Medafile.com

Donepezil Treatment and Alzheimer
Disease: Can the Results of Randomized
Clinical Trials Be Applied to Alzheimer
Disease Patients in Clinical Practice?
Jared R. Tinklenberg, M.D., Helena C. Kraemer, Ph.D.,
Kristine Yaffe, M.D., Leslie Ross, Ph.D.,
Javaid Sheikh, M.D., M.B.A., John W. Ashford, M.D., Ph.D.,
Jerome A. Yesavage, M.D., Joy L. Taylor, Ph.D.
Objective: To determine if results from randomized clinical trials of donepezil in Alzheimer disease
(AD) patients can be applied to AD patients in clinical practice by comparing the findings from a Nordic
one-year randomized AD donepezil trial with data from a one-year prospective, observational study of
AD patients. Methods: AD patients from a consortium of California sites were systematically followed
for at least one year. Their treatment regimens, including prescription of donepezil, were determined by
their individual physician according to his or her usual criteria. Results: The 148 California patients
treated with donepezil had a one-year decline of 1.3 (3.5 SD) points on the Mini-Mental State Exam
compared to a decline of 3.3 (4.4 SD) in the 158 AD patients who received no anti-Alzheimer drugs. The
Mini-Mental State Exam decline in Nordic sample was ϳ0.25 points for the 91 patients receiving
donepezil and ϳ2.2 for the 98 placebo patients. The overall effect sizes were estimated at about 0.49 in
both studies. The California data were further analyzed using propensity methods; after taking into
account differences that could bias prescribing decisions, benefits associated with taking donepezil
remained. Conclusion: A comparison of a randomized clinical trial of donepezil in AD patients and
this observational study indicates that if appropriate methodological and statistical precautions are
undertaken, then results from randomized clinical trials can be predictive with AD patients in clinical
practice. This California study supports the modest effectiveness of donepezil in AD patients having
clinical characteristics similar to those of the Nordic study. (Am J Geriatr Psychiatry 2007; 15:953–960)
Key Words: Propensity analyses, observational studies, donepezil effectiveness, clinical
practice, Alzheimer disease
Several randomized clinical trials (RCTs) have ing Alzheimer disease (AD) patients for periods of demonstrated the efficacy and safety of donepe- time up to six months. Two of these studies provided zil and other cholinesterase inhibitors (ChEI) in treat- the “pivotal” phase III data for U.S. Food and Drug Received December 7, 2006; revised March 15, 2007; accepted April 9, 2007. From the Sierra Pacific Mental Illness, Research, Education andClinical Center, Veterans Affairs Palo Alto Health Care System, Palo Alto, CA (JRT, HCK, JS, JWA, JAY, JLT); the Department of Psychiatry andBehavioral Sciences, Stanford University, Stanford, CA (JRT, HCK, JS, JAY, JLT); the Veterans Affairs San Francisco Health Care System and theDepartment of Psychiatry, University of California San Francisco, San Francisco, CA (KY); and the Institute for Health and Aging, University ofCalifornia San Francisco, San Francisco, CA (LR). The authors thank Sally Joseph, Pauline Luu, Art Noda, John Oehlert, and Jean Thompson fortheir technical assistance. Send correspondence and reprint requests to Jared R. Tinklenberg, M.D., Department of Psychiatry and BehavioralSciences, 401 Quarry Rd., C-301, Stanford University School of Medicine, Stanford, CA 94305. e-mail: [email protected] 2007 American Association for Geriatric Psychiatry Am J Geriatr Psychiatry 15:11, November 2007 Donepezil Treatment and Alzheimer Disease Administration (FDA) marketing approval of done- ket in the midst of the trial; after donepezil became pezil.1,2 However, for the practicing clinician, this available in clinical practice, there were more drop- research information is limited by both the short dura- outs from the Nordic double-blind trial.
tion and by the restrictive subject selection criteria that Donepezil was introduced to the California market excluded a number of patients from participating. Most in early 1997. Our state and Veteran’s Affairs (VA)- clinicians are interested in practical information regard- funded California study collected these data from ing the longer-term effects of donepezil on the patients January 1, 1998 through June 30, 2004. Thus, during the initial part of this study, practice guidelines for A significant step toward reducing this informa- donepezil were not available. Many physicians were tion gap was made by a Northern European RCT of cautious about prescribing donepezil because of ad- donepezil.3 This Nordic trial was longer (one year) verse experiences associated with tacrine (Cognex), than any previous RCTs. The issue of restrictive sub- the only other FDA-approved ChEI available at that ject selection remains, however, because the Nordic time. Donepezil use in California increased during trial retained some of the usual RCT subject exclu- the study period; analyses were done to factor in sions. Such limitations reduce the useful information these changes in prescribing patterns.
about donepezil for the many practitioners. For ex-ample, an earlier study of community-dwelling ADpatients evaluated at most of the California sitescollaborating in this present research study indicated that about 90% of them would be excluded fromtypical RCTs of anti-AD medications because of co- Study Design
morbid illnesses, concomitant medications, or otherclinical characteristics.4 Thus, many AD patients This study was designed to amass data from a pro- treated by community physicians would be excluded spective, longitudinal, multisite, observational study in California that could be directly compared to data from We constructed this present, prospective Califor- the Nordic RCT.3 In both studies, the diagnosis of AD nia study from a subset of AD patients who were was consistent with the National Institute of Neurolog- participating in ongoing, clinical observational stud- ical and Communicative Disorders and Stroke–AD and ies. Our aim was to determine how well information Related Disorders Association criteria for probable or from the Nordic RCT translates to AD patients who possible AD and Diagnostic and Statistical Manual of receive their care in more general clinical settings Mental Disorders, Fourth Edition criteria for AD6,7 Men and hence are treated with anti-AD drugs in a non- and women between 40 and 90 years of age were random fashion. An additional goal was to demon- included. Patients had to have mild to moderate AD strate the utility of signal detection– based propen- confirmed by a Mini-Mental State Exam (MMSE) score sity analyses in clinical observational studies. These of Ն10 and Յ26,8 sufficient physical abilities to partic- analyses help identify any biases for certain groups ipate in the initial outpatient diagnostic process, and a of patients to receive one treatment rather than another.
caregiver who agreed to participate in the research and Identifying such biases and determining their impact either lived with or closely monitored the patient. No on treatment outcomes permit more precise interpreta- patients could be taking donepezil or any other ChEI at tions of observational studies. To facilitate comparisons their baseline assessment or during the prior four with the Nordic findings, we limited this California study sample to AD patients who identified themselves After baseline assessment, each patient’s physician as white non-Hispanics. Recent California research in- determined treatment, including whether or not done- dicates that compared with white patients with AD, pezil was prescribed according to his or her usual Hispanics and other ethnic minority patients are less criteria. Patients were identified for the study after they completed a baseline assessment and a follow-up eval- The industry-sponsored Nordic study was con- uation two or more months later when their treatment ducted in the late 1990s and might have been im- status was reconfirmed. All patients were expected to pacted by the introduction of donepezil to the mar- participate in a structured clinic reassessment about Am J Geriatr Psychiatry 15:11, November 2007 one year after baseline. Depending on their clinical dic study as a secondary outcome. The MMSE has been status, some patients were seen more frequently during used extensively in dementia and drug research; it the study period. Patients who took any experimental provides a longitudinal “benchmark” that is under- drug, any other ChEI, or memantine throughout the stood by clinicians from different countries12 and has study period were excluded from the final analyses.
Study Sites
Statistical Analysis
The 11 study sites included eight Alzheimer’s Dis- For the primary outcome measure, a t test was done ease Research Centers of California (ARCCs): Stan- to test for differences between the donepezil and no- ford/Palo Alto VA (the coordinating site), University donepezil groups in one-year change. Supplementary of California Davis at Martinez, University of Califor- data analyses, based on propensity methods,14 were nia Davis at Sacramento, University of California Ir- carried out to address the observational nature of this vine, University of California Los Angeles, University study in which assignment to treatment is nonrandom.
of California San Diego, CA, University of California Propensity methods match patients on baseline charac- San Francisco, and University of Southern California at teristics that predict the propensity for an individual to Rancho, as well as three VA Mental Illness Research receive one treatment over another.15 If, for example, and Education Centers (MIRECC) in Northern Califor- patients with higher baseline MMSE scores are more nia: San Francisco, Martinez, and Palo Alto.
likely to be prescribed donepezil than patients with The ARCC sites have been closely collaborating lower scores, then it is important to match on baseline and using common research data collection protocols MMSE when contrasting treated versus untreated (Minimum Uniform Data Set [MUDS]) for over 10 groups, especially in view of past studies documenting years.9 Data are processed centrally through the In- that the initial MMSE value affects rate of decline in stitute for Health and Aging at the University of California in San Francisco. To increase intersite re- There are several methods for identifying which liability and accuracy, training and recalibration ex- baseline characteristics predict treatment group mem- ercises are held with case reports, videos, and au- bership, including logistic regression, discriminant topsy findings.10,11 The VA–MIRECC sites are also analysis, and signal detection theory (SDT). We chose directed by ARCC consortium investigators and also SDT because it can identify systematic differences in use MUDS protocols. Patients reside in the surround- baseline characteristics, does not assume variables are ing communities and many remain under the care of normally distributed, and will identify interactions among baseline characteristics.17,18 For example, the All sites follow some of their patients to autopsy degree to which initial MMSE predicts treatment status and systematically determine correlations between may depend on the patient’s age. Details of the results premorbid clinical diagnoses and neuropathological based on propensity matching are presented below, findings. Because of resource limitations and the re- following descriptions of the donepezil treatment and alities of clinical research, however, dropouts from nondonepezil groups, and the overall differences in longitudinal study efforts are often considerable. All one-year outcome. Data analyses were performed us- sites are experienced in conducting National Insti- ing SAS version 9.1. The SDT analyses were done with tutes of Health– and industry-sponsored collabora- receiver operating characteristic (ROC) procedures that tive trials of anti-AD medications. The study was are publicly available at http://mirecc.stanford.edu.
part of ongoing multisite research collaborations.
These are carried out in accordance with all applica-ble Institutional Review Board requirements.
Outcome Measure
A total of 502 patients (of whom 421 [83.9%] had The outcome measure of this California study was probable AD and 81 had possible AD) entered the the 30-point MMSE,8 which was also used in the Nor- study. Their baseline characteristics are summarized Am J Geriatr Psychiatry 15:11, November 2007 Donepezil Treatment and Alzheimer Disease TABLE 1.BASELINE CALIFORNIA PATIENT CHARACTERISTICS
DONEPEZIL TREATMENT
NO DONEPEZIL TREATMENT
COMPLETERS
NONCOMPLETERS
COMPLETERS
NONCOMPLETERS
in Table 1. At entry, concomitant medications were FIGURE 1.
Patient flow
being taken by 212 of 241 (88.0%) patients in done-pezil treatment group and 217 of 261 (83.1%) in no-donepezil group.
As summarized in Figure 1, 241 of the 502 patients were prescribed donepezil by their physician accord-ing to his or her usual criteria and 261 were not. Atthe 1-year follow-up period, 148 of the donepeziltreatment group and 158 of the nondonepezil grouphad completed the study (Figure 1). To be a studycompleter, the patient needed to have an MMSEscore 10 to 18 months after the baseline visit. Themost common reasons for patients to not completethe study protocol were: 1) they did not return for anin-clinic assessment until after the 18-month end ofstudy date; 2) they did have a clinic or phone assess-ment during the study period, but a MMSE scorewas not obtained because of resource or other limi-tations; 3) the donepezil treatment patients stoppedtaking donepezil, switched to another CHEI, oradded memantinel; and 4) the no-donepezil treat-ment patients started taking a CHEI or started anexperimental drug.
In investigating completer versus noncompleter biases using ROC analyses, we found that the biaseswere primarily due to time of study entry and sitefactors rather than clinical characteristics. Specifi- pleters, the remainder of the results will focus on the cally, one site had a disproportionate number of noncompleters in both donepezil and no-donepezilgroups; at this site, no-donepezil subjects were more Comparison of Donepezil and No-Donepezil
likely to be a study completer if their baseline visit occurred prior to 2001. Analyses indicated that inclu-sion of this particular site did not bias overall results.
As indicated in Figure 2, the 148 California pa- In view of the fact that there were no significant tients who completed the study and received done- clinical differences between completers and noncom- pezil treatment had an average one-year decline of Am J Geriatr Psychiatry 15:11, November 2007 In addition, patients were somewhat less likely to be FIGURE 2.
California Observational AD Study Versus the
prescribed donepezil regardless of their clinical charac- Nordic Randomized Clinical Trial
teristics in the earlier years of this study when donepe-zil prescribing was less widespread. As described be-low, a modest donepezil-associated benefit remainedwhen these and other baseline differences were takeninto account.
The first step of the propensity analysis was to identify baseline characteristics that significantly in-fluenced the likelihood or propensity of whether ornot donepezil was prescribed. We used ROC classi-fication tree methods with a p value of Ͻ0.01 toidentify variables suggested by the literature thatmight explain whether or not they were prescribeddonepezil. These 27 variables included both patientcharacteristics, such baseline cognitive status (MMSE),age of disease onset, comorbid illnesses, concomitantmedications, years of education, sex, marital status,relationship with caregiver, living arrangement, andveteran status, as well as nonpatient characteristics, Shown are Mini-Mental State Exam scores at baseline and such as date of baseline assessment and study site.
1 year for four groups. California patients prescribed donepezil who The ROC tree analysis indicated that the first were study completers (N ϭ 148, thin solid line), Nordic patients (strongest) branching variable predicting donepezil treated with donepezil who were study completers (N ϭ 91, thicksolid line), Nordic patients treated with placebo who were study prescription was the date of the baseline assessment; completers (N ϭ 98, thick broken line), and California patients not patients prescribed donepezil were more likely to prescribed donepezil who were study completers (N ϭ 158, thinbroken line).
have their baseline in 2000 or later. There was asecond branch within the group seen before 2000, inwhich patients who had no significant comorbid ill- 1.3 points (3.5 SD) on the MMSE compared to a nesses were more likely to be prescribed donepezil 3.3-point (4.4 SD) decline in the 158 completers who than those who did. The ROC identified no other sig- received no donepezil or other anti-AD drugs during nificant variables predicting donepezil prescription.
the study period. This difference was statistically We then stratified all patients into the above three significant (t304ϭϪ4.32, p Ͻ0.0001). By comparison, propensity groups that differed in characteristics as- the annualized MMSE decline in Nordic sample was sociated with donepezil prescribing; 1) patients seen about 0.25 for the 91 donepezil patients who com- in 2000 or later; 2) patients seen before 2000 who had pleted the trial and about 2.2 for 98 placebo compl- no comorbid illnesses; and 3) patients seen before eters. The overall effect sizes of these treatment- 2000 with one or more comorbid illnesses. A total of associated differences were estimated at about 0.49 59.3% (96 of 162) patients in propensity group 1 were in both the California and the Nordic studies.19 prescribed donepezil, 53% (28 of 53) in group 2, andonly 26% (24 of 91) in group 3.
In each propensity group, patients who received Propensity Analyses
donepezil declined less than patients who were not Despite the comparable effect sizes of the two stud- prescribed donepezil. The estimated treatment effect ies, it is important to identify biases in nonrandomized size was 0.37 for propensity group 1, 0.19 for propen- studies that might affect whether patients were in a sity group 2, and 0.79 for propensity group 3. A 2 ϫ 3 given treatment group or not. In this present study, for analysis of variance was used to compare differences in example, patients who were prescribed donepezil had MMSE decline across the six groups (donepezil versus baseline MMSE scores that were 1.8 points higher on no donepezil, stratified by propensity). The differences average than those who were not prescribed donepezil.
in rate of decline according to the propensity grouping Am J Geriatr Psychiatry 15:11, November 2007 Donepezil Treatment and Alzheimer Disease were not significant (F2,300ϭ2.10, pϭ0.12). The differ- from donepezil treatment. The Nordic trial excluded ence in decline associated with donepezil was signifi- certain medical conditions such as obstructive pulmo- cant (F1,300ϭ12.39, p Ͻ0.001). Third, the propensity ϫ donepezil interaction was not significant (Fϭ1.66, pϭ Some California patients had one or more interim 0.19), indicating that the benefit associated with done- clinic visits or phone follow-ups before their one-year pezil was comparable within groups of AD patients reassessment, but in general, they received less moni- matched for their likelihood of being prescribed done- toring and less frequent repeated administrations of the pezil. Taken together, these results indicate that the MMSE, with its known practice effects, than the Nordic modest benefit associated with taking donepezil re- patients.21 In the Nordic study, the patients agreed to mained after taking into account the lack of random- four scheduled in-clinic follow-up assessments and they knew they would receive randomly either done-pezil or placebo.3 With less scheduled attention, theCalifornia patients who were prescribed donepezilmay have been less compliant. Within the California DISCUSSION
study, patients prescribed donepezil were more likelyto have interim clinic visits than those who did not This one-year observational study supports the effec- receive donepezil. This could have contributed to their tiveness of donepezil treatment on cognitive function somewhat better 1-year scores. These various consid- in some California AD patients who were selected to erations may explain the differences found between the have clinical and ethnic characteristics similar to pa- tients in the Nordic randomized clinical trial.3 The These differences can also be considered in the con- magnitude of the positive effects with donepezil was text of a randomized, placebo-controlled one-year U.S.
smaller in the California study: the one-year MMSE clinical trial of donepezil that had more restrictive entry decline in California sample was ϳ1.3 for patients criteria than either the California or Nordic studies.22 prescribed donepezil and ϳ0.25 for Nordic patients The MMSE scores for both donepezil and placebo receiving donepezil. However, the effect size of the groups were generally better throughout this U.S. trial differences between donepezil versus no donepezil than in the Nordic or California studies. These MMSE in both studies was about 0.49, or one-half of a stan- variations underscore the importance of methodologi- dard deviation. Thus, our analyses indicated that cal details in interpreting study results.
results from the Nordic RCT can be predictive with One limitation of this California study was the AD patients in clinical practice if appropriate meth- possibility of biases due to the high dropout rate. In odological and statistical precautions are considered.
all, 39% of the California patients did not complete The differences in rate of MMSE decline between the the study. The Nordic study had a noncompletion two studies may result from a combination of factors rate of about 33%, with the largest portion of drop- such as differences in concomitant medications, co- outs occurring later in the trial when donepezil be- morbid conditions, frequency of monitoring, and came clinically available. In this California study, study completers and noncompleters did not differ In this California study, the patients and their care- significantly in their baseline clinical characteristics.
givers knew that donepezil and other drugs would be Although there were two nonclinical factors associ- prescribed in accordance with their physician’s usual ated with completing the study—site differences and clinical practices. This meant that some California pa- time of study entry—analyses indicated that these tients received medications with anticholinergic or factors did not bias overall results. Taken together, other psychoactive properties that could interfere with these findings indicate it is unlikely that the factor of the effects of donepezil.20 The Nordic patients knew noncompleters significantly impacted the California their study medications would be randomly assigned, results. We did not perform “intention-to-treat” anal- and their use of other medications would be restricted.
yses in this study because, in conditions like Alzhei- The California patients also were less restricted in mer disease where deterioration over time is likely, terms of comorbid medical problems; they could have carrying forward the last observation is problematic.
other illnesses if their physician felt they might benefit Interpreting differences between active and placebo Am J Geriatr Psychiatry 15:11, November 2007 groups is especially problematic if treated patients them. Thus, more representative patient samples in tend to drop out early because of unpleasant side observational studies may partially offset the inher- effects and relatively high cognitive scores are car- ent advantages of RCTs in which most biases are ried forward to the endpoint. The side effects re- ported by patients prescribed donepezil were similar Consequently, observational studies can provide to those reported in the Nordic and other studies: useful information on what a typical physician can mainly nausea or other gastrointestinal symptoms, expect in his or her clinical practice if appropriate mild depression, and vague anxiety or agitation.
procedures, such as propensity analyses, are uti- An additional limitation of this California study is lized to assure that nonrandom factors do not sig- nonrandomness of prescribing, in that each patient’s nificantly bias the findings. When we completed physician decided by his or her usual criteria these analyses, our one-year California study indi- whether or not to treat with donepezil. Nonrandom cated that donepezil treatment does have modest prescribing may result in two groups that differ sub- effectiveness in AD patients having clinical and stantially in either clinical or nonclinical factors, ethnic characteristics similar to those of the Nordic which may in turn lead to a biased estimate of a treatment effect. To gain a clearer picture, not biasedby baseline differences, we conducted propensityanalyses.14,15 Propensity analyses have similarlybeen used in cardiology and cancer research to ex- LIST OF CONTRIBUTORS
amine patient outcomes in clinical practice, whenmatching on key baseline characteristics is also im- Alzheimer’s Disease Research Centers of California: portant.23 When we performed these propensity University of California Davis–Martinez: B. Reed, C.
analyses, the statistical significance of the donepezil- Bibeau, J. Coleman, and J. Webb; University of Cal- associated benefit remained. The effect size of done- ifornia Davis–Sacramento: D. Mungas, C. DeCarli, pezil, which was estimated at about 0.49 in both the W. Jagust, B. Henk, and M. Verma; University of California and the Nordic studies, may be viewed as California Irvine: C. Cotman, R. Mulnard, M. Dick, small to moderate in terms relevant to clinical prac- C. Kawas, and H. Kim; University of California Los tice.24 A related limitation of this California study is Angeles: J. Cummings, J. Ringman, M. Carter, K.
that MMSE testing was not done blind to treatment Metz, and L. Brendt; University of California SanDiego: D. Galasko, D. Salmon, C. Robinson, and M.
status, as is done in randomized clinical trials. The Sundsmo; University of California San Francisco: B.
effect size is often exaggerated in observational stud- Miller, K. Yaffe, J. Johnson, J. Kramer, and R. Gear- ies, including nonblinded studies, compared to ran- heart; University of Southern California–Rancho Los domized clinical trials that address the same research Amigos: H. Chui, F. Segal-Gidan, B. Smith, and A.
question.25 Importantly, the effect sizes in this Cali- Ireland; Stanford/Palo Alto VA (coordinating site): fornia observational study were not larger those of H. Davies, J. Kim, H. Leutwyler, P. Luu, C.
the Nordic randomized clinical trial.
McFeeters, L. Newkirk, J. Oehlert, and J. Thompson; One advantage of this observational AD treatment Institute of Health and Aging & Alzheimer’s Disease study is that the results probably generalize into a Program of California: P. Fox, D. Tyrrell, C. Moty- typical clinical practice more directly than results lewski-Link, R. Chapman, M. Lackey, and P. Tang; from RCTs. RCTs often have exclusionary criteria Veteran Affairs Mental Illness Research and Educa- that screen out a number of patients typically seen in tion Centers in Northern California: Martinez: B.
ordinary clinical practice.4,26 In addition, some pa- Reed, C. Bibeau, and J. Webb; Palo Alto: J. Yesavage, tients who fulfill all entry criteria for a RCT are E. Gere, and R. O’Hara; San Francisco: K. Yaffe, E.
unwilling to participate in a RCT where they may not receive active medications.27 It is usually easier toenroll a more representative sample of patients intostudies where they are assured of getting whatever This work was supported by the State of California, treatment their individual physician thinks is best for Department of Health Services (grant 03-75273), VA Am J Geriatr Psychiatry 15:11, November 2007 Donepezil Treatment and Alzheimer Disease Sierra Pacific Mental Illness Research and Education Presented at the 10th International Conference on Centers, and by the National Institute on Aging Alzheimer’s Disease and Related Disorders, Madrid, References
1. Rogers SL, Doody RS, Mohs RC, et al: Donepezil improves cogni- development of a global functioning measure of dementia with tion and global function in Alzheimer’s disease: a 15-week, dou- linear measurement properties. Stat Med 2000; 19:1631–1644 ble-blind, placebo-controlled study. Arch Intern Med 1998; 158: 14. Rubin DB: Estimating causal effects from large data sets using propensity scores. Ann Intern Med 1997; 127:757–763 2. Rogers SL, Farlow MR, Doody RS, et al: A 24-week, double blind, 15. D’Agostino RB Jr: Propensity score methods for bias reduction in placebo-controlled trial of donepezil in patients with Alzheimer’s the comparison of a treatment to a non-randomized control 3. Winblad B, Engedal K, Soininen H, et al, and the Donepezil 16. Mendiondo MS, Ashford JW, Kryscio RJ, et al: Modeling Mini Nordic Study Group: a 1-year, randomized, placebo-controlled Mental State Examination changes in Alzheimer’s disease. Stat study of donepezil in patients with mild to moderate AD. Neu- 17. Kiernan M, Kraemer HC, Winkleby MA, et al: Do logistic regres- 4. Schneider LS, Olin JT, Lyness SA, et al: Eligibility of Alzheimer’s sion and signal detection identify different subgroups at risk? disease clinic patients for clinical trials. J Am Geriatr Soc 1997; Implications for the design of tailored interventions. Psychol 5. Mehta KM, Yin M, Resendez C, et al: Ethnic differences in ace- 18. Kraemer HC, Lowe KK, Kupfer DJ: To Your Health: How To tylcholinesterase inhibitor use for Alzheimer disease. Neurology Understand What Research Tells Us About Risk. New York, Ox- 6. McKhann G, Drachman D, Folstein M, et al: Clinical diagnosis of 19. Cohen J. Statistical Power Analysis for the Behavioral Sciences.
Alzheimer’s disease: report of the NINCDS-ADRDA Work Group under the auspices of the Department of Health and Human 20. Huey ED, Taylor JL, Luu PA, et al: Factors associated with use of Services Task Force on Alzheimer’s Disease. Neurology 1984; medications with potential to impair cognition or cholinesterase inhibitors among Alzheimer’s disease patients. Alzheimer De- 7. American Psychiatric Association: Diagnostic and Statistical Man- ual of Mental Disorders, 4th Ed. Washington, DC, American Psy- 21. Galasko D, Abramson I, Corey-Bloom J, et al: Repeated exposure to the Mini-Mental State Examination and the Information-Memory- 8. Folstein MF, Folstein SE, McHugh PR: “Mini-Mental State”: a prac- Concentration Test results in a practice effect in Alzheimer’s disease.
tical method for grading the cognitive state of subjects for the clinician. J Psychiatr Res 1975; 12:189–198 22. Mohs RC, Doody RS, Morris JC, et al: A 1-year placebo controlled 9. Edwards ER, Lindquist K, Yaffe K: Clinical profile and course of preservation of function survival study of donepezil in AD pa- cognitively normal patients evaluated in memory disorders clin- 23. Winkleby MA, Flora JA, Kraemer HC: A community-based heart 10. Chui HC, Victoroff JI, Margolin D, et al: Criteria for the diagnosis disease intervention: predictors of change. Am J Public Health of ischemic vascular dementia proposed by the State of California Alzheimer Disease Diagnostic and Treatment Centers. Neurology 24. Kraemer HC: Reporting the size of effects in research studies to facilitate assessment of practical or clinical significance. Psycho- 11. Chui HC, Mack W, Jackson JE, et al: Clinical Criteria for the Diagnosis of Vascular Dementia: A multicenter study of compa- 25. Ioannidis JP. Contradicted and initially stronger effects in highly rability and interrater reliability. Arch Neurology 2000; 57:191– cited clinical research. JAMA 2005;294:218–228 26. Humphreys K, Weisner C: Use of exclusion criteria in selecting 12. Salmon DP, Thal LJ, Butters N, et al: Longitudinal evaluation of research subjects and its effect on the generalizibility of alcohol dementia of the Alzheimer’s type: a comparison of three stan- treatment outcome studies. Am J Psychiatry 2000; 157:588–594 dardized mental status examinations. Neurology 1990; 40:1225– 27. Rush AJ, Trivedi MH, Wisniewski SR, et al: Bupropion-SR, sertra- line or venlafaxine-XR after failure of SSRIs for depression. N Engl 13. Mungas D, Reed BR: Application of item response theory for Am J Geriatr Psychiatry 15:11, November 2007

Source: http://www.medafile.com/jwa/Tink-2007.pdf