Potentially Inappropriate Medication,Anticholinergic Burden,and Mortality in People Attending Memory Clinics

Abstract

Background:

There is limited evidence regarding the association between potentially inappropriate medications (PIM) and mortality in older people with cognitive impairment.

Objective:

To examine whether use of medications considered to be potentially inappropriate in older people with cognitive impairment (PIMcog) and anticholinergic cognitive burden (ACB) were associated with mortality in people who attended memory clinics.

Methods:

Cross-sectional and longitudinal analyses of data from the Prospective Research In MEmory clinics (PRIME) study. Participants were community-dwelling people who attended nine memory clinics and had a diagnosis of mild cognitive impairment or dementia. PIMcog was defined as any medication considered potentially inappropriate for a person with cognitive impairment according to Beers or STOPP criteria. Anticholinergic burden was calculated using the ACB scale. Time-dependent Cox-proportional hazards regression was used to analyze associations between PIMcog use/ACB score and all-cause mortality over a three-year follow-up period. The regression model included the baseline variables: age, gender, education, cognitive diagnoses, total number of medications, disease-burden, cognition, physical function, and neuropsychiatric symptoms.

Results:

Of 964 participants, 360 (37.3%) used one or more PIMcog at some time during the study; most commonly anticholinergics and sedatives. 624 (64.7%) participants used a medication with potential or definite anticholinergic properties (ACB>0) at some point during the study. Both PIMcog use (adjusted hazard ratio: 1.42 95% CI: 1.12–1.80) and ACB score (adjusted hazard ratio: 1.18 95% CI: 1.06–1.32) were associated with mortality.

Conclusion:

Use of PIMcogs and medications with anticholinergic properties was common among memory clinic patients and both were associated with mortality.

Keywords

Alzheimer’s disease cholinergic antagonists cognitive dysfunction dementia inappropriate prescribing mortality potentially inappropriate medication list contstartabstract

INTRODUCTION

Older people are more susceptible to adverse drug events than younger adults due to factors including altered pharmacokinetics and pharmacodynamics, multiple comorbidities, polypharmacy, and drug interactions [1, 2]. Inappropriate medication use further increases the risk of adverse drug events including hospitalizations, morbidity, and mortality [1 , 3–9].

Cognitive impairment is common in older people, with around 5% of people aged ≥65 years having dementia [10] and a further 20% having mild cognitive impairment [11]. Older people with cognitive impairment have increased susceptibility to adverse central nervous system effects of medications, especially sedatives and anticholinergics [1, 2].

While there are tools for identifying potentially inappropriate medications (PIM) in people with advanced dementia [12], there are no specific tools for identifying PIM use in people with mild to moderate cognitive impairment. However, the Beers Criteria [13] and the Screening Tool of Older People’s Prescriptions (STOPP) [14] criteria list medications which should be avoided in older people with cognitive impairment.

Anticholinergic medications increase the risk of cognitive impairment, functional impairment, and mortality [15 –17]. There are a number of tools for measuring anticholinergic burden; however, there is no ‘gold standard’. The anticholinergic cognitive burden scale (ACB) [18, 19] is particularly useful for identifying medications that may have a negative impact on cognition [20]. Each one point increase in ACB score has been associated with a 26% increase in the risk of death in older people [15].

There is limited research regarding the association between PIM use and mortality in older people with cognitive impairment, and no data specifically in people attending memory clinics. Memory clinics in Australia are ambulatory assessment services for people with suspected memory and related cognitive disorders [21]. We have reported earlier that nearly one in four patients attending memory clinics use a PIM or have a clinically significant anticholinergic burden [22].

The aim of this study was to examine whether medications considered to be potentially inappropriate in older people with cognitive impairment (PIMcog) and anticholinergic burden were associated with mortality in older people who attended memory clinics.

METHODS

Design, setting, and participants

Participants were from the Prospective Research In MEmory clinics (PRIME) study (National Institute of Health Clinical Trials registry number: NCT00297271). PRIME was a three-year multi-center observational cohort study assessing the management of patients who attended nine memory clinics affiliated with secondary or tertiary care hospitals across four of the eight states/territories of Australia. Patients were recruited at their initial assessment visit or a subsequent follow-up appointment between April 2005 and July 2008, and then followed up by a research nurse and/or their specialist physician at 3, 6, 12, 24, and 36 months from baseline, or until their death or withdrawal from the study.

Patients were eligible for inclusion in the PRIME study if they had been diagnosed with dementia according to the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) criteria [23] or mild cognitive impairment (MCI) according to the Petersen criteria [24]. Participants had to be community dwelling with <40 h/week nursing care, be fluent in English, be able to provide written informed consent directly or through a legal guardian/proxy, and have a carer willing to provide consent. Further details of the PRIME study have been previously published [25].

Data collection and measures

Data were collected by clinic staff (physicians and nurses) and trained research nurses. Demographic and diagnostic data were collected at baseline. Medication data and clinical assessments were completed at baseline and at each follow-up appointment. Cognition was assessed using the Mini-Mental State Examination (MMSE) [26], functional ability was assessed using the Functional Autonomy Measurement System (SMAF) [27], and neuropsychiatric symptoms were assessed using the Neuropsychiatric Inventory (NPI) [28].

Medication use

Medication data were collected via patient and carer interviews conducted at the memory clinic. Secondary sources, including hospital/clinic records, patients’ medication packages, and medication lists, were examined if available. Current exposure to all medications, including prescription and non-prescription, regular and when required, was recorded. Medication names (brand and generic) were recorded, but no information regarding strength or dose was collected. Medications were coded using the Anatomical Therapeutic Chemical Classification System 2015 [29].

Medication-based disease burden index (MDBI)

Data regarding co-morbidities were not collected in the PRIME study. Therefore, an MDBI score was calculated for each participant based on their baseline medication regimen to give an estimation of their disease burden. The MDBI is a validated measure for quantifying disease burden using participant medication lists [30]. Chronic conditions, identified by the medication used to manage that condition, are scored based on their contribution to global deaths [30]. We excluded the score for ‘Alzheimer’s disease and other dementias’ from the total MDBI calculation given cognitive diagnoses were recorded separately.

PIMcog use

PIMcog was defined as any medication that was considered potentially inappropriate for use in older people with cognitive impairment according to the Beers 2012 [13] or STOPP 2014 criteria [14]. This included anticholinergics, sedatives, histamine-2 receptor antagonists, and systemic corticosteroids, which are associated with adverse central nervous system effects. Criteria unrelated to cognitive impairment were excluded. The PIMcog list consisted of 14 medication classes and 89 individual medications [22].

ACB score

The ACB scale assigns a score of zero for medications with no known anticholinergic activity, one for medication with possible anticholinergic properties (in vitro evidence of muscarinic receptor antagonism), two for medications with definite clinical anticholinergic properties, and three for medications with definite anticholinergic properties that may cause delirium [18, 31]. A score was given for each medication listed on the ACB 2012 scale [18, 32], and a total ACB score for each participant was calculated by adding the individual scores of different medications in a participant’s regimen.

Time-dependent measures of both PIMcog use and anticholinergic burden were defined using medication lists within one month of each follow up appointment. Participants were considered to be exposed to medications identified at each appointment until their next scheduled follow up, or until their date of death or date of withdrawal from the study. Medications that were started and ceased in between follow-up appointments were not known to the investigators.

Sensitivity analyses were conducted which excluded antipsychotics from the PIMcog list and ACB score as antipsychotic use has previously been associated with mortality [33].

Mortality

The primary outcome measure was all-cause mortality. Dates of death were recorded by memory clinic staff and research nurses, and confirmed using state registry records. Participant deaths that occurred within 90 days of completing or withdrawing from the study (determined by state registries search) were also included (hence total maximum duration of follow-up for the primary outcome was three years and 90 days).

Statistical analyses

Data analyses were conducted using the Statistical Package for Social Sciences (SPSS, version 22.0; IBM Corp. Armonk, USA).

Descriptive statistics were used to report the general demographics and features of medication use. Pearson χ ² test, Mann-Whitney U test and independent samples t-test were used, as appropriate, for exploratory bivariate analyses.

Cox proportional hazards regression analyses were conducted to quantify the association between baseline PIMcog use (yes or no) or ACB score and time to all-cause mortality with censoring at scheduled end of three-year follow-up or withdrawal/loss to follow-up. Both unadjusted and adjusted hazard ratios (HRs) were calculated. Variables associated with mortality or withdrawal from the study, as identified in exploratory analyses (p < 0.05), were adjusted for in the regression model. These variables included baseline age, gender, education, dementia/MCI diagnosis, total number of medications, MDBI score, MMSE, SMAF, and NPI score.

Time-dependent Cox proportional hazards regression analyses were used to determine the association between longitudinal PIMcog use (number of PIMcogs used at each time point) or ACB scores (at each time point) and mortality. From intraclass correlation coefficients (ICC), it was determined that between person variability was substantially higher than within-person variability in the medication exposure (i.e., ICC>0.80) and hence the time-dependent Cox models were appropriate [34, 35].

Results are presented as mean and standard deviation (SD), median and interquartile range (IQR), range, number and percentage and HR with 95% confidence interval. p-values less than 0.05 were considered to be statistically significant.

Ethical considerations

The PRIME study was approved by the research ethics committees of each participating institution. The analyses described in this manuscript received exemption from the Monash University Human Research Ethics Committee.

RESULTS

Participant baseline characteristics

There were 964 participants included at baseline with mean (SD) age 77.6 (7.4) years (Table 1). The cognitive diagnoses of participants included MCI (n = 185, 19.2%), Alzheimer’s disease (n = 521, 54.0%), vascular dementia (n = 51, 5.3%), dementia with Lewy bodies (n = 16, 1.7%), frontotemporal dementia (n = 31, 3.2%), mixed Alzheimer’s and vascular dementia (n = 129, 13.4%), and other dementia (n = 31, 3.2%).

Table 1

Characteristics of participants

					Withdrawn or
	Overall	Completed	Deceased		lost to follow up
Characteristic	(n = 964)	(n = 508)	(n = 146)	p-value^*	(n = 310)	p-value^∧
Age (years), mean (SD)	77.6 (7.4)	76.1 (7.7)	80.3 (6.4)	<0.001	78.7 (6.9)	<0.001
Sex (female), n (%)	456 (47.3)	244 (48.0)	54 (37.0)	0.023	158 (51.0)	0.458
Weight (kg), mean (SD)	71.1 (13.71)	71.0 (13.7)	72.0 (13.4)	0.485	70.8 (13.9)	0.861
Living alone, n (%)	121 (12.6)	58 (11.4)	18 (12.3)	0.876	45 (14.5)	0.235
Education (less than secondary level), n (%)	121 (12.6)	43 (8.5)	23 (15.8)	0.015	55 (17.7)	<0.001
Family history of dementia, n (%)	346 (35.9)	185 (36.4)	54 (37.0)	0.977	107 (34.4)	0.635
Cognitive diagnosis, n (%)
- MCI	185 (19.2)	124 (24.4)	10 (6.8)	<0.001	51 (16.5)	0.009
- Dementia	779 (80.8)	384 (75.6)	136 (93.2)		259 (83.5)
Total number of medications, median (IQR)	6 (4–8)	6 (4–8)	7 (5–10)	<0.001	6 (3–8)	0.906
Baseline disease burden (MDBI score), median (IQR)	0 (0–0.21)	0 (0–0.11)	0 (0–0.35)	<0.001	0 (0–0.21)	0.032
Baseline cognition (MMSE Score), median (IQR)	24 (20–27)	25 (21–27)	22 (16–25)	<0.001	23 (18–26)	<0.001
Baseline function (SMAF score), median (IQR)	–14	–11	–20.5	<0.001	–18	<0.001
	(–23 to –7)	(–19 to –4.5)	(–31.8 to –12.5)		(–27.3 to –9.5)
Baseline neuropsychiatric (NPI score), median (IQR)	7.5 (2–18)	6 (2–15)	9 (3–20)	0.004	10 (3–23)	<0.001
Baseline PIMcog use (yes), n (%)	206 (21.4)	100 (19.7)	45 (30.8)	0.006	61 (19.7)	1.000
PIMcog use at some point (yes), n (%)	360 (37.3)	182 (35.8)	80 (54.8)	<0.001	98 (31.6)	0.248
Baseline ACB score, median (IQR)	0 (0-1)	0 (0-1)	1 (0–2)	<0.001	0 (0-1)	0.110
ACB medication use at some point (yes), n (%)	624 (64.7)	318 (62.6)	120 (82.2)	<0.001	186 (60.0)	0.505

Bivariate p-values based on Pearson χ ², Mann-Whitney U test and independent samples t-test as appropriate. ^* p-values indicate difference between participants who died during the study or within 90 days of completing or withdrawing from the study (n = 146) versus those who completed the study (n = 508). ^∧ p-values indicate differences between participants who withdrew or were lost to follow up (n = 310) versus those who completed the study (n = 508). ACB, Anticholinergic Cognitive Burden; IQR, interquartile range; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; NPI, Neuropsychiatric Inventory; PIMcog, potentially inappropriate medication for a person with cognitive impairment; SMAF, Functional Autonomy Measurement System.

At baseline the median (IQR) MMSE score was 24/30 (20 to 27) (n = 957, missing data from seven participants), the median (IQR) SMAF score was –14 (–23 to –7, n = 905, missing data from 59 participants), and the median (IQR) NPI score was 7.5 (2 to 18, n = 826, missing data from 138 participants). The median (IQR) number of medications was 6 (4 to 8), range 0 to 20. The median (IQR) MDBI score was 0 (0 to 0.21), range 0 to 1.74.

Participant follow up

Of the 964 participants, 310 (32.2%) did not complete the study. Reasons included withdrawal of carer/proxy consent (n = 114), lost to follow up (n = 75), self-withdrawal (n = 55), admission to nursing home or high level care and unable to participate in follow up (n = 35), carer illness/death (n = 14), moved interstate/overseas (n = 10), participant illness (n = 4) and discharged from the memory clinic (n = 3).

One hundred and forty-six (15.1%) participants died during the study or within 90 days from the date of completion or withdrawal from the study. The mean (SD) length of participation in the study was 2.37 (0.93) years.

Demographic and baseline clinical features of the participants who completed, died, and did not complete the study are summarized in Table 1. Participants who did not complete the study were older, had a lower level of education, were more likely to be diagnosed with dementia, and had a higher disease burden, lower cognitive ability, lower functional ability, and more severe neuropsychiatric symptoms at baseline than those who completed the study. Participants who did not complete the study had similar PIMcog use and ACB scores compared to those who completed.

PIMcog use and mortality

Three hundred and sixty (37.3%) participants used at least one PIMcog (range 0 to 7) at some point over the three-year study period. Two hundred and six (22.4%) participants were using a PIMcog at baseline, and 211 (21.9%) had a PIMcog initiated after recruitment. The most common PIMcogs were anticholinergics and sedatives (Table 2).

Table 2

Potentially inappropriate medications for people with cognitive impairment (PIMcog) by medication class

PIMcog Class	Baseline^∧	Initiated	Total
		subsequently
Anticholinergics	104	120	224
Antipsychotics (with anticholinergic properties)	27	42	69
Tricyclic antidepressants	27	11	38
Antimuscarinics (urinary)	20	21	41
Antispasmodics	13	32	45
First generation antihistamines	7	6	13
Other antidepressants (with anticholinergic properties)	7	1	8
Antiparkinson agents	3	2	5
Other antihistamines (with anticholinergic properties)	0	5	5
Skeletal muscle relaxants	0	0	0
Sedatives and hypnotics	103	158	261
Benzodiazepines (BZD)	95	157	252
Non-BZD hypnotics	7	1	8
Barbiturates	1	0	1
Others medications that can induce delirium	50	43	93
Systemic corticosteroids	26	29	55
Histamine-₂ receptor antagonists	24	14	38
Total	257	321	578
Total participants ^*	206	211	360

^∧Details of baseline PIMcog use have been published previously [22]. ^*Total after removal of participants who took multiple PIMcogs.

Baseline PIMcog use was associated with mortality in the unadjusted analysis (unadjusted HR: 1.71, 95% CI: 1.20–2.43) but was not statistically significant in the adjusted analysis (Model 1, Table 3).

Table 3

Unadjusted and adjusted hazard ratios for the associations of PIMcog use with mortality over three years of follow up

Characteristic		Model 1	Model 2^∧
	Unadjusted HR	Adjusted HR	Adjusted HR
	(95% CI)	(95% CI)	(95% CI)
Age (HR per year)	1.07 (1.05–1.10)	1.05 (1.02–1.08)	1.05 (1.02–1.08)
Sex (female)	0.66 (0.47–0.92)	0.55 (0.38–0.81)	0.54 (0.37–0.79)
Education (less than secondary level)	1.57 (1.01–2.46)	1.05 (0.64–1.70)	0.99 (0.61–1.62)
Dementia diagnosis (not MCI)	3.83 (2.01–7.28)	2.18 (1.03–4.64)	2.16 (1.02–4.59)
Total number of medications (HR per medication)	1.08 (1.04–1.13)	1.01 (0.95–1.07)	0.99 (0.94–1.05)
Baseline disease burden (HR per 1 unit MDBI score)	4.16 (2.55–6.81)	2.08 (1.11–3.91)	2.27 (1.21–4.26)
Baseline cognition (HR per 1 unit MMSE score)	0.92 (0.90–0.95)	0.96 (0.92–0.99)	0.95 (0.92–0.99)
Baseline function (HR per 1 unit SMAF score)	0.95 (0.94–0.96)	0.97 (0.95–0.99)	0.97 (0.95–0.99)
Baseline neuropsychiatric symptoms (HR per 1 unit NPI score)	1.01 (1.00–1.02)	1.00 (0.98–1.01)	1.00 (0.98–1.01)
Baseline PIMcog use (yes)	1.71 (1.20–2.43)	1.26 (0.83–1.91)
PIMcog use over study period (HR per 1 unit PIMcog)^∧	1.52 (1.26–1.83)		1.42 (1.12–1.80)

^∧Time-dependent Cox-proportional hazards regression. CI, confidence interval; HR, hazard ratio; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; NPI, Neuropsychiatric Inventory; PIMcog, potentially inappropriate medication for a person with cognitive impairment; SMAF, Functional Autonomy Measurement System.

Time-dependent PIMcog use was associated with mortality in both the unadjusted (HR: 1.52, 95%, CI: 1.26–1.83) and adjusted (HR: 1.42, 95%, CI: 1.12–1.80) analyses (Model 2, Table 3).

Sensitivity analyses, which excluded antipsychotics from the PIMcog list, showed similar results to the full PIMcog list (baseline PIMcog use adjusted HR: 1.33, 95% CI: 0.87–2.03 and longitudinal time-dependent PIMcog use adjusted HR: 1.46, 95% CI: 1.14–1.87).

ACB score and mortality

According to the ACB scale, 624 participants (64.7%) took a medication with potential or definite anticholinergic properties at some point during the study, and 431 (44.7%) had a baseline ACB score greater than zero. The most common ACB medications were risperidone (n = 166), frusemide (n = 138), warfarin (n = 97), atenolol (n = 95), and digoxin (n = 81), which all have an ACB score of one. The most common medications with an ACB score of three were olanzapine (n = 30), oxybutynin (n = 27), quetiapine (n = 19), and amitriptyline (n = 17).

Baseline total ACB score was associated with mortality in the unadjusted (HR: 1.27, 95% CI: 1.15–1.40) and adjusted (HR: 1.15, 95% CI: 1.01–1.31) analyses (Model 1, Table 4).

Table 4

Unadjusted and adjusted hazard ratios for the association of anticholinergic use with mortality over three years follow up

Characteristic		Model 1	Model 2^∧
	Unadjusted HR	Adjusted HR	Adjusted HR
	(95% CI)	(95% CI)	(95% CI)
Age (HR per year)	1.07 (1.05–1.10)	1.05 (1.02–1.08)	1.05 (1.02–1.08)
Sex (female)	0.66 (0.47–0.92)	0.56 (0.38–0.82)	0.57 (0.39–0.83)
Education (less than secondary level)	1.57 (1.01–2.46)	1.08 (0.66–1.76)	1.02 (0.62–1.67)
Dementia diagnosis (not MCI)	3.83 (2.01–7.28)	2.09 (0.98–4.46)	2.08 (0.98–4.42)
Total number of medications (HR per medication)	1.08 (1.04–1.13)	0.99 (0.94–1.05)	1.00 (0.94–1.05)
Baseline disease burden (HR per 1 unit MDBI score)	4.16 (2.55–6.81)	2.05 (1.10–3.84)	1.99 (1.06–3.74)
Baseline cognition (HR per 1 unit MMSE Score)	0.92 (0.90–0.95)	0.96 (0.92–0.99)	0.96 (0.92–0.99)
Baseline function (HR per 1 unit SMAF score)	0.95 (0.94–0.96)	0.97 (0.95–0.99)	0.97 (0.95–0.99)
Baseline neuropsychiatric symptoms (HR per 1 unit NPI score)	1.01 (1.00–1.02)	1.00 (0.98–1.01)	1.00 (0.99–1.01)
Baseline ACB score (HR per 1 unit ACB score)	1.27 (1.15–1.40)	1.15 (1.01–1.31)
ACB scores over study period (HR per 1 unit ACB score)^∧	1.29 (1.19–1.39)		1.18 (1.06–1.32)

^∧Time-dependent Cox-proportional hazards regression. ACB, Anticholinergic Cognitive Burden; CI, confidence interval; HR, hazard ratio; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; NPI, Neuropsychiatric Inventory; SMAF, Functional Autonomy Measurement System.

Time-dependent ACB scores were also associated with mortality in both the unadjusted (HR: 1.29, 95% CI: 1.19–1.39) and adjusted (HR: 1.18, 95% CI: 1.06–1.32) analyses (Model 2, Table 4).

Sensitivity analyses, which excluded antipsychotics from the ACB score, showed similar results (baseline ACB score adjusted HR: 1.18, 95% CI: 1.02–1.36 and longitudinal time-dependent PIMcog use adjusted HR: 1.16, 95% CI: 1.02–1.31).

DISCUSSION

More than one-third of patients attending the memory clinics in our study used a PIMcog and almost two-thirds were exposed to potential or definite anticholinergic cognitive burden over a three-year period. Explicit prescribing tools such as the Beers and STOPP criteria, and the ACB score, do not take individual patients’ clinical picture or circumstances into consideration, so some of the usage of PIMcogs and ACB medications in our study population may have been clinically appropriate for individual patients. Nevertheless, their usage was high, and they were initiated both prior to the study and throughout the three years of follow up despite the patients’ diagnoses of MCI or dementia. This finding and the fact that their usage was associated with increased risk of mortality highlight a need to raise awareness among prescribers and pharmacists about the risks associated with these agents in people with cognitive impairment.

Baseline PIMcog use was not significantly associated with mortality over a three-year horizon when adjusting for other variables. However, longitudinal time-dependent PIMcog use was associated with a 40% increased risk of mortality. This may suggest that knowledge of longitudinal exposure to PIMcogs allows for better prediction of mortality than exposure at a single time point. The time-dependent exposure may also have demonstrated a stronger association with mortality due to its proximity to the occurrence of the outcome compared to baseline exposure.

A number of large population studies have shown that PIM use is associated with adverse drug events, hospitalizations, emergency department visits, and mortality in general populations of older people [7, 36]. A recent study found the risk of death in a community-based population of older adults was 44% higher in users of at least one PIM (measured using the 2012 Beers Criteria) [37]. However, there has been limited research regarding the association between PIM or PIMcog use and mortality in people with cognitive impairment. One study found a dose-response relationship between anticholinergic and sedative drug burden and mortality in people with Alzheimer’s disease [5]. Other studies, using the Swedish National Board of Health and Welfare indicators [7] and the 1997 Beers criteria [38] to define PIM use, found no significant association with mortality in people with dementia. Further research is needed to confirm the long-term effects of PIMcog use on mortality, and to test the potential benefits of deprescribing PIMcogs in this population.

In our study, higher baseline and longitudinal exposure to anticholinergic medications were both associated with mortality. In the time-dependent model, each one point increase in anticholinergic burden was associated with an 18% increase in risk of mortality for people attending memory clinics. This suggests that prescribers need to be cautious not only with the prescribing of individual anticholinergics, but also the potential cumulative anticholinergic burden of patients’ entire regimens. These results are consistent with past research using the ACB scale in the general older population, where the odds of dying increased by 26% for every additional point scored on the ACB [15]. Other measures of anticholinergic burden have shown inconsistent associations with mortality [17 , 39–41].

PIMcogs, particularly anticholinergic medications, are commonly prescribed for older people with cognitive impairment across many clinical settings [42]. Increasing polypharmacy, and recognition of potentially inappropriate prescribing, has led to a growing focus on deprescribing, the process of tapering or stopping medications, to minimize polypharmacy and improve patient outcomes [43]. Prescriber barriers to deprescribing include deficits in knowledge and skill (e.g., assessing benefits and harms of therapy, and recognizing adverse drug effects), lack of time, therapeutic inertia, fear of negative consequences, reluctance to stop medications prescribed by other physicians and patient resistance to change [44]. Cognitive impairment, and involvement of carers, also creates barriers to deprescribing as it can be challenging to balance carer and patient preferences, and carers may have difficulties being surrogate decision makers [45]. More physicians providing care and lack of access to a complete medication history are further barriers [46, 47]. The long-term effects of deprescribing on clinical, economical, and quality of life outcomes are yet to be established in large prospective studies.

In general, studies suggest that interdisciplinary care and improved communication between health professionals can improve appropriateness of prescribing [48, 49]. A recent systematic review suggested that clinical medication review is beneficial in improving the quality use of medications [50]. A small Australian study reported that placing a pharmacist in a specialist memory clinic improved the accuracy of medication histories and identification of medication-related problems [47]. Further research is needed to evaluate the long-term clinical impact of interventions to improve appropriateness of medication use in this population.

Our study had strengths and limitations. Strengths included the large sample size with a wide geographical distribution, longitudinal study design with six time points of data collection over three years, inclusion of participants from nine memory clinics with a range of cognitive diagnoses and mortality data confirmed by state registry records. Our study also included sensitivity analyses to show that the association of PIMcogs and anticholinergic burden with mortality was independent of antipsychotic use. Our PIMcog list was a composite of both the Beers and STOPP criteria, which enhanced the sensitivity in detecting adverse drug events [36].

A limitation of our study was that lack of information on medication doses prevented us from using dose-specific measures of PIM burden such as the Drug Burden Index [51]. A dose-specific measure could have differentiated the effect of high and low doses, and regular and when required (prn) use of medications. Limited information regarding patients’ co-morbidities (other than dementia) may have resulted in residual confounding that was not accounted for within the medication-based disease burden index. Residual confounding may also have been present due to analyses only adjusting for baseline covariates. Participant withdrawal or loss to follow up in the PRIME study was 32% at three years. Similar rates of attrition have been seen in other cohort studies of patients with dementia due to both patient and caregiver factors [52]. Although we adjusted for relevant covariates and censored all participants at their time of drop out from the study, participant withdrawal may still have affected the results. Finally, recruitment from specialist memory clinics, exclusion of people living in residential care or receiving 40 or more hours per week of nursing care, and the lower than expected proportion of women in our study may limit the extent to which the findings apply to the wider population of people with cognitive impairment.

Conclusion

More than one-third of patients attending Australian memory clinics used a PIMcog and almost two-thirds were exposed to anticholinergic medications over a three-year period. Both longitudinal PIMcog use and anticholinergic burden were independent predictors of mortality. Further research on interventions for improving the appropriateness of prescribing for people with cognitive impairment and their effect on cognitive and health outcomes are needed.

Footnotes

ACKNOWLEDGMENTS

We thank all the Australian investigators, study nurses, staff and hospitals who comprise the PRIME study group: Prince of Wales Hospital (Marika Donkin, Kim Burns, Katrin Seeher); The Queen Elizabeth Hospital (Shelley Casey, Trish Steventon); St George’s Hospital (Maree Mastwyk, Alissa Westphal, Nicola Lautenschlager, Olga Yastrubetskaya, Marilyn Kemp, Edmond Chiu and Jenifer Ames); Austin Health Repatriation Hospital (Irene Tan, Henry Zeimer, Leonie Johnston); Hornsby Ku Ring Gai Hospital (Sue Kurrle, Roseanne Hogarth, Judith Allan); Fremantle Hospital (Roger Clarnette, Janice Guy, Denae Clark); The Prince Charles Hospital (Chris Davis, Mary Wyatt, Katrina Brosnan, Margaret Morton); Rankin Park Hospital (John Ward, Jeanette Gatgens); Geelong Private Hospital (Alastair Mander, Bernadine Charles).

Data collection was funded by Janssen-Cilag Pty Limited. Janssen-Cilag had no input into the design, analysis, interpretation, or writing of this study.

Amanda Cross was supported by an Australian Government Research Training Program (RTP) scholarship.

Authors’ disclosures available online ().

References

Lau

, Mercaldo

, Shega

, Rademaker

, Weintraub

(2011) Functional decline associated with polypharmacy and potentially inappropriate medications in community-dwelling older adults with dementia. Am J Alzheimers Dis Other Demen 26, 606–615.

Field

, Gilman

, Subramanian

, Fuller

, Bates

, Gurwitz

(2005) The costs associated with adverse drug events among older adults in the ambulatory setting. Med Care 43, 1171–1176.

Akazawa

, Imai

, Igarashi

, Tsutani

(2010) Potentially inappropriate medication use in elderly Japanese patients. Am J Geriatr Pharmacother 8, 146–160.

Fick

, Mion

, Beers

, Waller

(2008) Health outcomes associated with potentially inappropriate medication use in older adults. Res Nurs Health 31, 42–51.

Gnjidic

, Hilmer

, Hartikainen

, Tolppanen

, Taipale

, Koponen

, Bell

(2014) Impact of high risk drug use on hospitalization and mortality in older people with and without Alzheimer’s disease: A national population cohort study. PLoS One 9, e83224.

Gosch

, Wortz

, Nicholas

, Doshi

, Kammerlander

, Lechleitner

(2014) Inappropriate prescribing as a predictor for long-term mortality after hip fracture. Gerontology 60, 114–122.

Skoldunger

, Fastbom

, Wimo

, Fratiglioni

, Johnell

(2015) Impact of inappropriate drug use on hospitalizations, mortality, and costs in older persons and persons with dementia: Findings from the SNAC study. Drugs Aging 32, 671–678.

Lau

, Kasper

, Potter

, Lyles

, Bennett

(2005) Hospitalization and death associated with potentially inappropriate medication prescriptions among elderly nursing home residents. Arch Intern Med 165, 68–74.

Dedhiya

, Hancock

, Craig

, Doebbeling

, Thomas

3rd (2010) Incident use and outcomes associated with potentially inappropriate medication use in older adults. Am J Geriatr Pharmacother 8, 562–570.

10.

Prince

, Bryce

, Albanese

, Wimo

, Ribeiro

, Ferri

(2013) The global prevalence of dementia: A systematic review andmetaanalysis. Alzheimers Dement 9, 63–75.e62.

11.

Ward

, Arrighi

, Michels

, Cedarbaum

(2012) Mild cognitive impairment: Disparity of incidence and prevalence estimates. Alzheimers Dement 8, 14–21.

12.

Disalvo

, Luckett

, Agar

, Bennett

, Davidson

(2016) Systems to identify potentially inappropriate prescribing in people with advanced dementia: A systematic review. BMC Geriatr 16, 114.

13.

(2012) American Geriatrics Society updated Beers Criteria for potentially inappropriate medication use in older adults. J Am Geriatr Soc 60, 616–631.

14.

O’ Mahony

, O’ Sullivan

, Byrne

, O’Connor

, Ryan

, Gallagher

(2015) STOPP/START criteria for potentially inappropriate prescribing in older people: Version 2. Age Ageing 44, 213–218.

15.

Fox

, Richardson

, Maidment

, Savva

, Matthews

, Smithard

, Coulton

, Katona

, Boustani

, Brayne

(2011) Anticholinergic medication use and cognitive impairment in the older population: The medical research council cognitive function and ageing study. J Am Geriatr Soc 59, 1477–1483.

16.

Sittironnarit

, Ames

, Bush

, Faux

, Flicker

, Foster

, Hilmer

, Lautenschlager

, Maruff

, Masters

, Martins

, Rowe

, Szoeke

, Ellis

(2011) Effects of anticholinergic drugs on cognitive function in older Australians: Results from the AIBL study. Dement Geriatr Cogn Disord 31, 173–178.

17.

Ruxton

, Woodman

, Mangoni

(2015) Drugs with anticholinergic effects and cognitive impairment, falls and all-cause mortality in older adults: A systematic review and meta-analysis. Br J Clin Pharmacol 80, 209–220.

18.

Boustani

, Campbell

, Munger

, Maidment

, Fox

(2008) Impact of anticholinergics on the aging bra, A review andractical application. Aging Health 4, 311–320.

19.

Salahudeen

, Hilmer

, Nishtala

(2015) Comparison of anticholinergic risk scales and associations with adverse health outcomes in older people. J Am Geriatr Soc 63, 85–90.

20.

Pasina

, Djade

, Lucca

, Nobili

, Tettamanti

, Franchi

, Salerno

, Corrao

, Marengoni

, Iorio

, Marcucci

, Violi

, Mannucci

(2013) Association of anticholinergic burden with cognitive and functional status in a cohort of hospitalized elderly: Comparison of the anticholinergic cognitive burden scale and anticholinergic risk scale: Results from the REPOSI study. Drugs Aging 30, 103–112.

21.

Woodward

, Woodward

(2009) A national survey of memory clinics in Australia. Int Psychogeriatr 21, 696–702.

22.

Cross

, George

, Woodward

, Ames

, Brodaty

, Ilomaki

, Elliott

(2016) Potentially inappropriate medications and anticholinergic burden in older people attending memory clinics in Australia. Drugs Aging 33, 37–44.

23.

American Psychiatric Association (2000) American Psychiatric Association Diagnostic and Statistical Manual of Mental Disorders, American Psyciatric Association, Washington DC.

24.

Winblad

, Palmer

, Kivipelto

, Jelic

, Fratiglioni

, Wahlund

, Nordberg

, Backman

, Albert

, Almkvist

, Arai

, Basun

, Blennow

, de Leon

, DeCarli

, Erkinjuntti

, Giacobini

, Graff

, Hardy

, Jack

, Jorm

, Ritchie

, van Duijn

, Visser

, Petersen

(2004) Mild cognitive impairment-beyond controversies, towards a consensus: Report of the International Working Group on Mild Cognitive Impairment. J Intern Med 256, 240–246.

25.

Brodaty

, Woodward

, Boundy

, Ames

, Balshaw

(2011) Patients in Australian memory clinics: Baseline characteristics and predictors of decline at six months. Int Psychogeriatr 23, 1086–1096.

26.

Folstein

, Folstein

, McHugh

(1975) “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

27.

Hebert

, Carrier

, Bilodeau

(1988) The Functional Autonomy Measurement System (SMAF): Description and validation of an instrument for the measurement of handicaps. Age Ageing 17, 293–302.

28.

Cummings

(1997) The Neuropsychiatric Inventory: Assessing psychopathology in dementia patients. Neurology 48, S10–S16.

29.

World Health Organization Collaborating Centre for Drug Statistics Methodology, The anatomical therapeutic chemical classifications system with defined daily doses (ATC/DDD) index., WHO, http://www.whocc.no/atc_ddd_index/, Accessed March.

30.

George

, Vuong

, Bailey

, Kong

, Marriott

, Stewart

(2006) Development and validation of the medication-based disease burden index. Ann Pharmacother 40, 645–650.

31.

Campbell

, Boustani

, Limbil

, Ott

, Fox

, Maidment

, Schubert

, Munger

, Fick

, Miller

, Gulati

(2009) The cognitive impact of anticholinergics: A clinical review. Clin Interv Aging 4, 225–233.

32.

Campbell

, Maidment

, Fox

, Khan

, Boustani

(2013) The 2012 update to the anticholinergic cognitive burden scale. J Am Geriatr Soc 61(Suppl.1), S142–S143.

33.

Maust

, Kim

, Seyfried

, Chiang

, Kavanagh

, Schneider

, Kales

(2015) Antipsychotics, other psychotropics, and the riskof death in patients with dementia: Number needed to harm. JAMA Psychiatry 72, 438–445.

34.

Kleinbaum

, Klein

(2012) Extension of the Cox proportional hazards model for time-dependent variables. In Survival Analysis: A Self-Learning Text, Third Edition. Springer New York, New York, pp. 241–288.

35.

Fisher

, Lin

(1999) Time-dependent covariates in the Cox proportional-hazards regression model. Annu Rev Public Health 20, 145–157.

36.

Brown

, Hutchison

, Li

, Painter

, Martin

(2016) Predictive validity of the Beers and Screening Tool of Older Persons’ Potentially Inappropriate Prescriptions (STOPP) criteria to detect adverse drug events, hospitalizations, and emergency department visits in the United States. J Am Geriatr Soc 64, 22–30.

37.

do Nascimento

, Mambrini

, Lima-Costa

, Firmo

, Peixoto

, de Loyola Filho

(2017) Potentially inappropriate medications: Predictor for mortality in a cohort of community-dwelling older adults. Eur J Clin Pharmacol 73, 615–621.

38.

Raivio

, Laurila

, Strandberg

, Tilvis

, Pitkala

(2006) Use of inappropriate medications and their prognostic significance among in-hospital and nursing home patients with and without dementia in Finland. Drugs Aging 23, 333–343.

39.

Lowry

, Woodman

, Soiza

, Hilmer

, Mangoni

(2012) Drug burden index, physical function, and adverse outcomes in older hospitalized patients. J Clin Pharmacol 52, 1584–1591.

40.

Lowry

, Woodman

, Soiza

, Mangoni

(2011) Associations between the anticholinergic risk scale score and physical function: Potential implications for adverse outcomes in older hospitalized patients. J Am Med Dir Assoc 12, 565–572.

41.

Mangoni

, van Munster

, Woodman

, de Rooij

(2013) Measures of anticholinergic drug exposure, serum anticholinergic activity, and all-cause postdischarge mortality in older hospitalized patients with hip fractures. Am J Geriatr Psychiatry 21, 785–793.

42.

Johnell

(2015) Inappropriate drug use in people with cognitive impairment and dementia: A systematic review. Curr Clin Pharmacol 10, 178–184.

43.

Scott

, Hilmer

, Reeve

, Potter

, Le Couteur

, Rigby

, Gnjidic

, Del Mar

, Roughead

, Page

, Jansen

, Martin

(2015) Reducing inappropriate polypharmacy: The process of deprescribing. JAMA Intern Med 175, 827–834.

44.

Anderson

, Stowasser

, Freeman

, Scott

(2014) Prescriber barriers and enablers to minimising potentially inappropriate medications in adults: A systematic review and thematic synthesis. BMJ Open 4, e006544.

45.

Reeve

, Bell

, Hilmer

(2015) Barriers to optimising prescribing and deprescribing in older adults with dementia: A narrative review. Curr Clin Pharmacol 10, 168–177.

46.

Reppas-Rindlisbacher

, Fischer

, Fung

, Gill

, Seitz

, Tannenbaum

, Austin

, Rochon

(2016) Anticholinergic drug burden in persons with dementia taking a cholinesterase inhibitor: The effect of multiple physicians. J Am Geriatr Soc 64, 492–500.

47.

Elliott

, Woodward

(2011) Medication-related problems in patients referred to aged care and memory clinics at a tertiary care hospital. Aust J Ageing 30, 124–129.

48.

Rojas-Fernandez

, Patel

, Lee

(2014) An interdisciplinarymemory clinic: A novel practice setting for pharmacists in primarycare. Ann Pharmacother 48, 785–795.

49.

Bregnhoj

, Thirstrup

, Kristensen

, Bjerrum

, Sonne

(2009) Combined intervention programme reduces inappropriate prescribing in elderly patients exposed to polypharmacy in primary care. Eur J Clin Pharmacol 65, 199–207.

50.

Jokanovic

, Tan

, van den Bosch

, Kirkpatrick

, Dooley

, Bell

(2016) Clinical medication review in Australia: A systematic review. Res Social Adm Pharm 12, 384–418.

51.

Hilmer

, Mager

, Simonsick

, Cao

, Ling

, Windham

, Harris

, Hanlon

, Rubin

, Shorr

, Bauer

, Abernethy

(2007) A drug burden index to define the functional burden of medications in older people. Arch Intern Med 167, 781–787.

52.

Coley

, Gardette

, Toulza

, Gillette-Guyonnet

, Cantet

, Nourhashemi

, Andrieu

, Grand

, Vellas

(2008) Predictive factors of attrition in a cohort of Alzheimer disease patients. The REAL.FR study. Neuroepidemiology 31, 69–79.