The Prevalence of Anticholinergic Drugs and Correlation with Pneumonia in Elderly Patients: A Population-Based Study in Taiwan

Anticholinergic drugs may increase the risk of serious respiratory infection, especially in the elderly. The study aims to investigate the prevalence of anticholinergic drugs and the correlation of incident pneumonia associated with the use of anticholinergic drugs among the elderly in Taiwan. The study population was 275,005 elderly patients aged ≥65 years old, selected from the longitudinal health insurance database (LHID) in 2016. Among all the elderly patients, about 60% had received anticholinergic medication at least once. Furthermore, the study selected elderly patients who had not been diagnosed with pneumonia and had not received any anticholinergic drugs in the past year in order to evaluate the correlation between pneumonia and anticholinergic drugs. The study excluded elderly patients who died or had received related drugs of incident pneumonia during the study period and selected elderly patients receiving anticholinergic drugs as the case group. Propensity score matching (PSM) on a 1:1 scale was used to match elderly patients that were not receiving any anticholinergic drugs as the control group, resulting in a final sample of 32,215 patients receiving anticholinergic drugs and 32,215 patients not receiving any anticholinergic drugs. Conditional logistic regression was used to estimate the association between anticholinergic drugs and pneumonia after controlling for potential confounders. Compared with patients not receiving anticholinergic drugs, the adjusted odds ratio of patients receiving anticholinergic drugs was 1.33 (95% confidence interval: 1.18 to 1.49). Anticholinergic medication is common among elderly patients in Taiwan. Elderly patients receiving anticholinergic drugs may increase their risk of incident pneumonia. The safety of anticholinergic drugs in the elderly should be of concern in Taiwan.


Introduction
Anticholinergic drugs act centrally and peripherally on muscarinic acetylcholine receptors. Cimetidine, metoclopramide, and ranitidine are commonly employed in the treatment of the urinary bladder, the respiratory tract, and gastroesophageal disease via the intervention of the muscarinic included detailed clinical records of the outpatient department and hospitalization, diagnostic codes, and prescribing information. The NHI program is nationwide social insurance that has enrolled up to 99% of citizens since 1995. Hence, the database is a nationally representative health database for Taiwan. HWDC provides scrambled random identification numbers for insured patients to protect the privacy of beneficiaries. This study protocol was approved as a completely ethical review by the Institutional Review Board of China Medical University Hospital, Taiwan (No. CMUH107-REC2-004). The database is anonymous; therefore, the requirement for informed consent was waived. Figure 1 lists the flowchart of the selected patients for inclusion. The study population was 275,005 elderly patients aged ≥65 years old, selected from LHID on 1 January 2016 to investigate the prevalence of anticholinergic drugs. Furthermore, the elderly patients who had not been diagnosed with pneumonia and had not received any anticholinergic drugs in the past year were enrolled in the study to evaluate the correlation between pneumonia and anticholinergic drugs. We also excluded 10,532 patients who died and the 33,189 patients who had received related drugs of incident pneumonia in the study period in order to improve the validity of the study results. The related drugs of incident pneumonia contained amiodarone, angiotensin-converting enzyme inhibitors (ACEIs), amantadine, and steroids [12][13][14][15][16][17]. Moreover, the study classified elderly patients according to their medication records in 2016 and selected elderly patients who were receiving anticholinergic drugs as the case group. In the study, there were a total of 49 drugs with the potential to cause anticholinergic adverse effects [17], including psychotropic medicines (e.g., paroxetine, quetiapine, trazodone), alimentary tract and metabolism medicines (e.g., cimetidine, metoclopramide, ranitidine, loperamide), respiratory system and allergy medicines (e.g., cetirizine, loratadine, pseudoephedrine), and neurological disorder medications (e.g., carbidopa-levodopa, levodopa) (see Table A1). Furthermore, to reduce the potential confounding caused by unbalanced covariates in nonexperimental settings, we used propensity score matching (PSM) on a 1:1 scale to match elderly patients not receiving any anticholinergic drugs as the control group, resulting in a final sample of 32,215 patients receiving anticholinergic drugs and 32,215 patients not receiving any anticholinergic drugs. The propensity score of the study is the probability of patients receiving anticholinergic drugs, calculated by gender, age, insured salary, urbanization, and the Charlson comorbidity index (CCI).

Study Design
The study is a retrospective study to investigate the prevalence of anticholinergic drugs among elderly patients in Taiwan. Furthermore, the study design is divided into three phases to assess the correlation between anticholinergic drugs and pneumonia. The first phase was the exclusion period, from 1 January to 31 December 2015, in which to select the study subjects. The second phase was the observation period, from 1 January to 30 November 2016, in which to estimate the correlation between anticholinergic drugs and pneumonia. Patients were followed-up 30 days after anticholinergic drug prescriptions in order to observe the incidence of pneumonia. The third phase was in December 2016 to ensure the follow-up period. Thus, the patients receiving anticholinergic drugs in this phase would not be enrolled as study subjects. The study design diagram is exhibited in Figure 2.
The definition of incident pneumonia in the study was according to the principal diagnosis code in J12-J18, based on the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM). The study investigated the association between anticholinergic drugs and pneumonia via conditional logistic regression. Control variables in the study contained gender, age, insured salary, urbanization, CCI scores, and comorbidities related to pneumonia. Comorbidities contained diabetes mellitus (DM; ICD-10-CM: E08-E13), Alzheimer's disease (AD; ICD-10-CM: G30, F00), stroke

Study Design
The study is a retrospective study to investigate the prevalence of anticholinergic drugs among elderly patients in Taiwan. Furthermore, the study design is divided into three phases to assess the correlation between anticholinergic drugs and pneumonia. The first phase was the exclusion period, from 1 January to 31 December 2015, in which to select the study subjects. The second phase was the observation period, from 1 January to 30 November 2016, in which to estimate the correlation between anticholinergic drugs and pneumonia. Patients were followed-up 30 days after anticholinergic drug prescriptions in order to observe the incidence of pneumonia. The third phase was in December 2016 to ensure the follow-up period. Thus, the patients receiving anticholinergic drugs in this phase would not be enrolled as study subjects. The study design diagram is exhibited in Figure 2.
The definition of incident pneumonia in the study was according to the principal diagnosis code in J12-J18, based on the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM). The study investigated the association between anticholinergic drugs and pneumonia via conditional logistic regression. Control variables in the study contained gender, age, insured salary, urbanization, CCI scores, and comorbidities related to pneumonia. diagnosis once a year.

Statistical Analysis
Descriptive statistics were used to summarize distributions of the prevalence of anticholinergic drugs in the elderly in Taiwan. We used the standardized mean difference to examine the balance of case and control groups after matching. We further used conditional logistic regression to estimate the association between anticholinergic drugs and pneumonia after controlling for potential confounders. SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA) was used for statistical analysis in the study, and the statistical significance was defined as p-value <0.05.  Table 1 presents the prevalence of anticholinergic drugs in elderly patients. There was a total of 967,887 prescriptions with anticholinergic drugs, about 12.96%, out of all prescriptions. The most frequency anticholinergic prescriptions were cimetidine (2.54%), cetirizine (1.19%), pseudoephedrine (1.15%), metoclopramide (1.13%), ranitidine (0.83%), quetiapine (0.80%), carbidopa-levodopa (0.66%), loratadine (0.65%), trazodone (0.54%), and loperamide (0.51%), sequentially. To elderly patients, there were 165,262 patients (60.09%) who had received anticholinergic medication at least once out of all elderly patients.

Statistical Analysis
Descriptive statistics were used to summarize distributions of the prevalence of anticholinergic drugs in the elderly in Taiwan. We used the standardized mean difference to examine the balance of case and control groups after matching. We further used conditional logistic regression to estimate the association between anticholinergic drugs and pneumonia after controlling for potential confounders. SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA) was used for statistical analysis in the study, and the statistical significance was defined as p-value < 0.05. Table 1 presents the prevalence of anticholinergic drugs in elderly patients. There was a total of 967,887 prescriptions with anticholinergic drugs, about 12.96%, out of all prescriptions. The most frequency anticholinergic prescriptions were cimetidine (2.54%), cetirizine (1.19%), pseudoephedrine (1.15%), metoclopramide (1.13%), ranitidine (0.83%), quetiapine (0.80%), carbidopa-levodopa (0.66%), loratadine (0.65%), trazodone (0.54%), and loperamide (0.51%), sequentially. To elderly patients, there were 165,262 patients (60.09%) who had received anticholinergic medication at least once out of all elderly patients.

The Baseline Characteristics Distribution of Study Subjects
After subject selection, there was a total of 64,430 elderly patients in the study, and both groups, with and without anticholinergic drugs, had 32,215 patients. The mean age of patients receiving anticholinergic drugs was 73.58 ± 7.35 years, 51.77% were female, and 72.74% were in highly urbanization areas. As in Table 2, baseline characteristics were well balanced between elderly patients with and without anticholinergic drugs after matching. Among patients receiving anticholinergic drugs, there were 8614 patients (26.74%) with DM, 3457 patients (10.73%) with stroke, 317 patients with Parkinson's disease (0.98%), 3684 patients with CKD (11.43%), 1928 patients with asthma (5.98%), 2781 patients with COPD (8.63%), 1076 patients with HF (3.34%), 18,688 patients with URI (58.01%), 862 patients with GERD (2.68%), and 120 patients with epilepsy (0.37%).   Table 3 indicates that among all participants, 1519 patients had occurred pneumonia, and the incidence rate was 23.58 cases per 1000 elderly patients. The incidence rate of patients receiving anticholinergic drugs was 33.23 cases per 1000 elderly patients. Compared with patients not receiving anticholinergic drugs, the risk ratio was 2.18 times of incident pneumonia. Compared with patients without DM, patients with DM had 1.31 times the risk of incident pneumonia. Additionally, compared with patients without comorbidities, patients with comorbidities had a higher risk of incident pneumonia, including stroke (2.21 times), PD (2.89 times), CKD (1.85 times), asthma (5.54 times), COPD (6.77 times), HF (2.75 times), URI (1.98 times), GERD (2.15 times), and epilepsy (4.61 times).

Discussion
Our study results indicate that about 60% of elderly patients had received anticholinergic medication at least once, and the most frequent anticholinergic prescriptions in Taiwan were cimetidine, cetirizine, pseudoephedrine, metoclopramide, and ranitidine. These drugs are used to treat URI and excessive secretions of gastric acid in primary care clinics in Taiwan. A previous study indicated that most frequent anticholinergic drugs were ranitidine, trazodone, paroxetine, oxybutynin, and nortriptyline [19]. Those were used to treat cardiovascular and neurological disorders. This indicates that there is a wide disparity in the prescription pattern of anticholinergic drugs between other countries and Taiwan. It may be related to the difference in medical care-seeking behavior caused by the health care system.
Pneumonia is a life-threatening infectious disease where age is one of the strongest risk factors. Our study demonstrates that anticholinergic drugs are related to incident pneumonia in the elderly. A population-based case-control study by Paul et al. [10] expressed that anticholinergic medication use is associated with pneumonia risk, compared with no use. Moreover, both acute and chronic use of anticholinergic drugs were associated with a higher risk for pneumonia, whereas there was no association with previous use. Another case-control study in America by Chatterjee et al. [6] estimated anticholinergic drug exposure and the risk of pneumonia. The overall use of anticholinergic drugs was associated with the risk of pneumonia, while the effect of anticholinergic drugs was not related to pneumonia. It indicated that even low-potency anticholinergic drugs may cause a significant risk of pneumonia. A nationwide study in Finland had a similar result [18]. Most of the low-potency anticholinergic drugs are cardiovascular drugs, and the risk may be partially explained by comorbidity. The main purpose of our study was to explore the potential relationship between the risk of pneumonia and anticholinergic drug use in the elderly in Taiwan. To enhance the accuracy of the study results, our study used PSM to obtain control groups for each elderly patient receiving anticholinergic drugs. A propensity score is a unit with certain characteristics that is assigned to each elderly patient receiving anticholinergic drugs. The scores can be used to reduce or eliminate selection bias in observational studies by the characteristics of elderly patients with and

Discussion
Our study results indicate that about 60% of elderly patients had received anticholinergic medication at least once, and the most frequent anticholinergic prescriptions in Taiwan were cimetidine, cetirizine, pseudoephedrine, metoclopramide, and ranitidine. These drugs are used to treat URI and excessive secretions of gastric acid in primary care clinics in Taiwan. A previous study indicated that most frequent anticholinergic drugs were ranitidine, trazodone, paroxetine, oxybutynin, and nortriptyline [19]. Those were used to treat cardiovascular and neurological disorders. This indicates that there is a wide disparity in the prescription pattern of anticholinergic drugs between other countries and Taiwan. It may be related to the difference in medical care-seeking behavior caused by the health care system.
Pneumonia is a life-threatening infectious disease where age is one of the strongest risk factors. Our study demonstrates that anticholinergic drugs are related to incident pneumonia in the elderly. A population-based case-control study by Paul et al. [10] expressed that anticholinergic medication use is associated with pneumonia risk, compared with no use. Moreover, both acute and chronic use of anticholinergic drugs were associated with a higher risk for pneumonia, whereas there was no association with previous use. Another case-control study in America by Chatterjee et al. [6] estimated anticholinergic drug exposure and the risk of pneumonia. The overall use of anticholinergic drugs was associated with the risk of pneumonia, while the effect of anticholinergic drugs was not related to pneumonia. It indicated that even low-potency anticholinergic drugs may cause a significant risk of pneumonia. A nationwide study in Finland had a similar result [18]. Most of the low-potency anticholinergic drugs are cardiovascular drugs, and the risk may be partially explained by comorbidity. The main purpose of our study was to explore the potential relationship between the risk of pneumonia and anticholinergic drug use in the elderly in Taiwan. To enhance the accuracy of the study results, our study used PSM to obtain control groups for each elderly patient receiving anticholinergic drugs. A propensity score is a unit with certain characteristics that is assigned to each elderly patient receiving anticholinergic drugs. The scores can be used to reduce or eliminate selection bias in observational studies by the characteristics of elderly patients with and without anticholinergic drugs [20][21][22][23]. Compared with previous studies, our study design and methodology are more rigorous. Our study was inappropriate for further analysis regarding the effect of different levels of anticholinergic drugs due to the elderly subjects in our study being matched based on whether they received anticholinergic drugs rather than different levels of anticholinergic drugs. To our knowledge, there has been no related population-based study to estimate the relationship between incident pneumonia and anticholinergic drug use in Taiwan. Our study was not aimed at exploring the mechanisms between anticholinergic drugs and increased pneumonia risk, but the association between anticholinergic drugs and pneumonia in the elderly was clear. Although the incident rate of pneumonia caused by anticholinergic drugs is relatively rare, future studies should develop more well-designed prospective population trials to be carried out to ascertain the relationship.
The comorbidities may also predispose elderly patients to an increased risk of pneumonia. Our study results were consistent with previous studies that noted that comorbidities, including DM, stroke, PD, CKD, asthma, COPD, HF, URI, and GERD, were risk factors of pneumonia. The elderly with COPD or asthma are a high-risk population of bacterial infection, and one of the major infections is caused by streptococcus pneumoniae bacteria. It can cause many types of illnesses that contained pneumonia, meningitis, and septicemia. Therefore, COPD and asthma are both associated with a greater risk of incident pneumonia in the elderly [24][25][26]. DM may also increase the pneumonia risk in the elderly due to poor glycemic control. Aging is associated with a progressive decline in respiratory system function, and poor glycemic control can cause microvascular complications of lung capillaries [27][28][29]. Moreover, previous studies have documented that the elderly with stroke, dementia, or Parkinson's disease have a higher risk of incident pneumonia [26,27,30]. It was possibly attributable to dysphagia, difficulty swallowing, and cough reflexes. Oropharyngeal dysphagia is also one of the risk factors for pneumonia in elderly patients [31][32][33]. Renal disease in the elderly is clinically important. Previous studies have described that there is a positive direction in the association between predialysis CKD and acute community-acquired infection [34]. Hypoimmunity of the elderly patient with CKD may predispose the patient to acute lower respiratory infections. Another study demonstrated that cardiovascular function declined in elderly patients with heart disease, which would affect the mucociliary clearance functions that trap and remove particulates and pathogens from the airways, leading to an increased risk of incident pneumonia [35]. Furthermore, a population-based cohort study indicated that GERD is associated with a long-term risk of pneumonia, especially in GERD with proton pump inhibitors (PPIs) for a longer treatment than four months [36]. There is a relationship between gastric acid suppressants and an increased risk of pneumonia [37]. Authors have suggested that the inhibition of gastric acid secretion by acid-suppressive therapy allows pathogen colonization from the upper gastrointestinal tract [24].
The major strength of the study is that it is based on a nationwide database, avoiding bias such as selection, nonresponse, and poor recall. Big data analysis is a new trend in modern healthcare research. LHID has completeness in recording prescriptions and clinical diagnosis. Moreover, the study was not only adjusted for potential confounding factors but also used PSM to avoid bias in the selection of study subjects. Therefore, the study indicates the correlation between anticholinergic drugs and pneumonia with a narrower and statistically significant confidence interval.
There were also a few limitations to the study. Some pneumonia-related variables, such as medication adherence, tobacco consumption behavior, and laboratory parameters, cannot be obtained from LHID. Additionally, the study only used the ICD code to define disease without any medical procedure codes. There may be overdiagnosis. Although our study indicated the correlation between anticholinergic drugs and pneumonia, the strength of the anticholinergic effect may also be related to pneumonia. We will explore the risk of incident pneumonia with different anticholinergic drugs in further study. Finally, this study is a type of observational study that analyzes data from a population database. The study result can only provide evidence to demonstrate that anticholinergic drugs are related to incident pneumonia. It is essential to obtain more information from other databases to analyze the cause-effect relation in future research.

Conclusions
The use of anticholinergic drugs is common in the elderly in Taiwan because these medications are prescribed for the symptomatic management of medical conditions. Elderly patients receiving anticholinergic drugs may increase their risk of incident pneumonia. Moreover, DM, stroke, PD, CKD, asthma, COPD, HF, URI, and GERD were also associated with incident pneumonia in the elderly. Anticholinergic drugs can have many beneficial effects, but these drugs need to be balanced against potential harm. Drugs with anticholinergic properties can be problematic, especially for the elderly population. Prescribers should consider dose reductions and monitoring when prescribing anticholinergic drugs to elderly patients to reduce the risks of adverse outcomes. Furthermore, randomized clinical trials are warranted to determine the effectiveness and safety of anticholinergic drugs in the elderly.