High exposure compared with standard exposure to metoclopramide associated with a higher risk of parkinsonism: a nationwide population‐based cohort study

Abstract

Aims

We conducted a cohort study utilizing a nationwide health insurance database to assess the European Medicines Agency's restrictions on using metoclopramide and its association with the risk of parkinsonism.

Methods

New oral metoclopramide users aged ≥20 years, and age‐ and gender‐matched non‐users were recruited between 2001 and 2011. Users were divided into high‐exposure (dose >30 mg day^–1 and/or duration >5 days) and standard‐exposure (dose ≤30 mg day^–1 and duration ≤5 days) groups. The adjusted hazard ratio (aHR) with 95% confidence interval (CI) estimated the risk of parkinsonism.

Results

During a 1‐year period, 122 of 218 931 (0.06%) users of metoclopramide vs. 56 of 218 931 (0.03%) non‐users developed parkinsonism (P < 0.001). Among the 122 cases of parkinsonism in users, 64 (0.04%) were from 168 566 standard‐exposure users and 58 (0.12%) from 50 365 high‐exposure users. Compared with non‐users, the risk of parkinsonism was higher in users (aHR 2.16; 95% CI 1.54, 3.02), including standard‐exposure (aHR 1.73; 95% CI 1.11, 2.70), and high‐exposure (aHR 3.15; 95% CI 1.78, 5.57) users. High‐exposure users had a higher risk of parkinsonism than standard‐exposure users (aHR 1.83; 95% CI 1.28, 2.63). Within the high‐exposure group, 45 233 of 50 365 (89.81%) users and 55 of 58 (94.83%) parkinsonism were from long‐duration exposure; 5 132 of 50 365 (10.19%) users and 3 of 58 (5.17%) parkinsonism were from high‐dose exposure and long‐duration + high‐dose exposure.

Conclusions

The risk of parkinsonism in metoclopramide users, although extremely low (0.06%), is 2.16‐fold greater than in non‐users. High‐exposure users have a 1.83‐fold higher risk than standard‐exposure users. As users in high‐exposure group had a higher risk of parkinsonism than in standard‐exposure group, and the majority of users and parkinsonism in high‐exposure group were from long‐duration exposure; thus, physician are advised to avoid prescribing metoclopramide for >5 days, even if the daily dose is ≤30 mg.

What is Already Known about this Subject

• Metoclopramide is a common cause of drug‐induced parkinsonism; however, the association between metoclopramide treatment and the risk of parkinsonism has never been assessed.

• The treatment doses and durations causing parkinsonism in reported metoclopramide users vary.

• In 2013, the European Medicines Agency (EMA) enacted restrictions for limiting the maximum metoclopramide dose and duration to 30 mg per day and 5 days, respectively, to reduce neurological side effects, including parkinsonism.

What this Study Adds

• We confirmed the positive association between the use of metoclopramide and a higher risk of parkinsonism, based on a nationwide population‐based cohort study.

• We verified the association between higher metoclopramide exposure and a greater risk of developing parkinsonism.

• We provided evidence to support the EMA's restrictions to avoid prescribing metoclopramide for longer than 5 days, even if the daily dose is no more than 30 mg.

Introduction

Parkinsonism is a clinical syndrome that encompasses a combination of different movement abnormalities, such as tremor, akinesia and rigidity 1, 2, 3. Among various causes of parkinsonism, drug‐induced parkinsonism (DIP) is the second most common, after idiopathic Parkinson's disease (IPD) 2, 4. DIP usually happens within the first 3 months after administering the causative drugs but may occur as late as 12 months later 1. Drugs that may be associated with the higher risk of parkinsonism include antipsychotic agents, calcium channel blockers, valproate, lithium and antiemetic agents 1, 2, 3, 4, 5, 6, 7.

http://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=241 is an antiemetic agent commonly used in patients experiencing gastro‐oesophageal reflux disease, receiving chemotherapy or undergoing adjunctive radiotherapy 8, 9; however, it has antidopaminergic properties and can be a cause of DIP 5, 10, 11, 12. In previous case reports, the intake of metoclopramide at doses of 20–40 mg per day for 2–10 months could cause parkinsonism 5, 10. In another case series study involving 10 cases (seven females, three males), aged 44–71 years (seven cases >60 years, three cases <60 years), eight and two patients developed parkinsonism within and beyond 1 year, respectively, after the administration of metoclopramide at a daily dose of 30 mg for 3 weeks to 19 months 11. After discontinuing use of the drug, all parkinsonian symptoms disappeared within 2 weeks to 12 months 5, 11. The incidence of parkinsonism related to metoclopramide may increase with age and differ between men and women. Older women, aged ≥60 years, are more likely to develop parkinsonism than age‐matched men after metoclopramide therapy 5, 11, 12.

In 2009, the Food and Drug Administration issued safety alerts concerning the hazards associated with high‐dose and/or long‐duration use of metoclopramide 13, 14. Accordingly, to reduce the potential risk of neurological side effects, such as parkinsonism, in 2013 the European Medicines Agency (EMA) enacted restrictions on using metoclopramide that limited the maximum dose to 30 mg per day and the duration to 5 days 14, 15. Based on the 2013 report of a postmarket surveillance system in Taiwan, metoclopramide has been listed among the top five causes of adverse drug reactions, including parkinsonism 16. However, most of the current data depicting the association between the intake of metoclopramide and the development of parkinsonism are derived from case reports or spontaneous reporting with small sample sizes 5, 10, 11, 17, 18, 19. Cohort studies focusing solely on the safety concerns of metoclopramide associated with parkinsonism are rare.

The present nationwide population‐based cohort study, centred on new oral metoclopramide users aged ≥20 years, aimed to examine the association between the use of metoclopramide and the risk of parkinsonism. Furthermore, based on the EMA's restrictions of high (dose >30 mg day^–1 and/or duration >5 days) and standard (dose ≤30 mg day^–1 and duration ≤5 days) exposure to metoclopramide, the study assessed whether high‐exposure users have a higher risk of parkinsonism compared with standard‐exposure users.

Methods

Data source

Data obtained from the Longitudinal Health Insurance Database 2005 (LHID 2005), a subset of the Taiwan National Health Insurance Research Database (NHIRD), formed the basis of the present retrospective cohort study. LHID 2005 includes data for 1 000 000 insured individuals sampled randomly from the 2005 Registry of Beneficiaries of the NHIRD and contains longitudinal healthcare utilization data for each patient from 1 January 2000 to 31 December 2012. Taiwan's National Health Research Institute established NHIRD in 1996 and reported that there were no statistically significant differences in the distributions of age, gender or healthcare costs between patients in the LHID 2005 and the original NHIRD. As one of the largest databanks of medical information in the world, NHIRD provides a unique opportunity to assess the risk factors for various diseases. The application of NHIRD in risk factor evaluation for Parkinson's disease (PD) has been validated in a number of previous studies 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34.

The study protocol (No: 201403012) was approved by the Institutional Review Board. Informed consent for each participant was waived because this study utilized the declassified secondary data that have been released to the public for research purposes.

Study population

The diagnostic codes assigned by the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD‐9‐CM) were used to define patients' diagnoses, and the Anatomical Therapeutic Chemical classification system was applied to ascertain the prescribed medications (Tables S1 and S2).

Adult subjects aged ≥20 years and administered metoclopramide for the first time between 1 January 2001 and 31 December 2011 were selected from the LHID 2005. Only new users who received their prescriptions of metoclopramide dispensed as oral solid dosage forms (e.g. tablets and capsules) from ambulatory care services were chosen. The index date was defined as the date when an eligible metoclopramide user was recruited. For each user included in the study cohort, one age‐ and gender‐matched non‐user was randomly selected from the remaining subjects in the LHID 2005 to be in the comparison cohort. The matched non‐users were assigned the same index date as their corresponding users, to ensure that users and non‐users alike were observed for the same duration. Users and non‐users were excluded if, 1 year prior to the index date, they had: (i) bipolar disorder; (ii) a schizophrenic disorder or other psychotic condition; (iii) cerebrovascular disease; (iv) dementia; (v) acute kidney failure or chronic kidney disease; (vi) PD or any other PD‐related disorders; (vii) prescriptions of metoclopramide or antiparkinsonian drugs (APDs) 1, 2, 4, 6, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 35; and/or (viii) been prescribed concomitant medications associated with a high to moderate risk of parkinsonism, including http://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=215 blockers (antipsychotic agents), dopamine depleters, calcium channel blockers, valproate and lithium (Table S2). APDs include levodopa, anticholinergic agents, amantadine, dopamine agonists (DAs), monoamine oxidase B inhibitors and catechol‐O‐methyltransferase inhibitors 3, 20, 21, 22, 30, 35, 36, 37.

High vs. standard exposure to metoclopramide

Based on the EMA's restrictions on the use of metoclopramide, users were divided into high‐exposure (daily dose >30 mg and/or prescription duration >5 days) and standard‐exposure (daily dose ≤30 mg and prescription duration ≤5 days) groups. High‐exposure users included long‐duration exposure (prescription duration >5 days and daily dose ≤30 mg), high‐dose exposure (daily dose >30 mg and prescription duration ≤5 days) and long‐duration + high‐dose exposure (prescription duration >5 days and daily dose >30 mg). Additionally, information concerning the administration of metoclopramide, including the prescription duration and the daily dose, were calculated. First, the prescription duration for metoclopramide = total tablets (capsules) in each prescription divided by tablets (capsules) in each day. Second, the daily dose of metoclopramide = active ingredient in each tablet (capsule) multiplied by tablets (capsules) in each day.

Endpoint event

The endpoint event was defined as during a 1‐year follow‐up, participants were diagnosed with parkinsonism in either an outpatient visit or a hospital admission that allowed a temporary prescription of at least one of the APDs. We used the ICD‐9‐CM codes and pharmacy data to identify and distinguish cases of parkinsonism and IPD 30, 34, 35, 36, 37. The ICD‐9‐CM diagnostic codes that could incorporate all possible parkinsonian features, including 332 (PD), 332.0 (paralysis agitans), 332.1 (secondary parkinsonism) and 333 (other extrapyramidal diseases and abnormal movement disorders), were used to indicate the diagnosis of parkinsonism 34, 35, 36. A temporary prescription of at least one of the APDs for a total duration of ≤4 weeks during a 1‐year follow‐up was allowed further to support the assertion that identified parkinsonism cases were DIP and make the diagnosis of IPD unlikely 37. All participants were observed from the index date to the date when the diagnosis of parkinsonism was made or until a full 1‐year period had been reached. Participants who failed to complete the full period of follow‐up were considered their participation in the study to end on the day of their last healthcare service visit, at the time of death or on 31 December 2012, which was the last day of the study.

Statistical analyses

Descriptive statistics were used to describe the baseline characteristics of participants enrolled in the study. A conditional Cox proportional hazards regression model was applied to compare the risks of parkinsonism between users, including high‐ and standard‐exposure users, and matched non‐users. The proportional hazards assumption was checked by using Schoenfeld residuals and estat phtest, and these were not violated for any covariates. An unconditional analysis was carried out to compare the difference in risk between high‐ and standard‐exposure users. Potential covariates in regard to the higher risk of developing parkinsonism, which consisted of hypertension, diabetes mellitus, hyperlipidaemia, ischaemic heart disease, epilepsy, traumatic brain injury and depression, were included in the regression analyses. The cumulative incidences of parkinsonism between users and non‐users and between high‐ and standard‐exposure users were evaluated by the Kaplan–Meier method and the log‐rank test.

All statistical analyses were performed using SAS statistical software version 9.3 (SAS Institute, Cary, NC, USA) and STATA 13 software (Stata Corp, College Station, TX, USA). A value of P < 0.05 in a two‐tailed test was considered statistically significant.

Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY 38, and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 39.

Results

From 1 January 2001 to 31 December 2011, 226 848 new metoclopramide users were identified. After excluding 7917 users owing to unknown prescription information, a total of 218 931 metoclopramide users and 218 931 age‐ and gender‐matched non‐users were recruited for analysis (Table 1). Of the participants in the user and non‐user cohorts, 126 728 (57.88%) were female, the median age was 41 years (interquartile range 30–53 years) and 182 578 (83.39%) were <60 years of age. Compared with non‐users, the prevalence of comorbid medical conditions, including hypertension, diabetes mellitus, hyperlipidaemia, ischaemic heart disease, epilepsy, traumatic brain injury and depression, was higher in users (all P < 0.001).

Table 1.

Comparisons of characteristics and comorbidities between metoclopramide users and non‐users

Characteristicsa	Users (n = 218 931)		Non‐users (n = 218 931)		P‐value
	n	%	n	%
Gender					1.00
Female	126 728	57.88	126 728	57.88
Male	92 203	42.12	92 203	42.12
Median age (IQR), years	41	(30–53)	41	(30–53)	1.00
Age groups					1.00
20–39 years	102 702	46.91	102 702	46.91
40–59 years	79 876	36.48	79 876	36.48
60–79 years	32 306	14.76	32 306	14.76
≥80 years	4047	1.85	4047	1.85
Comorbidities
Hypertension	26 527	12.12	19 462	8.89	<0.001
Diabetes mellitus	13 848	6.33	9279	4.24	<0.001
Hyperlipidaemia	16 069	7.34	11 017	5.03	<0.001
Ischaemic heart disease	8514	3.89	5038	2.30	<0.001
Epilepsy	428	0.20	302	0.14	<0.001
Traumatic brain injury	5331	2.44	3651	1.67	<0.001
Depression	3806	1.74	1654	0.76	<0.001

^aValues expressed as a number (n) and percentage (%) unless otherwise indicated. IQR, interquartile range

During a 1‐year follow‐up period (Table 2), 122 (0.06%) users and 56 (0.03%) non‐users were diagnosed with parkinsonism. Users were more likely to develop parkinsonism than non‐users (P < 0.001). For parkinsonism cases in users, compared with non‐users, the median age was slightly older (72.90 vs. 69.80 years), and the median onset time was slightly shorter (4.39 vs. 6.23 months), although the differences were statistically insignificant. Among all parkinsonism cases, 64 of 122 (52.46%) users and 25 of 56 (44.64%) non‐users were prescribed APDs. Although the difference was not statistically significant, users with parkinsonism were more likely to be administered APDs than non‐users with this condition. Among APDs, levodopa was the most commonly prescribed, accounting for more than half of APD‐prescribed cases in users (37 of 64; 57.81%) and non‐users (15 of 25; 60.00%). Anticholinergic agents, amantadine and DAs were prescribed with decreasing frequency, respectively, after levodopa both for users and non‐users. When comparing APDs prescribed for users vs. non‐users, anticholinergic agents were prescribed more frequently for users (31.25% vs. 28.00%). By contrast, levodopa, amantadine and DAs were more commonly given to non‐users (57.81% vs. 60.00%; 23.44% vs. 28.00%; and 4.69% vs. 8.00%, respectively). However, these comparisons remained statistically insignificant.

Table 2.

Comparisons of parkinsonism cases and antiparkinsonian drugs between metoclopramide users and non‐users

Outcomea	Users		Non‐users		P‐value
	(n = 218 931)		(n = 218 931)
Parkinsonism cases (n, %)	122	0.06	56	0.03	<0.001
Median age (IQR), years	72.90	(65.20–79.70)	69.80	(57.55–78.75)	0.28
Median onset time (IQR), months	4.39	(2.75–8.39)	6.23	(2.93–7.57)	0.41
APD‐prescribed cases b (n, %)	64	52.46	25	44.64	0.33
Levodopa c (n, %)	37	57.81	15	60.00	0.85
Anticholinergic agents c (n, %)	20	31.25	7	28.00	0.76
Amantadine c (n, %)	15	23.44	7	28.00	0.65
Dopamine agonists c (n, %)	3	4.69	2	8.00	0.54

APD, antiparkinsonian drug; IQR, interquartile range

^aValues were expressed as number and percentage unless otherwise indicated

^bNumber and percentage of cases in which APDs (including levodopa, anticholinergic agents, amantadine and dopamine agonists) were prescribed was determined on the basis of parkinsonism cases in users and in non‐users

^cThe number and percentage of cases prescribed individual APDs were determined on the basis of APD‐prescribed cases in users and in non‐users

The incidence of parkinsonism was higher in users than in non‐users (Table 3), indicating that users had a greater risk of parkinsonism compared with non‐users after adjusting for gender, age and comorbidities [adjusted hazard ratio (aHR) 2.16; 95% confidence interval (CI) 1.54, 3.02; P < 0.001]. When stratified by exposure level, standard‐exposure users were associated with an elevated risk of parkinsonism (aHR 1.73; 95% CI 1.11, 2.70; P = 0.02), and this risk was even more elevated in high‐exposure users (aHR 3.15; 95% CI 1.78, 5.57; P < 0.001) when compared with the corresponding non‐users. When stratified by gender, the risk of parkinsonism was increased similarly in female (aHR 2.20; 95% CI 1.36, 3.58; P = 0.001) and male (aHR 2.05; 95% CI 1.27, 3.32; P = 0.004) users when compared with gender‐corresponding non‐users. Moreover, users had a higher risk of parkinsonism than non‐users in the 60–79‐year (aHR 2.48; 95% CI 1.51, 4.08; P < 0.001) and ≥80‐year (aHR 2.08; 95% CI 1.06, 4.08; P = 0.03) age groups, as well as in the multimorbidity group (aHR = 3.00; 95% CI 1.47, 6.14; P = 0.003).

Table 3.

Incidences and adjusted hazard ratios of parkinsonism stratified by gender, age and comorbidities in metoclopramide users and non‐usersa

	Users (n = 218 931)			Non‐users (n = 218 931)			Users compared with non‐usersb
Outcome	P (n) 122	PYs 218 713	IR 0.56	P (n) 56	PYs 218 102	IR 0.26	HR	95% CI	P value	aHR	95% CI	P value
Overall	122	218, 713	0.56	56	218 102	0.26	2.18	1.59–2.99	<0.001	2.16	1.54–3.02	<0.001
Standard exposure	64	168 414	0.38	38	167 948	0.23	1.68	1.13–2.52	0.01	1.73	1.11–2.70	0.02
High exposure	58	50 299	1.15	18	50 154	0.36	3.22	1.90–5.47	<0.001	3.15	1.78–5.57	<0.001
Gender
Female	61	126 608	0.48	27	126 255	0.21	2.26	1.44–3.55	<0.001	2.20	1.36–3.58	0.001
Male	61	92 105	0.66	29	91 847	0.32	2.10	1.35–3.27	0.001	2.05	1.27–3.32	0.004
Age groups
20–39 years	9	102 626	0.09	8	102 282	0.08	1.13	0.43–2.92	0.81	0.88	0.32–2.41	0.80
40–59 years	16	79 814	0.20	7	79 692	0.09	2.29	0.94–5.56	0.07	2.03	0.67–6.18	0.21
60–79 years	68	32 242	2.11	27	32 172	0.84	2.52	1.61–3.93	<0.001	2.48	1.51–4.08	<0.001
≥80 years	29	4031	7.19	14	3956	3.54	2.07	1.09–3.92	0.03	2.08	1.06–4.08	0.03
Multimorbidity
No	49	169 803	0.29	32	185 053	0.17	1.53	0.86–2.72	0.15	1.53	0.86–2.72	0.15
Yes	73	48 910	1.49	24	33 049	0.73	3.00	1.47–6.14	0.003	3.00	1.47–6.14	0.003

CI, confidence interval; PY, person‐year; IR, incidence rate per 1000 PYs; HR, hazard ratio; aHR, adjusted HR

^aP (n) = parkinsonism cases (number)

^bHR and 95% CI were estimated by a conditional Cox proportional hazards regression model. aHR and 95% CI were adjusted for gender, age and comorbidities of hypertension, diabetes mellitus, hyperlipidaemia, ischaemic heart disease, epilepsy, traumatic brain injury and depression

Out of 218 931 metoclopramide users, 50 365 (23.00%) were in the high‐exposure group and 168 566 (77.00%) were in the standard‐exposure group (Table 4). Among the high‐exposure users, 45 233 (89.81%) had long‐duration exposure (prescription duration 11.37 ± 7.23 days and daily dose 13.05 ± 5.30 mg), 4876 (9.68%) had high‐dose exposure (daily dose 36.43 ± 6.21 mg and prescription duration 2.79 ± 0.66 days) and 256 (0.51%) had long‐duration + high‐dose exposure (prescription duration 9.28 ± 5.06 days and daily dose 47.57 ± 12.11 mg). In standard‐exposure users, the prescription duration was 2.93 ± 0.63 days and the daily dose was 15.34 ± 6.17 mg. Among the 122 parkinsonism cases in metoclopramide users, 58 were in the high‐exposure group (58 of 50 365; 0.12%) and 64 were in the standard‐exposure group (64 of 168 566; 0.04%). Compared with standard‐exposure users, high‐exposure users had a higher risk of parkinsonism (aHR 1.83; 95% CI 1.28, 2.63; P = 0.001). Furthermore, among the 58 parkinsonism cases in high‐exposure users, 55 (94.83%) had long‐duration exposure, two (3.45%) had high‐dose exposure and one (1.72%) had long‐duration + high‐dose exposure. Long‐duration exposure accounted for the majority of parkinsonism cases (55 of 58; 94.83%) in high‐exposure users. Parkinsonism cases in the high‐dose exposure and long‐duration + high‐dose exposure users were relatively rare (3 of 58; 5.17%).

Table 4.

Parkinsonism cases (n = 122) in groups and subgroups of metoclopramide users (n = 218 931)a

Groups	High exposure			Standard exposure
[n (%)]b	50 365 (23.00)			168 566 (77.00)
Subgroups	Long duration	High dose	Long duration + High dose
[n (%)] c	45 233, (89.81)	4876 (9.68)	256 (0.51)
Duration, days d	11.37 ± 7.23	2.79 ± 0.66	9.28 ± 5.06	2.93 ± 0.63
Daily dose, mg d	13.05 ± 5.30	36.43 ± 6.21	47.57 ± 12.11	15.34 ± 6.17
Parkinsonism cases [n (%)] e		58 (0.12)		64 (0.04)
[n (%)] f	55 (94.83)	2 (3.45)	1 (1.72)

^aGroups included high‐exposure and standard‐exposure groups of metoclopramide users. Subgroups included long‐duration, high‐dose and long‐duration + high‐dose users in high‐exposure group. [n (%)] = number and percentage; Duration = prescription duration

^bNumber and percentage of subjects in each group of metoclopramide users

^cNumber and percentage of subjects in each subgroup of high‐exposure group

^dValues of prescription duration (days) and daily dose (mg) were expressed as mean ± standard deviation

^eNumber and percentage of parkinsonism cases in each group of metoclopramide users. The risks of parkinsonism were higher in users in the high‐exposure group compared with the standard‐exposure group (adjusted hazard ratio 1.83; 95% confidence interval 1.28, 2.63; P = 0.001) based on an unconditional analysis and adjusting for age, gender and comorbidities of hypertension, diabetes mellitus, hyperlipidaemia, ischaemic heart disease, epilepsy, traumatic brain injury and depression

^fNumber and percentage of parkinsonism cases in each subgroup of the high‐exposure group

The cumulative incidence of parkinsonism was higher in metoclopramide users than in non‐users and was also higher in users with high exposure than in users with standard exposure to the drug (log‐rank test, both P < 0.001) (Figure 1).

Figure 1.

Cumulative incidence of parkinsonism between metoclopramide users and non‐users (A) and between high‐exposure (H‐EXP) users and standard‐exposure (S‐EXP) users (B) during a 12‐month follow‐up period

A sensitivity analysis using dental trauma (ICD‐9‐CM: 873.63 and 873.73) as a negative tracer was performed to strengthen our results. The numbers of users and non‐users who developed dental trauma were 90 (0.04%) and 79 (0.04%), respectively, indicating that the use of metoclopramide was not significantly associated with a higher risk of dental trauma (P = 0.397).

Discussion

The present retrospective cohort study was based on the data for oral metoclopramide users and non‐users from a large longitudinal database. It led to the conclusion that, although the absolute risk is exceeding small (0.06%), users are associated with a 2.16‐fold higher risk of parkinsonism than non‐users in a population of Taiwanese adults aged ≥20 years. Further assessment verified the association between higher metoclopramide exposure and a greater risk of developing parkinsonism. Compared with non‐users, high‐ and standard‐exposure users had a 3.15‐fold and 1.73‐fold higher risk of parkinsonism, respectively. High‐exposure users had a 1.83‐fold higher risk for parkinsonism compared with standard‐exposure users.

In high‐exposure users, either long‐duration exposure or high‐dose exposure may play an important role in increasing the risk of parkinsonism. In the present study, long‐duration exposure accounted for the majority of high‐exposure users (89.81%), while high‐dose exposure and long‐duration + high‐dose exposure were less frequent (10.19%). Additionally, parkinsonism cases in high‐exposure users were mostly from long‐duration exposure (94.83%), with few cases from high‐dose exposure and long‐duration + high‐dose exposure (5.17%). Accordingly, our results support the EMA's restrictions that limit the maximum duration of metoclopramide treatment to 5 days. Long‐duration exposure (11.37 ± 7.23 days and 13.05 ± 5.30 mg day^–1) mainly accounts for a greater risk of parkinsonism in high‐exposure users; the elevated risk of parkinsonism from high‐dose exposure (36.43 ± 6.21 mg day^–1 and 2.79 ± 0.66 days) and long‐duration + high‐dose exposure (9.28 ± 5.06 days and 47.57 ± 12.11 mg day^–1) was not evident because of the relatively small sample size. In Taiwan's pharmaceutical market, the amount of active ingredient in most metoclopramide preparations is less than 5 mg 40. This fact may explain why most users in the present population‐based study were administered low‐dose drugs; high‐dose administration was relatively infrequent. In addition, because the administration of metoclopramide is among the top five leading causes of adverse drug reactions in Taiwan 16, physicians may be aware of this information and avoid high‐dose administration; however, they may be unaware that a long‐duration prescription (>5 days) of metoclopramide, even using the standard daily dose (≤30 mg), may also be associated with a higher risk of parkinsonism.

Metoclopramide is a dopamine receptor antagonist. Unlike the presynaptic dopamine deficiency in IPD, DIP that is related to metoclopramide is believed to be caused by the pharmacological blockage of the postsynaptic dopamine (D2) receptor 5, 8, 9, 13 and may respond poorly to dopaminergic therapy 41. Anticholinergic agents and amantadine have anticholinergic properties that balance the action between dopamine and acetylcholine and are therefore the first‐line therapy for DIP in metoclopramide users; the administration of levodopa and/or DAs is controversial 2, 4, 6, 42. In the present study, although anticholinergic agents were prescribed more frequently for users than non‐users, levodopa was the most commonly prescribed drug both for users and non‐users. This is because DIP that is caused by metoclopramide may mimic IPD and may be frequently misdiagnosed and treated as IPD 42. Moreover, parkinsonism cases in the present study occurred mostly in older subjects, with a median onset age of 72.90 years in users and 69.80 years in non‐users. In older individuals, levodopa is usually the drug of choice for initial IPD therapy 6, 21. Additionally, various combinations of parkinsonian symptoms among individual DIP cases 42, and various prescription patterns for APDs among different specialists may affect how physicians choose appropriate drugs 43.

In previous case reports, DIP related to metoclopramide occurred more commonly in older female patients 5, 11, 12. In the present population‐based large cohort study, however, the incidence of parkinsonism cases in metoclopramide users did not differ between men and women. Moreover, although the majority of participants (182 578 of 218 931; 83.39%) were <60 years of age, most parkinsonism cases (97 of 122; 79.50%) occurred in metoclopramide users aged ≥60 years and in those with multiple comorbidities. This is because dopaminergic neurones are progressively lost during the ageing process. In addition, the comorbidities may increase the risk of IPD, and some may even precede the onset of the motor features of parkinsonism 44. Therefore, people aged 60 and older and with multimorbidity are more susceptible to the side effects of metoclopramide and prone to develop parkinsonism when administered the drug 6, 44. However, because the prevalence of multimorbidity was higher in metoclopramide users than in non‐users, the existence of potential indication bias cannot be ignored 45.

The scope of the present study extended beyond that of previous case reports, which have lacked covariate adjustment and been concerned solely with parkinsonism as a potential side effect of metoclopramide 46. By contrast, the present study included age‐ and gender‐matched non‐users for comparison, and adjusted for medical comorbidities to avoid over‐detecting cases of parkinsonism in metoclopramide users, who may be more likely than non‐users to develop this condition as a result of having more comorbidities and seeking more healthcare services. To obtain more accurate information regarding metoclopramide prescription duration, this was calculated by dividing the total metoclopramide amount in one prescription by the daily amount used, rather than using the number of drug prescription days listed in ambulatory and pharmacy claims. As the number of drug prescription days represents the longest duration of drug treatment within one prescription, using the calculated duration of the metoclopramide prescription may prevent overestimation of the duration for short‐term users. Moreover, in preclinical IPD, nonmotor gastrointestinal symptoms, such as constipation, can be evident before the occurrence of classical motor symptoms. Use of metoclopramide in these cases may unmask the manifestations of preclinical IPD; parkinsonism cases after the use of metoclopramide could be IPD rather than DIP. IPD needs long‐term treatment with APDs. To avoid the possibility of protopathic bias 47, we therefore focused on short‐duration (total duration ≤4 weeks within 1 year) administration of APDs, to exclude IPD and thus diagnose DIP with more certainty 30, 35, 36, 37. Discontinuation of metoclopramide may have been due to other reasons, such as improvement in gastrointestinal symptoms and loss to follow‐up, rather than the development of parkinsonism; thus, this was not considered as an endpoint event 48.

Several limitations may have affected the results of the present study. As the declassified secondary data utilized could not be linked to the medical records of healthcare services, personal medical data on personal history, genetic profile and family history of PD, as well as medication compliance and renal function, were not available in the database. The absence of these data may have negatively influenced physicians' diagnoses of DIP related to metoclopramide therapy 2, 4, 6, 49. Moreover, the absolute risk of parkinsonism associated with the use of metoclopramide in the current study was extremely low (0.06%) and most parkinsonism cases occurred in users aged ≥60 years and with multimorbidity. These facts may have made it difficult to establish a direct causal relationship between the use of metoclopramide and development of parkinsonism. Despite the fact that metoclopramide is an antidopaminergic agent, additional factors that could have affected individuals’ risk of parkinsonism, such as genetic predisposition 49, the presence of multimorbidity 44 and the degree of dopaminergic neurone loss while taking metoclopramide 27, may have been present. Furthermore, there was no assessment of data on metoclopramide users’ use of over‐the‐counter medicines, inpatient care services and parenteral administrations, or of users who developed parkinsonism beyond 1 year after drug administration. These unobtainable data may have caused underestimation of the occurrence of parkinsonism in metoclopramide users. Finally, most participants included in the study were ethnic Taiwanese, which may have limited the applicability of the results to other populations 31.

Conclusions

The present nationwide population‐based cohort study showed a positive association between use of oral metoclopramide and a higher risk of parkinsonism. Parkinsonism is a potential side effect of metoclopramide in both standard‐ and high‐exposure users. High‐exposure users have a higher risk compared with standard‐exposure users. Although the absolute risk is very small, users of older age and with multimorbidity are at a higher risk. As users in high‐exposure group had a higher risk of parkinsonism than in standard‐exposure group, and the majority of users and parkinsonism in high‐exposure group were from long‐duration exposure; thus, physicians are advised to avoid prescribing metoclopramide for >5 days, even if the daily dose is ≤30 mg.

Competing Interests

There are no competing interests to declare.

The study was supported in part by a grant (102 SHH‐HCP‐07) from Shuang Ho Hospital, Taipei Medical University. The authors thank Professor C.C. Chiueh for editing this manuscript and Professor. C.Y. Yeh for assisting with statistical analyses. The interpretations and conclusions presented in this study are those of the authors and do not represent the views of the Bureau of National Health Insurance, Department of Health or National Health Research Institute, Taiwan.

Contributors

S.‐C.T., R.‐Y.Y., S.‐Y.S. and L.‐N.C. conceived and designed the study. S.‐C.T., R.‐Y.Y. and H.‐C.L. were responsible for data acquisition. S.‐C.T., R.‐Y.Y., L.‐N.C. and H.‐C.L. analysed and interpreted the data. S.‐C.T., R.‐Y.Y., S.‐Y.S. and E.J.‐S.Y. drafted the article. R.‐Y.Y. and S.‐C.T. critically revised the article for important intellectual content. R.‐Y.Y., S.‐C.T., S.‐Y.S., L.‐N.C., H.‐C.L. and E.J.‐S.Y. gave final approval for the article.

Supporting information

Tables S1 International Classification of Diseases, Ninth Revision, Clinical Modification (ICD‐9‐CM) diagnostic codes

Table S2 Anatomical Therapeutic Chemical (ATC) classification system codes

Tsai, S.‐C. , Sheu, S.‐Y. , Chien, L.‐N. , Lee, H.‐C. , Yuan, E. J.‐S. , and Yuan, R.‐Y. (2018) High exposure compared with standard exposure to metoclopramide associated with a higher risk of parkinsonism: a nationwide population‐based cohort study. Br J Clin Pharmacol, 84: 2000–2009. 10.1111/bcp.13630.