Resident-Level Predictors of Dementia Pharmacotherapy at Long-Term Care Admission: The Impact of Different Drug Reimbursement Policies in Ontario and Saskatchewan: Prédicteurs de la pharmacothérapie de la démence au niveau des résidents lors de l’hospitalisation dans des soins de longue durée : l’impact de différentes politiques de remboursement des médicaments en Ontario et en Saskatchewan

Abstract

Objectives:

Cholinesterase inhibitors (ChEIs) and memantine are approved for Alzheimer disease in Canada. Regional drug reimbursement policies are associated with cross-provincial variation in ChEI use, but it is unclear how these policies influence predictors of use. Using standardized data from two provinces with differing policies, we compared resident-level characteristics associated with dementia pharmacotherapy at long-term care (LTC) admission.

Methods:

Using linked clinical and administrative databases, we examined characteristics associated with dementia pharmacotherapy use among residents with dementia and/or significant cognitive impairment admitted to LTC facilities in Saskatchewan (more restrictive reimbursement policies; n = 10,599) and Ontario (less restrictive; n = 93,331) between April 1, 2009, and March 31, 2015. Multivariable logistic regression models were utilized to assess resident demographic, functional, and clinical characteristics associated with dementia pharmacotherapy.

Results:

On admission, 8.1% of Saskatchewan residents were receiving dementia pharmacotherapy compared to 33.2% in Ontario. In both provinces, residents with severe cognitive impairment, aggressive behaviors, and recent antipsychotic use were more likely to receive dementia pharmacotherapy; while those who were unmarried, admitted in later years, had a greater degree of frailty, and recent hospitalizations were less likely. The direction of the association for older age, rural residency, medication number, and anticholinergic therapy differed between provinces.

Conclusions:

While more restrictive criteria for dementia pharmacotherapy coverage in Saskatchewan resulted in fewer residents entering LTC on dementia pharmacotherapy, there were relatively few differences in the factors associated with use across provinces. Longitudinal studies are needed to assess how differences in prevalence and characteristics associated with use impact patient outcomes.

Cholinesterase inhibitors (ChEIs) and memantine are the only medications approved for the symptomatic treatment of Alzheimer disease (AD) and Parkinson disease dementia in Canada. Current Canadian guidelines recommend that ChEIs be considered for persons at all stages of AD (including those with severe AD or coexisting cerebrovascular disease) and with mild to moderate Parkinson disease dementia and that memantine be considered for those at a moderate to severe stage.^1,2 In both Canada and the United States, these medications are widely used in long-term care (LTC), with approximately 30% to 40% of residents with dementia prescribed these agents.^3–6

Systematic reviews and meta-analyses have shown modest improvements in cognition and global functioning with ChEI use but also more adverse effects,^7–9 including gastrointestinal symptoms, bradycardia, syncope, falls, and fractures.^10–13 Much of this evidence has been generated in community-dwelling populations with less known about the long-term benefits and risks of ChEI and memantine use or discontinuation among institutionalized populations.^9,14 These older, frailer populations with diminished reserve may be particularly susceptible to the adverse effects associated with ChEI use.^9,15

In addition to patient- and prescriber-level factors, regional drug reimbursement policies are an important determinant of dementia pharmacotherapy use.¹⁶ As there is no national drug benefit program in Canada, there are interprovincial differences in publicly funded coverage of medications, criteria for restricted drug benefits, and out-of-pocket costs.^17,18 While public drug benefit programs in all Canadian provinces reimburse ChEIs for persons with mild to moderate AD,¹⁹ variation exists in terms of the criteria used for determining coverage, compliance with these criteria by prescribers, and in the enforcement of these criteria.^19–21 Memantine is not reimbursed by Canadian provincial drug benefit plans with the exception of Quebec, where coverage is provided under certain conditions.²² If not covered by a public plan, dementia pharmacotherapy can still be prescribed with the patient either paying for the medication or obtaining coverage through a private insurance plan. Earlier Canadian studies showed large interprovincial variation in the proportion of older persons with dementia receiving pharmacotherapy,^19,23 likely driven, in part, by the differing provincial policies governing reimbursement.

Recently, variation in the utilization of ChEIs and memantine has garnered increased attention across Canada and internationally given the growing number of potential recipients, differences in drug coverage that affect access and costs, uncertainty about the balance between benefits and risks, and concerns whether use aligns with clinical practice guideline recommendations.^24,25 For example, though current treatment guidelines recommend against the use of anticholinergic medications in persons with dementia due to their adverse impact on cognitive functioning, these medications continue to be coprescribed with ChEIs.²⁶ Clinical practice guidelines do not recommend discontinuation of treatment on entry to LTC or another institutional setting,²⁷ however, they do suggest that consideration be given to discontinuing therapy for those who have progressed to a stage where there is no meaningful benefit.²

In Canada, comparable linked administrative and clinical databases across provinces provide unique opportunities to perform population therapeutic studies that address the relative impact of varying drug policies on patterns and predictors of medication use. An important question is whether a more (vs. less) restrictive policy approach to the reimbursement of dementia pharmacotherapy leads to use that is more consistent with clinical practice guidelines at the individual patient level.

In the current study, we examine predictors of dementia pharmacotherapy use among newly admitted LTC residents with dementia and/or significant cognitive impairment in two provinces (Ontario and Saskatchewan) with significantly different ChEI reimbursement policies. The LTC setting allowed us to explore patterns and predictors of ChEI use in a frail older population, who often undergo medication review with accompanying changes at this important transition point in their care. We hypothesized that residents with increased frailty, more severe cognitive impairment, and recent use of psychotropic and anticholinergic medications would be less likely to receive dementia pharmacotherapy (reflecting current treatment guidelines)² and that these associations would differ between the two provinces given differences in their drug reimbursement policies.^20,21,23 We also hypothesized that Saskatchewan LTC residents would be less likely to newly start dementia pharmacotherapy in the year following admission.

Methods

We conducted a retrospective cohort study using comparable linked clinical and health administrative databases available in each province. Ontario data were linked using unique encoded identifiers and analyzed at ICES. Use of Ontario data in this project was authorized under section 45 of Ontario’s Personal Health Information Protection Act, which does not require review by a Research Ethics Board. Data from Saskatchewan were linked and analyzed at the Saskatchewan Health Quality Council. The University of Saskatchewan Biomedical Research Ethics Board (Bio-REB) approved the Saskatchewan component of the study.

Data Sources

LTC admission information was obtained from the Continuing Care Reporting System. This database includes clinical information captured using the Resident Assessment Instrument Minimum Dataset (RAI-MDS), Version 2.0. The RAI-MDS full assessment is administered on all residents in both provinces shortly after initial admission, annually, and following a significant change in health status. The assessment captures information on cognition, mood, function, behaviors, and comorbidities. These clinical data were linked to population-based health administrative databases in each province, which included information on demographic characteristics, health system, and drug utilization (see Supplemental Table S1 for list of databases).

Information on dispensed prescription medications for individuals aged 65+ years in Ontario was available from the Ontario Drug Benefit Database. Information on memantine and other medications either paid for out of pocket or reimbursed by private insurance was not available in Ontario. Dispensed prescription medication information for Saskatchewan residents was available from the Prescription Drug Plan Database, which includes prescriptions reimbursed by the publicly funded drug plan that covers approximately 91% of the Saskatchewan population.²⁸ This database also contains information on prescriptions reimbursed by private health insurance or paid for out of pocket. The Saskatchewan private drug reimbursement data included information on memantine use.

Study Cohort

All LTC residents aged 66+ years in Ontario (and 50+ years in Saskatchewan) newly admitted between April 1, 2009, and March 31, 2015, with a diagnosis of dementia and/or evidence of cognitive impairment were eligible for cohort inclusion. Younger age groups were included in Saskatchewan because drug data were available. The presence of dementia was ascertained by either a validated health administrative data algorithm²⁹ or documentation of its presence in the RAI-MDS. Cognitive impairment was based on the Cognitive Performance Scale (CPS score ≥2) based on items in the RAI-MDS.³⁰ As in previous work, individuals with significant cognitive impairment were included as they may not have received a formal diagnosis of dementia but could be prescribed ChEIs/memantine.⁵ Persons who did not have a RAI-MDS assessment within 30 days of admission were excluded, as recent information on resident health status was required for our analyses.

Dementia Pharmacotherapy

Detailed information on dementia pharmacotherapy reimbursement policies in both provinces can be found in Supplemental Appendix S1. Briefly, ChEI reimbursement policy in Saskatchewan requires detailed documentation of diagnosis, cognitive impairment, and functional status for public reimbursement. Anticholinergic medications are required to be discontinued prior to dementia pharmacotherapy initiation. In Ontario, patients are required to meet eligibility criteria based on diagnosis and cognitive impairment, but these criteria are not strictly enforced.

In Ontario, individuals were classified as receiving dementia pharmacotherapy (ChEI only) at LTC admission if the service date plus days supplied overlapped the admission date. As in previous work in Saskatchewan, the estimate for days supplied was based on the quantity obtained and recommendations for usual administration of the agent because information on days supplied was not provided.²¹ Individuals receiving multiple ChEIs on admission were excluded. Residents were followed for 1 year from admission to determine their pattern of dementia pharmacotherapy use, which was categorized as follows: never (did not receive any dementia pharmacotherapy at or following LTC admission), maintained (received dementia pharmacotherapy both at and following admission), discontinued (received dementia pharmacotherapy at admission but discontinued during the following year), and newly started (did not receive dementia pharmacotherapy at admission but newly started in the following year). Consistent with previous work, discontinuation was defined as a gap greater than 30 days between the end of one prescription (based on days supplied) and the dispensation of another prescription.⁵

Resident Characteristics

Resident demographic characteristics derived from administrative databases (and assessed on admission date) included age, sex, urban/rural residence, and neighborhood income quintile. Rural residence was defined as areas with <1,000 population based on Statistics Canada’s population center and rural area classification (POPCTRRAclass).³¹ Marital status was obtained from the RAI-MDS. The number of chronic conditions was defined using a summary measure of items from the RAI-MDS diagnoses list. A previously validated 70-item frailty index (based on data available in both provinces) was used to assess resident frailty using a cumulative deficits approach.³² The frailty index was categorized as nonfrail (<0.2), prefrail (0.2 to 0.3), and frail (>0.3). Validated clinical assessment scales derived from items in the RAI-MDS were used to describe residents’ aggressive behaviors (Aggressive Behavior Scale)³³ and cognitive impairment status (CPS).³⁰ Hospitalizations in the year prior to admission as well as any hospitalizations requiring an extended stay (i.e., alternate level of care [ALC]) were defined as measures of health system utilization. The number of distinct drugs (excluding ChEIs and memantine) and use of specific psychotropic drug subclasses (antipsychotics, benzodiazepines, antidepressants) as well as other drugs with anticholinergic properties not already included within these subclasses (defined using the 2015 Beers list of potentially inappropriate medications in persons with dementia) were examined.³⁴ Drug use was assessed on admission and in the 30 days prior to LTC admission (see Supplemental Table S2 for drug lists).

Statistical Analysis

Characteristics of Ontario and Saskatchewan residents receiving dementia pharmacotherapy were compared using z-tests for proportions and t-tests for continuous variables. Logistic regression models were constructed within each province to examine predictors of dementia pharmacotherapy use. Age- and sex-adjusted models as well as fully adjusted models were estimated and accounted for clustering within LTC homes. Fully adjusted models included relevant covariates identified in previous research.^4,21

A sensitivity analysis utilizing only public drug plan data in Saskatchewan (including ChEI use only) was conducted to determine whether predictors were similar to the primary analysis including public and private drug data. All analyses were completed using SAS Version 9.4 (SAS Institute, Cary, North Carolina). Results were considered statistically significant at P < 0.05 (two-tailed).

Results

There were 147,039 older adults admitted to Ontario LTC facilities during the study period and 19,775 to LTC facilities in Saskatchewan. After identifying persons with dementia and/or cognitive impairment and applying our exclusion criteria (including those without an RAI assessment within 30 days of admission; 18.5% [n = 3,660] in Saskatchewan and 5.0% [n = 7,378] in Ontario), there were 93,331 admissions in Ontario and 10,599 in Saskatchewan in the final analytic sample (Supplemental Table S3). Among these individuals, 8.1% (n = 854) of Saskatchewan residents were receiving dementia pharmacotherapy (ChEIs or memantine) compared with 33.2% (n = 30,995) in Ontario (ChEIs only). In Saskatchewan, 93% (n = 794) of treated residents received a ChEI only, 4.4% (n = 38) memantine only, and 2.6% (n = 22) both a ChEI and memantine. Of the residents who received dementia pharmacotherapy, 36.5% (n = 312) received a ChEI through the Saskatchewan public drug benefit program.

Newly admitted LTC residents in Saskatchewan were more likely to have never received dementia pharmacotherapy on admission or in the year following admission (90% vs. 60% in Ontario; Figure 1). Ontario residents were more likely to maintain use of dementia pharmacotherapy following admission to LTC (28% vs. 5% in Saskatchewan). While discontinuation and newly starting dementia pharmacotherapy occurred infrequently in both provinces, Saskatchewan residents were less likely to newly start dementia pharmacotherapy following admission (3% vs. 6% in Ontario).

Figure 1.

Pattern of dementia pharmacotherapy^a use at admission to long-term care and in the following year among newly admitted residents with dementia and/or cognitive impairment, by province (admitted between 2009 and 2015). ^aDementia pharmacotherapy defined as cholinesterase inhibitor use in Ontario (public drug data only) and cholinesterase inhibitor or memantine use in Saskatchewan (public and private drug data).

Relative to Saskatchewan residents, Ontario residents receiving dementia pharmacotherapy on admission were more likely to be female, widowed, and from a lower income neighborhood but less likely to reside in a rural area (Table 1). Ontario residents receiving dementia pharmacotherapy were more likely to be frail and have severe aggressive behaviors, whereas those receiving dementia pharmacotherapy in Saskatchewan were more likely to be moderately to severely cognitively impaired and to have had a recent hospital admission (including one with an ALC stay). Saskatchewan residents were also more likely to have recently received an antipsychotic, whereas Ontario residents were more likely to have received an antidepressant. There was no statistically significant difference in the use of drugs with other anticholinergic effects among residents in the two provinces.

Table 1.

Sociodemographic and Health Characteristics of Newly Admitted LTC Residents with Dementia and/or Cognitive Impairment Receiving Dementia Pharmacotherapy^a at Admission, by Province.

Resident Characteristicsb	Saskatchewan (n = 854)	Ontario (n = 30,995)	P Value
Sociodemographic
Sex—female	517 (60.5%)	20,361 (65.7%)	0.002
Age, years
Mean ± SD	83.5 ± 7.0	83.7 ± 6.3	0.41
50 to 74c	75 (8.8%)	2,708 (8.7%)	0.96
75 to 84	367 (43.0%)	13,412 (43.3%)	0.86
85+	412 (48.2%)	14,875 (48.0%)	0.88
Rural residence	366 (42.9%)	4,588 (14.8%)	<0.001
Marital status
Married	304 (35.6%)	11,513 (37.1%)	0.36
Widowed	375 (43.9%)	16,168 (52.2%)	<0.001
Never married/divorced/separated	175 (20.5%)	3,314 (10.7%)	<0.001
Neighborhood income quintile
1 (lowest)	161 (18.9%)	6,790 (21.9%)	0.03
2	178 (20.8%)	5,948 (19.2%)	0.23
3	137 (16.0%)	6,251 (20.2%)	0.003
4	155 (18.1%)	6,147 (19.8%)	0.22
5 (highest)	173 (20.3%)	5,674 (18.3%)	0.15
Missing	50 (5.9%)	185 (0.60%)	<0.001
Health status
Number of chronic conditions
0 to 5	766 (89.7%)	24,610 (79.4%)	<0.001
>5	88 (10.3%)	6,385 (20.6%)	<0.001
Frailty index
Nonfrail (<0.2)	225 (26.3%)	5,325 (17.2%)	<0.001
Prefrail (0.2 to 0.3)	340 (39.8%)	11,791 (38.0%)	0.29
Frail (≥0.3)	289 (33.8%)	13,879 (44.8%)	<0.001
Aggressive Behavior Scale
Not aggressive (0)	577 (67.6%)	16,845 (54.4%)	<0.001
Mild to moderate (1 to 4)	226 (26.5%)	11,019 (35.6%)	<0.001
Severe (5+)	51 (6.0%)	3,131 (10.1%)	<0.001
Cognitive Performance Scale
Intact/borderline intact/mild impairment (0 to 2)	212 (24.8%)	10,919 (35.2%)	<0.001
Moderate impairment (3)	408 (47.8%)	13,714 (44.3%)	0.04
Moderately severe to very severe impairment (4 to 6)	234 (27.4%)	6,362 (20.5%)	<0.001
# hospitalizations in year prior to LTC admission
0	369 (43.2%)	19,434 (62.7%)	<0.001
1	251 (29.4%)	8,077 (26.1%)	0.03
2+	234 (27.4%)	3,484 (11.2%)	<0.001
Acute care admission with ALC in year prior to LTC admission	227 (26.6%)	5,525 (17.8%)	<0.001
Medications (in 30 days prior to or on LTC admission)
Number of distinct drugsd
0 to 5	281 (32.9%)	9,326 (30.1%)	0.08
6 to 9	270 (31.6%)	10,120 (32.7%)	0.52
10 to 19	279 (32.7%)	10,597 (34.2%)	0.36
20+	24 (2.8%)	952 (3.1%)	0.66
Psychotropic medications
Antipsychotics	253 (29.6%)	6,975 (22.5%)	<0.001
Antidepressants	120 (14.1%)	8,549 (27.6%)	<0.001
Benzodiazepines	79 (9.3%)	2,635 (8.5%)	0.44
Other drugs with anticholinergic effects	31 (3.6%)	1,192 (3.9%)	0.75

Note. Admitted Between 2009 and 2015. ALC = alternate level of care; LTC = long-term care; SD = standard deviation.

^a Dementia pharmacotherapy defined as cholinesterase inhibitor use in Ontario (public drug data only) and cholinesterase inhibitor or memantine use in Saskatchewan (public and private drug data).

^b Assessed at admission unless otherwise noted.

^c Aged 66 to 74 years in Ontario, 50 to 74 years in Saskatchewan.

^d Excluding cholinesterase inhibitors and memantine

Among users of dementia pharmacotherapy, Ontario residents showed longer total durations of use prior to admission; however, Saskatchewan residents were more likely to receive dementia pharmacotherapy in the 3- and 6-month periods immediately prior (Table 2). Donepezil was the most commonly used ChEI in both provinces, followed by galantamine and rivastigmine. Memantine was used in relatively few Saskatchewan residents (7% of dementia pharmacotherapy users).

Table 2.

Characteristics of Dementia Pharmacotherapy^a Use Among Newly Admitted LTC Residents with Dementia and/or Cognitive Impairment Who Received Dementia Pharmacotherapy at Admission, by Province.

Dementia Pharmacotherapy Use Characteristicsa	Saskatchewan	Ontario	P Value
N	854	30,995
Total duration of ChEI/memantine use prior to admission (months)b
No previous use	35 (4.1%)	1,758 (5.7%)	0.049
<6	150 (17.6%)	3,951 (12.8%)	<0.001
6 to <12	93 (10.9%)	2,977 (9.6%)	0.21
12 to <24	209 (24.5%)	5,445 (17.6%)	<0.001
≥24	367 (43.0%)	16,864 (54.4%)	<0.001
Mean ± SD, days	800.4 ± 689.9	977.3 ± 815.1	<0.001
Median, days (IQR)	641 (294 to 1,158)	822 (301 to 1,467)
Received at least one ChEI/memantine during time periods prior to admission
3 Months prior	731 (85.6%)	19,345 (62.4%)	< 0.001
6 Months prior	766 (89.7%)	26,639 (86.0%)	0.002
1 Year prior	774 (90.6%)	28,463 (91.8%)	0.21
ChEI/memantine type at admission
Donepezil only	578 (67.7%)	19,285 (62.2%)	0.001
Galantamine only	178 (20.8%)	9,124 (29.4%)	<0.001
Rivastigmine only	38 (4.4%)	2,586 (8.3%)	<0.001
Memantine only	38 (4.4%)	N/A	N/A
Memantine and any ChEI	22 (2.6%)	N/A	N/A

Note. Admitted between 2009 and 2015. ChEI = cholinesterase inhibitor; IQR = interquartile range; LTC = long-term care; N/A= not applicable; SD = standard deviation.

^a Dementia pharmacotherapy defined as cholinesterase inhibitor use in Ontario (public drug data only) and cholinesterase inhibitor or memantine use in Saskatchewan (public and private drug data).

^b Total duration of previous use prior to LTC admission calculated back to April 1, 2003. Note that this duration is not necessarily continuous.

In fully adjusted models, several predictors of use were similar for both provinces (Table 3). In both provinces, higher levels of cognitive impairment and aggressive behaviors as well as recent receipt of antipsychotics were associated with increased odds of receiving dementia pharmacotherapy. Residents in both provinces who were unmarried, prefrail or frail, with a recent hospitalization, and/or admitted during the later years of the study were less likely to receive dementia pharmacotherapy. Older individuals (aged 85+ years) in Ontario were less likely to receive dementia pharmacotherapy at admission (odds ratio [OR] = 0.94; 95% confidence interval [95% CI], 0.88 to 0.99), while older Saskatchewan residents showed a nonstatistically significant positive association (OR = 1.25; 95% CI, 0.90 to 1.75). In Ontario, individuals previously residing in the highest income neighborhoods were more likely to receive dementia pharmacotherapy (OR = 1.09; 95% CI, 1.04 to 1.14). A similar (but not statistically significant) association was observed in Saskatchewan (OR = 1.29; 95% CI, 0.99 to 1.69). Receipt of higher numbers of medications was associated with increased odds of dementia pharmacotherapy in Saskatchewan (OR = 1.30; 95% CI, 1.09 to 1.55) but decreased odds in Ontario (OR = 0.93; 95% CI, 0.89 to 0.96). In Saskatchewan, recent receipt of an anticholinergic medication was associated with decreased odds of dementia pharmacotherapy (OR = 0.47; 95% CI, 0.31 to 0.69) in contrast to Ontario where receipt was associated with increased odds (OR = 1.09; 95% CI, 1.01 to 1.18). Receipt of antidepressants was associated with increased odds of dementia pharmacotherapy in Ontario (OR = 1.52; 95% CI, 1.46 to 1.57). A similar trend was evident but not statistically significant in Saskatchewan. Likewise while receipt of benzodiazepines was associated with decreased odds of dementia pharmacotherapy in Ontario (OR = 0.80; 95% CI, 0.76 to 0.84), this did not reach statistical significance in Saskatchewan.

Table 3.

Age/Sex Adjusted and Fully Adjusted Odds Ratios (95% Confidence Intervals) for Dementia Pharmacotherapy Use^a at Admission to LTC, Associated with Resident Sociodemographic and Health Characteristics for Each Province.

Covariateb	Saskatchewan		Ontario
	Age- and Sex-Adjusted Odds Ratio (95% CI)	Fully Adjusted Odds Ratio (95% CI)	Age- and Sex-Adjusted Odds Ratio (95% CI)	Fully adjusted odds ratio (95% CI)
Sociodemographic
Sex—female	0.99 (0.85 to 1.15)	1.11 (0.94 to 1.31)	1.05 (1.02 to 1.08)	1.09 (1.05 to 1.13)
Age, years
50 to 74c(ref)	1.00	1.0	1.0	1.0
75 to 84	2.15 (1.64 to 2.81)	1.83 (1.35 to 2.48)	1.35 (1.28 to 1.42)	1.33 (1.26 to 1.41)
85+	1.36 (1.04 to 1.77)	1.25 (0.90 to 1.75)	0.88 (0.84 to 0.93)	0.94 (0.88 to 0.99)
Rural residence	1.34 (1.16 to 1.54)	1.24 (1.05 to 1.48)	0.92 (0.89 to 0.96)	0.95 (0.91 to 1.00)
Marital status
Married (ref)	1.00	1.0	1.0	1.0
Widowed	0.69 (0.58 to 0.83)	0.68 (0.56 to 0.82)	0.77 (0.74 to 0.79)	0.79 (0.76 to 0.82)
Never married/divorced/separated	0.49 (0.40 to 0.61)	0.49 (0.41 to 0.59)	0.63 (0.60 to 0.66)	0.66 (0.63 to 0.69)
Neighborhood income quintile
1 (lowest, ref)	1.0	1.0	1.0	1.0
2	1.30 (1.03 to 1.63)	1.13 (0.88 to 1.45)	1.08 (1.03 to 1.12)	1.04 (0.99 to 1.09)
3	1.08 (0.85 to 1.37)	0.95 (0.74 to 1.22)	1.09 (1.04 to 1.13)	1.03 (0.99 to 1.08)
4	1.10 (0.87 to 1.39)	0.97 (0.74 to 1.28)	1.15 (1.10 to 1.19)	1.05 (1.01 to 1.10)
5 (highest)	1.51 (1.20 to 1.90)	1.29 (0.99 to 1.69)	1.18 (1.13 to 1.23)	1.09 (1.04 to 1.14)
Missing	0.94 (0.67 to 1.32)	0.87 (0.61 to 1.23)	0.91 (0.76 to 1.08)	0.92 (0.77 to 1.11)
Fiscal year of admission to LTC
2009 (ref)	1.0	1.0	1.0	1.0
2010	0.85 (0.67 to 1.07)	0.85 (0.65 to 1.11)	0.97 (0.92 to 1.02)	0.97 (0.92 to 1.02)
2011	0.91 (0.72 to 1.15)	0.88 (0.70 to 1.12)	0.93 (0.89 to 0.98)	0.93 (0.88 to 0.98)
2012	0.67 (0.52 to 0.85)	0.68 (0.54 to 0.86)	0.90 (0.86 to 0.95)	0.90 (0.85 to 0.95)
2013	0.59 (0.46 to 0.76)	0.58 (0.45 to 0.75)	0.84 (0.80 to 0.88)	0.82 (0.78 to 0.87)
2014	0.63 (0.49 to 0.81)	0.63 (0.48 to 0.82)	0.73 (0.70 to 0.77)	0.71 (0.67 to 0.75)
Health status
Number of chronic conditions
0 to 5 (ref)	1.0	1.0	1.0	1.0
>5	0.70 (0.56 to 0.89)	1.00 (0.79 to 1.28)	0.78 (0.75 to 0.80)	0.96 (0.93 to 1.00)
Frailty index
Nonfrail (<0.2) (ref)	1.0	1.0	1.0	1.0
Prefrail (0.2 to 0.3)	0.75 (0.63 to 0.90)	0.67 (0.55 to 0.83)	0.89 (0.85 to 0.92)	0.78 (0.75 to 0.81)
Frail (≥ 0.3)	0.54 (0.45 to 0.66)	0.47 (0.38 to 0.58)	0.72 (0.69 to 0.75)	0.55 (0.52 to 0.58)
Cognitive Performance Scale
Intact/borderline intact/mild impairment (0 to 2), ref	1.0	1.0	1.0	1.0
Moderate impairment (3)	1.51 (1.26 to 1.79)	1.63 (1.35 to 1.96)	1.69 (1.64 to 1.74)	1.69 (1.64 to 1.75)
Moderately severe to very severe impairment (4 to 6)	1.40 (1.15 to 1.71)	1.65 (1.36 to 2.01)	1.72 (1.65 to 1.79)	1.73 (1.65 to 1.82)
Aggressive behavior scale
Not aggressive (0) (ref)	1.0	1.0	1.0	1.0
Mild to moderate (1 to 4)	1.15 (0.98 to 1.36)	1.13 (0.94 to 1.35)	1.29 (1.25 to 1.32)	1.17 (1.14 to 1.21)
Severe (5+)	1.40 (1.03 to 1.92)	1.32 (0.91 to 1.91)	1.71 (1.63 to 1.80)	1.39 (1.32 to 1.47)
# of hospitalizations in year prior to LTC admission
0 (ref)	1.0	1.0	1.0	1.0
1	0.57 (0.48 to 0.67)	0.62 (0.51 to 0.76)	0.57 (0.55 to 0.59)	0.65 (0.63 to 0.67)
2+	0.30 (0.25 to 0.35)	0.39 (0.31 to 0.48)	0.40 (0.38 to 0.41)	0.49 (0.46 to 0.51)
Medications (in 30 days prior to or on admission)
Number of distinct drugsd
0 to 5 (ref)	1.0	1.0	1.0	1.0
6 to 9	1.21 (1.01 to 1.45)	1.27 (1.03 to 1.56)	1.01 (0.97 to 1.04)	1.10 (1.06 to 1.14)
10+	1.02 (0.86 to 1.22)	1.30 (1.09 to 1.55)	0.79 (0.76 to 0.82)	0.93 (0.89 to 0.96)
Psychotropic medications
Antipsychotics	1.73 (1.47 to 2.03)	1.41 (1.18 to 1.67)	1.97 (1.90 to 2.04)	1.62 (1.56 to 1.68)
Antidepressants	1.15 (0.93 to 1.41)	1.04 (0.81 to 1.32)	1.57 (1.52 to 1.62)	1.52 (1.46 to 1.57)
Benzodiazepines	1.02 (0.80 to 1.32)	0.83 (0.66 to 1.05)	0.96 (0.91 to 1.01)	0.80 (0.76 to 0.84)
Other drugs with anticholinergic effects	0.52 (0.36 to 0.77)	0.47 (0.31 to 0.69)	1.15 (1.07 to 1.24)	1.09 (1.01 to 1.18)

Note. Admitted between 2009 and 2015. Bolded estimates are statistically significant at P < 0.05. Fully adjusted model is adjusted for all variables shown in table in addition to clustering in LTC homes. CI = confidence interval; LTC = long-term care; ref = reference.

^a Dementia pharmacotherapy defined as cholinesterase inhibitor use in Ontario (public drug data only) and cholinesterase inhibitor or memantine use in Saskatchewan (public and private drug data).

^b Assessed at admission unless otherwise noted.

^c Aged 66 to 74 years in Ontario, 50 to 74 years in Saskatchewan.

^d Excluding cholinesterase inhibitors and memantine.

In sensitivity analyses using only publicly funded drug data in Saskatchewan (capturing ChEI use only), predictors were generally similar to the primary analysis that included both publicly and privately reimbursed drug data other than some attenuation of effect estimates (Supplemental Table S4).

Discussion

Using comparable linked health administrative data, we confirmed the previously noted disparity in dementia pharmacotherapy use between Ontario and Saskatchewan.^20,21,23 Further, we observed several similarities as well as some differences in resident-level characteristics associated with pharmacotherapy use in these two provinces. We found that the more restrictive reimbursement policy in Saskatchewan was associated with a 4-fold lower rate of dementia pharmacotherapy than among Ontario LTC residents, even after including private pay data in Saskatchewan. Several of the observed predictors of use were similar in both provinces, though we found divergent associations for a few, including increased age, rural residency, number of concurrent medications, and use of anticholinergics. The finding for anticholinergic use is perhaps not surprising as it aligns with specific eligibility criteria in Saskatchewan.³⁵ However, the finding of several similar predictors in both provinces was unexpected given their vastly different drug reimbursement policies. The more restrictive policy in Saskatchewan appears to have lowered use overall rather than altering use among specific targeted populations. We cannot directly comment on whether use in Saskatchewan, with its more restrictive policy, is more appropriate at an individual or population level than the more liberal use found in Ontario, as we have not examined treatment outcomes among LTC residents in these provinces.

The utilization of ChEIs in newly admitted LTC residents in the present study highlights a concern that there may be barriers to the use of these agents in Saskatchewan for appropriate candidates of dementia pharmacotherapy. Earlier studies examining dementia pharmacotherapy use and costs in Saskatchewan and Ontario found similarly large differences. One that examined variation across Canadian provinces found that publicly funded use ranged from a low of 10 users per 1,000 older population in Saskatchewan to a high of 33 per 1,000 in New Brunswick in 2013, with similarly high use in Ontario.¹⁹ In Canada overall, dispensing of ChEIs/memantine was found to increase by 17% over 5 years from 2009 to 2014. Of the total CDN$249 million spent on dementia pharmacotherapy at retail pharmacies in 2012 to 2013, the majority of the costs of dementia pharmacotherapy in Canada (74%) were covered by provincial drug plans, followed by out of pocket expenses (17%) and private insurance (9%).²³ Ontario had the highest per capita spending on dementia medications (CDN$63 and CDN$53 per person for women and men aged 65+ years) while Saskatchewan had the lowest (CDN$17 per person for women and men aged 65+ years).

Previous work examining the impact of provincial policies on access and utilization of drug therapies has found that variations exist even among drugs approved for public coverage.^23,36 In a study examining cancer pharmacotherapy utilization and costs across 6 Canadian provinces (including Saskatchewan and Ontario), more than 5-fold variation was observed in utilization rates of commonly used cancer drugs.³⁶ This variation is likely due to differences in eligibility requirements and coverage. American studies have demonstrated that drug policies requiring prior authorization may increase the rates of use of preferred antipsychotic agents, decrease the use of nonpreferred agents, and also increase risks of treatment discontinuity, highlighting both intended and unintended consequences of drug policy changes.^37,38 In British Columbia, the initiation of provincial coverage for ChEIs increased median persistence from 9.4 months in the period prior to 17.6 months during the period following coverage,³⁹ indicating a strong influence of provincial reimbursement policy on utilization. In Canada, coordinated national pharmaceutical policies under a national pharmacare program may help to reduce inequities in access to dementia pharmacotherapy and other prescription medications, to improve the appropriateness of prescribing, and to reduce costs including those associated with out of pocket expenditures.^40,41 However, much will depend on how policies are defined and implemented.

The inverse association between frailty and dementia pharmacotherapy in both provinces may reflect prescriber uncertainty regarding the usefulness of these medications in vulnerable LTC residents with limited life expectancies and possibly greater susceptibility to adverse events. ChEIs are now recommended for all stages of AD based on evidence from more recent clinical trials showing efficacy in persons with severe AD.² The association between severity of cognitive impairment and increased odds of receiving dementia pharmacotherapy in both provinces may indicate prescribing in residents with severe dementia, who may be less likely to benefit from these medications. Notably, ChEI and memantine deprescribing guidelines recommend that individuals at severe dementia be considered candidates for a trial of discontinuation.⁴² Some clinicians may prescribe dementia pharmacotherapy in the hope that they will treat behavioral and psychological symptoms of dementia and reduce the need for psychotropic medications.⁴³ The inverse association shown between benzodiazepine use and dementia pharmacotherapy may indicate that prescribers are avoiding benzodiazepines among persons clearly suffering from dementia due to concerns related to worsening of cognitive function.⁴⁴ The association between dementia pharmacotherapy and increased antipsychotic use was also found in a recent study of LTC residents with dementia⁴⁵ and may be due to prescribing to manage behavioral challenges.

The strong inverse association found between other drugs with anticholinergic properties and dementia pharmacotherapy use in Saskatchewan (and opposite direction of association in Ontario) indicates that policies requiring the discontinuation of anticholinergic medications prior to the initiation of a ChEI are generally being followed in that province, even for private paid and out-of-pocket drug use⁴⁶ though we note there is still some use of anticholinergic medications in this population. The use of anticholinergic medications in persons with dementia is discouraged due to potential adverse effects including impaired cognitive function, confusion, and delirium.^47–49 Anticholinergics have opposing effects to those of ChEIs, and their use raises questions among those receiving a ChEI.⁵⁰ Previous work has shown that use of ChEIs was associated with concurrent use of anticholinergic medications despite potential risks including worsening of cognitive impairment in persons with dementia.^26,51 A recent Australian study found that use of anticholinergic medications increased following initiation of dementia pharmacotherapy, with one-third of anticholinergic medications newly prescribed after dementia pharmacotherapy initiation.⁵² It is possible that anticholinergic medications are being prescribed to deal with adverse effects arising from the use of ChEIs.⁵³ Nevertheless, our findings point to the ability of reimbursement policies to impact prescribing patterns, though it is unclear whether these policies may reduce potential harms associated with dementia pharmacotherapy or its interactions with other medications.

Differing associations between rural residence and the receipt of dementia pharmacotherapy between the provinces may point to population differences as well as differences in health system access. More than one-third of the Saskatchewan population resides in rural areas, while fewer than one-fifth of Ontarians do so.⁵⁴ Rural residence was associated with an increased likelihood of receiving a ChEI among persons with incident dementia in Germany,⁵⁵ a finding similar to what we observed in Saskatchewan. Previous work examining the association between rural residence and ChEI discontinuation found no significant association in either Saskatchewan²¹ or Ontario.⁵ Nonclinical factors, including patient-, physician-, and practice-related factors, can have a strong influence on prescribing decisions.⁵⁶ Previous work has shown that physician sex is associated with ChEI prescribing patterns.⁵⁷ Other work has shown that physicians are influenced by family wishes when deciding whether to prescribe ChEIs.⁵⁸ Further work is required to understand the nonclinical factors that may influence prescribing of dementia pharmacotherapy and whether these vary by region.

The inverse association between use of dementia pharmacotherapy and year of admission to LTC may be due to several factors. It is possible that changes in the health and social care systems have resulted in persons with mild to moderate dementia living in the community longer. This would result in LTC facilities being reserved for those with more severe dementia who may have had dementia pharmacotherapy discontinued prior to admission. It is also possible that overall use of ChEIs among those admitted to LTC is decreasing over time. Additional work is needed to explore this finding.

Strengths of the current study include the large population-based sample of LTC residents in two Canadian provinces with comparable databases, measures, and analyses but differing drug reimbursement policies. This setting enabled the evaluation of predictors of dementia pharmacotherapy use under different policy environments while adjusting for health status confounders collected using standardized clinical assessments. In addition, we were able to explore the impact of including nonpublicly reimbursed prescribing of these agents on predictors of dementia pharmacotherapy use in Saskatchewan. Limitations include the lack of information on privately reimbursed prescription medications and use by persons <65 years in Ontario. As we were unable to capture memantine use in Ontario, we may have underestimated the use of dementia pharmacotherapy in the province. In addition, the small sample size of treated individuals in Saskatchewan limited our power to detect associations. Although the health administrative data algorithm used to identify persons with dementia in this study was validated in Ontario,²⁹ previous work using a similar algorithm in Saskatchewan generated prevalence and incidence estimates comparable to Ontario.⁵⁹ Our measure of severity of cognitive impairment was limited to the CPS as other measures of clinical dementia severity were not available. While the CPS may not completely capture the spectrum of dementia severity among residents, it has been shown to be moderately correlated with Mini-Mental State Examination scores.⁶⁰ As with all observational studies, there may have been unmeasured confounders that may account for some of the associations observed.

Conclusion and Implications

Consistent with the relative restrictiveness of Saskatchewan’s drug policies, the receipt of dementia pharmacotherapy among newly admitted LTC residents with dementia and/or cognitive impairment was 4-fold lower in Saskatchewan as compared with Ontario. We found that the two provinces generally demonstrated similar predictors of use, with some exceptions such as rural residence and recent anticholinergic prescriptions. Receipt of anticholinergic medications was associated with lower odds of dementia pharmacotherapy in Saskatchewan, where the drug policy specifies discontinuation of these medications prior to ChEI initiation. This study illustrates that reimbursement policies have the potential to influence both the frequency and the factors associated with medication use. Future studies should examine the relative impacts on access and resident outcomes of more versus less restrictive reimbursement policies for dementia pharmacotherapy. A national pharmacare policy may help alleviate disparities in access, but the implications for individuals already covered under provincial plans will differ depending on the restrictiveness of the dementia pharmacotherapy reimbursement policy adopted.

Authors’ Note

The data set from this study is held securely in coded form at ICES. While data sharing agreements prohibit ICES from making the data set publicly available, access may be granted to those who meet prespecified criteria for confidential access, available at www.ices.on.ca/DAS. The full data set creation plan and underlying analytic code are available from the authors upon request, understanding that the computer programs may rely upon coding templates or macros that are unique to ICES and are therefore either inaccessible or may require modification. For any supplemental information related to the study such as the study protocol, analysis plan, or analytic code, please contact author.