Anticholinergic Drug Burden and Delirium: A Systematic Review

Objectives: To investigate the association between anticholinergic drug burden (ADB), measured with anticholinergic drug scales, and delirium and delirium severity. Design: Systematic review. Setting and Participants: All available studies. Methods: A systematic literature search was performed in Medline, Embase, PsycINFO, Web of Science, CINAHL, Cochrane library, and Google Scholar. Studies evaluating the association between ADB (measured as a total score) and delirium or delirium severity, published in English, were eligible for inclusion. Results: Sixteen studies, including 148,756 persons, were included. Fifteen studies investigated delirium. ADB was measured with the Anticholinergic Risk Scale (ARS, n 1⁄4 5), the Anticholinergic Cognitive Burden Scale (ACB, n 1⁄4 6), the list of Chew (n 1⁄4 1), the Anticholinergic Drug Scale (ADS, n 1⁄4 5), a modified version of the ARS (n 1⁄4 1), and a modified version of the ACB (n 1⁄4 1). A high ADB, measured with the ARS, was associated with delirium (5/5). Also with the modified version of the ARS and ACB, an association was found between a high ADB and delirium during 3-month (1/1) and 1-year follow-up (1/1), respectively. When ADB was assessed with other scales, the results were inconclusive, with only 1 positive association for the ACB (1/6) and ADS (1/5) each. The possible association between ADB and delirium severity has also been investigated (ADS n 1⁄4 2, Summers Drug Risk Number n 1⁄4 1). One study found an association between a high ADB, measured with the ADS, and an increase in severity of delirium. Conclusions and Implications: ADB assessed with the ARS is consistently associated with delirium. The association found between the modified versions of the ARS and ACB and delirium needs confirmation. When ADB was assessed with other scales, the findings were inconclusive. The current findings suggest that the ARS might be a useful tool to identify patients at increased risk for delirium. 2020 AMDA e The Society for Post-Acute and Long-Term Care Medicine. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Delirium is very common among older patients and is associated with poor outcomes, such as functional and cognitive decline and increased mortality.1 Despite the high prevalence and clinical impact, this syndrome is still poorly understood. Knowledge about the underlying pathophysiology and identification of modifiable risk factors are of paramount interest. agencies in the public, commD, PhD, Section of Geriatric C University Medical Center, Netherlands. ts). te and Long-Term Care Medicine. T The neurotransmitter acetylcholine is implicated in several processes that are impaired during delirium, such as attention, sleep, and memory, and this has led to the hypothesis that a cholinergic deficiency might be involved in the pathogenesis of delirium.2,3 Drugs with anticholinergic properties are commonly prescribed in older persons, and the use of these drugs can cause some degree of cholinergic deficiency by blocking the effects of acetylcholine.4 Therefore, use of anticholinergic drugs could be a risk factor for delirium. Previous studies have investigated the possible association between anticholinergic drugs and delirium, but the findings are conflicting.5,6 Discrepancies in the results might be caused by the methods used to assess anticholinergic drug use, which differ his is an open access article under the CC BY license (http://creativecommons.org/ A. Egberts et al. / JAMDA xxx (2020) 1e9 2 substantially among studies.7 In some studies, anticholinergic drug use is assessed with crude measures such as “exposed or not exposed” or the total number of anticholinergic drugs taken. Other studies use the anticholinergic drug burden (ADB), which takes into account the specific anticholinergic load of the drugs used by a person. The ADB can be calculatedwith anticholinergic drug scales and is defined as the sum of scores assigned to the drugs. In the last decade, different anticholinergic drug scales have been developed, but these scales differ substantially from each other in number and ranking of drugs, and the question rises whether the use of all these scales results in comparable associations with delirium. Therefore, the aim of the present review was to investigate the possible association between ADB, measured with anticholinergic drug scales and delirium.

substantially among studies. 7 In some studies, anticholinergic drug use is assessed with crude measures such as "exposed or not exposed" or the total number of anticholinergic drugs taken. Other studies use the anticholinergic drug burden (ADB), which takes into account the specific anticholinergic load of the drugs used by a person. The ADB can be calculated with anticholinergic drug scales and is defined as the sum of scores assigned to the drugs. In the last decade, different anticholinergic drug scales have been developed, but these scales differ substantially from each other in number and ranking of drugs, and the question rises whether the use of all these scales results in comparable associations with delirium. Therefore, the aim of the present review was to investigate the possible association between ADB, measured with anticholinergic drug scales and delirium.

Methods
This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The checklist is added as Supplementary Data (Appendix 1).

Data Sources and Search Strategy
A systematic literature search was performed in Medline Ovid, Embase, PsycINFO, Web of Science, CINAHL, Cochrane library, and Google Scholar covering the period up until January 31, 2020 using relevant terms for anticholinergic drugs and delirium. The search queries were developed with the assistance of an experienced biomedical information specialist and can be found in the Supplementary Data, Appendix 2. Reference lists of review articles and included studies were manually screened to identify additional eligible studies.

Eligibility Criteria
Studies that met each of the following criteria were eligible for inclusion: (1) the association between ADB and delirium or delirium severity was investigated; (2) ADB was measured with an anticholinergic drug scale and expressed as a total score; and (3) the study was published in the English language. In case full text articles were not available, the corresponding authors were contacted and whenever answers were not obtained despite reminders, articles were excluded. Case studies, case series, review articles, commentaries, letters, editorials, conference abstracts, and studies that used the Drug Burden Index without stratification into the anticholinergic and sedative components were excluded.

Study Screening and Selection
All references identified by the search queries were downloaded in Endnote X9 (Clarivate Analytics, Philadelphia, PA) and duplicates were removed. Three reviewers (A.E., R.M.G., and H.A.) independently screened titles and abstracts for potentially eligible studies, and assessed full-text articles against the eligibility criteria. Disagreements at any stage were resolved through consensus.

Data Extraction
Data from all studies that met the inclusion criteria were independently extracted by 2 authors (A.E. and F.M.R.) using a predefined extraction table, including author, year of publication, study design, population and setting, sample size, participant age and sex, number of persons with delirium, methods of measuring ADB, tools used to assess delirium and delirium severity, type of delirium (prevalent or incident), information on the statistical analyses, and the results with regard to the possible association between ADB and delirium (odds ratios, hazard ratios, relative risks, differences in proportions or regression coefficients). When studies used multiple models to investigate the association between ADB and delirium, the results of the model including the most covariates were extracted. Authors were contacted when study details were missing and data were considered unattainable if no answer was obtained despite several reminders. Any uncertainties were resolved through discussion.

Quality Assessment
Two reviewers (A.E. and G.Z.) independently assessed the methodological quality of the included studies using the Newcastle-Ottawa Scale for cohort and case-control studies. 8 The scale used for casecontrol studies was additionally used for cross-sectional studies. 9 The Newcastle-Ottawa Scale evaluates 3 aspects of the study methodology: the selection of study groups (score range 0-4), comparability of the groups (score range 0-2), and the quality of outcome/ exposure ascertainment (score range 0-3). The total score ranges from 0 to 9 (highest quality). Disagreements were resolved through discussion.

Data Synthesis
The association between ADB and delirium was investigated separately for prevalent and incident delirium. To explore any variations in results across the studies and anticholinergic drug scales, the results were additionally grouped based on the clinical settings where the studies were performed, regardless of delirium type. Furthermore, we planned to perform subgroup analyses to explore the influence of potential confounding factors, such as dementia and severity of illness. 3 Furthermore, the association between ADB and delirium severity was investigated.

Study Selection
The primary literature search resulted in 3085 records. After exclusion of duplicates, 1960 records remained; of these, 1829 were excluded based on titles and abstracts. In total, 131 records were assessed for eligibility. Sixteen studies met the inclusion criteria and were included in the final review. An overview of the study selection process is presented in Figure 1.

Study Characteristics
The characteristics of the sixteen included studies are presented in Table 1. There were 8 prospective cohort studies, 10e13,16,17,21,26 5 retrospective cohort studies, 14,18,23e25 1 nested case-control study, 19 and 2 retrospective cross-sectional studies. 20,22 A total of 148,756 persons were studied (sample size range 90e118,750; mean ¼ 9297.25; median ¼ 420.5). Thirteen studies were conducted in the hospital setting, 10e14,17e22,25,26 of which 10 on a medical ward 10,12e14,17e20,22,26 and 3 on a surgical ward, 11,21,25 1 study was performed in nursing homes, 16 1 study was performed in community-dwelling patients with dementia, 23 and 1 study was performed in the general population. 24 Delirium was assessed with the Diagnostic and Statistical Manual of Mental Disorders (4 th and 5 th edition) criteria, 19,20 the Confusion Assessment Method, 10,11,13,16,17,25,26 the Nursing Delirium Scale, 25 the Delirium Rating Scale, 12 codes for delirium according to the International Classification of Diseases, 9 th and 10 th edition, 18,23,24 the 4 'A's test, 22 and a validated chart-based instrument developed by Inouye et al 15 for the identification of delirium 14 and documented diagnoses of delirium in discharge summaries. 21 Delirium severity was assessed with the Delirium Index 10 and the Memorial Delirium Assessment Scale. 12 ADB was measured with the Anticholinergic Risk Scale (ARS), 14,16e18,20 the Anticholinergic Cognitive Burden scale (ACB), 13,19e22,26 the list of Chew, 20 the Anticholinergic Drug Scale (ADS), 10e12,17,19,25 a modified version of the ARS, 24 and a modified version of the ACB. 23 Characteristics of these anticholinergic drug scales are outlined in Table 2. Studies were performed in the United States, 12e14,18 the Netherlands, 17,20 Italy, 16,22 Canada, 11,21 Korea, 23,24 Norway, 10 the United Kingdom, 19 Germany, 25 and Portugal. 26

Quality of the Studies
Details on the methodological quality of the included studies according to the Newcastle-Ottawa Scale are provided as Supplementary Data (Appendix 3). Quality scores ranged from 5 to 9 stars (median 8 stars). One study scored the maximum 4 stars for the study selection criteria. Fourteen out of 16 studies scored the maximum 2 stars for the comparability of the study groups, and 14 studies achieved the maximum 3 stars for the outcome/exposure criteria.

Anticholinergic Drug Burden and Delirium
The possible association between ADB and delirium was investigated in 15 of the 16 studies. 11e14,16e26 Four studies investigated prevalent delirium, 11,19,20,22 7 studies incident delirium, 11e14,17,21,25 2 studies delirium at some point during the hospital stay (combination of prevalent and incident delirium), 18,26 1 study delirium during 3 months follow-up, 24 and 2 studies delirium during 1-year followup. 16,23 The studies investigating incident delirium, delirium at some point during the hospital stay, and delirium during follow-up are combined in this review.

Prevalent Delirium
Four studies reported on the possible association between ADB and prevalent delirium (delirium on admission in 3 studies 19,20,22 and preoperative delirium in 1 study 11 ). A total of 1993 persons were studied (659 with delirium). Three studies were performed in acutely ill patients admitted to the hospital 19,20,22 and 1 study in patients admitted with a hip fracture. 11 In all 4 studies, the median or mean age was >80 years. Delirium was assessed with the Diagnostic and Statistical Manual of Mental Disorders (4 th and 5 th edition), 19,20 the 4 'A's test, 22 and the Confusion Assessment Method. 11 ADB was assessed with the ARS, 20 the ACB, 19,20,22 the ADS, 11,19 and the list of Chew. 20 The study results are presented in Table 3. Only a moderate and high ADB, measured with the ARS, was associated with delirium on admission. 20 No associations were found with the other anticholinergic drug scales.

Incident Delirium
Twelve studies reported on the possible association between ADB and incident delirium. 11e14,16e18,21,23e26 A total of 146,849 persons were studied (1703 with delirium; in 1 study, the number of patients with delirium was not defined 16 ). Nine studies were performed in patients admitted to the hospital (palliative inpatients, 14 patients with cognitive impairments, 13 patients admitted to the Intensive Care Unit, 17 patients with a hip fracture undergoing surgery, 11 patients with Parkinson's disease, 18 patients undergoing transcatheter aortic valve implantation, 21 patients with malignancies, 12 cancer patients undergoing surgery, 25 and acutely ill patients 26 ), 1 study in nursing home patients, 16 1 study in community-dwelling patients with dementia, 23 and 1 study in the general population. 24 In 9 studies, the median or mean age was >70 years, 11,13,14,16,18,21,24e26 in 1 study 60.9% of the patients were 75 years or older, 23 in 1 study the mean age was 60 years, 17 and in 1 study the mean age was 41.5 years. 12 Delirium was assessed with the Confusion Assessment Method, 11,13,16,17,25,26 the Nursing Delirium Scale, 25 the Delirium Rating Scale, 12 International Classification of Diseases, 9 th and 10 th edition codes for delirium, 18,23,24 24 and a modified version of the ACB. 23 The study results are presented in Table 4. In all 4 studies using the ARS, an association was found between ADB and incident delirium. A moderate and high ADB as well as an increase in burden during the hospital stay, measured with the ARS, was associated with an increased risk of developing delirium. In addition, with the modified versions of the ARS and ACB, an association was found between a high ADB and delirium during follow-up. Conflicting results were found when the ADB was assessed with the ACB or ADS.

Subgroup Analyses
The studies included in this review are performed in different patient populations, which might influence the association. The outcomes of the studies were therefore additionally grouped based on the clinical setting (Supplementary Data, Appendix 4). Only in acutely ill hospitalized patients was the association investigated more than 2 times (6 studies in total 13,18e20,22,26 ): the ARS was used in 2 studies, 18,20 the ACB in 5, 13,19,20,22,26 and the ADS 19 and Chew 20 in 1. Both studies that used the ARS found an association, 18,20 and only 1 study that used the ACB did. 26 In addition, the included studies used a wide range of variables in the multivariate models. Dementia and severity of acute illness might have a large impact on the association between ADB and delirium. 3 Only 2 studies adjusted for dementia 22,26 and 3 for severity of acute illness as defined by the Acute Physiology and Chronic Health Evaluation (APACHE) score and the American Society of Anesthesiologists Physical (ASA) status. 14,17,25 Because of the small number of studies adjusting for these variables, no subgroup analyses could be performed.

Anticholinergic Drug Burden and Delirium Severity
Two studies reported on the possible association between ADB and delirium severity. 10,12 A total of 368 persons were studied (323 with delirium). One study was performed in acutely ill patients admitted to the hospital 10 and 1 study in patients with malignancies. 12 Mean ages in these studies were 84.3 years and 41.5 years, respectively. Delirium was diagnosed with the Confusion Assessment Method 10 and the Delirium Rating Scale. 12 The study results can be found in the Supplementary Data, Appendix 5. Both studies used the ADS and 1 study 10 found an association between an increase in ADB and an increase in delirium severity.

Discussion
The findings of this systematic review demonstrate consistent evidence that ADB measured with the ARS is associated with delirium. In addition, with a modified version of the ARS and ACB an association was found between high ADB and delirium. The findings were conflicting when ADB was assessed with other scales, with more negative than positive studies.
This systematic review has evaluated the association between anticholinergic drugs and delirium in more depth than previous reviews. 5,6 In the present review, we specifically included studies in which the ADB score was calculated with a scale and this has highly increased the ability to compare the findings. Previous reviews have reported conflicting findings and these discrepancies can be caused by the fact that the included studies were quite heterogeneous in their quantification of the anticholinergic load. Moreover, the review of Welsh et al included only other systematic reviews about ADB tools and was not designed to investigate the association between anticholinergic drugs and delirium. 32 The 16 studies included in the present review have used 6 different anticholinergic drug scales (ie, the ARS, ACB, ADS, the list of Chew, a modified version of the ARS, and a modified version of the ACB), and only the ARS was consistently associated with delirium (5 out of 5 studies found a positive association). Also, in the 2 studies that used a modified version of the ARS and ACB, an association was found between a high ADB and delirium during the 3-month 24 and 1-year follow-up, 23 respectively. The modified version of the ARS includes 60 more drugs, 24 and the modified version of the ACB includes 79 more drugs 23 than the original ARS 27 and ACB scale. 29 Moreover, in both studies, the authors also took into account the daily drug dose and, therefore, the findings cannot be compared with findings found with the original scales. When ADB was assessed with other scales, the results were inconclusive, with only 1 positive association for the ACB 26 (1 out of 6 studies) and 1 for the ADS 25 (1 out of 5 studies). An explanation for the discrepancies in findings might be the large differences in the total number and ranking of drugs between the available anticholinergic drug scales as well as the different methods used to develop the scales. A previous study has evaluated the agreement between the ARS, ACB, ADS, and the anticholinergic subscale of the Drug Burden Index for measuring ADB, and found a poor agreement between the 4 scales. Only the ACB and ADS showed a good agreement, 33 and these findings were confirmed in another study. 34 Previous systematic reviews have already highlighted that the association between anticholinergic drug scales and outcomes, such as mortality and physical function, can be different depending on which scale is used. 35e37 Therefore, the large differences in the measurement of the ADB among the available anticholinergic drug scales can also have a high impact on finding an association with delirium.
In addition, the ARS attempts to predict both peripheral and central effects, 27 in contrast to the ACB in which the grading of drugs is based on the potential to cause cognitive effects. 28 It might be possible that in delirium not only central, but also peripheral anticholinergic effects may play a role. Blurred vision, urinary retention, and constipation, known peripheral adverse effects of anticholinergic drugs, 4 are risk factors for delirium 38 and might explain why the ARS was associated with delirium. However, because the individual studies did not report on adverse effects, this remains speculative. Furthermore, it might be possible that the differences in findings among the anticholinergic drug scales are caused by the variety in patient populations and the diversity in variables for which has been adjusted in multivariate models. Unfortunately, no conclusions can be drawn because some patient populations have only been studied once. Only in acutely ill older patients has the association between ADB and delirium been investigated several times. 13,18e20,22,26 Five studies used the ACB (with almost comparable mean age and delirium prevalence), 13,19,20,22,26 and only 1 study found an association. 26 Moreover, the included studies did not adjust for the same confounding factors. Factors that might influence the association, such as dementia and baseline severity of illness, 3 were not always included in the analyses and, therefore, no conclusions can be drawn for the effect of possible confounders.
Based on the findings of the present review, it can be concluded that the ARS could be a suitable instrument to identify patients at increased risk of delirium. Previous studies have shown that medication reviews can be effective in reducing ADB scores (based on the ARS) in persons age 65 years and older. 39,40 Therefore, it would be interesting to investigate whether regular medication reviews with the ARS as an additional tool, in both the community and hospital setting, will reduce delirium.

Limitations and Strengths
This systematic review has some limitations. First, our search was limited to articles published in the English language. As far as we are aware, there is 1 study published in Spanish in which the association between the ARS, ACB, and ADS and delirium was investigated in patients admitted to a geriatric ward of a hospital. 41 The results are in line with our findings: a significant association was found between the ARS and incident delirium and no association was found with the other anticholinergic drug scales. Second, one might speculate that publication bias could have played a role, considering that 50 conference abstracts were excluded. Of these 50 abstracts, 8 abstracts explicitly described that they have investigated the association between ADB, measured with a scale, and delirium. Two of these abstracts are included as full-text studies in the present review. Of the remaining abstracts, 3 have used the ARS, of which 2 have found an association and 1 not; 4 abstracts have used the ACB and none have found an association; and the ADS was used in 1, and also this abstract found no association. These findings are in line with the results of the present review, and therefore, we think that publication bias has not influenced the results. Third, there was considerable heterogeneity among the studies. However, considering that the evidence for the ARS is consistent among the studies despite the different settings and populations, we do not believe that this has influenced our findings. Fourth, the studies included in this review used the ARS, ACB, ADS, and the list of Chew. Although these are the most frequently used scales in research, other scales exist and it is not known whether these scales are associated with delirium. Moreover, the list of Chew and the modified versions of the ACB and ARS were only used in 1 study each 20,23,24 ; therefore, confirmation of the findings is warranted. Fifth, this review identified only 2 studies investigating the possible association between ADB and the severity of delirium, 10,12 which hampers the ability to draw conclusions. More studies in this field are needed.
Major strengths of this review are the comprehensive search, which was performed in multiple databases, and the inclusion, which was limited to studies in which the ADB score was calculated.

Conclusions and Implications
The findings of this systematic review demonstrate consistent evidence that ADB measured with the ARS is associated with delirium. Also, with the modified versions of the ARS and ACB, an association was found between high ADB and delirium, but these findings need confirmation. The current findings suggest that the ARS might be a useful tool to identify persons at increased risk for delirium. Future studies are needed to investigate whether regular medication reviews with the ARS in both the community and hospital settings will reduce delirium.

Study selection 17
Give numbers of studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at each stage, ideally with a flow diagram. 8, Figure 1 Study characteristics 18 For each study, present characteristics for which data were extracted (eg, study size, PICOS, follow-up period) and provide the citations.
8-9, Table 1 Risk of bias within studies 19 Present data on risk of bias of each study and, if available, any outcome level assessment (see item 12).

9, appendix 3
Results of individual studies 20 For all outcomes considered (benefits or harms), present, for each study: (a) simple summary data for each intervention group (b) effect estimates and confidence intervals, ideally with a forest plot.

9-12, Tables 3 and 4, appendix 4-5
Synthesis of results 21 Present results of each meta-analysis done, including confidence intervals and measures of consistency.

Not applicable
Risk of bias across studies 22 Present results of any assessment of risk of bias across studies (see Item 15). Not applicable Additional analysis 23 Give results of additional analyses, if done (eg, sensitivity or subgroup analyses, meta-regression [see Item 16]).

Summary of evidence 24
Summarize the main findings including the strength of evidence for each main outcome; consider their relevance to key groups (eg, healthcare providers, users, and policy makers).

13-15
Limitations 25 Discuss limitations at study and outcome level (eg, risk of bias), and at reviewlevel (eg, incomplete retrieval of identified research, reporting bias).

Conclusions 26
Provide a general interpretation of the results in the context of other evidence, and implications for future research.

Funding 27
Describe sources of funding for the systematic review and other support (eg, supply of data); role of funders for the systematic review.