Increase of Meningitis Risk in Stroke Patients in Taiwan

Background and purpose The blood–brain barrier (BBB) not only provides a physical obstruction but also recruits and activates neutrophils in cases of infection. Hemorrhagic or ischemic stroke reportedly induces the disruption of the BBB. However, few studies have reported a correlation between the incidence of meningitis in patients with a history of stroke. This study tested the hypothesis that patients with a history of stroke may be more vulnerable to meningitis. Methods Stroke and age-matched comparison (n = 29,436 and 87,951, respectively) cohorts were recruited from the Taiwan National Health Insurance database (2000–2011). Correlations between the two cohorts were evaluated by Cox proportional hazard regression model, Kaplan–Meier curve, and log-rank tests. Results The incidence of meningitis was higher in the stroke cohort compared to that in the comparison cohort [hazard ratio (HR), 2.89; 95% confidence interval (CI), 2.23–3.74, p < 0.001]. After adjusting for age, sex, and comorbidities, the estimated HR in the stroke cohort was 2.55-fold higher than that in the comparison cohort (CI, 1.94–3.37; p < 0.001). Notably, patients who had experienced hemorrhagic stroke had a higher incidence rate of meningitis than those with a history of ischemic stroke, except for patients older than 75 years (incidence rates in hemorrhagic/ischemic stroke patients, 3.14/1.48 in patients younger than 45 years, 1.52/0.41 in 45- to 64-year group, 1.15/0.90 in 65- to 74-year group, 0.74/0.93 in patients older than 75 years). Moreover, stroke patients who had undergone head surgery had the highest meningitis risk (adjusted HR, 8.66; 95% CI, 5.55–13.5; p < 0.001) followed by stroke patients who had not undergone head surgery (adjusted HR, 2.11; 95% CI, 1.57–2.82; p < 0.001). Conclusion Our results indicated that stroke patients have higher risks of meningitis. Compromised BBB integrity in stroke patients may lead to increased vulnerability to infectious pathogens. In summary, our study provided new evidence of the clinical relationship between stroke and meningitis, and our findings suggest the need for precautions to prevent meningitis in stroke patients.

inTrODUcTiOn According to the statistics from the World Health Organization, stroke is the second leading cause of death worldwide (1). Meanwhile, cerebrovascular disease (including stroke) was the leading cause of death in Taiwan in 2016 (2). Moreover, poststroke patient management imposes a heavy financial burden on the families as well as the health care system (3,4). Age, hypertension, diabetes mellitus, obesity, atrial fibrillation, and head injury are significant risk factors for stroke (5,6). It has been reported that the risk of stroke is related to several infectious diseases, including meningitis (7). Bacterial meningitis is a significant cause of morbidity and mortality worldwide (8). Attention has been devoted to the neurological complications of meningitis, including stroke and other neurovascular events (9)(10)(11)(12)(13). Various studies have also addressed the incidence of stroke in adult patients with meningitis (10)(11)(12)14). Stroke is a serious complication among adult bacterial meningitis patients and is associated with long-term sequelae and also possibly death (15). Previous studies have reported that hemorrhagic and ischemic stroke not only impair neuronal function but also affect the cerebral vasculature, as indicated by the loss of blood-brain barrier (BBB) integrity (16,17). Brain proteins might leak into the bloodstream, the dural venous sinuses, or the lymph nodes as a result of the compromised BBB in stroke patients (18). BBB dysfunction may cause several neurological diseases, e.g., multiple sclerosis, Alzheimer's disease, and meningitis (19). To our knowledge, no study has explored the prevalence of meningitis in stroke patients compared to a healthy population; nevertheless, stroke is considered a common complication among meningitis patients.
This study analyzed the medical records of 29,436 stroke patients and 87,951 age-matched controls acquired from Taiwan National Health Insurance (TNHI) database from 2000 to 2011. The risk factors for stroke, including hypertension, diabetes, hyperlipidemia, atrial fibrillation, and head injury were considered. The correlation between meningitis incidences in stroke patients and the comparison group was also assessed.

Data source
This retrospective cohort study used data from a Longitudinal Health Insurance Database (LHID) derived from the TNHI program set up by the Taiwan Bureau of National Health Insurance. The TNHI program has a coverage ratio above 99% because all Taiwanese compulsorily join the program. One million insurants were randomly selected from the year 2000 Registry from among beneficiaries in the LHID. The LHID included all inpatient and outpatient medical records for each insurant from 1996 to 2011. Disease was identified in the LHID based on the International Classification of Diseases, ninth Revision, and Clinical Modification (ICD-9-CM). To avoid researchers obtaining information from specific patients, the identification of insurants was re-coded. This study was also approved by the Institutional Review Board of China Medical University Hospital [CMUH104-REC2-115(CR-2)].

study subjects
We collected data 30,019 patients with a new diagnosis of stroke (ICD-9-CM 430-438) at admission between 2000 and 2011 and defined the date of stroke diagnosis at the index date. Stroke patients with meningitis history were excluded (ICD-9-CM 047 and 320-322, n = 583). Finally, we selected 29,436 stroke patients as the stroke cohort. The stroke patients were grouped into hemorrhagic stroke (ICD-9-CM 430-432) and ischemic stroke (ICD-9-CM 433-438) based on the first stroke diagnosis. The comparison cohort (n = 87,951) was selected among individuals without stroke and meningitis history and was frequency matched for age (5-year strata; for example, 0-4, 5-9, and 10-14 years of age), sex, and index year at an approximately 3:1 rate.
end Point, comorbidity, and Operation All study subjects were followed up from the index date to the date of meningitis occurrence. Those who did not develop meningitis were followed up until the date that they withdrew from the program or the end of 2011. The comorbidities included hypertension (ICD-9-CM 401-405), diabetes (ICD-9-CM 250), hyperlipidemia (ICD-9-CM 272), atrial fibrillation (ICD-9-CM 427.31), and head injury (ICD-9-CM 850-854, 959.01). All comorbidities were defined before the index date. Head surgery, a kind of stroke treatment (ICD-9 operation code 01 and 02), was used to define the severity of stroke in this study. Patients with head surgery treatment were defined as serious stroke and others were defined as mild stroke. Stroke patients with head surgery within 7 days after stroke occurred were defined as head surgery treatment group.

statistical analysis
Chi-square tests were used to assess the significance of the differences in age group, sex, and comorbidity between stroke and comparison cohorts. The incidences of meningitis were calculated for each cohort. Hazard ratios (HRs) and 95% confidence intervals (CIs) of meningitis in the stroke cohort compared with those of the comparison cohort were calculated using Cox proportional hazard regression analysis. The adjusted Cox model was adjusted for age, sex, and all comorbidities (including hypertension, diabetes, hyperlipidemia, atrial fibrillation, and head injury). The association between meningitis and stroke type was also assessed. Because the interaction test between age and stroke showed a significant difference (p = 0.03 in adjusted model), we estimated the age-specific risk of meningitis in the adjusted model. In further analyses, we estimated the association between meningitis and head surgery (stroke severity). Kaplan-Meier analysis was used to plot the cumulative incidence of meningitis, and log-rank tests were used to assess the differences in cumulative incidence between the two cohorts. All analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC, USA) with a statistical significance level of p < 0.05.

characteristics of the subject Population
A total of 117,387 study subjects (29,436 stroke patients and 87,951 comparison subjects) were included in this study. The

cumulative incidences
The estimated risk of meningitis obtained with the Kaplan-Meier estimator is shown in Figure 1. The cumulative incidence rate of meningitis in the stroke cohort was approximately threefold higher than that in the comparison cohort after a follow-up duration of 12 years (median follow-up, 3.22 years) (0.63 vs. 0.21%, log-rank test, p < 0.001).

incidence and risk of Different Meningitis Types between stroke and comparison cohort
We explored the impacts of stroke on developing different types of meningitis (e.g., viral and bacterial meningitis) (

The Prognosis and Outcome of stroke Patients with or without Meningitis
The mortality rates in stroke patients with and without meningitis development were 50.9 and 42.6%, respectively. Compared to stroke patients without meningitis development, those with meningitis development had a modest increased death risk using Poisson regression after adjusted for age, sex, and comorbidity (relative risk, 1.33; 95% CI, 1.03-1.73, p < 0.05) ( Table 5).

DiscUssiOn
Although ischemic stroke is the predominant form of stroke in our study and worldwide, the incidence of meningitis was higher in hemorrhagic stroke patients. The degree of BBB disruption is positively correlated with the severity of intracranial hemorrhage (ICH) (20). We did not find any reports that assessed the association between the severity of ischemic stroke and the degree of BBB damage. However, one study explored the degree of BBB damage in patients with acute ischemic stroke (AIS) who received thrombolytic therapy (21). ICH is the worst complication of intravenous tissue plasminogen activator therapy for AIS. The authors reported that the permeability of the BBB was significantly correlated with the degree of ICH (parenchymal hematoma > hemorrhagic infarction > no hemorrhage). Our findings of a higher meningitis incidence in hemorrhagic stroke patients could be supported by their results. Hemorrhagic stroke patients may have increased risks of meningitis due to the increased permeability of the BBB. One possible underlying mechanism of our study was the upregulation of matrix metalloproteinase-9 (MMP-9) in the brain tissue. Both hemorrhagic and ischemic stroke induced the expression of MMP-9 in the brain tissue; meanwhile, MMP-9 mediated the breakdown of the BBB (22)(23)(24). Interestingly, it was reported that the disruption of the BBB in bacterial meningitis was mediated by MMP-9 (25). Since the integrity of the BBB is required for the physical defense against pathogens, higher meningitis risk was observed in stroke patients compared to the comparisons ( Table 2). The other group's study indicated that the plasma concentrations of MMP-9 were significantly higher in either hemorrhagic or ischemic stroke patients than control participants (26). The upregulation of MMP-9 was an early response of stroke; however, we still do not fully understand the impact of MMP-9 on BBB recovery and how the BBB recovers (27). Moreover, another group found microvascular remodeling and disappearances of vasculature close by the injury site of stroke in a long-term in vivo investigation (28). These changes may compromise the integrity of BBB permanently. Even though there were no significantly different on the duration between index and end point of meningitis occurrence in stroke patients and comparisons (Table S2 in Supplementary Material), higher meningitis risk was observed in stroke patients compared to the comparisons ( Table 2). Our data showed that stroke patients younger than 45 years had the highest incidence of meningitis. This finding could be explained by our observation that the highest prevalence of hemorrhagic stroke was found in the age group of <45 years. Compared to the other age groups, the meningitis incidence among ischemic stroke patients was slightly higher than that among hemorrhagic stroke patients in those older than 75 years. However, this observation may not be accurate since there were only two meningitis patients with history of hemorrhagic stroke. Since hemorrhagic stroke patients showed a higher meningitis incidence than that in ischemic stroke patients, it was surprising that ischemic stroke patients undergoing head surgery had higher risks of meningitis. The detailed mechanisms should be explored in the future.

strengths and limitations
The major strengths of this study were the large cohort sizes, the long follow-up duration, and the integrated medical records. Nevertheless, this study has several limitations. For example, there were few cases of meningitis patients who suffered hemorrhagic stroke in the 65-to 74-year and older than 75-years age groups. Head surgery, a kind of stroke treatment, was applied to determine the severity of stroke in this study. We did not consider head surgery as a risk factor for meningitis in stroke patients, whereas there was one study exploring the incidence and risk factors of meningitis in patients with craniotomy (29). We did not have the data of MMP-9 expression levels in stroke and comparison cohorts. Due to this limitation, we cannot evaluate the effects of MMP-9 on the development of virus or bacterial meningitis in stroke and comparison cohort. Moreover, we assumed the opportunities for exposure to meningitis pathogens were equal among stroke patients and comparisons in this study. However, it was possibly not the case in the daily life.
cOnclUsiOn Our data indicated that stroke patients had higher risks of developing meningitis. Patients younger than 45 years with history of hemorrhagic stroke showed the highest risk of meningitis compared to the other groups. Ischemic stroke patients who had undergone head surgery showed an elevated meningitis incidence. In conclusion, our study provides new insight into the clinical relevance of meningitis in stroke patients.