Long‐term anticholinergic, benzodiazepine and Z‐drug use in community‐dwelling older adults: What is the impact on cognitive and neuropsychological performance?

Long‐term use of anticholinergics, benzodiazepines and related drugs (or “Z‐drugs”) have been associated with cognitive impairment and dementia. However, the relationship of these medications with cognitive function and domain‐specific neuropsychological performance in older adults without dementia, is unclear.


| INTRODUCTION
The use of medications with anticholinergic properties have been associated with the later development of dementia in older adults. [1][2][3][4][5] Medications with such anticholinergic properties, either as an intended/potent side effect, are used to treat a wide range of urological, psychiatric and cardiovascular conditions and are used by significant proportions of older adults. 6,7 Several longitudinal studies have demonstrated a significantly increased risk of incident Mild Cognitive Impairment (MCI) and dementia in those regularly using these medications. 2,8 Less well explored is the specific relationship between these medications and detailed cognitive and neuropsychological performance in community-dwelling older adults, free from a diagnosis of dementia.
One of the most convincing longitudinal studies on anticholinergic medication use and incident dementia, involving just-under 300,000 individuals, demonstrated significant associations between the use of several types of strong anticholinergic drugs (those with an anticholinergic cognitive burden score of 2 or 3) and the later risk of dementia. 9 This study follows earlier studies that demonstrated an effect of strong anticholinergic medication use on cognitive impairment and dementia diagnosis, in addition to accelerated cognitive decline in those with established dementia using anticholinergic medications. 10 Comparatively fewer studies have focused on specific domains of cognitive function and the studies that do exist have typically focused on a relatively small number of general cognitive tests. 11,12 The impact of anticholinergic medication use on specific domains of detailed neuropsychological and cognitive function has been less well evaluated.
Regular benzodiazepine use has also been linked to an increased risk of cognitive impairment. Despite recommendations against their use in older adults, benzodiazepines and related medications (BDZRs: Bendoziazepine and Related Medications) are commonly used as anxiolytics, sedatives, hypnotics and anticonvulsants. 13 BDZRs consist of benzodiazepine and non-benzodiazepine sedative hypnotics (also termed "Z-drugs"). Several published studies have examined the impact of BDZRs on dementia diagnosis but, whilst a number of studies support an association, other studies do not. [14][15][16][17] Some studies examining BDZRs have demonstrated an association with poorer cognitive function and/or with the rate of cognitive decline. 18,19 Again, the evidence is somewhat limited in the range of cognitive tests used and studies examining the association between these medication classes and detailed tests of neuropsychological function are lacking.
The Trinity-Ulster-Department of Agriculture (TUDA) study enrolled over five-thousand adults of >60 years, and free from a diagnosis of dementia, from two separate jurisdictions on the island of Ireland. The purpose of the current study was to explore the TUDA data to investigate the relationship of anticholinergic and BDZR medication use with neuropsychological performance in this cohort, based on global cognition and detailed assessments of specific cognitive domains.

| Medication use
Participants were asked to provide a list of current medications and information about duration of use. Medications were coded using the Anatomic Therapeutic Classification (ATC) System. For the current study, we only included medications which were taken daily (not on an "as needed"/"prn" basis) and had been used continuously for at least six months at the time of assessment.
In order to characterise anticholinergic medication use, we used the Anticholinergic Cognitive Burden (ACB) score. 23 The ACB score assigns a score of 1 (potential anticholinergic medication), two or three (definite evidence of clinically significant anticholinergic effect) and the total score for each participant was calculated by combining an arithmetic sum of the potential/definite anticholinergic medications used. Further, a binary variable for potential and definite anticholinergic medications were used in order to assess the association of these medication classes individually with cognitive impairment.
Use of benzodiazepines and related drugs (BDZR) was identified using the ATC codes: N03AE, N05BA, N05CD and N05CF relating to benzodiazepine drugs and benzodiazepine-related sedative hypnotics ("Z-Drugs").

| Cognitive assessment
The Mini-Mental State Examination (MMSE) was used as an assessment of general cognitive function in the current study. 26 To assess frontal lobe function/executive function, the Frontal Assessment Battery (FAB) was used which assesses conceptualization (assessing similarities), mental flexibility (verbal fluency), motor programming ('Luria' test), resistance to interference (conflicting instructions), inhibitory control (via a go-no go paradigm) and environmental autonomy (prehension behaviour). 27 The Repeatable Battery for Assessment of Neuropsychological Status (RBANS) was used to measure performance overall and on specific domains including immediate memory (Index I), visual-spatial (Index II), language (Index III), attention (Index IV) and delayed memory (Index V), to provide a comprehensive neuropsychological assessment. 28

| Statistical analysis
All analysis was carried out using STATA v15.0. Descriptive statistics are provided as means with standard deviations and proportions with percentages as appropriate. Between group differences were analysed using ANOVA and chi-square tests as appropriate. After examining data for normality (using Q-q plots and histograms), data which did not conform to a normal distribution was log-transformed prior to analysis.
In order to assess the relationship between anticholinergic  Analyses were performed in the first instance using total ACB score as the predictor variable. Given that studies have demonstrated that it may only be definite anticholinergic medications that are linked to potential adverse cognitive effects, we repeated analysis estimating effects for definite and potential anticholinergic medication use separately (with no regular anticholinergic use as the reference group in both instances). Analysis were then repeated using both regular benzodiazepine and finally with regular Z-drug use as the independent variable of interest.
Sensitivity analyses were conducted in order to further evaluate observed associations. In the first instance, we excluded individuals with intermittent use ("as required"/"prn") of either anticholinergic or BDZR medication from both the exposure and reference group, in order to only assess the impact of regular use vs non-use. Analyses were also repeated excluding individuals using both anticholinergic and BDZR medication, in order to avoid potential interactions between these medication classes and to estimate the effect of each medication class individually. In order to evaluate the association in those without cognitive impairment, given that these medications may be prescribed in some instances for early symptoms of cognitive DYER ET AL.
impairment and that established cognitive impairment may bias results, we used a MMSE score cut-off of 24, consistent with the population under study to define potential cognitive impairment. 29 This was carried out in order to assess those with MMSE scores within the normal range.
Finally, in order to further evaluate the association between definite anticholinergic medication use and cognitive function, we divided definite anticholinergics into: (i) antipsychotics; (ii) urologicals and; (iii) antidepressants which accounted for the majority of definite anticholinergic use in order to examine if any observed effects were specific to a particular subclass of anticholinergic medication or represent an overall class effect. Results of mixed-effects models are reported as beta coefficients and 95% confidence intervals (CIs). For all analysis p < 0.05 was considered the threshold for statistical significance.

| Study participants and medication usage
In total, of 5186 recruited to TUDA, 5153 participants were included in the current study (aged 74.0 ± 8.3 years; 67.4% female) following exclusion of those with missing data for all three cognitive assessments (n = 33). Use of any anticholinergic medication (potential or definite) was reported by 44.0% (2259/5153) of participants whilst 9.7% (500/5,153) regularly used a definite anticholinergic medication. Of those using any anticholinergic medication, the median ACB score was 2. 1-3 A minority, 7.1% (363/ 5135) were regular users of a benzodiazepine medication whilst 7.5% (387/5135) regularly used a "Z-drug". Baseline characteristics by medication use are presented in tabular format below (See Table 1). The

| Anticholinergic use and neuropsychological performance
Under unadjusted models, increasing ACB score in addition to both potential and definite anticholinergic medication use was associated with significantly poorer cognitive performance on the MMSE, FAB and RBANS in addition to all RBANS indices (

| Anticholinergic medication subclass and cognitive performance
We divided definite anticholinergic use into regular (i) antipsychotic; (ii) urological and; (iii) antidepressant use. Associations for antipsychotic use and poorer performance on all total cognitive scores and RBANS indices were observed and persisted following adjustment.
Regular use of urologicals was associated with poorer performance on the MMSE and RBANS indices I and III after adjustment. Associations for anticholinergic antidepressant use were somewhat weaker, with poorer performance on the total RBANS score persisting following robust covariate adjustment in addition to indices I, II and V (See Table 3).
On excluding those using both anticholinergic and benzodiazepines, associations for total ACB score, definite anticholinergics and benzodiazepine use and poorer MMSE score were attenuated.
Associations for definite anticholinergic use and poorer performance on the FAB persisted. Finally, associations for total ACB score, definite anticholinergic use and benzodiazepine use persisted on the total RBANS and RBANS index I and RBANS index IV whilst associations for poorer performance on index II, III and V were seen for ACB score and definite anticholinergic use only (Table S2).

RBANS index II (Visuo-spatial)
ACB score (total)   In order to analyse the association of anticholinergic/benzodiazepine and related medication usage with poorer cognitive function, we re-ran analysis excluding those with a MMSE score < 24. Using this analysis, associations for ACB score and definite anticholinergics, but not benzodiazepines persisted for the FAB whilst associations for ACB score, definite anticholinergic and benzodiazepine persisted for total RBANS score. Additionally, associations between benzodiazepine use and poorer performance on RBANS domains I and V were not observed on excluding those with probable cognitive impairment based on MMSE score (Table S3).

| DISCUSSION
In the current study of over 5000 community-dwelling older adults, we demonstrated a significant association between the regular use of strong anticholinergic medications (for the preceding 6 months) and benzodiazepines and poorer neuropsychological performance. These effects were observed for strong anticholinergic use across all domains of cognitive function, and for immediate memory and attention were most pronounced for benzodiazepine use. This is one of the first studies to characterise in detail the association between use of these medications and neuropsychological performance in older adults.
In the current study, associations between definite (but not potential) anticholinergic medication use and cognitive function per- anticholinergic and benzodiazepine use in those with no objective signs of established cognitive impairment on the MMSE.
In line with preceding studies, the effect sizes observed in the current study are relatively small. Whilst our findings highlight potential adverse cognitive effects of strong anticholinergic medications and benzodiazepines, the risk vs benefit of prescribing these medications in day-to-day practice should be assessed on an individual basis. It is also worth bearing in mind that to date, there is limited evidence that decreasing anticholinergic burden in high risk individuals may not improve cognitive function. 30 However, it must be noted that many studies in this regard have examined this in individuals with established cognitive impairment with typically short follow-up period. 30 Trials aimed at reducing overall anticholinergic burden are needed in older adults without established cognitive impairment to examine the impact of discontinuation on cognitive and neuropsychological performance.
Not least, our findings would encourage prescribers to consider lessanticholinergic alternatives where available (for instance with antidepressant medications). Further, the other potential effects of medications in older adults must be taken into account. For instance, benzodiazepines and Z-drugs have been associated with increased risk of both falls and delirium, which may be greatest in those with established cognitive impairment. 31 A notable strength of the current study is the large sample size of community-dwelling older adults and the detailed cognitive and neuropsychological assessments used. An important limitation however is its cross-sectional design which precludes inference about causality. Further longitudinal studies will be required to replicate and further characterise the specific findings in relation to medication use and neuropsychological performance in older adults. Like any study examining the links between medication use and cognition, we are also unable to exclude confounding by indication/protopathic bias. Whilst we were able to perform a detailed assessment of screened anxiety and depressive symptoms in addition to other important confounders, it is important to reflect that anticholinergic/ benzodiazepines may be used for prodromal symptoms of cognitive impairment/dementia such as anxiety and sleep disturbance.

| CONCLUSIONS AND IMPLICATIONS
In conclusion, we demonstrated an association between strong anticholinergic medication/benzodiazepine use and cognitive/neuropsychological performance in older adults free from a diagnosis of dementia. Further longitudinal studies will be required to establish the longitudinal effects of anticholinergic medications and benzodiazepine use on specific domains of neuropsychological performance, in order to identify potential targets for the promotion of brain health in older adults at risk of cognitive impairment.