ǂ Follow-Up Used for the Multivariate Analysis

Supplementary table 1: Summary of the studies assessing factors associated with ChEI’s non persistency (conducting with a new user design and using multivariate analyses).

Author Date / Country / Data source / Study inclusionperiod / Timing of the main outcome measurementǂ / Sample size / Population (dementia criteria if available) / Antidementia drugs
studied / Definition of incident use / Outcome and management of memantine therapy in the outcome definition / Statistical analysis / Ajustement on dementia severity/ tested variables / Non persistency rate / Factors associated with non persistency / Comment
Saleh 2013 [27] / Canada (Saskatchewan) / Memory clinic / 2004 / 6 months / 63 / Ambulatory patients (AD: 84%, LBD: 8%, mixed dementia: 5%, vaD : 3%) / Donepezil
Rivastigmine
Galantamine / Not defined / Discontinuation: (not defined)
Switching not considered / Logistic regression / Yes
Cognitive
(MMSE) and functional assessments (FAQ, BADL, IADL), BPSD (NPI, CESD),
age, sex, marital status, education years, ethnicity, nb disease,
smoking status, alcohol consumption, physical exercise,
quality of life, life concern scale / 6-month discontinuation: 30.2% / Low education / -
Haider 2013 [28] / Austria / Reimbursement database (10 out of 13 Austrian Health Insurances : 9 Federal District Health Insurance Funds) / January 2008-December 2009 / 6 and 12 months / 15809 / > 50 yocommunity subjects with dementia (diagnosis by a specialist)
initiating ChEI or memantine / Donepezil
Rivastigmine
Galantamine
Memantine / 12 months / Non persistency: discontinuation or switch
Discontinuation: gap>90 days
Switching : change between ChEI or for memantine at any time point / Logistic regression / No
(age, sex, index ChEI) / 6-month discontinuation: 34%
12-month discontinuation: 58.5%
6-month switch: 6.9%
12-month switch: 12% / None identified / ChEI are restricted to mild-to-moderate AD, memantine for moderate to severe dementia
Dementia should be first diagnosed by a specialist (ICD F00 F02.3) . General practitioners are not allowed to initiate antidementia drugs
Brewer 2013 [10] / Ireland / Reimbursement database(Irish Health Service Executive- Primary Care Reimbursement Services) / January 2007-December 2010 / 6 months / 14197 / >70 yocommunity subjects with ≥ 2 monthly prescription of Chei or memantine between 2007 and 2010 / Donepezil
Rivastigmine
Galantamine
Memantine / 12 months / Discontinuation : gap>63 days after previous filled prescription
Switching not considered (excluded from the analysis) / Survival analysis using time fixed covariables / No
(age, sex,antidementia drug, year of antidementia drug initiation,maximumnb of drug prescribed monthly, coprescription of an antidepressant or AP drug,coprescription of >1
antidementiadrugs) / 6-month discontinuation : 30.1 %
12-month discontinuation :43.8 % / Factors associated with 6-month discontinuation:
older age,
index drug,
first (2007) vs last (2010) year of the study,
single vs multiple antidementia drugs therapy / The 4 antidementia drugs are subsidized by the Primary care reimbursement services
Amuah 2010 [9] / Canada
(Saskatchewan) / Reimbursement database(Saskatchewan Administrative health data, covering 91% of the province) linked to EDS special request file, population registry, hospital separations, physician services, outpatient prescription drug claims, special care home (long-term care), and vital statistics / December 2000- December 2001 / Up to 40 months / 1080 / Mild-to-moderate AD (DSM-IV criteria) (mostly in the community) / Donepezil
Rivastigmine / Not defined.
(ChEI received during the first year of coverage) / Non persistency : gap>60 days from estimated depletion date
Switches were not considered as a discontinuation except if they occurred after 60 days
Memantine unavailable / Survival analysis using time varying covariables / Yes
Cognitive and functional domains (recent MMSE and FAQ scores),
age sex, residence (urban/rural), social assistance/income supplements, residence in a long-term care facility at baseline, share of ChEI cost (>65%) Chronic Disease Score, number of physicians visits / 12-month discontinuation : 66.4% / Female gender,
low MMSE scores, absence of social assistance,
paying at least 65% of total prescription costs,
rare physician visits,
low Chronic Disease Score, low FAQ / Stringent initiation and renewal criteria for ChEI coverage.
ChEIs are covered in mild-to-moderate AD (DSM IV).
For coverage to be continued, patients must not have shown both a greater than two-point reduction in MMSE and a one-point increase in FAQ over the preceding 6 months.
Pariente 2010 [19] / France / Reimbursement database(EchantillonGeneraliste de Beneficiaires) / January 2004-December 2005 / 12 months / 942 / ChEI in monotherapy
initiators,
(mostly ambulatory) / Donepezil
Rivastigmine
Galantamine / 6 months / Discontinuation:
gap>60 days between refills
ITT approach :
switches (to ChEI or memantine) were not considered as discontinuation except if they occurred after >60 days / Logistic regression / No
(age, sex, initial prescriber specialty for the index drug, other drugs) / 12-month discontinuation : 54.7% / Older age, no antidepressant use / -
Kroger 2010 [15] / Netherlands / Administrative database(PHARMO Record Linkage System) / July 1998-january 2008 / 6 months
Follow-up up to 3 years (median follow-up 267 days) / 3396 / > 50 yo
ChEI initiators / Rivastigmine
Galantamine / 12 months / Discontinuation:
gap >90 days from estimated depletion date using dosage instruction
Switching was not considered as non persistency
Memantine not considered / Survival analysis using time fixed covariables / No
(age,sex, chronic disease score, index ChEI, ChEIdose comedication with atropinics, with CNS properties, other drug use) / 6-month discontinuation : 30.8%
12-month discontinuation : 42.3% / Early (<6-month) discontinuation: female gender, chronic disease score, ChEI type and dosage, no benzodiazepine use, no SSRI use, no PD drug, no propulsive drug use, no cardiac drug use
Late (>6-month) discontinuation: ChEI type and dosage, no SSRI use, no AP drug, no propulsive use / -
Gadzhanova 2010 [11] / Australia / Reimbursement database(Veteran Affairs pharmacy claims) / January 2003-December 2006 / 12 months / 10088 / >65 yocommunity and long term careChEI initiators / Donepezil
Rivastigmine
Galantamine / 12 months / Non persistency:discontinuation or switch
Discontinuation : not defined
Switching : not defined
Memantine: not précised / Survival analysis using time fixed covariables / No
(sex, age, index ChEI drug, residential status) / 6-month non persistency : 47%
12-month non persistency : 58% / ChEI index drug, residential status / ChEI subsidized under the Pharmaceutical Benefits Scheme
Sun 2008[22] / Taiwan / Reimbursment database (Bureau of National Health Insurance (BNHI)) / 2001-2004 / Up to 48 months / 9877 / Mild-to-moderate (MMSE 10-26, CDR 1-2) AD (ICD9, DSM III or NINCDS-ADRDA) started on ChEI / Donepezil
Rivastigmine
Galantamine / Probably mostly incident use considering the rules for initiation, but unclear and not defined / Not defined
Memantine unavailable / Survival analysis using time fixed covariables / No
(age, sex, index ChEI) / Mean duration of use: 432 days.
6-month discontinuation rate: 35.4%
12 month discontinuation rate: 56% / Older age > 76 yo / Donepezil availability: october 1998, Galantamine : January 2000, Rivastigmine : April 2000
Stringent rules for ChEI coverage (mild to moderate AD) and discontinuation (MMSE score worsened > 2 points or CDR worsened by ≥ 1 grade in the follow-up every half year).
Massoud 2008[17] / Canada (Québec) / Reimbursementdatabase (Régie de l’Assurance Maladie du Québéc) / April 2001-june 2003 / 12 months / 18748 / > 65 yo
mild-to -moderate AD initiating ChEI (criteria for ChEI coverage) / Donepezil
Rivastigmine
Galantamine / Not defined / Non persistency: non renewal of any ChEI (gap> 45 days after last fill or >15 days from estimated depletion date
Memantine unavailable / Survival analysis using time fixed covariables / No
(age, gender,
specialty of the physician initiating ChEI, nb of drug dispensations, health care
services use per person , chronic disease score) / 12-month non persistency : 60% / Older age, ChEI type, prescriber specialty (general practitioner) / Stringent inclusion criteria for ChEI coverage : diagnosis
of mild to moderate AD (MMSE between 10-26).
Umegaki 2008[26] / Japan / Geriatric outpatient clinic (Nagoya university hospital / July 2003 June 2005 / Up to 24 months / 264 / Community AD (DSM-IV) / Donepezil / Not defined / Not defined / Logistic regression / Yes
(CDR, age, sex, living arrangements) / Discontinuation cumulative incidence rate : 53% / More severe impairment (CDR) / Donepezil is the single ChEI licensed in Japan
Mucha 2008 [18] / USA / Reimbursement database (MarketScan Medicare) / January 2001-December 2002 / 12 months / 3177 / > 65 yo subjects with AD (ICD9 331.0) in the community and long-term care / Donepezil
Rivastigmine
Galantamine / 6 months / Discontinuation: gap>30 days from estimated depletion date
Switching: a
different ChEI at any point during follow-up
Memantine unavailable / Logistic regression / No
(age, sex, region, managed care plan type, Charlsoncomorbidity score, index ChEI) / 12-month discontinuation:
Donepezil : 63.5%
Galantamine:63.7
Rivastigmine:68.0% / Discontinuation: index ChEI
Switching : older age, index ChEI / -
Frankfort 2005 [24] / Netherlands / Geriatric outpatient department of a general hospital / 1998-2004 / 6 months / 154 / Mild to moderately severe AD (NINCDS ADRDA criteria) started on rivastigmine / Rivastigmine / Probably mostly incident use, but unclear (especially for the patients included after 1999) and not defined / Not defined
Memantine unavailable / Logistic regression / Yes
( (baseline MMSE score, age, sex, education, place of living, nb concomitant drugs, titration schedule, maximum achieved dose, nurse support) / 6-month discontinuation : 39.6% / Low level of education, lower maximum achieved dose, less nurse support / Donepezilunavailable in the Netherlands
Discontinuation criteria: if one of the domains showed major
deterioration or if minor decline in two domains without
improvement in the third domain was shown ( based on comparison with a historical control cohort of AD patients before antidementia drugs licensing)
Abugosh 2008 [8] / USA / Reimbursement database (Rhode Island Medicaid) / July 2001- June 2003 / At least 6 months, up to 30 months / 1564 / ≥50 yocommunity and long-term care
ChEI initiators / Donepezil
Rivastigmine
Galantamine / 6 months / Discontinuation:
gap>180 daysbetween refills
Memantine unavailable / Survival analysis using time fixed covariables / No
(age, sex, race, care setting, ChEI index drug, co-morbid) / 12-month discontinuation : 42.7% / Non Caucasian race / -
Suh
2005 [21] / USA / Reimbursement database (MarketScan Medicare) / July 2000-June 2001 / 12 months / 783 / > 65 yocommunity incident
AD (ICD9 codes 331.0 ; 290.0 to 290.3) ChEI initiators / Donepezil
Rivastigmine / 18 months / Non persistency: discontinuation or switch
Discontinuation : gap >60 days from estimated depletion date
Switch : change for any other antidementia drug strategy
Memantine unavailable / Survival analysis using time fixed covariables / No
(age, sex, CNS drugs use, co-morbid conditions, patterns of medical services use (hospitalisation, nb of physician office visits) / 12-month non persistency: 53% / CNS drug use, absence of hospitalization, absence of physician visits
Singh
2005 [20] / USA
(California) / Reimbursement database (Medicaid) / Avril 2000 / Up to 20 months (mean 431 days) / 17 742 / > 40 yo
ChEI initiators / Donepezil
Rivastigmine / 3 months / Discontinuation: switch
or gap ≥60 days
from estimated depletion date
Memantine unavailable / Survival analysis using time fixed covariables / No
(age,sex,
Indexdrug) / Median time to discontinuation : 120 days (donepezil) and 135 days (rivastigmine) / None identified / -
Kogut 2005 [14] / USA / Reimbursment database (Rhode Island Medicaid) / January 2000-June 2002 / 6 months / 1183 / >45 yocommunity and long term careChEI initiators / Donepezil
Rivastigmine
Galantamine
Tacrine / 2 months / ≥ 5 prescriptions for a 1-month supply of the same ChEI, without an extended gap (<124 days between 3th and 5th refill)
Switches were excluded
Memantine unavailable / Logistic regression / No
(age, sex,
race, care setting,
drugs impairing cognition) / 6-month discontinuation : 26% / Drugs impairing cognition, non Caucasian race, community setting / -

ǂ Follow-up used for the multivariate analysis

Legend : AD: Alzheimer’s Disease; ADL: Activities of daily living; AP :Antipsychotic; BADL: Bristol Activities of Daily Living; BPSD : Behavioral and Psychological Disturbances in Dementia ; CDR: clinical dementia rating ; CES-D: Center for Epidemiologic Studie- Depression scale; ChEI : Cholinesterase inhibitors; CNS : Central Nervous System; DSM : Diagnostic and Statistical Manual of Mental Disorders; FAQ: Functional Assessment Questionnaire; IADL: Instrumental Activities of Daily Living; ICD: International Classification of Diseases; LBD: Lewy Body Dementia; MMSE: Mini Mental Score Examination; nb: number; NINCDS-ADRDA: National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association; NPI: Neuro Psychological Inventory; PD : Parkinson Disease; PY : person- year; QOL: quality of life; SSRI : selective serotonin reuptake inhibitors; vaD : Vascular Dementia; yo : year-old.