Mark Marsico

Exploring novel cognitive outcomes for clinical trials in MCI and Alzheimer's disease compared with the ADAS-Cog

of accurately predicting prodromes of different dementia syndromes, we examined whether empirically-derived cognitive subtypes of MCI were consistent with consensus diagnoses (i.e., amnestic and non-amnestic singleand multi-domain; Winblad et al., 2004) and whether resulting neuropsychological profiles vary based on the criteria used for MCI diagnosis. Methods: 135 participants were diagnosed with MCI using conventional criteria (i.e., 1.5 SDs below normal on one test within a cognitive domain; Petersen & Morris, 2005) and 81 participants were diagnosed with MCI using comprehensive criteria developed to increase diagnostic specificity (i.e., 1 SD below normal on two tests within one domain; Jak et al., 2009). To examine tests and underlying cognitive constructs that constitute subgroup profiles, we conducted a hierarchical cluster analysis using 13 measures of language, memory, executive function, visuospatial function, and attention. Results: The conventional criteria for MCI yielded the following three distinct groups: An Amnestic subgroup (n 1⁄4 57) mildly impaired primarily on memory measures; a Mixed or multi-domain subgroup (n 1⁄4 14) impaired on memory, executive function, language, and visuospatial measures; and a Relatively Normal subgroup (n 1⁄4 64) that performed within normal limits. Of the MCI participants diagnosed via comprehensive criteria, four distinct groups emerged: An Amnestic MCI subgroup (n 1⁄4 30), a Dysexecutive MCI subgroup (n 1⁄4 13), a Mixed MCI subgroup (n1⁄4 19), and a Visuospatial subgroup (n1⁄4 19) impaired on a specific visuospatial measure. Conclusions: Findings revealed that neuropsychological profiles vary depending on classification scheme used to define MCI. The conventional criteria produced subgroups consistent with single and multi-domain amnestic MCI diagnoses. However, one subgroup performed largely within normal limits suggesting that using an isolated poor performance for diagnosis may increase risk for false positives. Comprehensive criteria revealed more nuanced subgroups characterized by memory as well as dysexecutive and visuoconstructional impairments. Both schemes revealed a Mixed subgroup consistent with a prodromal AD profile and more severe impairments that may reflect a later stage of MCI.

HEALTH CARE USE AMONG NEWLY DIAGNOSED ALZHEIMER’S DISEASE PATIENTS IN A U.S. DATABASE OF EMPLOYER-SPONSORED MEDICARE SUPPLEMENT INSURANCE PLANS

impairment in consciousness and/or attention is present, the disturbance is recognized as post-operative delirium (POD). Postoperative cognitive decline (POCD), on the other hand, is a subacute codition measured using changes in performance with serial cognitive assessment. While POD and POCD are both areas of interest for researchers and concern for older surgical patients, the relationship between these two conditions has not been well-examined. Methods:Participants were from the Successful Aging after Elective Surgery study, a prospective study of 560 older adults ( 70 years) without dementia at baseline who underwent elective surgery. Deliriumwas evaluated each postoperative day using the Confusion Assessment Method (CAM) supplemented with medical chart review. Cognition was assessed with a battery of neuropsychological tests. POCDwas defined using the approach developed by the International Study of Postoperative Cognitive Dysfunction. Results: POD occurred in nearly a quarter (n1⁄4 134, 24%) of patients. About 47%of patients satisfied criteria for POCDat 1month following surgery, but this proportionwas lower atmonths 2, 6 and 12 (23%, 16%, and 24%, respectively) following surgery. At each follow-up, the level of agreement (on the basis of a kappa agreement statistic) between post-operative delirium (POD) and POCDwas poor (kappa < .08) and the (tetrachoric) correlations were small (r < .16). The risk for POCD was significantly elevated for patients with a history of POD at 1 month (RR1⁄4 1.4, 95% CI 1.2, 1.7, P< .001) but not at month 2 (RR1⁄4 1.1, P1⁄4.64),month 6 (RR1⁄4 0.95, P1⁄4.82), ormonth 12 (RR 1⁄4 1.2, P 1⁄4 .40). Conclusions:POD does not appear to be a risk factor for POCD. The association of POD and POCD at 1 month likely reflects unresolved delirium. This study suggests that POD and POCD are distinct entities, raising the strong possibility that specific underlying causal mechanisms are at work. Distinct and targeted interventions may be required to improve the cognitive outcomes of older surgical patients.

HEALTH CARE USE AMONG NEWLY DIAGNOSED ALZHEIMER’S (AD) PATIENTS IN A U.S. COMMERCIAL MEDICARE ADVANTAGE INSURANCE PLAN

Background:Leveraging real-world data to better understand health care utilization patterns prior to and after incident AD diagnosis may help inform the clinical management of the disease. Methods: Medical and pharmacy claims data from a large, geographically diverse, commercial, US Medicare Advantage health plan, for the period 2009-2014, were used to identify cases of newly diagnosed AD. The following beneficiaries were eligible for inclusion in the analysis: aged 65 with medical and prescription drug coverage, a claims history of either 2 AD diagnostic codes or one diagnosis code and a prescription fill for a medication indicated to treat AD, and 24 months of continuous coverage in the health plan prior to and after a diagnosis. Descriptive statistics were used to estimate mean monthly health care utilization and total health plan costs in the two years prior to incident AD diagnosis. A control cohort of elderly individuals without any dementia type was constructed using a 1:1 exact match for age, gender, and geographic region. Results:A cohort of 4,039 beneficiaries was identified for inclusion in the analysis, 72% of whom were female. Health care utilization and monthly Medicare expenditures trended upward approximately 6-12 months prior to the diagnosis, with a clear peak in the 30 days immediately prior to the incident diagnosis (i.e. index month). Increases in the mean number of monthly outpatient visits (80%), emergency room visits (502%), inpatient stays (829%) and medication use (22%) were observed in the index month, compared to the 12 months prior to index. Mean per-person expenditures were greater among the AD cohort relative to controls in the indexmonth (

LEVERAGING MEDICARE CLAIMS TO CHARACTERIZE DIAGNOSTIC PATHWAYS IN ALZHEIMER’S DISEASE

1,446 v

Profiles of functional and cognitive impairment in Alzheimer's disease and mild cognitive impairment compared to elderly community controls

376) and in the 12 months prior to index (

Discriminating healthy elderly, MCI, mild and moderate Alzheimer's disease: The Computerized Neuropsychological Test Battery (CNTB)

3,993 v

P4-385: Development and validation of new measures of long-term memory consolidation

3,449). Conclusions:These findings demonstrate that the demand for health care begins to rise modestly in the year prior to the first appearance of a diagnostic code. An improved understanding of health care patterns immediately preceding an AD diagnosis may allow health care providers to implement strategies that help patients and caregivers manage the disease. P2-558 COMPARISON OF HEALTHCARE COSTS FOR ALZHEIMER’S DISEASE (AD) PATIENTS DIAGNOSED IN INPATIENT VERSUS NON-INPATIENT SETTINGS Howard M. Fillit, Rezaul Karim Khandker, Lin Xie, Baishali M. Ambegaonkar, Furaha Kariburyo, Onur Baser, Huseyin Yuce, Christopher M. Black, The Alzheimer’s Drug Discovery Foundation, New York, NY, USA; Merck&Co., Inc., Kenilworth, NJ, USA; STATinMED Research, Ann Arbor, MI, USA. Contact e-mail: christopher. black2@merck.com

P3-111: Cognitive assessment: Discrimination of impairment and detection of decline in Alzheimer's disease and mild cognitive impairment

Background: Real-world, administrative claims data is uniquely suited to help characterize the journey of Alzheimer’s disease (AD) patients from the onset of disease symptoms through diagnosis and beyond. The primary aims of this analysis were to: (1) characterize the temporal relationship between diagnostic and treatment codes associated with AD; and (2) explore the pre-diagnostic claims record for the appearance of codes suggestive of ADrelated symptoms. Methods: Medical and pharmacy claims data from a large, geographically diverse, commercial, US Medicare Advantage health plan, for the period 2009-2014, was used to identify cases of newly diagnosed AD. Beneficiaries with claims for AD-specific diagnostic codes and prescription fills for a medication indicated to treat AD were included in the analysis. Results:Among the 4,039 beneficiaries identified for inclusion in the analysis: 31% received a treatment prior to their first Alzheimer’s-specific diagnostic code, 14% received an AD diagnostic code prior to their first treatment; and 23% received their treatment and diagnostic codes within 60 days of one another. Thirty-percent of the diagnosed cohort had multiple diagnostic codes but no evidence of treatment at any time during the 48-month observation period (24 months prior to and after incident diagnosis). Notable gaps between treatment and diagnosis were observed. The median number of days between treatment and diagnosis among those receiving a treatment first was 577; the median number of days between diagnosis and treatment in those receiving a diagnosis first was 397. Sixty-three percent of the combined cohort had an average of 6.1 claims suggestive of AD-related symptoms (e.g. memory loss, mild cognitive impairment) before their incident diagnostic code. Conclusions: These data suggests there is considerable heterogeneity in the AD diagnostic and treatment pathways. While the first appearance of a diagnosis code is commonly used as the index date from which disease related health care use is attributed, measurable careseeking behavior and therapy associated with AD begins much earlier. Further research is needed to better understand the factors influencing pathway variability and the risks associated with diagnostic and treatment delays.

P3-120: Episodic memory test constructs affect discrimination between healthy elderly and cases with mild cognitive impairment and Alzheimer's disease

Gender and number of APOE e 4 alleles included as covariates. Results: We tested a total of 11 SNPs and the APOE e 4 allele for association with AD in our African American case control series (106 AD, 208 controls). None of the SNPs tested achieved nominally significant association with AD diagnosis. Analysis of APOE identified significant increased risk with increasing number of APOE e 4 alleles (OR1⁄43.92, p<1e-04).Conclusions: The novel LOAD susceptibility loci identified in recent years through GWAS analysis of Caucasian LOAD patients and controls do not achieve nominally significant association in our African American case control series. This may be due to the small sample size of our case control series, resulting in the study being underpowered, or may reflect a true fundamental difference in the association of these SNP’s with AD in the two populations. Analysis with cognitive quantitative phenotypes collected in these subjects will be carried out. To increase the power of these studies, recruitment/collection of biological samples of AD patients and controls in the African American population is of increasing importance.

Cognitive and Functional Profiles in Mild-to-Moderate Alzheimer’s Disease and Mild Cognitive Impairment Compared to Healthy Elderly

Background: It is important to know whether subtle cognitive decline shown by older adults is an early sign of dementia or a healthy aging process. Methods: The database of Prevalence Study 1998 and those of Incidence Study 2003 were retrospectively analyzed. The 200 adults who were assessed as Clinical Dementia Rating (CDR) 0 in the two surveys during a 5-year period (1998 to 2003) were classified into four age groups based on the age at the baseline in 1998, i.e. 65-69 years, 70-74 years, and 75 years and over groups. Cognitive functions were assessed using the Cognitive Abilities Screening Instrument (CASI). It consists of nine cognitive domains, i.e. remote memory, recent memory, attention, concentration and mental manipulation, orientation, figure copying, abstraction and judgment, list-generating fluency, and language. Two-way ANOVA with the covariance of educational level was performed to show the time differences of each cognitive domain. Results: All CASI domains showed no time differences between 1998 and 2003 except for orientation. Conclusions: We found that the healthy older adults assessed as CDR 0 showed stable cognitive performances except for time orientation. This means that recent memory impairment, even a slight decline, may be an early sign of dementia which could be discriminated from the healthy aging process.