- Email: [email protected]
A Double-Blind Comparison of Citalopram and Risperidone for the Treatment of Behavioral and Psychotic Symptoms Associated With Dementia
A Double-Blind Comparison of Citalopram and Risperidone for the Treatment of Behavioral and Psychotic Symptoms Associated With Dementia Bruce G. Pollock, M.D., Ph.D., Benoit H. Mulsant, M.D., Jules Rosen, M.D., Sati Mazumdar, Ph.D., Richard E. Blakesley, B.S., Patricia R. Houck, M.S.H., Kimberly A. Huber, M.P.H.
Objective: To compare citalopram and risperidone for the treatment of psychotic symptoms and agitation associated with dementia, with a priori hypotheses that risperidone would be more efficacious for psychosis and citalopram for agitation. Methods: A 12-week randomized, controlled trial in nondepressed patients with dementia hospitalized because of behavioral symptoms (N ⫽ 103) was conducted at the University of Pittsburgh Medical Center. Participants were consecutively recruited on an inpatient unit if they had at least one moderate to severe target symptom (aggression, agitation, hostility, suspiciousness, hallucinations, or delusions). Once they improved sufficiently, they were discharged to nursing homes, personal care homes, or residential homes for continued treatment. Planned pre-post and mixed model analyses of the main outcome measures of Neurobehavioral Rating Scale and Side Effect Rating Scale at baseline and at weekly/biweekly intervals were conducted. Results: Completion rates did not differ for citalopram and risperidone (overall completion rate: 44%). Agitation symptoms (aggression, agitation, or hostility) and psychotic symptoms (suspiciousness, hallucinations, or delusions) decreased in both treatment groups but the improvement did not differ significantly between the two groups. There was a significant increase in side effect burden with risperidone but not with citalopram such that the two groups differed significantly. Conclusion: No statistical difference was found in the efficacy of citalopram and risperidone for the treatment of either agitation or psychotic symptoms in patients with dementia. These findings need to be replicated before citalopram or other serotonergic antidepressants can be recommended as alternatives to antipsychotics for the treatment of agitation or psychotic symptoms associated with dementia. (Am J Geriatr Psychiatry 2007; 15:942–952) Key Words: Dementia, Alzheimer disease, behavioral disturbance, psychosis, agitation, pharmacotherapy, citalopram, risperidone
Received March 23, 2007; revised April 13, 2007; accepted April 16, 2007. From the Rotman Research Institute at Baycrest (BGP) and the Geriatric Mental Health Program, Centre for Addiction and Mental Health (BGP, BHM), University of Toronto, Ontario, Canada; and the Department of Psychiatry (BGP, BHM, JR, PRH, KAH) and the Department of Biostatistics (SM, REB), Graduate School of Public Health, the University of Pittsburgh, Pittsburgh, PA. Send correspondence and reprint requests to Bruce G. Pollock, The Rotman Research Institute at Baycrest, 3560 Bathurst St., Toronto, ON M6A 2E1, Canada. e-mail: [email protected] © 2007 American Association for Geriatric Psychiatry
942
Am J Geriatr Psychiatry 15:11, November 2007
Pollock et al.
N
oncognitive symptoms such as agitation or psychosis are among the most distressing manifestations of dementia.1 Optimal management of these symptoms includes the search for treatable physical and environmental precipitants; nevertheless, pharmacotherapy is frequently used.2 Current guidelines recommend the use of second-generation (“atypical”) antipsychotics.2– 4 Nonetheless, there are concerns regarding both their safety5,6 and effectiveness7 in patients with dementia. Sparse and inconclusive evidence support the use of alternative agents such as antidepressants or cognitive enhancers.2 In a previous randomized controlled trial (RCT), citalopram was more efficacious than placebo and as efficacious as, but better tolerated than, perphenazine in patients with dementia hospitalized for the treatment of agitation or psychosis.8 Therefore, we conducted an RCT to compare the efficacy and safety of citalopram and risperidone. Based on the existing literature when the study was designed9,10 and our prior work,8 we hypothesized that risperidone would be more efficacious for psychotic symptoms, citalopram would be more efficacious for agitation, and citalopram would be associated with fewer side effects.
METHODS Setting and Participants Participants of this 12-week clinical trial were recruited upon admission to the geropsychiatric unit of an academic hospital providing treatment to a large urban and suburban population. If they improved sufficiently, participants were discharged to nursing homes, personal care homes, or residential homes for continued treatment under double-blind conditions. Admissions were eligible if they had a dementia of the Alzheimer type (DAT), vascular dementia, dementia with Lewy bodies, mixed dementia, or dementia not otherwise specified. After a warning was issued to Canadian physicians in 2002 regarding cerebrovascular events associated with risperidone, we excluded patients with vascular dementia. Target symptoms had to be of moderate or higher severity as evidenced by the need for hospitalization and a rating of 3 or higher (moderate to severe) on at least one of the agitation items (aggression, agitation, hos-
Am J Geriatr Psychiatry 15:11, November 2007
tility) or psychosis items (suspiciousness, hallucinations, or delusions) of the Neurobehavioral Rating Scale (NBRS).11 Patients were excluded if they had a current or past diagnosis of schizophrenia, schizoaffective disorder, delusional disorder, psychotic disorders not otherwise specified, bipolar disorder, mental retardation, cognitive deficits following head trauma, or a current diagnosis of delirium, substance-induced persisting dementia, Parkinson disease, drug/alcohol abuse, or dependence. Other exclusion criteria were a major depressive episode within the past 6 months or clinically significant depressive symptoms with a rating of 12 or higher on the Cornell Scale for Depression in Dementia (CSDD)12; unstable physical illness; creatinine ⱖ2.0 mg/100 mL; aspartate aminotransferase or bilirubin more than twice the upper limit of normal; potentially reversible cause of dementia (e.g., untreated hypothyroidism or B12 deficiency); treatment with a depot neuroleptic drug within 2 months of screening or fluoxetine within 4 weeks of screening; or a history of allergy or intolerance to citalopram or risperidone. As required by the Institutional Review Board, participants’ authorized representatives provided written informed consent with participants’ verbal assent. From February 2000 to June 2005, 408 consecutive admissions were prescreened. The three most common reasons patients were deemed ineligible after prescreening were: 1) unstable or excluded medical illness (e.g., seizure disorder or Parkinson disease); 2) insufficient severity of target symptoms; 3) an excluded psychiatric diagnosis (e.g., mood disorder) or previous participation in the trial. Less than 20% of prescreened admissions did not sign consent because they, their family, or their psychiatrist preferred treatment outside of the study. Written consent was obtained for 111 patients and 103 were randomized (Fig. 1). Treatment Protocol Participants could continue cholinesterase inhibitors or memantine if they had been taking them for at least 12 weeks at the same dose for at least 4 weeks. Other psychotropics (except for lorazepam, see below) were discontinued prior to randomization. After baseline ratings, participants were randomized to citalopram or risperidone. The project biostatisticians (PRH and SM) generated the randomization sched-
943
Citalopram and Risperidone for Dementia
FIGURE 1.
Participants’ Flow
ule at the beginning of the trial. Randomization was stratified according to the presence or absence of psychotic symptoms (defined as a moderate to severe score on NBRS items rating delusions, suspicious thoughts, or hallucinations) and blocked to balance cell sizes over the study period. Only the research pharmacist knew the treatment assignments; participants, caregivers, investigators, and assessors remained blind throughout the study. Initial and target dosages of risperidone were chosen based on the results of one large study conducted in nursing home patients.9 Initial and target dosages
944
of citalopram were chosen based on our own previous placebo-controlled study.8 Study medications were administered once a day using identical capsules containing 10 mg of citalopram or 0.5 mg of risperidone, starting with one capsule at bedtime for 3 days and then two capsules per day. After at least two weeks, two additional increases (up to a maximum of four capsules per day) were allowed based on assessment of response and treatment-emergent side effects. These increases were separated by at least two weeks. At any time, dosages could also be titrated downward in the presence of adverse effects
Am J Geriatr Psychiatry 15:11, November 2007
Pollock et al. or intercurrent physical problems. Participants were discontinued in the event of new medical symptoms that endangered their safety or health; adverse events that could not be managed by a dose reduction; a significant worsening of agitation or psychosis; or an administrative reason (e.g., moving away, withdrawal of consent; Fig. 1). Lorazepam up to 2 mg/day could be used throughout the trial for acute treatment of extreme agitation or aggression. The lowest possible dosage for the shortest period of time was used. Assessments Diagnostic and laboratory assessments were completed upon admission to evaluate for reversible causes of dementia, agitation, or psychosis. Participants were also assessed at the time of enrollment; baseline (i.e., time of randomization); after receiving study medication for 3 days, 7 days, then weekly for 5 weeks, then every 2 weeks; and at the time of discharge from the hospital or termination if the study was terminated early. Information was obtained from direct observation of the participants, interviews with participants and nursing staff, and chart review of all notes for the 3 days preceding the rating. In addition to the NBRS, instruments included the CSDD,12 Neuropsychiatric Inventory (NPI),13 Udvalg for Kliniske Undersogelser (UKU) side effect scale,14 Mini-Mental State Examination (MMSE),15 Severe Impairment Battery (SIB),16 and Cumulative Illness Rating Scale–Geriatrics.17 The NBRS is a 28-item observer-rated instrument combining the breadth of psychopathology covered by the Brief Psychiatric Rating Scale with more comprehensive assessment of impairments seen in dementia. Scoring is on a 7-point scale (0: not present; 1: very mild; 2: mild; 3: moderate; 4: moderately severe; 5: severe; 6: extremely severe).12 All raters participated in initial and yearly training sessions during which interrater reliability was established and monitored for the NBRS, UKU, SIB, and MMSE. Throughout the study interrater reliability was good to excellent. Diagnoses were established according to Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM–IV) criteria considering all available information during a consensus conference attended by three investigator geropsychiatrists and the re-
Am J Geriatr Psychiatry 15:11, November 2007
search staff. Of the randomized participants, 86 received a DSM–IV consensus diagnosis of DAT with late onset; 5 DAT with early onset; one of vascular dementia; and 9 dementia not otherwise specified. Following their randomization, two participants were judged to have a diagnosis that should have led to their exclusion (one with late-onset DAT with delirium randomized to citalopram and one with substance-induced persisting delirium randomized to risperidone). Following the intent-to-treat principle, these two participants are included in all analyses. Diagnoses were distributed similarly in the two treatment groups. Statistical Analysis We prospectively estimated sample size based on our previous study conducted in the same inpatient setting and expected percent changes in NBRS agitation or psychosis scores with each drug.8 A priori, a difference between the two treatments of less than 15% was judged not to be clinically significant. Therefore, a sample size of 103 was selected because it is sufficient to detect a 15% or larger difference with a power of 80% and a type I error of 0.05. All analyses were conducted using SAS18 based on the randomized assignment. First, we compared the baseline characteristics of the treatment groups using t or 2 tests. Then, we compared the dropout rates using log-rank statistics. We then performed two sets of analyses including all randomized participants who received at least one dose of study medication and who completed the baseline and at least one post baseline evaluation. All statistical tests were twosided. A first set of pre-post analyses compared ratings at baseline (pre) and at the time of termination from the study (post) for each treatment, and the changes observed with each treatment. A second set of analyses consisted of a mixed model analysis of the main efficacy outcomes (NBRS agitation and psychosis scores), comparing the two treatments over the total follow-up period using all randomized participants. Mixed model analyses use all the data that are collected longitudinally and also account for the within-subject correlations between measures. As routine application of these models can lead to erroneous conclusions when data are not missing at random, we investigated the reasons for dropping out and effects of different dropout times on the param-
945
Citalopram and Risperidone for Dementia eter estimates using a pattern-mixture model.19,20 This approach stratifies the observed data according to the dropout patterns (number and timing of missing values), models the response process within each pattern, and combine the parameter estimates. Using SAS-PROC MIXED, the mixed models included treatment (group), week (time), and group-by-time interaction terms. The intercept and the week terms were assumed to be random effects. As proposed in our a priori data-analytic plan, we used the NBRS because it was sensitive to change in agitation and psychosis in our previous study.8 However, since the NPI is used in some trials, we also performed the same mixed model analysis using NPI scores. Results were qualitatively similar and thus we present only the planned NBRS analyses.
RESULTS Of the randomized participants, 53 were randomized to citalopram and 50 to risperidone. The sex distribution and baseline NBRS total scores differed significantly between the two groups (Table 1). Thirty
TABLE 1.
Baseline Participant Characteristics
Characteristic N Age (years) Female, % (N)a White, % (N) Cumulative Illness Rating Scale–Geriatrics score Mini Mental State Examination score Median Interquartile range Severe Impairment Battery Cornell Scale for Depression in Dementia Delusions or hallucinations, % (N) NBRS total scoreb NBRS agitation score (items 8, 11, and 14) NBRS psychosis score (items 16, 18, and 20) NPI total score UKU total score
Citalopram
Risperidone
53 82.0 (7.3) 47.2 (25) 81.1 (43) 9.6 (3.0)
50 81.5 (9.2) 76.0 (38) 82.0 (41) 9.5 (4.5)
9.8 (8.5)
11.9 (8.4)
7 2–18 63.9 (35.5) 10.7 (11.8)
11.5 4.5–19.5 71.4 (29.3) 13.3 (16.2)
35.8 (19) 60.3 (16.8) 10.1 (4.8)
40.0 (20) 53.6 (15.6) 8.9 (4.5)
5.9 (3.5)
6.0 (4.5)
40.8 (23.1) 13.7 (5.2)
34.4 (21.1) 12.1 (4.1)
participants continued a stable dose of a cholinesterase inhibitor: 16 (30.2%) randomized to citalopram and 14 (28.0%) randomized to risperidone (Fisher exact p⫽0.83). No participant was on memantine during the study. Mean (SD) maximum doses were 31.1 (8.7) mg/day of citalopram and 1.36 (0.47) mg/ day for risperidone, corresponding to 3.1 (0.9) and 2.7 (1.0) capsules, respectively (t⫽ 2.19, df⫽101, p⫽ 0.031). Mean (SD) final doses were 29.4 (9.9) mg/day of citalopram and 1.25 (0.51) mg/day of risperidone, corresponding to 2.9 (1.0) and 2.5 (1.0) capsules, respectively (t⫽ 2.25, df⫽101, p⫽0.027). Similar numbers of participants in each treatment group received lorazepam (citalopram: N⫽43; risperidone: N⫽41; Fisher exact p ⫽1.00). Mean (SD) lorazepam daily doses and treatment durations did not differ significantly between the two groups: citalopram was 0.6 (0.2) mg/day for 7.8 (7.0) days and risperidone was 0.6 (0.7) mg/day for 8.0 (8.3) days (dose: t⫽0.04, df⫽ 82, p⫽0.96; duration: t⫽⫺0.12, df ⫽82, p⫽0.90). Dropouts Overall, 43.7% of the participants completed the 12-week trial: 47.2% in the citalopram group and 40.0% in the risperidone group (Fig. 2). Time to dropout for all reasons did not differ significantly between the two groups (log-rank 21 ⫽ 0.85, p⫽0.36). The distribution of the causes to which dropouts were attributed did not differ either (Fig. 1; early dropouts: Fisher exact p ⬍0.32; late dropouts: Fisher exact
FIGURE 2.
Kaplan-Meier Plot of Time to Dropout for All Reasons
Notes: Data are means (SD) unless noted. a Significant difference between treatment groups: 2 ⫽ 9.00, df ⫽ 1, p ⫽ 0.003. b Significant difference between treatment groups: t ⫽ 2.04, df ⫽ 94, p ⫽ 0.044.
946
Am J Geriatr Psychiatry 15:11, November 2007
Pollock et al. p ⬍0.73). Numbers and specific reasons for dropouts were as follows for early citalopram dropouts: intercurrent medical problems, 2 (seizure, ileus); adverse events, 3 (gait disturbance, other extrapyramidal symptoms [EPS], bruising); psychiatric worsening, 9 (increased agitation, 7; onset of psychosis, onset of depression); and administrative reasons, 1. For late citalopram dropouts, the reasons included: intercurrent medical problems, 3 (infection, 2; hypoglycemia); adverse events, 1 (sedation); psychiatric worsening, 7 (increased agitation, 5; readmission, 2); and administrative reasons, 2. Reasons for early risperidone dropouts included: intercurrent medical problems, 4 (gastrointestinal bleeding, intracranial bleeding, pneumonia, ileus); adverse events, 7 (gait disturbance, 3; other EPS, 3; hypotension); psychiatric worsening, 5 (increased agitation, 4; onset of psychosis); and administrative reasons, 3. For late risperidone dropouts, reasons included: intercurrent medical problems, 2 (pneumonia, unresponsiveness); adverse events, 2 (elevated liver function tests, unresponsiveness); psychiatric worsening, 7 (increased agitation, 3; readmission, 3; suicide attempt). Pre-Post Analysis for Agitation and Psychosis Outcomes Changes in NBRS agitation or psychosis scores did not differ significantly between the two groups (Table 2). There was a significant decrease in agitation score with citalopram (⫺12.5%) but not with risperidone (⫺8.2%). Psychosis scores decreased significantly both with citalopram (⫺32.3%) and risperidone (⫺35.2%). Pre-Post Analysis for the Side Effect Outcomes Changes in UKU total score and psychic subscore differed significantly between the two groups (Table 2). UKU total scores increased significantly with risperidone (19.2%) but not with citalopram (⫺3.6%). Because some of these measures were skewed, comparisons were repeated using nonparametric Kruskal-Wallis tests; the same statistical differences were detected (data not shown). Inspection of specific items suggested that these differences were primarily driven by a 26.2% decrease in sedation with citalopram and an 83.3% increase with risperidone [UKU somnolence mean (SD) score: citalopram pre, 0.42 (0.72); citalopram post, 0.30 (0.64); risperidone pre, 0.24 (0.60); risperidone post, 0.45 (0.79)]. Comparable increases in EPS were
Am J Geriatr Psychiatry 15:11, November 2007
seen with both medications: 87.5% with citalopram and 102.5% with risperidone [UKU rigidity and tremor items score: citalopram pre, 0.32 (0.61); citalopram post, 0.60 (0.86); risperidone pre, 0.40 (0.90); risperidone post, 0.81 (1.31)]. Efficacy: Mixed Model Analysis Investigation of the specific reasons for dropping out revealed that they were mostly based on measured values with only a few due to administrative reasons (Fig. 3). The analyses, using a pattern-mixture model where data were stratified by dropouts before six weeks (when most of the dropouts occurred), after six weeks, and completers, were in agreement with the results of the mixed model analyses. Thus, the requirement of the missing at random (MAR) assumption was considered to be satisfied and we present the results of the mixed models that used all observed data of the 103 participants and compared the two treatments over the total follow-up time. A significant time effect was found for both efficacy outcomes (agitation: F1,67 ⫽10.4, p ⬍0.002; psychosis: F1,60 ⫽ 28.9, p ⬍0.001). However, there was no significant group effect or group-by-time interaction for either outcome. The same mixed model analysis excluding the patients without a diagnosis of Alzheimer disease produced similar results, showing only significant time effects for both efficacy outcomes. Because of the baseline difference in sex distribution, another model included sex as an effect. As in the original model, there was no group effect or group-by-time interaction for either outcome. However, both a time (F1,61.9 ⫽ 6.57, p ⬍0.013) and timeby-sex interaction (F1,61.9 ⫽8.95, p ⬍0.004) were observed for psychosis (such that women improved faster than men) but not for agitation. Finally, because recruitment took place over 6 years, a last model included the year of recruitment as an effect. The results of this model were similar to the original model, showing only a time effect for both outcomes.
DISCUSSION In patients with dementia hospitalized for the treatment of behavioral or psychotic symptoms, citalopram and risperidone had similar efficacy but cita-
947
Citalopram and Risperidone for Dementia
TABLE 2.
Efficacy and Tolerability Analysis
NBRS agitation score (items 8, 11, and 14) Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI) NBRS psychosis score (items 16, 18, and 20) Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI) UKU total score Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI) UKU psychic subscale score Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI) UKU neurologic subscale score Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI) UKU autonomic subscale score Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI) UKU other subscale score Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI)
Citalopram
Risperidone
Citalopram vs. Risperidone
10.08 (4.81) 8.81 (5.63) ⫺1.26 (4.58) ⫺2.01 (52), 0.050 (⫺2.527, ⫺0.001) 0.28 (0.0002, 0.55)
8.94 (4.54) 8.20 (5.08) ⫺0.73 (4.91) ⫺1.05 (48), 0.30 (⫺2.145, 0.676) 0.15 (⫺0.13, 0.43)
⫺0.56 (100), 0.57 (⫺2.394, 1.336) 0.11 (⫺0.28, 0.50)
5.88 (3.51) 3.98 (3.90) ⫺1.90 (4.49) ⫺3.02 (50), ⬍0.004 (⫺3.165, ⫺0.639) 0.42 (0.13, 0.71)
6.13 (4.45) 3.98 (3.99) ⫺2.16 (4.68) ⫺3.09 (44), ⬍0.004 (⫺3.560, ⫺0.751) 0.46 (0.15, 0.77)
0.27 (94), 0.79 (⫺1.606, 2.113) 0.06 (⫺0.35, 0.46)
13.74 (5.15) 13.25 (6.11) ⫺0.49 (4.81) ⫺0.74 (52), 0.46 (⫺1.816, 0.835) 0.10 (⫺0.17, 0.37)
12.14 (4.12) 14.47 (5.86) 2.33 (6.05) 2.69 (48), ⬍0.010 (0.588, 4.065) 0.38 (0.09, 0.67)
⫺2.61 (100), ⬍0.011 (⫺4.957, ⫺0.677) 0.52 (0.12, 0.91)
9.89 (3.26) 8.98 (4.06) ⫺0.91 (3.42) ⫺1.93 (52), 0.059 (⫺1.847, 0.036) 0.27 (⫺0.01, 0.54)
8.24 (3.17) 9.31 (3.54) 1.06 (3.77) 1.97 (48), 0.054 (⫺0.021, 2.143) 0.28 (⫺0.01, 0.57)
⫺2.77 (100), ⬍0.007 (⫺3.378, ⫺0.556) 0.55 (0.15, 0.94)
2.38 (2.21) 2.23 (2.29) ⫺0.15 (1.42) ⫺0.77 (52), 0.44 (⫺0.542, 0.240) 0.11 (⫺0.16, 0.38)
1.98 (1.63) 2.27 (2.46) 0.29 (2.36) 0.85 (48), 0.40 (⫺0.393, 0.964) 0.12 (⫺0.16, 0.40)
⫺1.12 (77.4), 0.27 (⫺1.213, 0.340) 0.22 (⫺0.17, 0.61)
1.19 (1.61) 1.21 (1.66) 0.02 (1.73) 0.08 (52), 0.94 (⫺0.457, 0.495) 0.01 (⫺0.26, 0.28)
1.47 (1.85) 1.57 (1.89) 0.10 (2.34) 0.31 (48), 0.76 (⫺0.570, 0.774) 0.04 (⫺0.24, 0.32)
⫺0.20 (87.9), 0.84 (⫺0.897, 0.731) 0.04 (⫺0.35, 0.43)
0.28 (0.63) 0.85 (1.08) 0.57 (1.08) 3.80 (52), ⬍0.001 (0.267, 0.865) 0.52 (0.23, 0.81)
0.45 (1.16) 1.20 (1.58) 0.76 (1.45) 3.64 (48), ⬍0.001 (0.338, 1.172) 0.52 (0.22, 0.82)
⫺0.74 (88.5), 0.46 (⫺0.696, 0.318) 0.15 (⫺0.24, 0.54)
Notes: Data are means (SD) unless noted. Pre, baseline measurement; post, last nonmissing measurement; CI, confidence interval.
lopram was associated with a lower burden of side effects. To our knowledge, this study is the first head-to-head comparison of a selective serotonin reuptake inhibitor (SSRI) and a second-generation
948
(atypical) antipsychotic in the treatment of noncognitive symptoms associated with dementia. Contrary to our hypothesis and conventional beliefs that an antipsychotic would be superior for the treatment of
Am J Geriatr Psychiatry 15:11, November 2007
Pollock et al.
FIGURE 3.
Agitation and Psychosis Mean Scores over 12 Weeks of Treatment
psychotic symptoms, a similar improvement was observed with both citalopram (⫺32.3%) and risperidone (⫺35.2%). The absence of statistical difference between medications does not appear to be due to a lack of power: the 2.9% difference observed was considerably less than 15%, a change that had been defined a priori as clinically meaningful and that this study was powered to detect. Such a small difference
Am J Geriatr Psychiatry 15:11, November 2007
would have required a sample size of 8,300 to be statistically significant at the 0.05 level. The high rates of dropouts observed across multiple classes of medications in trials involving patients with dementia highlights the challenges of treating these frail patients with pharmacotherapy. This reinforces the need for continual assessment of benefit to harm. New concerns have recently emerged regard-
949
Citalopram and Risperidone for Dementia ing both antipsychotics and SSRIs in older patients. Antipsychotics have been associated with cerebrovascular events and excess mortality in patients with dementia.5,6 Although there is no comparable database on the safety of SSRIs in patients with dementia, SSRIs have been associated with a risk of hyponatremia, bleeding, and suicide attempts in older depressed patients.21–23 We observed a comparable proportion of dropouts attributed to adverse events or physical problems that could not be directly attributed to the study medications. The relatively small study group and the 12-week duration of our trial prevent us from drawing valid conclusions regarding the relative safety of citalopram and risperidone. However, the overall burden of side effects and in particular “psychic” side effects (including sedation, tension, and apathy) were significantly lower with citalopram. A comparable increase in EPS was observed with both drugs. Risperidone causes EPS in a clear dose-dependent fashion.9 By contrast, SSRIs are not typically associated with EPS even though they may cause them in some elderly patients.24 Because most clinicians would not expect an SSRI to cause EPS, this unexpected finding supports that the blind was successfully maintained in our study. This study had both strengths and limitations. Strengths include a parallel treatment design, prospectively defined hypotheses and data analytical plan, participants’ careful diagnostic and symptomatic characterization, and the rigorous maintenance of the double-blind design. An additional strength is the enrolment of participants who were representative of demented patients with severe target symptoms and whom clinicians would usually treat with antipsychotics. Five factors contributed to this representativeness: the initial inpatient setting, absence of a placebo, use of proxy consent, the short study duration, and flexible dosages. The target dosage of risperidone was congruent with dosages recommended by experts3 and those used by clinicians when they treat agitation or psychosis associated with dementia. The target dosages of both risperidone and citalopram were the same as dosages used in other studies, for instance the recently published Clinical Antipsychotic Trials of Intervention Effectiveness–Alzheimer’s Disease (CATIE-AD) study.7 As in this study, the research psychiatrists were al-
950
lowed to carefully adjust dosages in the presence of potential side effects or if the observed improvement was insufficient. Both the maximum and final numbers of study medication capsules were significantly lower with risperidone than with citalopram, confirming the difference in tolerability captured by the UKU ratings. A major limitation is the absence of a placebo. When this trial was designed, the efficacy of risperidone in the treatment of these symptoms was considered to have been established by two trials of registration quality.9,10 Both federal sponsors and our institutional review board viewed a placebocontrolled trial involving patients with dementia hospitalized for the treatment of agitation or psychosis as unethical. While the absence of placebo increased the study group representativeness, it limits the interpretation of our findings. Rather than equivalent efficacy of the two drugs, the absence of differences may reflect their similar lack of efficacy. The high dropout rate was another potential limitation. However, our intent-to-treat analysis using a mixed model was based on all data, not only the completers. In our simulation studies with missing data, comparable to the present level of missing data, a mixed-model analysis estimates the parameters well.19,25 Also, even though we recruited inpatients with severe symptoms, our dropout rate is similar to the dropout rates reported in studies conducted in long-term care facilities9,26 or in the community.7 These high dropout rates reflect the vulnerability of patients with dementia and the limited efficacy of treatment.2,5,7 Nonetheless, the similar dropout rates (and use of rescue lorazepam) confirm that from a pragmatic point of view the clinical impact of citalopram or risperidone was comparable.7,27 Risperidone is considered a first-line treatment for the noncognitive symptoms of dementia given the extent of relevant published data.2,3,28 Its efficacy over placebo was demonstrated in three out of four published RCTs involving a total of 1,787 long-term care setting patients.9,10,29,30 In CATIE-AD, time to discontinuation due to “lack of efficacy” favored risperidone over placebo (26.7 versus 9.0 weeks, p ⬍0.01) even though tolerability of atypical antipsychotics was judged to offset their efficacy.5 A recent meta-analysis has shown that studies conducted in institutionalized patients are more likely to demonstrate a superiority of
Am J Geriatr Psychiatry 15:11, November 2007
Pollock et al. antipsychotics over placebo.28 Most of the participants in our trial were institutionalized and were comparable to the participants in the positive trials conducted in long-term care facilities.9,10,29,30 In contrast to risperidone, the efficacy of citalopram in treating agitation and psychotic symptoms in demented patients has been supported by only two published RCTs: citalopram was superior to placebo in our previous 17-day trial in 52 hospitalized patients.8 Another positive placebo-controlled trial was conducted in 98 patients with dementia.31 However, it did not report the treatment setting and included patients with significant depressive symptoms or with mild behavioral disturbances. Our finding that a serotonergic antidepressant and an antipsychotic had similar effects on psychotic symptoms of dementia may contribute to the understanding of their pathophysiology.32 Neurochemical data consistently point to dopaminergic deficits in dementia casting doubt about the rationale for using D2-blocking agents in these patients.33 In contrast, serotonergic deficits are strongly associated with impulse-control disorders and aggression.34 Similarly, agitation and aggression may be linked to serotonergic dysregulation observed in patients with DAT.35,36
CONCLUSION We urge caution in generalizing the results of a single trial or in extrapolating them to other drugs. Nonetheless, considering the better tolerability of citalopram and concerns regarding increased mortality associated with antipsychotics,5,6 our findings should encourage the conduct of additional trials of citalopram and other agents in the treatment of behavioral and psychotic
symptoms associated with dementia. Given the current view of questionable risk-benefit ratio of second-generation antipsychotics in the treatment of patients with dementia,7 the inclusion of a placebo should be considered in future trials. This work was supported by grants from the U.S. Public Health Service (MH59666, MH65416, MH69430, M01RR0056) and the Sandra A. Rotman Program in Neuropsychiatry (Toronto). Presented in part at the 159th Annual Meeting of the American Psychiatric Association, Toronto, Ontario, Canada, May 20 to 25, 2006. The authors thank Jennifer Maurer of the University of Pittsburgh Medical Center for database management, and Christina Pataky of the Rotman Research Institute at Baycrest for assistance in preparation of the manuscript. This trial is registered at Clinical Trials.gov (http:// www.clinicaltrials.gov/); identifier:NCT00073658. Richard Blakesley, Kimberly Huber, and Patricia Houck have no conflicts of interest or financial interests to disclose. Sati Mazumdar has directly purchased stocks of Forest. Benoit Mulsant has received grants or Research Support from the National Institute of Health, Eli Lilly, Janssen, and Pfizer. He has been a consultant for Lundbeck and Pfizer. He is on the speakers’ bureau of AstraZeneca and Pfizer. He has directly purchased stocks (all below $10,000) of Akzo-Nobel, Alkermes, AstraZeneca, Biogen Idec, Celsion, Elan, Eli Lilly, Forest, General Electric, Immune Response, Pfizer. Bruce Pollock has received grants or research support from the National Institute of Health and Janssen Pharmaceuticals. He has served on the advisory board of Forest Laboratories and is a faculty member of the Lundbeck Institute. He is on the speakers’ bureau of Forest and Lundbeck. Jules Rosen is on the speakers’ bureau of Forest, Janseen, and Pfizer. He has directly purchased stocks of Forest.
References 1. Ropacki SA, Jeste DV: Epidemiology of and risk factors for psychosis of Alzheimer’s disease: a review of 55 studies published from 1990 to 2003. Am J Psychiatry 2005; 162:2022–2030 2. Sink KM, Holden KF, Yaffe K: Pharmacological treatment of neuropsychiatric symptoms of dementia: a review of the evidence. JAMA 2005; 293:596–608 3. Alexopoulos GS, Streim J, Carpenter D, et al: Using antipsychotic agents in older patients. J Clin Psychiatry 2004; 65(suppl 2):5– 104 4. Lyketsos CG, Colenda CC, Beck C, et al: Position statement of the American Association for Geriatric Psychiatry regarding princi-
Am J Geriatr Psychiatry 15:11, November 2007
ples of care for patients with dementia resulting from Alzheimer disease. Am J Geriatr Psychiatry 2006; 14:561–573 5. Schneider LS, Dagerman KS, Philip Insel MS: Risk of death with atypical antipsychotic drug treatment for dementia: meta-analysis of randomized placebo-controlled trials. JAMA 2005; 294:1934–1943 6. Wang PS, Schneeweiss S, Avorn J, et al: Risk of death in elderly users of conventional vs. atypical antipsychotic medications. N Engl J Med 2005; 353:2335–2341 7. Schneider LS, Tariot PN, Dagerman KS, et al: Effectiveness of atypical antipsychotic drugs in patients with Alzheimer’s disease. N Engl J Med 2006; 355:1525–1538
951
Citalopram and Risperidone for Dementia 8. Pollock BG, Mulsant BH, Rosen J, et al: Comparison of citalopram, perphenazine and placebo for the acute treatment of psychosis and behavioral disturbances in hospitalized, demented patients. Am J Psychiatry 2002; 159:460–465 9. Katz IR, Jeste DV, Mintzer JE, et al: Comparison of risperidone and placebo for psychosis and behavioral disturbances associated with dementia: a randomized, double-blind trial. J Clin Psychiatry 1999; 60:107–115 10. De Deyn PP, Rabheru K, Rasmussen A, et al: A randomized trial of risperidone, placebo, and haloperidol for behavioral symptoms of dementia. Neurology 1999; 53:946–955 11. Levin HS, High WM, Goethe KE, et al: The neurobehavioral rating scale. J Neurol Neurosurg Psychiatry 1987; 50:183–193 12. Alexopoulos GS, Abrams RC, Young RC, et al: Cornell scale for depression in dementia. Biol Psychiatry 1988; 23:271–284 13. Cummings JL, Mega M, Gray K, et al: The Neuropsychiatric Inventory: comprehensive assessment of psychopathology in dementia. Neurology 1994; 44:2308–2314 14. Lingjaerde O, Ahlfors UG, Bech P, et al: The UKU side effect rating scale. Acta Psychiatr Scand 1987; 334(suppl):1–100 15. Folstein MF, Folstein SF, McHugh PR: Mini-Mental State Exam: a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975; 12:189–198 16. Saxton J, Swihart AA, McGonigle-Gibson KL, et al: Assessment of the severely impaired patient: description and validation of a new neuropsychological test battery. Psychol Assess 1990; 2:298–303 17. Miller MD, Paradis CF, Houck PR, et al: Rating chronic medical illness burden in geropsychiatric practice and research: application of the cumulative illness rating scale (CIRS). Psychiatry Res 1992; 41:237–248 18. SAS/STAT Software: Version 8. Cary, NC, SAS Institute, 2000 19. Mazumdar S, Tang G, Houck PR, et al: Statistical analysis of longitudinal psychiatric data with dropouts. J Psychiatr Res 2006 Nov 7; [Epub ahead of print] 20. Little RJA, Rubin DB: Statistical analysis with missing data. New York, Wiley & Sons, 1987 21. Dalton SO, Sorensen HT, Johansen C: SSRIs and upper gastrointestinal bleeding: What is known and how should it influence prescribing? CNS Drugs 2006; 20:143–151 22. Fabian TJ, Amico JA, Kroboth PD, et al: Paroxetine-induced hyponatremia in older adults: A twelve-week prospective study. Arch Intern Med 2004; 164:327–332 23. Juurlink DN, Mamdani MM, Kopp A, et al: The risk of suicide with selective serotonin reuptake inhibitors in the elderly. Am J Psychiatry 2006; 163:813–821
952
24. Damsa C, Bumb A, Bianchi-Demicheli F, et al: “Dopamine-dependent” side effects of selective serotonin reuptake inhibitors: a clinical review. J Clin Psychiatry 2004; 65:1064–1068 25. Begley AE, Tang G, Mazumdar S, et al: Use of OSWALD for analyzing longitudinal data with informative dropout. Comput Methods Programs Biomed 2007; 85:109 –114 26. Street JS, Clark WS, Gannon KS, et al: Olanzapine treatment of psychotic and behavioral symptoms in patients with Alzheimer disease in nursing care facilities: a double-blind, randomized, placebo-controlled trial. The HGEU Study Group. Arch Gen Psychiatry 2000; 57:968–967 27. Schneider LS, Ismail MS, Dagerman KS, et al: Clinical antipsychotic trials of intervention effectiveness (CATIE): Alzheimer’s disease trial. Schizophr Bull 2003; 29:57–72 28. Schneider LS, Dagerman K, Insel PS: Efficacy and adverse effects of atypical antipsychotics for dementia: Meta-analysis of randomized, placebo-controlled trials. Am J Geriatr Psychiatry 2006; 14:191–210 29. Brodaty H, Ames D, Snowdon J, et al: A randomized placebocontrolled trial of risperidone for the treatment of aggression, agitation, and psychosis of dementia. J Clin Psychiatry 2003; 64:134–143 30. Mintzer J, Greenspan A, Caers I, et al: Risperidone in the treatment of psychosis of Alzheimer disease: results from a prospective clinical trial. Am J Geriatr Psychiatry 2006; 14:280–291 31. Nyth A-L, Gottfries C-G: The clinical efficacy of citalopram in treatment of emotional disturbances of dementia disorders. A Nordic multicentre study. Br J Psychiatry 1990;157:894–901 32. Jeste DV, Finkel SI: Psychosis of Alzheimer’s disease and related dementias. Am J Geriatr Psychiatry 2000; 8:29–34 33. Colloby SJ, O’Brien JT, Fenwick JD, et al: The application of statistical parametric mapping to 123I-FP-CIT SPECT in dementia with Lewy bodies, Alzheimer’s disease and Parkinson’s disease. Neuroimage 2004; 23:956–966 34. Lesch KP, Merschdorf U: Impulsivity, aggression, and serotonin: a molecular psychobiological perspective. Behav Sci Law 2000; 18:581–604 35. Sweet RA, Pollock BG, Sukonick DL, et al: The 5HTTLPR-polymorphism confers liability to a combined phenotype of psychotic and aggressive behavior in Alzheimer’s disease. Int Psychogeriatr 2001; 13:401–409 36. Mintzer J, Brawman-Mintzer O, Mirski DF, et al: Fenfluramine challenge test as a marker of serotonin activity in patients with Alzheimer’s dementia and agitation. Biol Psychiatry 1998; 44: 918–921
Am J Geriatr Psychiatry 15:11, November 2007
Objective: To compare citalopram and risperidone for the treatment of psychotic symptoms and agitation associated with dementia, with a priori hypotheses that risperidone would be more efficacious for psychosis and citalopram for agitation. Methods: A 12-week randomized, controlled trial in nondepressed patients with dementia hospitalized because of behavioral symptoms (N ⫽ 103) was conducted at the University of Pittsburgh Medical Center. Participants were consecutively recruited on an inpatient unit if they had at least one moderate to severe target symptom (aggression, agitation, hostility, suspiciousness, hallucinations, or delusions). Once they improved sufficiently, they were discharged to nursing homes, personal care homes, or residential homes for continued treatment. Planned pre-post and mixed model analyses of the main outcome measures of Neurobehavioral Rating Scale and Side Effect Rating Scale at baseline and at weekly/biweekly intervals were conducted. Results: Completion rates did not differ for citalopram and risperidone (overall completion rate: 44%). Agitation symptoms (aggression, agitation, or hostility) and psychotic symptoms (suspiciousness, hallucinations, or delusions) decreased in both treatment groups but the improvement did not differ significantly between the two groups. There was a significant increase in side effect burden with risperidone but not with citalopram such that the two groups differed significantly. Conclusion: No statistical difference was found in the efficacy of citalopram and risperidone for the treatment of either agitation or psychotic symptoms in patients with dementia. These findings need to be replicated before citalopram or other serotonergic antidepressants can be recommended as alternatives to antipsychotics for the treatment of agitation or psychotic symptoms associated with dementia. (Am J Geriatr Psychiatry 2007; 15:942–952) Key Words: Dementia, Alzheimer disease, behavioral disturbance, psychosis, agitation, pharmacotherapy, citalopram, risperidone
Received March 23, 2007; revised April 13, 2007; accepted April 16, 2007. From the Rotman Research Institute at Baycrest (BGP) and the Geriatric Mental Health Program, Centre for Addiction and Mental Health (BGP, BHM), University of Toronto, Ontario, Canada; and the Department of Psychiatry (BGP, BHM, JR, PRH, KAH) and the Department of Biostatistics (SM, REB), Graduate School of Public Health, the University of Pittsburgh, Pittsburgh, PA. Send correspondence and reprint requests to Bruce G. Pollock, The Rotman Research Institute at Baycrest, 3560 Bathurst St., Toronto, ON M6A 2E1, Canada. e-mail: [email protected] © 2007 American Association for Geriatric Psychiatry
942
Am J Geriatr Psychiatry 15:11, November 2007
Pollock et al.
N
oncognitive symptoms such as agitation or psychosis are among the most distressing manifestations of dementia.1 Optimal management of these symptoms includes the search for treatable physical and environmental precipitants; nevertheless, pharmacotherapy is frequently used.2 Current guidelines recommend the use of second-generation (“atypical”) antipsychotics.2– 4 Nonetheless, there are concerns regarding both their safety5,6 and effectiveness7 in patients with dementia. Sparse and inconclusive evidence support the use of alternative agents such as antidepressants or cognitive enhancers.2 In a previous randomized controlled trial (RCT), citalopram was more efficacious than placebo and as efficacious as, but better tolerated than, perphenazine in patients with dementia hospitalized for the treatment of agitation or psychosis.8 Therefore, we conducted an RCT to compare the efficacy and safety of citalopram and risperidone. Based on the existing literature when the study was designed9,10 and our prior work,8 we hypothesized that risperidone would be more efficacious for psychotic symptoms, citalopram would be more efficacious for agitation, and citalopram would be associated with fewer side effects.
METHODS Setting and Participants Participants of this 12-week clinical trial were recruited upon admission to the geropsychiatric unit of an academic hospital providing treatment to a large urban and suburban population. If they improved sufficiently, participants were discharged to nursing homes, personal care homes, or residential homes for continued treatment under double-blind conditions. Admissions were eligible if they had a dementia of the Alzheimer type (DAT), vascular dementia, dementia with Lewy bodies, mixed dementia, or dementia not otherwise specified. After a warning was issued to Canadian physicians in 2002 regarding cerebrovascular events associated with risperidone, we excluded patients with vascular dementia. Target symptoms had to be of moderate or higher severity as evidenced by the need for hospitalization and a rating of 3 or higher (moderate to severe) on at least one of the agitation items (aggression, agitation, hos-
Am J Geriatr Psychiatry 15:11, November 2007
tility) or psychosis items (suspiciousness, hallucinations, or delusions) of the Neurobehavioral Rating Scale (NBRS).11 Patients were excluded if they had a current or past diagnosis of schizophrenia, schizoaffective disorder, delusional disorder, psychotic disorders not otherwise specified, bipolar disorder, mental retardation, cognitive deficits following head trauma, or a current diagnosis of delirium, substance-induced persisting dementia, Parkinson disease, drug/alcohol abuse, or dependence. Other exclusion criteria were a major depressive episode within the past 6 months or clinically significant depressive symptoms with a rating of 12 or higher on the Cornell Scale for Depression in Dementia (CSDD)12; unstable physical illness; creatinine ⱖ2.0 mg/100 mL; aspartate aminotransferase or bilirubin more than twice the upper limit of normal; potentially reversible cause of dementia (e.g., untreated hypothyroidism or B12 deficiency); treatment with a depot neuroleptic drug within 2 months of screening or fluoxetine within 4 weeks of screening; or a history of allergy or intolerance to citalopram or risperidone. As required by the Institutional Review Board, participants’ authorized representatives provided written informed consent with participants’ verbal assent. From February 2000 to June 2005, 408 consecutive admissions were prescreened. The three most common reasons patients were deemed ineligible after prescreening were: 1) unstable or excluded medical illness (e.g., seizure disorder or Parkinson disease); 2) insufficient severity of target symptoms; 3) an excluded psychiatric diagnosis (e.g., mood disorder) or previous participation in the trial. Less than 20% of prescreened admissions did not sign consent because they, their family, or their psychiatrist preferred treatment outside of the study. Written consent was obtained for 111 patients and 103 were randomized (Fig. 1). Treatment Protocol Participants could continue cholinesterase inhibitors or memantine if they had been taking them for at least 12 weeks at the same dose for at least 4 weeks. Other psychotropics (except for lorazepam, see below) were discontinued prior to randomization. After baseline ratings, participants were randomized to citalopram or risperidone. The project biostatisticians (PRH and SM) generated the randomization sched-
943
Citalopram and Risperidone for Dementia
FIGURE 1.
Participants’ Flow
ule at the beginning of the trial. Randomization was stratified according to the presence or absence of psychotic symptoms (defined as a moderate to severe score on NBRS items rating delusions, suspicious thoughts, or hallucinations) and blocked to balance cell sizes over the study period. Only the research pharmacist knew the treatment assignments; participants, caregivers, investigators, and assessors remained blind throughout the study. Initial and target dosages of risperidone were chosen based on the results of one large study conducted in nursing home patients.9 Initial and target dosages
944
of citalopram were chosen based on our own previous placebo-controlled study.8 Study medications were administered once a day using identical capsules containing 10 mg of citalopram or 0.5 mg of risperidone, starting with one capsule at bedtime for 3 days and then two capsules per day. After at least two weeks, two additional increases (up to a maximum of four capsules per day) were allowed based on assessment of response and treatment-emergent side effects. These increases were separated by at least two weeks. At any time, dosages could also be titrated downward in the presence of adverse effects
Am J Geriatr Psychiatry 15:11, November 2007
Pollock et al. or intercurrent physical problems. Participants were discontinued in the event of new medical symptoms that endangered their safety or health; adverse events that could not be managed by a dose reduction; a significant worsening of agitation or psychosis; or an administrative reason (e.g., moving away, withdrawal of consent; Fig. 1). Lorazepam up to 2 mg/day could be used throughout the trial for acute treatment of extreme agitation or aggression. The lowest possible dosage for the shortest period of time was used. Assessments Diagnostic and laboratory assessments were completed upon admission to evaluate for reversible causes of dementia, agitation, or psychosis. Participants were also assessed at the time of enrollment; baseline (i.e., time of randomization); after receiving study medication for 3 days, 7 days, then weekly for 5 weeks, then every 2 weeks; and at the time of discharge from the hospital or termination if the study was terminated early. Information was obtained from direct observation of the participants, interviews with participants and nursing staff, and chart review of all notes for the 3 days preceding the rating. In addition to the NBRS, instruments included the CSDD,12 Neuropsychiatric Inventory (NPI),13 Udvalg for Kliniske Undersogelser (UKU) side effect scale,14 Mini-Mental State Examination (MMSE),15 Severe Impairment Battery (SIB),16 and Cumulative Illness Rating Scale–Geriatrics.17 The NBRS is a 28-item observer-rated instrument combining the breadth of psychopathology covered by the Brief Psychiatric Rating Scale with more comprehensive assessment of impairments seen in dementia. Scoring is on a 7-point scale (0: not present; 1: very mild; 2: mild; 3: moderate; 4: moderately severe; 5: severe; 6: extremely severe).12 All raters participated in initial and yearly training sessions during which interrater reliability was established and monitored for the NBRS, UKU, SIB, and MMSE. Throughout the study interrater reliability was good to excellent. Diagnoses were established according to Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM–IV) criteria considering all available information during a consensus conference attended by three investigator geropsychiatrists and the re-
Am J Geriatr Psychiatry 15:11, November 2007
search staff. Of the randomized participants, 86 received a DSM–IV consensus diagnosis of DAT with late onset; 5 DAT with early onset; one of vascular dementia; and 9 dementia not otherwise specified. Following their randomization, two participants were judged to have a diagnosis that should have led to their exclusion (one with late-onset DAT with delirium randomized to citalopram and one with substance-induced persisting delirium randomized to risperidone). Following the intent-to-treat principle, these two participants are included in all analyses. Diagnoses were distributed similarly in the two treatment groups. Statistical Analysis We prospectively estimated sample size based on our previous study conducted in the same inpatient setting and expected percent changes in NBRS agitation or psychosis scores with each drug.8 A priori, a difference between the two treatments of less than 15% was judged not to be clinically significant. Therefore, a sample size of 103 was selected because it is sufficient to detect a 15% or larger difference with a power of 80% and a type I error of 0.05. All analyses were conducted using SAS18 based on the randomized assignment. First, we compared the baseline characteristics of the treatment groups using t or 2 tests. Then, we compared the dropout rates using log-rank statistics. We then performed two sets of analyses including all randomized participants who received at least one dose of study medication and who completed the baseline and at least one post baseline evaluation. All statistical tests were twosided. A first set of pre-post analyses compared ratings at baseline (pre) and at the time of termination from the study (post) for each treatment, and the changes observed with each treatment. A second set of analyses consisted of a mixed model analysis of the main efficacy outcomes (NBRS agitation and psychosis scores), comparing the two treatments over the total follow-up period using all randomized participants. Mixed model analyses use all the data that are collected longitudinally and also account for the within-subject correlations between measures. As routine application of these models can lead to erroneous conclusions when data are not missing at random, we investigated the reasons for dropping out and effects of different dropout times on the param-
945
Citalopram and Risperidone for Dementia eter estimates using a pattern-mixture model.19,20 This approach stratifies the observed data according to the dropout patterns (number and timing of missing values), models the response process within each pattern, and combine the parameter estimates. Using SAS-PROC MIXED, the mixed models included treatment (group), week (time), and group-by-time interaction terms. The intercept and the week terms were assumed to be random effects. As proposed in our a priori data-analytic plan, we used the NBRS because it was sensitive to change in agitation and psychosis in our previous study.8 However, since the NPI is used in some trials, we also performed the same mixed model analysis using NPI scores. Results were qualitatively similar and thus we present only the planned NBRS analyses.
RESULTS Of the randomized participants, 53 were randomized to citalopram and 50 to risperidone. The sex distribution and baseline NBRS total scores differed significantly between the two groups (Table 1). Thirty
TABLE 1.
Baseline Participant Characteristics
Characteristic N Age (years) Female, % (N)a White, % (N) Cumulative Illness Rating Scale–Geriatrics score Mini Mental State Examination score Median Interquartile range Severe Impairment Battery Cornell Scale for Depression in Dementia Delusions or hallucinations, % (N) NBRS total scoreb NBRS agitation score (items 8, 11, and 14) NBRS psychosis score (items 16, 18, and 20) NPI total score UKU total score
Citalopram
Risperidone
53 82.0 (7.3) 47.2 (25) 81.1 (43) 9.6 (3.0)
50 81.5 (9.2) 76.0 (38) 82.0 (41) 9.5 (4.5)
9.8 (8.5)
11.9 (8.4)
7 2–18 63.9 (35.5) 10.7 (11.8)
11.5 4.5–19.5 71.4 (29.3) 13.3 (16.2)
35.8 (19) 60.3 (16.8) 10.1 (4.8)
40.0 (20) 53.6 (15.6) 8.9 (4.5)
5.9 (3.5)
6.0 (4.5)
40.8 (23.1) 13.7 (5.2)
34.4 (21.1) 12.1 (4.1)
participants continued a stable dose of a cholinesterase inhibitor: 16 (30.2%) randomized to citalopram and 14 (28.0%) randomized to risperidone (Fisher exact p⫽0.83). No participant was on memantine during the study. Mean (SD) maximum doses were 31.1 (8.7) mg/day of citalopram and 1.36 (0.47) mg/ day for risperidone, corresponding to 3.1 (0.9) and 2.7 (1.0) capsules, respectively (t⫽ 2.19, df⫽101, p⫽ 0.031). Mean (SD) final doses were 29.4 (9.9) mg/day of citalopram and 1.25 (0.51) mg/day of risperidone, corresponding to 2.9 (1.0) and 2.5 (1.0) capsules, respectively (t⫽ 2.25, df⫽101, p⫽0.027). Similar numbers of participants in each treatment group received lorazepam (citalopram: N⫽43; risperidone: N⫽41; Fisher exact p ⫽1.00). Mean (SD) lorazepam daily doses and treatment durations did not differ significantly between the two groups: citalopram was 0.6 (0.2) mg/day for 7.8 (7.0) days and risperidone was 0.6 (0.7) mg/day for 8.0 (8.3) days (dose: t⫽0.04, df⫽ 82, p⫽0.96; duration: t⫽⫺0.12, df ⫽82, p⫽0.90). Dropouts Overall, 43.7% of the participants completed the 12-week trial: 47.2% in the citalopram group and 40.0% in the risperidone group (Fig. 2). Time to dropout for all reasons did not differ significantly between the two groups (log-rank 21 ⫽ 0.85, p⫽0.36). The distribution of the causes to which dropouts were attributed did not differ either (Fig. 1; early dropouts: Fisher exact p ⬍0.32; late dropouts: Fisher exact
FIGURE 2.
Kaplan-Meier Plot of Time to Dropout for All Reasons
Notes: Data are means (SD) unless noted. a Significant difference between treatment groups: 2 ⫽ 9.00, df ⫽ 1, p ⫽ 0.003. b Significant difference between treatment groups: t ⫽ 2.04, df ⫽ 94, p ⫽ 0.044.
946
Am J Geriatr Psychiatry 15:11, November 2007
Pollock et al. p ⬍0.73). Numbers and specific reasons for dropouts were as follows for early citalopram dropouts: intercurrent medical problems, 2 (seizure, ileus); adverse events, 3 (gait disturbance, other extrapyramidal symptoms [EPS], bruising); psychiatric worsening, 9 (increased agitation, 7; onset of psychosis, onset of depression); and administrative reasons, 1. For late citalopram dropouts, the reasons included: intercurrent medical problems, 3 (infection, 2; hypoglycemia); adverse events, 1 (sedation); psychiatric worsening, 7 (increased agitation, 5; readmission, 2); and administrative reasons, 2. Reasons for early risperidone dropouts included: intercurrent medical problems, 4 (gastrointestinal bleeding, intracranial bleeding, pneumonia, ileus); adverse events, 7 (gait disturbance, 3; other EPS, 3; hypotension); psychiatric worsening, 5 (increased agitation, 4; onset of psychosis); and administrative reasons, 3. For late risperidone dropouts, reasons included: intercurrent medical problems, 2 (pneumonia, unresponsiveness); adverse events, 2 (elevated liver function tests, unresponsiveness); psychiatric worsening, 7 (increased agitation, 3; readmission, 3; suicide attempt). Pre-Post Analysis for Agitation and Psychosis Outcomes Changes in NBRS agitation or psychosis scores did not differ significantly between the two groups (Table 2). There was a significant decrease in agitation score with citalopram (⫺12.5%) but not with risperidone (⫺8.2%). Psychosis scores decreased significantly both with citalopram (⫺32.3%) and risperidone (⫺35.2%). Pre-Post Analysis for the Side Effect Outcomes Changes in UKU total score and psychic subscore differed significantly between the two groups (Table 2). UKU total scores increased significantly with risperidone (19.2%) but not with citalopram (⫺3.6%). Because some of these measures were skewed, comparisons were repeated using nonparametric Kruskal-Wallis tests; the same statistical differences were detected (data not shown). Inspection of specific items suggested that these differences were primarily driven by a 26.2% decrease in sedation with citalopram and an 83.3% increase with risperidone [UKU somnolence mean (SD) score: citalopram pre, 0.42 (0.72); citalopram post, 0.30 (0.64); risperidone pre, 0.24 (0.60); risperidone post, 0.45 (0.79)]. Comparable increases in EPS were
Am J Geriatr Psychiatry 15:11, November 2007
seen with both medications: 87.5% with citalopram and 102.5% with risperidone [UKU rigidity and tremor items score: citalopram pre, 0.32 (0.61); citalopram post, 0.60 (0.86); risperidone pre, 0.40 (0.90); risperidone post, 0.81 (1.31)]. Efficacy: Mixed Model Analysis Investigation of the specific reasons for dropping out revealed that they were mostly based on measured values with only a few due to administrative reasons (Fig. 3). The analyses, using a pattern-mixture model where data were stratified by dropouts before six weeks (when most of the dropouts occurred), after six weeks, and completers, were in agreement with the results of the mixed model analyses. Thus, the requirement of the missing at random (MAR) assumption was considered to be satisfied and we present the results of the mixed models that used all observed data of the 103 participants and compared the two treatments over the total follow-up time. A significant time effect was found for both efficacy outcomes (agitation: F1,67 ⫽10.4, p ⬍0.002; psychosis: F1,60 ⫽ 28.9, p ⬍0.001). However, there was no significant group effect or group-by-time interaction for either outcome. The same mixed model analysis excluding the patients without a diagnosis of Alzheimer disease produced similar results, showing only significant time effects for both efficacy outcomes. Because of the baseline difference in sex distribution, another model included sex as an effect. As in the original model, there was no group effect or group-by-time interaction for either outcome. However, both a time (F1,61.9 ⫽ 6.57, p ⬍0.013) and timeby-sex interaction (F1,61.9 ⫽8.95, p ⬍0.004) were observed for psychosis (such that women improved faster than men) but not for agitation. Finally, because recruitment took place over 6 years, a last model included the year of recruitment as an effect. The results of this model were similar to the original model, showing only a time effect for both outcomes.
DISCUSSION In patients with dementia hospitalized for the treatment of behavioral or psychotic symptoms, citalopram and risperidone had similar efficacy but cita-
947
Citalopram and Risperidone for Dementia
TABLE 2.
Efficacy and Tolerability Analysis
NBRS agitation score (items 8, 11, and 14) Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI) NBRS psychosis score (items 16, 18, and 20) Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI) UKU total score Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI) UKU psychic subscale score Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI) UKU neurologic subscale score Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI) UKU autonomic subscale score Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI) UKU other subscale score Pre Post Post – pre t (df), p Post – pre 95% CI Effect size (95% CI)
Citalopram
Risperidone
Citalopram vs. Risperidone
10.08 (4.81) 8.81 (5.63) ⫺1.26 (4.58) ⫺2.01 (52), 0.050 (⫺2.527, ⫺0.001) 0.28 (0.0002, 0.55)
8.94 (4.54) 8.20 (5.08) ⫺0.73 (4.91) ⫺1.05 (48), 0.30 (⫺2.145, 0.676) 0.15 (⫺0.13, 0.43)
⫺0.56 (100), 0.57 (⫺2.394, 1.336) 0.11 (⫺0.28, 0.50)
5.88 (3.51) 3.98 (3.90) ⫺1.90 (4.49) ⫺3.02 (50), ⬍0.004 (⫺3.165, ⫺0.639) 0.42 (0.13, 0.71)
6.13 (4.45) 3.98 (3.99) ⫺2.16 (4.68) ⫺3.09 (44), ⬍0.004 (⫺3.560, ⫺0.751) 0.46 (0.15, 0.77)
0.27 (94), 0.79 (⫺1.606, 2.113) 0.06 (⫺0.35, 0.46)
13.74 (5.15) 13.25 (6.11) ⫺0.49 (4.81) ⫺0.74 (52), 0.46 (⫺1.816, 0.835) 0.10 (⫺0.17, 0.37)
12.14 (4.12) 14.47 (5.86) 2.33 (6.05) 2.69 (48), ⬍0.010 (0.588, 4.065) 0.38 (0.09, 0.67)
⫺2.61 (100), ⬍0.011 (⫺4.957, ⫺0.677) 0.52 (0.12, 0.91)
9.89 (3.26) 8.98 (4.06) ⫺0.91 (3.42) ⫺1.93 (52), 0.059 (⫺1.847, 0.036) 0.27 (⫺0.01, 0.54)
8.24 (3.17) 9.31 (3.54) 1.06 (3.77) 1.97 (48), 0.054 (⫺0.021, 2.143) 0.28 (⫺0.01, 0.57)
⫺2.77 (100), ⬍0.007 (⫺3.378, ⫺0.556) 0.55 (0.15, 0.94)
2.38 (2.21) 2.23 (2.29) ⫺0.15 (1.42) ⫺0.77 (52), 0.44 (⫺0.542, 0.240) 0.11 (⫺0.16, 0.38)
1.98 (1.63) 2.27 (2.46) 0.29 (2.36) 0.85 (48), 0.40 (⫺0.393, 0.964) 0.12 (⫺0.16, 0.40)
⫺1.12 (77.4), 0.27 (⫺1.213, 0.340) 0.22 (⫺0.17, 0.61)
1.19 (1.61) 1.21 (1.66) 0.02 (1.73) 0.08 (52), 0.94 (⫺0.457, 0.495) 0.01 (⫺0.26, 0.28)
1.47 (1.85) 1.57 (1.89) 0.10 (2.34) 0.31 (48), 0.76 (⫺0.570, 0.774) 0.04 (⫺0.24, 0.32)
⫺0.20 (87.9), 0.84 (⫺0.897, 0.731) 0.04 (⫺0.35, 0.43)
0.28 (0.63) 0.85 (1.08) 0.57 (1.08) 3.80 (52), ⬍0.001 (0.267, 0.865) 0.52 (0.23, 0.81)
0.45 (1.16) 1.20 (1.58) 0.76 (1.45) 3.64 (48), ⬍0.001 (0.338, 1.172) 0.52 (0.22, 0.82)
⫺0.74 (88.5), 0.46 (⫺0.696, 0.318) 0.15 (⫺0.24, 0.54)
Notes: Data are means (SD) unless noted. Pre, baseline measurement; post, last nonmissing measurement; CI, confidence interval.
lopram was associated with a lower burden of side effects. To our knowledge, this study is the first head-to-head comparison of a selective serotonin reuptake inhibitor (SSRI) and a second-generation
948
(atypical) antipsychotic in the treatment of noncognitive symptoms associated with dementia. Contrary to our hypothesis and conventional beliefs that an antipsychotic would be superior for the treatment of
Am J Geriatr Psychiatry 15:11, November 2007
Pollock et al.
FIGURE 3.
Agitation and Psychosis Mean Scores over 12 Weeks of Treatment
psychotic symptoms, a similar improvement was observed with both citalopram (⫺32.3%) and risperidone (⫺35.2%). The absence of statistical difference between medications does not appear to be due to a lack of power: the 2.9% difference observed was considerably less than 15%, a change that had been defined a priori as clinically meaningful and that this study was powered to detect. Such a small difference
Am J Geriatr Psychiatry 15:11, November 2007
would have required a sample size of 8,300 to be statistically significant at the 0.05 level. The high rates of dropouts observed across multiple classes of medications in trials involving patients with dementia highlights the challenges of treating these frail patients with pharmacotherapy. This reinforces the need for continual assessment of benefit to harm. New concerns have recently emerged regard-
949
Citalopram and Risperidone for Dementia ing both antipsychotics and SSRIs in older patients. Antipsychotics have been associated with cerebrovascular events and excess mortality in patients with dementia.5,6 Although there is no comparable database on the safety of SSRIs in patients with dementia, SSRIs have been associated with a risk of hyponatremia, bleeding, and suicide attempts in older depressed patients.21–23 We observed a comparable proportion of dropouts attributed to adverse events or physical problems that could not be directly attributed to the study medications. The relatively small study group and the 12-week duration of our trial prevent us from drawing valid conclusions regarding the relative safety of citalopram and risperidone. However, the overall burden of side effects and in particular “psychic” side effects (including sedation, tension, and apathy) were significantly lower with citalopram. A comparable increase in EPS was observed with both drugs. Risperidone causes EPS in a clear dose-dependent fashion.9 By contrast, SSRIs are not typically associated with EPS even though they may cause them in some elderly patients.24 Because most clinicians would not expect an SSRI to cause EPS, this unexpected finding supports that the blind was successfully maintained in our study. This study had both strengths and limitations. Strengths include a parallel treatment design, prospectively defined hypotheses and data analytical plan, participants’ careful diagnostic and symptomatic characterization, and the rigorous maintenance of the double-blind design. An additional strength is the enrolment of participants who were representative of demented patients with severe target symptoms and whom clinicians would usually treat with antipsychotics. Five factors contributed to this representativeness: the initial inpatient setting, absence of a placebo, use of proxy consent, the short study duration, and flexible dosages. The target dosage of risperidone was congruent with dosages recommended by experts3 and those used by clinicians when they treat agitation or psychosis associated with dementia. The target dosages of both risperidone and citalopram were the same as dosages used in other studies, for instance the recently published Clinical Antipsychotic Trials of Intervention Effectiveness–Alzheimer’s Disease (CATIE-AD) study.7 As in this study, the research psychiatrists were al-
950
lowed to carefully adjust dosages in the presence of potential side effects or if the observed improvement was insufficient. Both the maximum and final numbers of study medication capsules were significantly lower with risperidone than with citalopram, confirming the difference in tolerability captured by the UKU ratings. A major limitation is the absence of a placebo. When this trial was designed, the efficacy of risperidone in the treatment of these symptoms was considered to have been established by two trials of registration quality.9,10 Both federal sponsors and our institutional review board viewed a placebocontrolled trial involving patients with dementia hospitalized for the treatment of agitation or psychosis as unethical. While the absence of placebo increased the study group representativeness, it limits the interpretation of our findings. Rather than equivalent efficacy of the two drugs, the absence of differences may reflect their similar lack of efficacy. The high dropout rate was another potential limitation. However, our intent-to-treat analysis using a mixed model was based on all data, not only the completers. In our simulation studies with missing data, comparable to the present level of missing data, a mixed-model analysis estimates the parameters well.19,25 Also, even though we recruited inpatients with severe symptoms, our dropout rate is similar to the dropout rates reported in studies conducted in long-term care facilities9,26 or in the community.7 These high dropout rates reflect the vulnerability of patients with dementia and the limited efficacy of treatment.2,5,7 Nonetheless, the similar dropout rates (and use of rescue lorazepam) confirm that from a pragmatic point of view the clinical impact of citalopram or risperidone was comparable.7,27 Risperidone is considered a first-line treatment for the noncognitive symptoms of dementia given the extent of relevant published data.2,3,28 Its efficacy over placebo was demonstrated in three out of four published RCTs involving a total of 1,787 long-term care setting patients.9,10,29,30 In CATIE-AD, time to discontinuation due to “lack of efficacy” favored risperidone over placebo (26.7 versus 9.0 weeks, p ⬍0.01) even though tolerability of atypical antipsychotics was judged to offset their efficacy.5 A recent meta-analysis has shown that studies conducted in institutionalized patients are more likely to demonstrate a superiority of
Am J Geriatr Psychiatry 15:11, November 2007
Pollock et al. antipsychotics over placebo.28 Most of the participants in our trial were institutionalized and were comparable to the participants in the positive trials conducted in long-term care facilities.9,10,29,30 In contrast to risperidone, the efficacy of citalopram in treating agitation and psychotic symptoms in demented patients has been supported by only two published RCTs: citalopram was superior to placebo in our previous 17-day trial in 52 hospitalized patients.8 Another positive placebo-controlled trial was conducted in 98 patients with dementia.31 However, it did not report the treatment setting and included patients with significant depressive symptoms or with mild behavioral disturbances. Our finding that a serotonergic antidepressant and an antipsychotic had similar effects on psychotic symptoms of dementia may contribute to the understanding of their pathophysiology.32 Neurochemical data consistently point to dopaminergic deficits in dementia casting doubt about the rationale for using D2-blocking agents in these patients.33 In contrast, serotonergic deficits are strongly associated with impulse-control disorders and aggression.34 Similarly, agitation and aggression may be linked to serotonergic dysregulation observed in patients with DAT.35,36
CONCLUSION We urge caution in generalizing the results of a single trial or in extrapolating them to other drugs. Nonetheless, considering the better tolerability of citalopram and concerns regarding increased mortality associated with antipsychotics,5,6 our findings should encourage the conduct of additional trials of citalopram and other agents in the treatment of behavioral and psychotic
symptoms associated with dementia. Given the current view of questionable risk-benefit ratio of second-generation antipsychotics in the treatment of patients with dementia,7 the inclusion of a placebo should be considered in future trials. This work was supported by grants from the U.S. Public Health Service (MH59666, MH65416, MH69430, M01RR0056) and the Sandra A. Rotman Program in Neuropsychiatry (Toronto). Presented in part at the 159th Annual Meeting of the American Psychiatric Association, Toronto, Ontario, Canada, May 20 to 25, 2006. The authors thank Jennifer Maurer of the University of Pittsburgh Medical Center for database management, and Christina Pataky of the Rotman Research Institute at Baycrest for assistance in preparation of the manuscript. This trial is registered at Clinical Trials.gov (http:// www.clinicaltrials.gov/); identifier:NCT00073658. Richard Blakesley, Kimberly Huber, and Patricia Houck have no conflicts of interest or financial interests to disclose. Sati Mazumdar has directly purchased stocks of Forest. Benoit Mulsant has received grants or Research Support from the National Institute of Health, Eli Lilly, Janssen, and Pfizer. He has been a consultant for Lundbeck and Pfizer. He is on the speakers’ bureau of AstraZeneca and Pfizer. He has directly purchased stocks (all below $10,000) of Akzo-Nobel, Alkermes, AstraZeneca, Biogen Idec, Celsion, Elan, Eli Lilly, Forest, General Electric, Immune Response, Pfizer. Bruce Pollock has received grants or research support from the National Institute of Health and Janssen Pharmaceuticals. He has served on the advisory board of Forest Laboratories and is a faculty member of the Lundbeck Institute. He is on the speakers’ bureau of Forest and Lundbeck. Jules Rosen is on the speakers’ bureau of Forest, Janseen, and Pfizer. He has directly purchased stocks of Forest.
References 1. Ropacki SA, Jeste DV: Epidemiology of and risk factors for psychosis of Alzheimer’s disease: a review of 55 studies published from 1990 to 2003. Am J Psychiatry 2005; 162:2022–2030 2. Sink KM, Holden KF, Yaffe K: Pharmacological treatment of neuropsychiatric symptoms of dementia: a review of the evidence. JAMA 2005; 293:596–608 3. Alexopoulos GS, Streim J, Carpenter D, et al: Using antipsychotic agents in older patients. J Clin Psychiatry 2004; 65(suppl 2):5– 104 4. Lyketsos CG, Colenda CC, Beck C, et al: Position statement of the American Association for Geriatric Psychiatry regarding princi-
Am J Geriatr Psychiatry 15:11, November 2007
ples of care for patients with dementia resulting from Alzheimer disease. Am J Geriatr Psychiatry 2006; 14:561–573 5. Schneider LS, Dagerman KS, Philip Insel MS: Risk of death with atypical antipsychotic drug treatment for dementia: meta-analysis of randomized placebo-controlled trials. JAMA 2005; 294:1934–1943 6. Wang PS, Schneeweiss S, Avorn J, et al: Risk of death in elderly users of conventional vs. atypical antipsychotic medications. N Engl J Med 2005; 353:2335–2341 7. Schneider LS, Tariot PN, Dagerman KS, et al: Effectiveness of atypical antipsychotic drugs in patients with Alzheimer’s disease. N Engl J Med 2006; 355:1525–1538
951
Citalopram and Risperidone for Dementia 8. Pollock BG, Mulsant BH, Rosen J, et al: Comparison of citalopram, perphenazine and placebo for the acute treatment of psychosis and behavioral disturbances in hospitalized, demented patients. Am J Psychiatry 2002; 159:460–465 9. Katz IR, Jeste DV, Mintzer JE, et al: Comparison of risperidone and placebo for psychosis and behavioral disturbances associated with dementia: a randomized, double-blind trial. J Clin Psychiatry 1999; 60:107–115 10. De Deyn PP, Rabheru K, Rasmussen A, et al: A randomized trial of risperidone, placebo, and haloperidol for behavioral symptoms of dementia. Neurology 1999; 53:946–955 11. Levin HS, High WM, Goethe KE, et al: The neurobehavioral rating scale. J Neurol Neurosurg Psychiatry 1987; 50:183–193 12. Alexopoulos GS, Abrams RC, Young RC, et al: Cornell scale for depression in dementia. Biol Psychiatry 1988; 23:271–284 13. Cummings JL, Mega M, Gray K, et al: The Neuropsychiatric Inventory: comprehensive assessment of psychopathology in dementia. Neurology 1994; 44:2308–2314 14. Lingjaerde O, Ahlfors UG, Bech P, et al: The UKU side effect rating scale. Acta Psychiatr Scand 1987; 334(suppl):1–100 15. Folstein MF, Folstein SF, McHugh PR: Mini-Mental State Exam: a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975; 12:189–198 16. Saxton J, Swihart AA, McGonigle-Gibson KL, et al: Assessment of the severely impaired patient: description and validation of a new neuropsychological test battery. Psychol Assess 1990; 2:298–303 17. Miller MD, Paradis CF, Houck PR, et al: Rating chronic medical illness burden in geropsychiatric practice and research: application of the cumulative illness rating scale (CIRS). Psychiatry Res 1992; 41:237–248 18. SAS/STAT Software: Version 8. Cary, NC, SAS Institute, 2000 19. Mazumdar S, Tang G, Houck PR, et al: Statistical analysis of longitudinal psychiatric data with dropouts. J Psychiatr Res 2006 Nov 7; [Epub ahead of print] 20. Little RJA, Rubin DB: Statistical analysis with missing data. New York, Wiley & Sons, 1987 21. Dalton SO, Sorensen HT, Johansen C: SSRIs and upper gastrointestinal bleeding: What is known and how should it influence prescribing? CNS Drugs 2006; 20:143–151 22. Fabian TJ, Amico JA, Kroboth PD, et al: Paroxetine-induced hyponatremia in older adults: A twelve-week prospective study. Arch Intern Med 2004; 164:327–332 23. Juurlink DN, Mamdani MM, Kopp A, et al: The risk of suicide with selective serotonin reuptake inhibitors in the elderly. Am J Psychiatry 2006; 163:813–821
952
24. Damsa C, Bumb A, Bianchi-Demicheli F, et al: “Dopamine-dependent” side effects of selective serotonin reuptake inhibitors: a clinical review. J Clin Psychiatry 2004; 65:1064–1068 25. Begley AE, Tang G, Mazumdar S, et al: Use of OSWALD for analyzing longitudinal data with informative dropout. Comput Methods Programs Biomed 2007; 85:109 –114 26. Street JS, Clark WS, Gannon KS, et al: Olanzapine treatment of psychotic and behavioral symptoms in patients with Alzheimer disease in nursing care facilities: a double-blind, randomized, placebo-controlled trial. The HGEU Study Group. Arch Gen Psychiatry 2000; 57:968–967 27. Schneider LS, Ismail MS, Dagerman KS, et al: Clinical antipsychotic trials of intervention effectiveness (CATIE): Alzheimer’s disease trial. Schizophr Bull 2003; 29:57–72 28. Schneider LS, Dagerman K, Insel PS: Efficacy and adverse effects of atypical antipsychotics for dementia: Meta-analysis of randomized, placebo-controlled trials. Am J Geriatr Psychiatry 2006; 14:191–210 29. Brodaty H, Ames D, Snowdon J, et al: A randomized placebocontrolled trial of risperidone for the treatment of aggression, agitation, and psychosis of dementia. J Clin Psychiatry 2003; 64:134–143 30. Mintzer J, Greenspan A, Caers I, et al: Risperidone in the treatment of psychosis of Alzheimer disease: results from a prospective clinical trial. Am J Geriatr Psychiatry 2006; 14:280–291 31. Nyth A-L, Gottfries C-G: The clinical efficacy of citalopram in treatment of emotional disturbances of dementia disorders. A Nordic multicentre study. Br J Psychiatry 1990;157:894–901 32. Jeste DV, Finkel SI: Psychosis of Alzheimer’s disease and related dementias. Am J Geriatr Psychiatry 2000; 8:29–34 33. Colloby SJ, O’Brien JT, Fenwick JD, et al: The application of statistical parametric mapping to 123I-FP-CIT SPECT in dementia with Lewy bodies, Alzheimer’s disease and Parkinson’s disease. Neuroimage 2004; 23:956–966 34. Lesch KP, Merschdorf U: Impulsivity, aggression, and serotonin: a molecular psychobiological perspective. Behav Sci Law 2000; 18:581–604 35. Sweet RA, Pollock BG, Sukonick DL, et al: The 5HTTLPR-polymorphism confers liability to a combined phenotype of psychotic and aggressive behavior in Alzheimer’s disease. Int Psychogeriatr 2001; 13:401–409 36. Mintzer J, Brawman-Mintzer O, Mirski DF, et al: Fenfluramine challenge test as a marker of serotonin activity in patients with Alzheimer’s dementia and agitation. Biol Psychiatry 1998; 44: 918–921
Am J Geriatr Psychiatry 15:11, November 2007