5 senting symptoms.6-9 In this report, we describe a case of neuropathologi- cally confirmed DLB, with early executive dysfunction Received February 20, 2002. Received revised June 18, 2002. Accepted and aphasia with a positive family history of dementia. for publication June 18, 2002. From the Department of Veterans Affairs Northwest Network Mental CASE PRESENTATION Illness Research, Education, and Clinical Center (MIRECC) (Drs. Bonner, Tsuang, Eugenio, Du, and Leverenz); Parkinson’s Disease Research, Education, and Clinical Center (Dr. Leverenz); and Geriatric Research, The proband, a 64-year-old right-handed male aerospace Education and Clinical Centers (Dr. Bird and Ms. Steinbart), University engineer, was referred to a local neurologist for evaluation of Washington, Seattle; Departments of Laboratory Medicine (Drs. of a 3-year history of mild depressive symptoms, memory Limprasert and La Spada), Neurology (Drs. La Spada, Bird, and Leverenz), Psychiatry and Behavioral Sciences (Drs. Bonner, Tsuang, Cherrier, and and speech disturbance, and deterioration in handwriting Leverenz), and Medicine (Medical Genetics) (Drs. Bird and La Spada), after cardiac bypass surgery.He reported new onset of dif- University of Washington, Seattle; and Tulane University School of ficulty recalling names of casual professional and social Medicine, Department of Psychiatry and Neurology and the Department of Veterans Affairs South Central MIRECC (Dr. Seltzer), New Orleans, acquaintances. Although these symptoms were continuous Louisiana. and troublesome to the patient, they were not initially appar- This work was supported by NIA RO1-AG18644, NIA P50 AG05136, and ent to others. He continued to function at work and at home the Department of Veterans Affairs. The authors wish to thank the family without substantial difficulties. More recently, his cogni- members for their valuable time and willingness to participate in research tive difficulties had become noticeable to his wife with loss and Lynne Greenup for her technical assistance. of language and decision-making skills. His wife noted rel- Reprint requests: James B. Leverenz, MD, VA Puget Sound Health Care atively intact recall for recent events, although he reported System, MIRECC (116 MIRECC), 1660 S. Columbian Way, Seattle, WA 98108; e-mail: [email protected]. using written notes to assist his memory.He had been able to maintain full-time employment as an aerospace engineer. DOI: 10.1177/0891988702250585 Past medical history was significant for coronary artery

Table 1. Summary of Neuropsychological Evaluations

Date of Neuropsychological Evaluation September 1983a January 1990

Age at time of testing (years) 64 70 Duration of symptoms at time of testing 2 years 8 years Cognitive domain/test Intellectual functioning WAIS (WAIS-R) full-scale IQ 124 (95 percentile) 109 (73 percentile) Verbal IQ 120 (91 percentile) 104 (61 percentile) Performance IQ 124 (95 percentile) 115 (84 percentile) Highest subtest Arithmetic Picture completion Mini-Mental Status Examination NA 24/30 Attention Trail-making test, part A WNL WNL Memory Verbal Story recall WMS/WMS-R logical memory Immediate recall WNL Impaired (8 percentile) 2/46 Delayed recall WNL Impaired (8 percentile) 0/46 Visual WMS/WMS-R visual reproduction Immediate recall WNL WNL (99 percentile) 11/14 Delayed recall b WNL (99 percentile) 10/14 Construction Aphasia screening test drawings Mild difficulties Mild difficulties WAIS-R block design subtest b WNL (84 percentile) SS = 13 Executive functions Trail-making test, part B Impaired (3 errors) Impaired (6 to 10 percentile) (3 errors) Halstead-Reitan Category Test WNL (with confusion during subtest 5) NA

WAIS = Wechsler Adult Intelligence Scale; NA = not administered; WNL = within normal limits; WMS = Wechsler Memory Scale. a. Details of neuropsychological evaluation were obtained from narrative report. Therefore, individual test scores and percentiles were available only as noted in the report. b. Test administered but actual score unavailable. disease and successfully treated pernicious anemia. The Evaluating neuropsychologists noted he had significant neurologist found no signs of parkinsonism on examination. depression with dysphoric mood, early morning awaking, Initial neuropsychological evaluation (see Table 1) and thoughts of death and euthanasia. On examination (see revealed constructional dyspraxia and mild difficulties Table 1), he was noted to be aphasic with a Folstein Mini- with executive functioning. He was noted to be fluent with Mental Status Exam (MMSE)10 score of 24/30, reflecting superior intellectual functioning and memory that was difficulties with serial 7s, delayed recall, and inability to “high for age” with no recall deficits. correctly copy intersecting pentagons. Examiners noted His prior evaluation by his primary care physician had impairments in executive functioning, language, and ver- been negative and included serum chemistry, sedimenta- bal memory with significant word-finding difficulties. tion rate, VDRL and serum B12. Head computed tomog- Periodic neurological examinations from age 71 to 77 raphy (CT) at that time demonstrated slight prominence documented continued cognitive decline. Aphasic symptoms of the sylvian fissure area at the left temporal tip with included increasingly severe difficulties with word finding, normal-sized ventricles. The neurologist attributed his naming, and word comprehension. He also exhibited evi- cognitive problems to age, whereas the neuropsychologist dence of apraxia. He developed behavioral disturbances attributed them to mixed anxiety and depression. including auditory and visual hallucinations, verbal out- He was subsequently referred to a psychiatrist and bursts, agitation, pacing, and frequent falls. He was treated underwent a series of treatment trials with antidepressants with multiple psychotropic medications including antipsy- including selective serotonin reuptake inhibitors and tri- chotics, antidepressants, anxiolytics, and anticonvulsants, cyclic antidepressants. However, his cognitive difficulties with poor clinical response. The proband maintained some continued to progress, leading to retirement at age 65. insight into his disorder stating, “My brain is going away At age 70, the patient was evaluated by his primary and I cannot convert thoughts to words.” care physician secondary to subjective reports of cognitive Parkinsonian symptoms including bradykinesia decline. The neurological evaluation revealed no parkin- appeared at approximately age 73 and were noted to sonian signs and symptoms. His primary care physician felt worsen after treatment with haloperidol with an increase that his cognitive functioning was normal. However, the in tremor, rigidity, abnormal gait, and posture. patient was referred for neuropsychological reassessment General neurological examination at age 76, 15 years in a dementia specialty clinic. During this evaluation, the after the onset of dementia symptoms, noted continued evi- patient reiterated his complaints of a progressive decline dence of clinical parkinsonism with cogwheel rigidity; in memory, speech, and cognition with depressed mood. slow, stooped shuffling gait; and positive snout reflex. Familial Dementia With Lewy Bodies With an Atypical Clinical Presentation / Bonner et al. 61

Figure 1. Pedigree of family with dementia.

Resting tremor, extraocular movement abnormalities, and The proband’s sister (III 7) first manifested symp- other parkinsonian signs were not appreciated. Magnetic toms of dementia when she was in her late 50s. Her fam- resonance imaging and computerized axial tomography ily reported difficulties with executive functioning and revealed mild to moderate cerebral atrophy. Electroen- memory with decline in ability to perform activities of cephalogram (EEG) demonstrated diffuse slowing. daily living. Neurological evaluation at age 61 found severe His functional status worsened, and his behavioral dis- and continued deterioration in functioning with normal lab- turbances including severe agitation and assaultive behav- oratory parameters and head CT demonstrating general- ior increased, eventually leading to nursing home ized atrophy. Reexamination at age 62 documented placement. Neurological examination at age 78 revealed inappropriate affect, with excessive laughing and joking a severe global aphasia and worsening parkinsonism, during the examination; conversation was noted to be although without tremor. “rich in puns.” Cognitive evaluation at that time revealed He died at age 79 from acute bronchopneumonia and delayed recall of 2 out of 3 items after 5 minutes. On neu- multiple pulmonary emboli. rological examination, she demonstrated parkinsonian symptoms with rigidity, cogwheeling, and shuffling gait. FAMILY HISTORY She was also noted to have a positive glabellar, bilateral grasp, slight Babinski, and poor upward gaze. Head CT Family history obtained from family members revealed a demonstrated severe cortical atrophy, normal lumbar clinical history of dementia in 7 relatives including his puncture, and EEG slowing of background activity. mother and sister (Figure 1). Her dementia continued to worsen with cognitive and functional decline, auditory and visual hallucinations, and First-Degree Relatives worsening parkinsonian symptoms. She subsequently The proband’s mother (II 4) was approximately 60 years developed language impairments, and her mental status old when she demonstrated difficulties with memory, fluctuated from episodes of coherence to periods of bizarre organization, planning, and visual spatial deficits. She behavior. Terminally, she was mute and bedridden, dying would wander and become lost in her neighborhood on mul- at age 65. tiple occasions. Approximately 2 to 3 years into the course of her illness, she demonstrated a marked decline in her Second-Degree Relatives ability to perform activities of daily living as well as slow- The proband’s maternal aunt (II 1) had memory probing of her movements, difficulty maintaining balance, and lems that began in her 70s. She died in her late 70s/early urinary incontinence. There was no history of fluctuating 80s. She did not demonstrate parkinsonism, hallucina- mental status. She died at the age of 65. tions, or bizarre behaviors. 62 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003

The proband’s maternal aunt (II 2) initially mani- of 27/30, reflecting a 0/3 score for delayed recall. Admin- fested symptoms of dementia during her 60s. Her family istration of the UPDRS–Motor Examination13 revealed a reported difficulties with memory, visuospatial and exec- score of 2, reflecting mild bilateral slowing and/or reduc- utive functioning. She would frequently wander the neigh- tion in leg agility. borhood, becoming lost. As her illness progressed, she became paranoid and developed aphasia and apraxia. NEUROPATHOLOGY Eventually,she was institutionalized at a state mental hospital secondary to extreme combativeness and resistive- At autopsy,the proband’s brain weighed 1225 grams with ness to care. At the time of death, she was virtually mute. gross evidence of moderate to severe fronto-temporal atro- The proband’s maternal uncle (II 7) had a history of phy and mild parietal atrophy. There was mild patchy excessive alcohol intake. His dementia symptoms began large vessel atherosclerosis. On gross sectioning, there in his 60s with his family reporting difficulties with mem- was severe ventriculomegaly and depigmentation of the ory, apraxia, and visuospatial and executive functioning. substantia nigra. Microscopically, Bielschowsky silver Toward the end of his life, he developed parkinsonian staining demonstrated moderate to severe changes of AD. symptoms with slow,stooped gait; bradykinesia; and rest- There was moderate to severe senile plaque formation in ing tremor. He also manifested symptoms of agitation, the cortex and hippocampal formation. Neurofibrillary paranoia, and visual and auditory hallucinations. Although tangles were moderate to severe in the medial temporal he continued to use alcohol until his death, his dementia lobe and hippocampus and mild to moderate in association and behavioral symptoms were generally not associated cortices (Braak Stage V).14 Hematoxylin and Eosin revealed with intoxication or withdrawal. neuronal loss in the substantia nigra with occasional LB inclusions in the remaining neurons (Figure 2a). Immuno- Third-Degree Relatives histochemistry for alpha-synuclein (LB 509, Zymed) demon- The proband’s cousin (III 2) is 80 years old and reports a strated widespread inclusions in the substantia nigra, 1-year history of increasing difficulty with short-term cingulate gyrus, parahippocampal gyrus, and amygdala memory that interferes with daily activities, visuospatial (Figure 2b). There were immunopositive neurites in these dysfunction with geographic disorientation, and slight regions and in the CA-2 subfield of the hippocampus. The impairment of executive functioning. She had a recent case fulfilled neuropathologic diagnostic criteria for DLB Clinical Dementia Rating (CDR) Scale11 score of 1.0, indi- (neocortical subtype)1 and high likelihood AD.15 There was cating mild dementia, and Neuropsychiatric Inventory no evidence grossly or microscopically of any vascular Score (NPI)12 of 0, both obtained via a telephone interview lesions. with a relative. The CDR and NPI are assessments of cognitive and neuropsychiatric symptomatology,respectively. MOLECULAR GENETICS Both are obtained by a semistructured interview with the caregiver and are easily and reliably performed. Neuro- Genomic DNA from the proband and 3 family members was logical examination revealed an MMSE10 of 27/30, reflect- screened for alpha-synuclein mutations. The complete ing incorrect date and 1/3 delayed recall. Administration genomic sequences encompassing the alpha-synuclein of the Unified Parkinson’s Disease Rating Scale gene is available through Genbank (http://www.ncbi.nlm. (UPDRS)–Motor Examination13 revealed a score of 8, nih.gov/Genbank). Polymerase chain reaction amplification reflecting mild hypomimia, mild bilateral slowing and/or of the 5 coding regions (exons) that encoded for the amino reduction in amplitude of both hand movements and rapid acid sequences was conducted. There were no genetic vari- alternating movements, mild bilateral slowing and/or ants in the gene, specifically,the previously reported A53T reduction in leg agility, and minimal bradykinesia and mutation.16 In addition, apolipoprotein E genotyping was hypokinesia. conducted using standard methods17 in the University of The proband’s cousin (III 5) is 81 years old. She reports Washington Alzheimer’s Disease Research Center Geno- “occasional forgetfulness.” A telephone interview with an typing Core (G. Schellenberg, core principal investigator). informant showed no substantial behavioral nor func- The proband’s APOE genotype was 3/4. tional disturbance (scores of 0 on both the CDR11 and the NPI12). Neurological examination revealed a MMSE10 of DISCUSSION 29/30 (she missed 1 out of 3 items on delayed recall). The UPDRS–Motor Examination13 score was 2, reflecting mild Our index case fulfilled both clinical and pathological cri- slowing and/or reduction in amplitude of left hand move- teria for DLB, although his early clinical presentation ments and left hand rapid alternating movements. made the diagnosis difficult. Ultimately, he exhibited 2 of The proband’s cousin (III 13) is 71 years old. He and the 3 core DLB symptoms1 including hallucinations and his family deny memory difficulties. Via telephone inter- motor symptoms of parkinsonism. Fluctuations in mental view with a relative, he has scores of 0 on both the CDR 11 status were not well documented, although his wife and NPI12. Neurological examination revealed an MMSE10 reported nighttime confusion. In addition, his history of Familial Dementia With Lewy Bodies With an Atypical Clinical Presentation / Bonner et al. 63

Figure 2. (a) Lewy body inclusion in pigmented substantia nigra neuron (large arrow) and microglial cell cluster, containing pigment, suggestive of neuronal degeneration (small arrow) (125X, hematoxylin and eosin). (b) Alpha-synuclein immunopositive inclusions in neurons of the deep layers of the parahippocampal gyrus (31X, alpha-synuclein immunohistochemistry). Images digitized with adjustment of brightness and contrast. repeated falls and exquisite neuroleptic sensitivity support report of an FTLD-like clinical presentation in a neu- a diagnosis of DLB. However, he did not exhibit any of these ropathologically verified DLB case. DLB symptoms during the first 10 years of his 18-year The family history of our case suggests an autosomal course. Pathologically,he exhibited both brainstem and cor- dominant inheritance pattern with similar symptoms in tical LBs consistent with the neocortical category of DLB.1 affected relatives. Although autopsy was not available for As with many DLB patients, he also exhibited concomitant other affected family members, they also demonstrated AD pathology with moderate to severe senile plaque and dementia characterized by symptoms of executive dys- neurofibrillary tangle changes.1 function, with the sister demonstrating additional lan- The initial clinical presentation of the proband included guage impairment. Seven of the proband’s relatives also clinical and neuropsychological evidence of visuospatial, demonstrated dementia, some characterized by symptoms language, and executive dysfunction. Despite early sub- suggestive of DLB including clinical parkinsonism and hal- jective memory complaints, initial neuropsychological test- lucinations. ing found superior memory function and argued against There have been multiple reports of familial parkin- a diagnosis of AD. His early language and executive dys- sonism with dementia.19-26 In the majority of these fami- function, with subsequent development of a severe global lies, clinical parkinsonism is the predominant presenting aphasia, suggested a cortical disorder similar to fronto- symptom with later development of a dementing disorder. temporal lobar degeneration (FTLD).18 In fact, his clinical Only a few families include individuals that presented with course included several core (insidious onset, gradual pro- dementia.20,22 However, there are no neuropathological gression, nonfluent spontaneous speech) and supportive examinations available in these affected individuals. Most features (age of onset less than 65 years of age, early preser- recently,Tsuang et al27 and Galvin et al28 described 2 fam- vation of word meaning, late mutism, early preservation ilies with 2 or more individuals with autopsy-proven DLB. of social skills, late behavioral changes, late akinesia, and Mutations in α-synuclein have also been found in rigidity) of progressive nonfluent aphasia, a proposed clin- families with clinical parkinsonism and dementia.29 Patho- ical subtype of FTLD.18 One exclusionary item for FTLD logically,most of these familial cases have brainstem, lim- present during the initial presentation was clinical and neu- bic, and cortical LB pathology.30,31 In this proband, there ropsychological evidence of visuospatial dysfunction. was no mutation in the α-synuclein gene. Other disorders DLB patients generally exhibit neuropsychological associated with a familial frontal dementia with parkin- deficits quite similar to that observed in AD except for rel- sonism include fronto-temporal dementia with parkin- atively more severe visuospatial deficits and less severe sonism associated with chromosome 17 (FTDP-17) and memory dysfunction.1 In several studies, language dys- hereditary dysphasic dementia.32,33 The former is charac- function has been found to be equivalent in AD and terized by -associated pathology and LBs are not char- DLB.7-9 In our proband, the only early symptom that sug- acteristic, while the latter is associated with inconsistent gested a diagnosis of DLB, versus FTLD, was the visu- parkinsonism and LB pathology. ospatial dysfunction. It was only in the second decade of This family is unique in that all 7 potentially affected his clinical course that the core symptoms of DLB became cases presented with dementia and the autopsy of the evident. To the best of our knowledge, this is the first case index case demonstrated cortical and brainstem LB pathol- 64 Journal of Geriatric Psychiatry and Neurology / Vol. 16, No. 1, March 2003 ogy.Our index case demonstrates the degree of clinical het- 14. Braak H, Braak E. Neuropathological staging of Alzheimer- erogeneity in DLB. DLB should be considered in the dif- related changes. Acta Neuropatholo (Berl) 1991; 82:239-259. ferential diagnosis of demented patients presenting with 15. National Institute on Aging and Reagan Institute Working Group on Diagnostic Criteria for the Neuropathological Assess- FTLD-like symptoms, especially if there is early evidence ment of Alzheimer’s Disease. Consensus recommendations for of visuospatial dysfunction. Other disorders character- the postmortem diagnosis of Alzheimer’s disease. The National ized by parkinsonian signs and dementia must also be con- Institute on Aging, and Reagan Institute Working Group on sidered in these patients including Parkinson’s disease with Diagnostic Criteria for the Neuropathological Assessment of Alzheimer’s Disease. Neurobiol Aging 1997; 18:S1-S2. dementia,34 progressive supranuclear palsy,34 FTDP-17,32 34 16. Polymeropoulos M, Lavedan C, Leroy E, et al. Mutation in the and AD. Further evaluation of this family and a search alpha-synuclein gene identified in families with Parkinson’s for additional families with autopsy-proven DLB will help disease. Science 1997; 276:2045-2047. to further elucidate the genetics and pathophysiology of 17. Hixson J, Vernier D. Restriction isotyping of human apolipopro- this important dementing disorder. tein E by gene amplification and cleavage with HhaI. J Lipid Res 1990; 31:545-548. 18. Neary D, Snowden JS, Gustafson L, et al. Frontotemporal lobar References degeneration: a consensus on clinical diagnostic criteria. Neu- 1. McKeith LG, Galasko D, Kosaka K, et al. Consensus guidelines rology 1998; 51:1546-1554. for the clinical and pathologic diagnosis of dementia with Lewy 19. Waters CH, Miller CA. Autosomal dominant Lewy body parkin- bodies (DLB): report of the consortium on DLB international sonism in a four-generation family. Ann Neurol 1994; 35:59-64. workshop. Neurology 1996; 47:1113-1124. 20. Ohara K, Takauchi S, Kokai M, et al. Familial dementia with 2. Leverenz J, Sumi M. Parkinson’s disease in patients with Lewy bodies (DLB). Clin Neuropathol 1999; 18:232-239. Alzheimer’s disease. Arch Neurol 1986; 43:662-664. 21. Golbe LI, Miller DC, Duvoisin RC. Autosomal dominant Lewy- 3. Lim A, Tsuang D, Kukull W, et al. Clinico-neuropathological body Parkinson’s disease. Adv Neurol 1990; 53:287-292. correlation of Alzheimer’s disease in a community-based cases series. J Am Geriatr Soc 1999; 47:564-569. 22. Wakabayashi K, Hayashi S, Ishikawa A, et al. Autosomal dominant diffuse Lewy body disease. Acta Neuropathol (Berl) 1998; 4. Hulette C, Mirra S, Wilkinson W,et al. The Consortium to Estab- 96:207-210. lish a Registry for Alzheimer’s Disease (CERAD). Part IX. A prospective cliniconeuropathologic study of Parkinson’s fea- 23. Ishikawa A, Takahashi H, Tanaka H, et al. Clinical features of tures in Alzheimer’s disease. Neurology 1995; 45:1991-1995. familial diffuse Lewy body disease. Eur Neurol 1997; 38:34-38. 5. Dickson D, Crystal H, Davies P, Hardy J. Cytoskeletal and 24. Denson MA, Wszolek ZK, Pfeiffer RF,et al. Familial parkinsonism, Alzheimer-type pathology in Lewy body disease. In: Perry R, McK- dementia, and Lewy body disease: study of family G. Ann Neu- eith I, Perry E, eds. Dementia with Lewy bodies: clinical, patho- rol 1997; 42:638-643. logical, and treatment issues. New York: Cambridge University 25. Muenter MD, Forno LS, Hornykiewicz O, et al. Hereditary form Press, 1996:224-237. of parkinsonism—dementia. Ann Neurol 1998; 43:768-781. 6. Gibb WR, Luthert PJ, Janota I, Lantos PL. Cortical Lewy body 26. Golbe LI, Lazzarini AM, Schwarz KO, et al. Autosomal dominant dementia: clinical features and classification. J Neurol Neuro- parkinsonism with benign course and typical Lewy-body pathol- surg Psychiatry 1989; 52:185-192. ogy. Neurology 1993; 43:2222-2227. 7. Salmon D, Galasko D. Neuropsychological aspects of Lewy body 27. Tsuang D, Dalan A, Eugenio C, et al. Familial dementia with Lewy dementia. In: Perry R, McKeith I, Perry E, eds. Dementia with bodies: a clinico-neuropathological study of two families. Arch Lewy bodies. Cambridge, UK: Cambridge University Press, Neurol. In press. 1996:99-133. 28. Galvin J, Lee S, Perry A, et al. Familial dementia with Lewy bod- 8. Gnanalingham K, Byrne E, Thornton A, et al. Motor and cogni- ies: clinicopathologic analysis of two kindreds. Neurology 2002; tive function in Lewy body dementia: comparison with 58:A251. Alzheimer’s and Parkinson’s diseases. J Neurol Neurosurg Psy- 29. Kruger R, Kuhn W, Leenders KL, et al. Familial parkinsonism chiatry 1997; 62:243-252. with synuclein pathology: clinical and PET studies of A30P 9. Hansen L, Salmon D, Galasko D, et al. The Lewy body variant mutation carriers. Neurology 2001; 56:1355-1362. of Alzheimer’s disease: a clinical and pathologic entity. Neurol- 30. Golbe L, Di Iorio G, Bonavita V,et al. A large kindred with auto- ogy 1990; 40:1-8. somal dominant Parkinson’s disease. Ann Neurol 1990; 27:276- 10. Folstein MF,Folstein SE, McHugh PR. Mini-Mental State: a prac- 282. tical method for grading the cognitive state of patients for the 31. Spira PJ, Sharpe DM, Halliday G, et al. Clinical and patholog- clinician. J Psychiatr Res 1975; 12:189-198. ical features of a parkinsonian syndrome in a family with an 11. Hughes CP, Berg L, Danziger WL, et al. A new clinical scale for Ala53Thr alpha-synuclein mutation. Ann Neurol 2001; 49:313- the staging of dementia. Br J Psychiatry 1982; 140:566-572. 319. 12. Cummings JL, Mega M, Gray K, et al. The Neuropsychiatric 32. Reed LA, Wszolek ZK, Hutton M. Phenotypic correlations in Inventory: comprehensive assessment of psychopathology in FTDP-17. Neurobiol Aging 2001; 22:89-107. dementia. Neurology 1994; 44:2308-2314. 33. Morris JC, Cole M, Banker BQ, Wright D. Hereditary dyspha- 13. Fahn S, Elton R, and members of the UPDRS Development sic dementia and the Pick-Alzheimer spectrum. Ann Neurol Committee. Unified Parkinson’s disease rating scale. In: Fahn 1984; 16:455-466. S, Marsden C, Calne D, Goldstein M, eds. Recent development in 34. Cummings JL, Benson DF. Dementia: a clinical approach. Boston: Parkinson’s disease. Florham Park, NJ: Macmillan Health Care Butterworth-Heinemann, 1992. Information, 1987:153-163.