The Incidence of Benzodiazepine and Benzodiazepine‐Related Drug Use in People With and Without Parkinson's Disease—A Nationwide Cohort Study

ABSTRACT

Background

Non‐motor symptoms of Parkinson's disease (PD), including sleep problems and anxiety, are common and can appear before motor symptoms. Benzodiazepines and benzodiazepine‐related drugs (BZDR) may be used to treat these symptoms, although they are associated with an increased fall risk. It is unknown how the incidence of BZDR use changes in relation to PD diagnosis.

Objective

To investigate the incidence of BZDR use in people with and without PD from 6 years before to 10 years after the PD diagnosis.

Methods

A nationwide register‐based FINPARK cohort includes 18 651 persons diagnosed with PD in 1996–2015 and their matched comparison persons (n = 127 806). BZDR incidence was ascertained from a national prescription register. The study adheres to RECORD‐PE reporting guideline.

Findings

People with PD initiated BZDR use more frequently than the comparison people (39.0% vs. 25.9%). The incidence rate was higher in the PD group throughout the study period, and the incidence began to increase 2 years before and peaked at 6 months before the diagnosis.

Conclusion

The higher incidence of BZDR use in persons with PD already before the PD diagnosis may indicate non‐motor symptoms before the PD diagnosis. The use of fall‐risk increasing BZDR for people with PD is concerning, as the disease itself increases the risk of falling.

1. Introduction and Background

Parkinson's disease (PD) is a rapidly increasing neurodegenerative disease, and in 2016 there were 6.1 million people with PD in the world [1, 2]. Higher age is the biggest risk factor for developing PD, as the disease is uncommon in people under 50 years old [1]. In 2016, the global prevalence was highest between 85 and 89 years. Non‐motor symptoms such as depression, olfactory dysfunction and autonomic nervous system problems like constipation can precede the motor symptoms characteristic of PD [3, 4].

Disruptions in circadian rhythm and sleep disorders like rapid eye movement (REM) sleep behaviour disorder can also precede the PD diagnosis [3, 5, 6]. In the early stages of the disease, people with PD suffer from reduced REM sleep, reduced sleep efficiency and longer sleep latency [7]. In addition, disrupted sleep, early morning awakenings and excessive daytime sleepiness are common among people with PD [8, 9]. In a questionnaire study, 81.5% of PD patients reported sleep disruption, and 36.9% reported having a chronic inability to sleep [8].

Anxiety symptoms are also common among people with PD, and they can manifest already before motor symptoms set in [10]. According to a meta‐analysis, nearly a third (31%) of persons with PD had some anxiety disorder, and among them approximately half had a generalized anxiety disorder [11]. In a case series study, 25% of PD patients (n = 79) were diagnosed with anxiety and anxiety was associated with PD severity [12].

Indications for benzodiazepines include both insomnia and anxiety, whereas benzodiazepine‐related drugs are indicated for insomnia only [13]. According to Finnish guidelines for care, benzodiazepine‐related drugs and medium‐acting benzodiazepines can be used to treat severe acute insomnia, but they should not be used for longer than 2 weeks [14]. The STOPP/START criteria for prescribing drugs for older adults recommend limiting the use of benzodiazepines and benzodiazepine‐related drugs for insomnia to 2 weeks, and the use of benzodiazepines for other reasons to a maximum of 4 weeks [15]. BZDR use is associated with an increased risk of falls and fractures in older people [16, 17]. In a Danish cohort study, people with PD had increased morbidity and an increased risk of injurious falls in the 3 years prior to PD diagnosis compared with comparison people without PD [18]. As PD itself increases the risk of falling [19, 20], BZDR use can pose a problem for people with PD.

The aim of this study was to investigate the incidence of BZDR use in community‐dwelling people with and without PD from 6 years before to 10 years after the PD diagnosis.

2. Materials and Methods

2.1. Cohort Description and Data Sources

The register‐based nationwide Finnish Study on Parkinson's disease (FINPARK) cohort includes 22 189 people who received clinically confirmed PD diagnosis during 1996–2015 and were community‐dwelling at the time of diagnosis. The persons with PD diagnosis were identified from the Special Reimbursement Register maintained by the Social Insurance Institution of Finland (SII) (special reimbursement code 110). In order to apply for the special reimbursement, a medical statement of the patient's anamnesis and clinical symptoms typical of PD (e.g., rest tremor, bradykinesia and rigidity) is needed [21]. The diagnosis has to be confirmed by a neurologist. In addition to PD, also dopa‐responsive dystonia and other basal ganglia degenerative diseases can also entitle people to have this special reimbursement code. The original study sample included 29 942 people with special reimbursement code 110, but we excluded persons who did not have PD diagnosis (International Statistical Classification of Diseases and Related Health Problems (ICD) code other than G20) or were younger than 35 years at the time of diagnosis. As false diagnoses are common in the beginning of PD [22], a list of exclusion diagnoses was compiled [23], and persons with these exclusion diagnoses within ± 2 years from PD diagnosis were excluded.

Comparison persons matched by age (± 1 year), sex and region were identified from a nationwide SII register including all residents. The index date was the date of diagnosis for the matched person with PD. The comparison persons were not allowed to have purchased PD medicines (Anatomical therapeutic chemical classification [ATC]) (Code N04) or have a special reimbursement number for PD (110) in 12 months after the matched persons' PD diagnosis month. The comparison persons had to be alive and community‐dwelling during the month of the index date. The criteria for exclusion were the same as for the PD cohort, with the addition of PD dementia (ICD code F02.3). There were 148 009 comparison persons in the study.

The incidence of BZDR use was investigated from 6 years before to 10 years after the PD diagnosis. Only the first purchase was taken into consideration. The data on BZDR purchases in 1996–2016 was obtained from the prescription register. The register includes information on reimbursed drug purchases but not the data on medication used in hospitals or nursing homes. BZDRs include drugs that belong in ATC classes N05BA, N05CD and N05CF. They are further classified as benzodiazepines (N05BA and N05CD) and benzodiazepine‐related drugs, so‐called Z‐drugs (N05CF). Diazepam, alprazolam, nitrazepam, chlordiazepoxide, potassium chlorazepate and clobazam were categorized as long‐acting benzodiazepines, while oxazepam, lorazepam and temazepam were categorized as medium‐acting and midazolam and triazolam as short‐acting benzodiazepines. Zopiclone and zolpidem were categorized as benzodiazepine‐related drugs. All BZDR drugs or different package sizes were not reimbursed during the study period. This concerns mainly originator brands and small package sizes. A detailed list of non‐reimbursed products is provided in the Supporting Information table.

New BZDR users were recognized using a 1‐year washout period that began 7 years before the PD diagnosis. For the persons who were diagnosed in 2002 or before, the washout period was year 1995. The persons who had purchased BZDR during the washout period or who had been in hospital for more than 6 months or for the last 90 days of the washout period were excluded from the analyses (Figure 1).

FIGURE 1.

Flow chart. PD: Parkinson's disease; BZDR: benzodiazepine and benzodiazepine‐related drug.

The data on hospitalizations was obtained from the Care Register for Health Care. The co‐morbidities and drug use were gathered from different registers according to diagnoses and special reimbursement codes. The data on comorbidities from 1972 until the index date were obtained from the Special Reimbursement Register: asthma/chronic obstructive pulmonary disease (COPD, special reimbursement code 203), cardiovascular diseases including heart failure (201), hypertension (205), coronary artery disease (206, 213 and 280) and rheumatoid arthritis and connective tissue diseases (202). Diabetes was defined according to the special reimbursement code 103 or the purchase of diabetes medication (since 1995, ATC A10, excluding guar gum A10BX01). The history of mood disorders and schizophrenia from 1987 until the beginning of follow‐up was identified from the Care Register for Health Care with ICD‐9‐codes 296, 3004A, 3011D and 295, and ICD‐10‐codes F30–F39 and F20. In addition, the use of antidepressants (ATC‐code N06A), antipsychotics (N05A, excluding prochlorperazine N05AB04) and lithium (N05AN01) during the washout period was identified from the prescription register.

2.2. Statistical Analysis

The incidence rates (IR) of BZDR per hundred person‐years were calculated in 6‐month intervals beginning 6 years before and continuing 10 years after the index date. Incidence rate ratios (IRR) comparing the persons with and without PD were calculated from the IRs using Poisson regression. The persons who were in hospital for 120 days or longer during the 6 months were excluded from the 6‐month period because the prescription register does not have information about medications used in hospitals. The follow‐up ended with BZDR initiation, death, end of follow‐up (10 years from the diagnosis) or end of data linkage (31.12.2016). The comparison people were censored if they were diagnosed with PD. The differences in comorbidity and other categorical variables between patients with PD and comparison persons and between BZDR initiators and non‐initiators were calculated with the χ ² test. For the normally distributed continuous variable (age at the time of BZDR initiation), the t test was used. The statistical analyses were performed using STATA MP 14.0.

The data were gathered from the registers using individualized identifiers, and the researchers handled only pseudonymized data; therefore, based on Finnish legislation, ethical approval or informed consent were not required. The study was conducted in accordance with the Basic and Clinical Pharmacology and Toxicology policy for experimental and clinical studies [24]. We adhere to the reporting of studies conducted using the observational routinely collected health data statement for pharmacoepidemiology (RECORD‐PE) reporting guideline [25].

3. Results

There were altogether 146 457 people in the final study sample: 18 651 persons with PD and 127 806 comparison persons. The majority of the study population with or without PD were men (57.1%), and the mean age for the population at index date was 70 years.

Initiation of BZDR was more common in people with PD: altogether 7280 (39.0%) people with PD and 33 066 (25.9%) comparison persons initiated BZDR use during the follow‐up (Table 1). In both PD and comparison groups, the proportion of women was higher in the initiators than in the non‐initiators. BZDR initiators were more likely to have a history of mood disorders and schizophrenia as well as the use of antidepressants or antipsychotics during the washout period in both groups.

TABLE 1.

The characteristics, comorbidities and medications of benzodiazepine and benzodiazepine‐related drug (BZDR) initiators and non‐initiators with and without Parkinson's disease (PD).

	PD n = 18 651		p	No PD n = 127 806		p	p between initiators with and without PD
	Initiators n (%)	Non‐initiators n (%)		Initiators n (%)	Non‐initiators n (%)
	7280 (39.0)	11 371 (61.0)		33 066 (25.9)	94 740 (74.1)		< 0.001
Age
At index date, mean (95% CI)	70.7 (70.5–71.0)	70.2 (70.0–70.3)	< 0.001	71.5 (71.4–71.6)	69.4 (69.4–69.5)	< 0.001
At BZDR initiation, mean (95% CI)	70.9 (70.7–71.1)			71.7 (71.6–71.9)			< 0.001
Sex			< 0.001			< 0.001	< 0.001
Women	3552 (48.8)	4423 (38.9)		16 981 (51.3)	37 902 (40.0)
Men	3728 (51.2)	6948 (61.1)		16 085 (48.7)	56 838 (60.0)
Comorbidities
Asthma/COPD	542 (7.5)	712 (6.3)	0.002	2862 (8.7)	6055 (6.4)	< 0.001	0.001
Diabetes	859 (11.8)	1489 (13.1)	0.009	4185 (12.7)	11 111 (11.7)	< 0.001	0.045
Cardiovascular disease	2941 (40.4)	4004 (35.2)	< 0.001	14 182 (42.9)	32 135 (33.9)	< 0.001	< 0.001
Rheumatoid arthritis	232 (3.2)	364 (3.2)	0.957	1381 (4.2)	3108 (3.3)	< 0.001	< 0.001
Comorbidities before follow‐up
Mood disorders	192 (2.6)	171 (1.5)	< 0.001	803 (2.4)	1078 (1.1)	< 0.001	0.298
Schizophrenia	47 (0.7)	35 (0.3)	0.001	148 (0.5)	259 (0.3)	< 0.001	0.027
Drug use during the washout period
Antidepressants	473 (6.5)	354 (3.1)	< 0.001	1827 (5.5)	2127 (2.2)	< 0.001	0.001
Antipsychotics	238 (3.3)	157 (1.4)	< 0.001	635 (1.9)	871 (0.9)	< 0.001	< 0.001
Lithium	18 (0.3)	19 (0.2)	0.230	30 (0.1)	40 (0.04)	0.001	< 0.001

Abbreviations: CI: confidence interval; COPD: chronic obstructive pulmonary disease.

Comparison of BZDR initiators with and without PD showed that the proportion of men was higher in the PD group. Use of antidepressants, antipsychotics and lithium during the washout period was more common in initiators with PD, but asthma/COPD, cardiovascular diseases and rheumatic diseases were more common among initiators without PD.

Of the BZDR initiators, people with PD initiated benzodiazepine more frequently (53.3% of BZDR initiations) than the comparison persons (42.2%, Table 2). Medium‐acting benzodiazepines were the most frequently initiated benzodiazepines in both groups, with oxazepam being the most common one (21.5% of initiators with PD and 14.4% of initiators without PD). Altogether, 47.8% of BZDR initiators with PD and 59.0% without PD initiated with benzodiazepine‐related drugs. Zopiclone was more frequently initiated than zolpidem. In both cohorts, 1.7% of initiators had purchased two or more different BZDRs at initiation.

TABLE 2.

Initiated benzodiazepines and related drugs (BZDR) in persons with or without Parkinson's disease (PD).

	BZDR initiators with PD (n = 7280)		BZDR‐initiators without PD (n = 33 066)		p
	n	%	n	%
Benzodiazepines	3881	53.3	13 944	42.2	< 0.001
Long‐acting benzodiazepines	1118	15.4	4242	12.8	< 0.001
Diazepam	602	8.3	2378	7.2
Alprazolam	412	5.7	1222	3.7
Nitrazepam	55	0.8	357	1.1
Other	55	0.8	294	0.9
Medium‐acting benzodiazepines	2744	37.7	9576	29.0	< 0.001
Oxazepam	1562	21.5	4754	14.4
Lorazepam	243	3.3	649	2.0
Temazepam	961	13.2	4244	12.8
Short‐acting benzodiazepines	40	0.6	244	0.7	0.082
Benzodiazepine‐related drugs	3477	47.8	19 502	59.0	< 0.001
Zopiclone	2850	39.2	15 784	47.7
Zolpidem	628	8.6	3721	11.3
Number of purchased BZDR at initiation					0.108
1	7156	98.3	32 490	98.3
≥ 2	124	1.7	576	1.7

The rate of BZDR initiation was higher in the PD group throughout the entire follow‐up (Figure 2a). In the comparison group, the initiation rate declined consistently, whereas in the PD group, the decline ended 2 years before the index date, and the IR peaked 6 months before the diagnosis.

FIGURE 2.

Incidence of benzodiazepines and benzodiazepine‐related drugs in relation to the Parkinson's disease (PD) diagnosis. Incidence rate (IR) and 95% confidence intervals of (a) benzodiazepine and benzodiazepine‐related drugs combined, (b) benzodiazepines and (c) benzodiazepine‐related drugs in six‐month intervals 6 years before and 10 years after the PD diagnosis per 100 person years (PY). PD: people with Parkinson's disease; no PD: matched comparison people.

Similar pattern was observed for benzodiazepines (Figure 2b), while the difference in benzodiazepine‐related drug initiations was smaller, although a similar peak around the diagnosis was observed (Figure 2c).

4. Discussion

In this study, BZDR initiation was more frequent in people with PD compared to a matched cohort of people without PD already 6 years before to 10 years after the PD diagnosis. The peak in the BZDR initiation rate in people with PD occurred 6 months before the diagnosis of PD.

To our knowledge, the incidence of BZDR use in relation to PD diagnosis has not been studied before. In our study, the BZDR incidence rates in people with PD began to rise 2 years before and peaked 6 months before the index date. This is in line with the findings from a previous register‐based study on prediagnostic symptoms in which people with PD had a higher incidence of insomnia than controls already 2 years but not yet 5 years before the diagnosis [26]. On the other hand, people with non‐apnea sleep disorders had a higher risk for PD within 10 years than the matched controls [5]. Sleep disorders have been associated with the neurodegenerative process [27]. They can also have a negative effect on the glymphatic system, a mechanism with which the extracellular proteins are cleared from the brain, as the system seems to work most effectively during deep sleep [28, 29, 30]. Disruptions in the sleep–wake rhythm could prevent protein clearance and thereby accelerate the neurodegenerative process. Studies on animal models of PD have also shown that dopamine deficiency can negatively affect the circadian rhythm. When primates were treated with the neurotoxin MPTP causing parkinsonism, the animals' rest–activity cycles and cognitive performance changed even before motor symptoms occurred [31]. Transgenic mice with dopaminergic neurodegeneration also showed impaired circadian cycles [32].

A previous case–control study found that the incidence of anxiety was higher in PD patients than in controls already 5 years before the diagnosis [26]. In our study, BZDR IRs did not begin to rise as early as that, probably because BZDR are mainly used for insomnia and benzodiazepine‐related drugs are not indicated for anxiety [13]. Nowadays, antidepressants such as selective serotonin reuptake inhibitors (SSRIs) and serotonin and norepinephrine reuptake inhibitors (SNRIs) are the first‐line medication choice for treating anxiety, and benzodiazepines are recommended to be avoided especially when treating older adults [33, 34, 35]. In a previous study, people with PD initiated SSRIs more frequently than people without PD, and the incidence began to increase 7 years before the PD diagnosis [23]. The different incidences for benzodiazepines and benzodiazepine‐related drugs in our study might partly be due to the double indication of benzodiazepines, as the same person can have both sleep problems and anxiety.

There was a difference in benzodiazepine initiation rates between people with PD and comparison people during the entire study period, and the IRs in people with PD even increased after the diagnosis of PD. Our finding of higher benzodiazepine initiation rates in the PD population before and after the PD diagnosis is in agreement with a study of 175 patients with early untreated PD and 166 controls, where both anxiety (16.4% vs. 2.4%) and sleep disturbances (17.8% vs. 9.1%) were more common in PD patients than in controls [36]. Similarly, in another study where people with early untreated PD were investigated, anxiety was more than twice as common in people with PD as in controls [37].

In an older Norwegian study, 40.2% of people with PD and 23.0% of controls reported using sleeping pills [38]. In a Danish register study investigating the effect of psychotropic medication use on all‐cause mortality, people with PD used benzodiazepines or benzodiazepine‐related drugs as single medication less frequently than the comparison people (benzodiazepines 6.2 vs. 7.5%, benzodiazepine‐related drugs 4.1 vs. 7.5%), but the use of psychotropic drug combinations including BZDR was more common in people with PD [39].

Several guidelines recommend avoiding BZDR in older adults due to adverse effects like cognitive impairment, delirium, falls and fractures [15, 34, 40, 41]. According to the American Geriatrics Society Beers Criteria, both long‐acting and short‐ and intermediate‐acting BZDR should be avoided in people at least 65 years of age, and they should be used for people with a history of falls and fractures only if there are no safer alternatives [34]. If benzodiazepines are prescribed, it is recommended they should be used only for a short time [40] and not for longer than 4 weeks [15]. In our study, medium‐acting oxazepam and temazepam, as well as zopiclone, were the most frequently initiated BZDR in people with or without PD. This is in line with Finnish guidelines, which advise choosing only medium‐acting benzodiazepines or zopiclone when prescribing BZDR to older adults [40]. Although the therapeutic intensity of benzodiazepines and related drugs (defined daily doses sold per 1000 inhabitants per day) decreased in Finland between 2013 and 2020, and a decline in the prevalence of BZDR use was observed during the same period, the prevalence of benzodiazepine use did not decline similarly [42]. Concerningly, the highest prevalences were observed in the oldest age groups.

The World Falls Guidelines recommend assessing the risk of falls before prescribing potentially fall‐risk increasing drugs to older adults, and also recommend reviewing medication and de‐prescribing if necessary [43]. Meta‐analyses have found an increased risk for falls among older users of benzodiazepines (Odds ratio 1.42; 95% confidence interval 1.22–1.65) as well as among users of benzodiazepine‐related drugs [16]. In another meta‐analysis, both benzodiazepine and benzodiazepine‐related drug use increased the risk of hip fracture in older adults (> 50 years old), the risk being highest within 2 weeks from the beginning of use [44]. This finding underscores the importance of careful consideration before prescribing BZDR for older adults even for a short time use. This is especially important in persons with PD as the disease itself increases the risk of falling threefold compared to controls [20]. The reasons behind the fall risk include slow gait, poor postural control, and disease severity as well as orthostatic hypotension, which is common in people with PD [45, 46, 47].

In this study, we studied only the incidence of BZDR use, and the duration of the use was not investigated. However, other studies have shown that despite recommendations, BZDR are still often prescribed for long‐term use. In a Finnish cohort study, 55% of new BZDR users 65 years or older became long‐term users [48]. Long‐term use or otherwise inappropriate use of benzodiazepines is common, especially in older adults and people with chronic illness [49, 50].

The results of this study represent the Finnish community dwellers with PD well, as the study is based on a nationwide cohort. The risk of selection bias was also minimized. The limitation of this study was that not all the benzodiazepines were not reimbursed during the entire study period, and thus, the coverage is incomplete. Based on a previous study, the coverage of benzodiazepine consumption in the applied register was 90% in 2006–2012, decreasing to 76% in 2013–2014 following the changes in temazepam, nitrazepam, and chlordiazepoxide reimbursability. [51] The coverage gap concerns mainly small package sizes and specific product names for which a generic, reimbursable alternative existed. Most of the products are captured in our data, and it is unlikely that the prescription rates of nonreimbursed products compared to the reimbursed ones would be different between the cohorts in such a magnitude that it would affect the main conclusions. Because of the missing data on nonreimbursed products, the absolute incidence rates are underestimated. This does not undermine our conclusions, but rather emphasizes them.

5. Conclusions

In conclusion, the incidence of BZDR use was more frequent in people with PD than people without it. The difference was highest 6 months before the PD diagnosis. This is concerning as BZDR are known to increase the risk of falls and fall‐related injuries in older adults. As people with PD are already more prone to falling than people without PD, careful consideration and weighing of benefits and harms are needed before prescribing BZDR to people with PD.

Author Contributions

Conception of the research project: V.L., S.H., M.T. and A.M.T.; organization and execution of the research project: V.L. and A.M.T.; design of the statistical analysis: V.L., S.H. and A.M.T.; execution of the statistical analysis: V.L.; review and critique of the statistical analysis: V.L. and A.M.T.; writing the first draft of the manuscript: V.L.; review and critique of the manuscript: V.L., S.H., M.T., M.K. and A.M.T.

Ethics Statement

All methods were carried out in accordance with relevant guidelines and regulations. Register maintainers have approved the FINPARK study plan. Data were pseudonymized before submission to the research team, and study participants were not contacted. Therefore, according to Finnish legislation, including the Data Protection Act 1050/2018 (and the previous Personal Data Act 523/1999), the Act on the Openness of Government Activities 621/1999 and the Act on the Secondary Use of Health and Social Data 552/2019 (and the previous Act on the National Healthcare registers [not an official English translation as this is not available] 556/1989), the study has been granted an exemption from requiring ethics approval or informed consent.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: List of benzodiazepines and benzodiazepine‐related drugs which were not reimbursed and thus not observable during the study period (1996–2016). The list is based on Pharmaca Fennica, which is a national medication information system listing products and reimbursability status on an annual basis. Therefore, products available for a shorter time may not be captured.

Lonka V., Hartikainen S., Tiihonen M., Koponen M., and Tolppanen A.-M., “The Incidence of Benzodiazepine and Benzodiazepine‐Related Drug Use in People With and Without Parkinson's Disease—A Nationwide Cohort Study,” Basic & Clinical Pharmacology & Toxicology 137, no. 5 (2025): e70123, 10.1111/bcpt.70123.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author, but restrictions apply to the availability of these data, and so they are not publicly available. Data are, however, available from the authors upon reasonable request and with permission of the register maintainers.