Association Between Z Drugs Use and Risk of Cognitive Impairment in Older Patients With Chronic Insomnia

Background Benzodiazepines (BZDs) and Z drugs are widely used for patients with chronic insomnia. The long term use of BZDs in older patients can cause cognitive impairment and potentially increase the risk of dementia. However, evidence for an association with Z drugs is limited. This study aimed to investigate the association between the risk of cognitive impairment and exposure to Z drugs in older patients with chronic insomnia. Methods We recorded older patients with chronic insomnia who visited the outpatient department of neurology, Beijing Friendship Hospital, and assessed the global cognitive function (MoCA) and ve cognitive domains (CVLT, TMT-B, BNT-30, CDT and DST). Multiple regression analysis was conducted to determine the independent factors of cognition, and evaluate the effect of Z drugs use (zolpidem and zopiclone) on cognition. Results A total of 88 subjects were identied. BZDs use (P=0.03, and BZDs exposure density (P=0.01, were independent risk factors of cognitive impairment in older patients with chronic Neither Z drugs use (P=0.11) nor Z drugs exposure density correlated with global cognitive function. Moreover, compared with BZDs users, there were positive associations between Z drugs use and memory (P=0.00), attention (P=0.00) and verbal function (P=0.04). Additionlly, education level (P=0.02, 95% duration of insomnia (P=0.04, OR=1.05, and of (P=0.03, were also independent factors of global cognitive function.


Background
In recent years, with the increasing competition in modern life and the quickening pace of life, the incidence rate of insomnia is increasing, which seriously affects people's life. About 10-30% of the world's people have insomnia, and about 50% of them suffer from chronic insomnia [1]. It is reported that in healthy subjects, sleep disorders increase levels of amyloid-beta (aβ) in the cerebrospinal uid (CSF), leading to the progression of neurodegeneration and the emergence of mild cognitive impairment [2][3][4].
Animal studies have shown that sleep restriction increases the susceptibility to aβ induced memory impairment in mice, accompanied by the increase of plasma corticosterone and pro-in ammatory cytokines in the brain lead to memory impairment and synaptic damage [5]. Therefore, sleep disorders have become the major risk factors for the development of Alzheimer's disease (AD) [6]. Treatment of sleep disorders may reduce the risk of probable future AD [7].
The treatment of chronic insomnia mainly includes cognitive behavioral therapy, drug therapy, light therapy, complementary and alternative drug therapy. Although cognitive behavioural therapy is the recommended rst-line treatment [8], there are many limitations in its implementation. Therefore, Benzodiazepines (BZDs) and their receptor agonist Z drugs are commonly used for insomnia. However, long-term use of BZDs in elderly causes serious side effects, which not only lead to drug dependence and tolerance but also increase the risk of fall and fracture [9]. The effect of BZDs on cognition in the elderly is highly debated. Some studies have reported that long term use increase the risk of cognitive impairment [10][11][12], others have shown the contrary results [9,[13][14][15][16]. Although the association between BZDs and cognitive decline remains uncertain, it was determined that the use of BZDs in elderly should be avoided or limited for the short term.
Z drugs as an agonist of the BZD receptor component of the GABA A receptor complex are commonly used for insomnia. Z drugs usually include Zolpidem, Zopiclone, Eszopiclone and Zaleplon. There are limited clinical studies on the association between Z drugs and cognitive impairment. Cheng HT et al. [17] reported that the use of a high cumulative dose of zolpidem increases the risk of AD among the elderly living in Taiwan. Shih HI et al. [18] suggested that the effect of zolpidem on cognition in patients with AD remained uncertain. The association between the Z drugs use and the risk of developing cognitive impairment in older patients with chronic insomnia is unknown. The aim of this study was to evaluate the effect of Z drugs on cognition among elderly patients with chronic insomnia. (such as AD, Parkinson's disease, Lewy body dementia, frontotemporal dementia). 7. The patient had a history of stroke, symptoms and signs of neurological de cit, and responsible lesion. 8. Patients have brain trauma, epilepsy, encephalitis, hydrocephalus, brain tumours and other neurological diseases that can cause cognitive impairment. 9. Patients had abnormal laboratory indexes: moderate anaemia, fasting blood glucose more than two times of the upper limit of normal, creatinine level more than 1.5 times of the upper limit of normal, liver enzymes (aspartate aminotransferase or alanine aminotransferase levels) more than two times of the upper limit of normal. 10. Jet lag or work shift cause chronic insomnia. 11. Systolic blood pressure > 180 mmHg or < 90 mmHg, or diastolic blood pressure > 120 mmHg or < 60 mmHg. 12. Patients who take drugs to improve cognition, such as donepezil, galanthamine, kabbalatin, memantine, etc., or take drugs that have negative effects on cognition, such as anticholinergic drugs, antipsychotic drugs, antiepileptic drugs, glucocorticoids, etc.

Assessment of cognition and insomnia
The patients were divided into normal cognition group (Montreal Cognitive Assessment ,MOCA score ≥ 26) and mild cognitive impairment group (MOCA score < 26). If the education period ≤ 12 years, 1 point will be added. California verbal learning test (CVLT), Trail making test-B (TMT-B), Boston naming test-30 (BNT-30), Clock drawing test (CDT) and Digit span test (DST) were used to assess the memory, executive function, verbal function, visuospatial ability and attention. The total score of 5 times of memory recall in CVLT ≤ 35 is considered memory impairment. TMT-B score ≥ 135.5 seconds, BNT-30 score ≤ 22, CDT score ≤ 3 and DST score ≤ 4 are respectively considered abnormal. We used the PSQI scale to evaluate the severity of insomnia. Age at onset of insomnia (years), duration of insomnia (years), use of sedativehypnotics and family history of insomnia were also recorded to analyse.

BZDs and Z drugs use
BZDs and BZD-related drugs, called Z drugs, are categorised according to the World Health Organization's Anatomical Therapeutic Chemical classi cation system. Estazolam, Lorazepam, Diazepam, and Clonazepam were categorised as BZDs. Zolpidem and Zopiclone were as Z drugs. Duration of drugs use (years), frequency of drugs use (times/week), BZDs and Z drugs use, BZDs and Z drugs exposure density (mg/d) were recorded. The rst two indexes were obtained from patients; the latter two indexes were calculated from prescription register data since the one year before recruitment. The exposure density of each BZD or Z drug, which corresponded to an average one-day exposure, equals the total doses in milligrams divided by the duration of use in days. Each drug use and exposure density was calculated separately, and the overlapping drugs use were calculated respectively, according to each drug.

Confounders
We recorded known potential confounders for cognitive impairment, including marriage, living alone, economy, education level, insomnia and some comorbidities, such as hypertension, diabetes mellitus, coronary heart disease, dyslipidemia. Depression is an important mental differential diagnosis of cognitive impairment. HAMD-17 items and HAMA scales were used to evaluate depression and anxiety disorder. Daily physical activity was assessed by International Physical Activity Questionnaires (IPAQ) (short format). The collected laboratory indexes include fasting blood glucose (mmol/l), triglyceride (mmol/l), total cholesterol (mmol/l), Low-density lipoprotein cholesterol (LDL-C, mmol/l), high-density lipoprotein cholesterol (HDL-C, mmol/l), serum uric acid (umol/l) and serum albumin (g/l).

Statistical analysis
All statistical analysis was conducted using SPSS, version 22.0. Continuous variables were expressed as mean ± SD and categorical variables as numbers and percentages. The independent sample t-test or Mann-Whitney U test was performed to compare continuous variables. The χ 2 test or Fisher exact test was performed for categorical variables. Correlation analysis was used to analyse the associations between cognition and drugs use or other social-demographic, clinical and laboratory variables. Multiple logistic regression analysis or linear regression analysis was conducted to determine the independent factors of cognition. Statistical signi cance was de ned by P-value < 0.05 (two-tailed).
Correlation analysis showed that there were a negative association between PSQI score and MoCA score (r =-0.22, P = 0.04), a signi cant positive association between PSQI score and TMT-B seconds (r =-0.31, P = 0.00). However, no correlation was found between PSQI score and CVLT score, BNT-30 score, CDT score and DST score.

Correlation between global cognitive function and social-demographic and clinical characteristics
Among the 88 patients, they were divided into normal cognition group (n = 20, 22.73%) and mild cognitive impairment group (n = 68, 77.27%) according to the MoCA score. Table 1 shows the socio-demographic and clinical characteristics and the comparison between the two groups. Compared to the mild cognitive impairment group, patients with normal cognition had better economy level (P = 0.01), a higher education level (P = 0.00), lower PSQI score (P = 0.01), less insomnia duration (P = 0.05) and less history of diabetes mellitus (P = 0.03). However, there is no signi cant difference in sex, BMI, marriage, living with others, HAMA and HAMD scores, age at onset of insomnia and use of sedative-hypnotics.  and PSQI score (P = 0.03,odds ratio, 1.27, 95% CI, 1.02-1.57;) were independently associated with global cognitive function (see Table 2).

Correlation between sedative-hypnotics and global cognitive function
In the BZDs and Z drugs users, Table 3 shows the comparison between normal cognition group and mild cognitive impairment group. The patients with normal cognition had less BZDs use (P = 0.01) and BZDs exposure density (P = 0.01), lower PSQI score (P = 0.00), higher education level (P = 0.00), higher serum albumin level (P = 0.00) and lower serum uric acid level (P = 0.02). In multiple linear regression analysis, there were three factors that independently predicted global cognitive function, including BZDs use, BZDs exposure density and education level. The coe cients of these factors were shown in Table 4. The results indicate that patients with normal cognition were likely to have less BZDs use and lower BZDs exposure density. No signi cant correlations were found between global cognitive function and Z-drugs use, Zdrugs exposure density, duration and frequency of drugs use. To better investigate the effect of sedative-hypnotics on cognition, the associations between sedativehypnotics and ve major cognitive domains were analysed. According to the scores of CVLT, the patients were divided into normal memory group (> 35) and impaired memory group (≤ 35). Table 5 shows the factors with statistical differences between the two groups. There were signi cant correlations between memory and duration of insomnia (P = 0.00), duration of sedative-hypnotics use (P = 0.00), BZDs use (P = 0.01), BZDs exposure density (P = 0.00), Z drugs use (P = 0.00), male (P = 0.05), age (P = 0.00) and economy level (P = 0.00). In multiple linear regression analysis, Z drugs use (P = 0.00, B value, -0.54, 95% CI, -0.89--0.20;), duration of insomnia (P = 0.02, B value, -0.01, 95% CI, -0.02--0.00;), male (P = 0.00, B value, 0.32, 95% CI, 0.11-0.52;), and economy level (P = 0.00, B value, 0.40, 95% CI, 0.21-0.59;) were independently associated with memory (Table 5). Patients with better memory were likely to be female, less worried about their nances, had short insomnia duration and prefer to Z drugs. Correlation analysis showed signi cant positive associations between Z drugs use and DST score (r =-0.39, P = 0.00), BNT-30 score (r =-0.26, P = 0.04). However, no correlation was found between Z drugs and TMT-B seconds and CDT score. There was only a signi cant negative correlation between BZDs use and DST score (r =-0.37, P = 0.00). The Z drugs exposure density was not associated with the scale of each cognitive domain.

Discussion
As far as we know, this is the rst study to evaluate the effect of Z drugs on cognitive function in older patients with chronic insomnia. In all patients with chronic insomnia, we found that duration and severity (PSQI) of insomnia, education level were independent factors associated with global cognitive function (MoCA). Patients with long duration of insomnia, high PSQI score and low education level were likely to have a worse global cognitive function. Among the sedative-hypnotics users, we observed the independent associations between BZDs use, BZDs exposure density, education level and global cognitive function. It suggests that patients with normal cognition would have less BZDs use and lower BZDs exposure density. We found that neither Z drugs use nor Z drugs exposure density correlated with global cognitive function. However, Z drugs were independently associated with memory, signi cantly associated with attention and verbal function. Additionally, we did not detect the associations between Z drugs exposure density and each cognitive domain.
The correlation of Z drugs and cognition in this study is not aligned with that of previous studies. Cheng HT et al. [17] and Lee J et al. [19] reported that use of Zolpidem was associated with an increased risk of Alzheimer's disease among older people, especially a high cumulative dose, while Shih HI et al. [18] found that the effect of Zolpidem on cognition in patients with AD remained uncertain. Our ndings of BZDs use were consistent with several previous studies that reported long term use of BZDs increases risk of cognitive incline [10][11][12], though there were also some studies that reported no association between BZDs use and cognitive impairment [9,[13][14][15][16]. For retrospective studies, lacking well-controlling confounders, such as education level, marriage, depression, stroke, diabetes, hypertension, coronary heart disease, physical activity, smoking, medicines affected cognition, especially insomnia, is the possible cause of these inconsistent ndings. As known, insomnia increases the risk of developing AD. If we do not adjust the confounder of insomnia, the correlation between sedative-hypnotics and cognitive decline may not be reliable.
The mechanism of how sedative-hypnotics could increase the risk of cognitive decline is still unclear.
There is a hypothesis that the cognitive reserve capacity of the elderly is limited after taking BZDs for a long time. Because BZDs and Z drugs are positive regulators of the GABA A receptor, they will reduce brain activation, decrease synaptic plasticity, and affect the ability of patients to create new memories [20]. Second, it has been reported that the binding of BZDs and α5GABA A subunit, which is mainly expressed in the hippocampus, impairs monkeys' memory of context information. However, Zolpidem does not impair the performance of visual cue based tasks due to the a nity for α1GABA A rather than α5GABA A [21]. Those results suggest that α5GABA A receptor plays a special role in BZDs related cognitive impairment. It is speculated that BZDs increase the risk of cognitive impairment through α5GABA A in the hippocampus, while Z drugs, which bind for α1GABA A has a lower risk. However, another study showed that the activation of α1GABA A receptor might in uence the spatial learning ability of rodents [22]. Our study found no association between Z drugs and cognitive impairment, can be partly explained by its a nity for α1GABA A subunit rather than α5GABA A . The molecular mechanism of cognitive impairment caused by BZDs and Z drugs needs further prospective study to con rm.
In this study, the use of BZDs and Z drugs in patients aged 50-75 years old with chronic insomnia was up to 72.73%. The high prevalence of drugs may vary due to doctors and patients' ignorance of the risk of drug-induced falls, fractures and cognitive decline, as well as the lack of attention to behavioural cognitive therapy. Besides, the exclusion of patients without sedative-hypnotics who have not visited the outpatient department may partly contribute to the high prevalence of drugs. Due to age-related pharmacokinetic and pharmacodynamics changes, the elderly are more sensitive to the effects of BZDs on the central nervous system [23]. Therefore, the use of BZDs in this population may lead to daytime sedation and decreased alertness [24]. Our study found that BZDs are an independent factor of cognitive impairment in patients with chronic insomnia. These evidence indicate that the use of BZDs, especially long-term use, is not encouraged in the elderly.
This study had several strengths. First, at each time of patient visit, doctors used a standardized cognitive test to collect data to ensure the reliability of data. Second, this study excluded patients with schizophrenia, major depression, stroke, brain trauma, epilepsy ,brain tumour, alcohol dependence, and patients taking medicines that affect cognition, and recorded the patients' education and economic level, marriage, smoking, physical activity, concomitant illness, laboratory examination and sleep assessment, so that well-controlled the confounders of cognition. Thus, the reliability of the conclusion is guaranteed. Finally, this study assessed the cognitive status of patients in detail, including global cognitive function and ve cognitive domains function, to better evaluate the effect of BZDs and Z drugs on cognition.
This study also had some limitations. Insomnia is considered to be a prodromal symptom of dementia.
The use of BZDs may not be a cause of dementia, but insomnia as a clinical manifestation of the early stage of dementia. Our study can not completely exclude the possibility of reverse causation. Our study is a single centre study with small sample size. A large sample prospective study is needed to con rm the ndings of our study in the future.

Conclusion
We found that BZDs use and BZDs exposure density were independent risk factors of cognitive impairment in older patients with chronic insomnia, but no correlation was found between Z drugs use and cognitive impairment. Moreover, compared with BZDs users, our study showed positive associations between Z drugs use and memory, attention and verbal function. Additionlly, education level, duration and severity of insomnia were also independent factors of global cognitive function. Our ndings suggest that the cognitive status should be extensively evaluated and monitored in older patients with sedativehypnotics, the prescription of BZDs should be avoided or limited in the short term, Z drugs should be prescribed with extreme caution in elderly.

Consent for publication
The author con rmed that its publication has been approved by all co-authors.

Competing interest
The authors declared no potential con icts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The work was supported by grants from the National Natural Science Foundation of China(Country: China) IDs: 81771235. It plays a role in designing research, collecting data and paying for patients' cooperation with the questionnaire.
Authors' contributions YL and ZW participated in data collection. GF participated in data analysis and prepared manuscript, revised manuscript and tables; ZYB oversaw data analysis, manuscript revision and nancial support. All authors have read and approved the manuscript.