Prevalence and predictors of seizure in patients with Alzheimer’s disease at a tertiary care center in Riyadh, Saudi Arabia

Purpose: To assess the prevalence and predictors of seizures in patients with Alzheimer’s disease (AD) at a Saudi tertiary hospital. Methods: A retrospective, matched case-control study was conducted using the electronic medical records of patients with AD who had an unprovoked seizure, from October 2015 to May 2018. Results: Nineteen cases and 195 controls were identified. Statistically significant risk factors for an unprovoked seizure in patients diagnosed with AD were hypertension (p = 0.001), autoimmune disease, stroke and TIA (p = 0.001). The multivariate logistic regression analysis identified hypertension (OR = 2.89; p = 0.009) and autoimmune disease (OR = 19.6; p = 0.045) as predictors of unprovoked seizure in AD patients. Conclusion: The occurrence of unprovoked seizures is more likely in severe cases of AD. In addition, the risk of seizure in patients with AD increases with two co-morbid conditions, hypertension, and autoimmune disease. However, further studies are required to determine the underlying mechanism of the association between the two risk factors and AD.


INTRODUCTION
Alzheimer's disease (AD) is one of the most common neurodegenerative disorders known among older patients [1]. In the United States, AD is considered the sixth leading cause of mortality and 4.5 million patients are currently living with the disease [1]. In Saudi Arabia, there are no official statistics related to the prevalence of AD, but according to the Saudi Ministry of Health, the estimated number of AD cases are at least 50,000, most of whom are women [2]. depression and family history of AD. In general, the clinical manifestations and symptoms of AD develop slowly and are progressive [1]. The most prevalent clinical manifestations of AD include memory loss, confusion, impaired judgment, language disturbance and agitation, while some other clinical manifestations are unusual which include seizures and Parkinsonian features [1].
Some previous studies have linked the late stages of AD with unprovoked seizures [3,4]. It is estimated that the prevalence of seizures in AD patients can reach up to 64 % [5]. While the predictors of an unprovoked seizure in AD patients include a younger age, severe dementia, focal epileptiform findings from an electroencephalogram (EEG) and an African American ethnicity [6]. Amatniek et al found that the risk of developing unprovoked seizure was 87-fold higher in younger patients, while it was more than 3-fold higher in the older group compared to the known AD age group [6]. In addition, there is a reported relationship between AD and seizure disorders. Hauser et al found that patients with AD are at higher risk of developing seizures and epilepsy [4].
The aim of this study was to determine the prevalence and predictors of seizures in AD patients at King Abdulaziz Medical City (KAMC) in Riyadh, Saudi Arabia.

Study setting and participants
A case-control study was conducted retrospectively among AD patients who had seizures from October 2015 to May 2018 and were seen in the neurology clinic at KAMC, Ministry of National Guard-Health Affairs (MNGHA), in Riyadh, Saudi Arabia. Patients with the diagnosis of AD and older than 18 years, either males or females, were included in the study. Patients who had a seizure provoked by medications, hypoglycemia or electrolyte imbalance were excluded from the study. Institutional Review Board (IRB) approval (no. H01R005) was obtained from King Abdullah International Medical Research Center (KAIMRC), Riyadh, Saudi Arabia. In addition, the STROBE guidelines were followed in this study [7].

Identification of cases
The KAIMRC database was used to identify the cases that will be matching the inclusion criteria at the neurology clinic. Using an electronic chart review approach, all AD patients diagnosed with seizures from October 2015 to May 2018 were included (n = 214). Of the 214 identified AD patients, only 19 patients matched our inclusion criteria. For each case, the electronic medical record was reviewed to ensure that the case is eligible for inclusion. Cases were matched to controls based on predictors of seizure.

Seizure's predictors in AD Patients
Seizure's predictors assessed based on published findings, seizure's epidemiological knowledge, and patient's characteristics. The evaluated potential predictors are listed in Table  1.

Data collection and handling
The data was retrieved from the Best-Care System, an electronic medical records system used in Ministry of National Guard -Health Affairs (MNGHA), under the supervision of a neurology consultant and a statistician. Medical record information was collected using a prepared data collection form. The data collectors used the approved data collection form to ensure consistency of coding. The supervisors reviewed the data collected by the data collectors to ensure the accuracy of the data.

Data cleaning
The raw data was audited and cleaned prior to the statistical analysis. In order to accomplish this task, all interval variables were checked and summarized in terms of maximum and minimum values. Minimum and maximum values were checked and compared against a possible maximum and minimum value of each variable and variables with implausible values were flagged. A similar process was done to the categorical variables to identify any potential anomalies (miscodes) using a frequency analysis.

Data analysis
Descriptive statistical analyses were performed. Continuous variables were summarized as mean ± SD and median (range). Proportions were used for categorical variables. In addition, demographic and clinical information were summarized in frequency tables. The 95% confidence interval (CI) was used. The binary logistic regression model was used to estimate the regression equation between the outcome variable (AD) and selected significant explanatory variables, namely hypertension and auto-immune disease. The binary logistic regression model takes the formula: A P value of less than 5 % was considered as statistically significant. All statistical analyses were performed using SPSS 21.0 (Release 21.0.0.0, IBM, USA).

RESULTS
The total number of identified AD patients was 214 patients. Patients' age ranged from 56 years to 111 years. Almost half of the sample (46 %, n = 98) were males. In addition, majority of the patients (98 %) were Saudis.

Prevalence of seizures
Out of 214 patients, 19 (8.9 %) of AD patients developed an unprovoked seizure (95 % CI, 8.2 -9.5). Males constituted 47.4 % (n = 9). The mean age at diagnosis of AD was 73.2 + 8.8 years (range = 53 to 88 years), and the mean age at seizure onset was 74.3 + 13.8 years (range = 57 -91years). The mean duration from the AD diagnosis to the occurrence of the seizure was 19 + 12.8 months (range = 6 -60 months). Generalized tonic-clonic seizure was the most prevalent type of seizure accounting for 94.7 % (n = 18) of the patients, and one patient (5.2 %) had a focal seizure.   The analysis included 19 patients with AD and an incidental diagnosis of seizures, and 195 patients with AD diagnosis but without seizures. Characteristics of matched cases reported in Table 2. The majority of the patients (84.0 %, n = 16) had severe AD, and 3 patients (15.8 %) were classified as moderate AD, suggesting that the risk of a seizure is more likely to occur in severe cases of the disease.
The potential risks of developing a seizure in AD patients include hypertension, stroke or transient ischemic attack (TIA), head injury, current use of psychotropic medications, congestive heart disease and auto-immune disease. The confounders were analyzed using bi-variant analysis. Statistically significant risk factors of developing a seizure in AD patients were hypertension (p = 0.001), autoimmune disease (p = 0.040) as well as stroke and TIA (p = 0.001) ( Table 3).

DISCUSSION
Earlier studies reported a higher risk of unprovoked seizures among patients with AD [4][5][6]. In the current study, unprovoked seizure was reported in 19 of 214 AD patients with an incidence rate of 8.9 %. McAreavey et al conducted a study among AD patients, and found that out of 208 AD patients, unprovoked seizures were reported among 19 of them (9.1 %) [11]. In addition, Cheng et al reported that 44 of 937 AD patients (4.7 %) developed unprovoked seizures during the 10 -year followup period [12].
In terms of the type of seizure, different types are reported in literature [4,6]. In the current study, the majority (94.7 %, n = 18) were classified as generalized seizures. Similarly, Mendez et al reported generalized seizures in majority of AD cases (90 %) [13]. In contrast, Rao et al reported that 72 % of the seizures were complex partial seizures [9]. In addition, Mendez et al indicated that the onset of the seizures tends to occur during the later stages of the disease, on average 6.8 years after the diagnosis of AD [13]. In contrast, in the current study, the average seizure onset was 1.5 years after the diagnosis of the disease.
Furthermore, many studies suggested that the severe form of AD had a higher risk of seizure with frequencies ranging from 9 % to 64 % [13 -15]. In our study, 84 % of AD patients with seizure had a severe type of AD. In addition, hypertension also indicated an increased risk of seizure in AD patients. However, Bernardi et al did not find hypertension as a predictor of seizure in AD patients but considered it as a protective factor against seizure occurrence [16].
However, in the current study, younger age was not a risk factor for developing seizure in AD patients. This could be attributed to the lack of proper documentation as many of the files lacked the date of the initial diagnosis.

Limitations of the study
We acknowledge that our study has some limitations. This is a single center study. The research results may have been more generalizable if it was conducted in multiple centers with a larger population. In addition, due to the retrospective study design, no follow up data of patients were obtained. Furthermore, the data were collected from electronic medical records with some missing data; for example, the Mini-Mental State Examination scores (MMSE), Clinical Dementia Rating (CDR), Electroencephalogram (EEG), and the classification of AD.