Clinical Epidemiology, Risk Factors, and Outcomes of Encephalitis in Older Adults

Abstract

Background

Encephalitis is associated with significant morbidity and mortality, with unknown etiologies in the majority of patients. Large prognostic studies evaluating elderly patients are currently lacking.

Methods

We performed a retrospective cohort of encephalitis cases in 19 hospitals from New Orleans, Louisiana, and Houston, Texas, between the years 2000 and 2017.

Results

A total of 340 adult (aged ≥17 years) patients with confirmed encephalitis were enrolled, and 194 (57%) had unknown etiologies. A cerebrospinal fluid polymerase chain reaction (PCR) for herpes simplex virus (HSV) and varicella zoster virus was done in 237 (69%) and 82 (24%) patients, respectively. Furthermore, an arboviral serology was done in 169 (49%) patients and measurements of anti-N-methyl-D-aspartate receptor antibodies were taken in 49 (14%) patients. A total of 172 out of 323 patients (53%) had adverse clinical outcomes (ACOs) at discharge. Older individuals (>65 years of age) had a lower prevalence of human immunodeficiency virus, had a higher number of comorbidities, were less likely to receive adjuvant steroids, were more likely to have a positive arbovirus serology, were more likely to have a positive HSV PCR, were more likely to have abnormal computerized tomography findings, and were more likely to have to have an ACO (all P values < .05). Prognostic factors independently associated with an ACO were age ≥65, fever, Glasgow Coma Scale (GCS) score <13, and seizures (all P values ≤0.01).

Conclusions

Encephalitis in adults remain with unknown etiologies and adverse clinical outcomes in the majority of patients. Independent prognostic factors include age ≥65 years, fever, GCS score <13, and seizures.

Encephalitis continues to be surrounded by many unanswered questions and numerous clinical challenges. Its global incidence is conservatively estimated as 3.5–7.4 cases per 100 000 patient-years, and is often referenced with caution that accurate diagnoses and reporting are highly variable [1–3]. In 2010, Vora et al [4] estimated the total cost of encephalitis-associated hospitalizations in the United States as approximately $2 billion per year. However, this dollar figure cannot encompass the entire burden of the disease, which leaves behind a striking rate of severe morbidity (50%) and mortality (15%) in those that are diagnosed [5].

With encephalitis, as well as other central nervous system (CNS) infections, individuals of an age greater than 65 have continually been shown to be at an increased risk of contracting these infections and developing associated poor outcomes, compared to younger individuals [5, 6]. Yet, despite this correlation, only a small number of studies have looked closely at the older adult population to better understand this behavior. Those that have, have showed that age is often confounded with many other risk factors and that it is difficult to delineate areas in which meaningful interventions can be used to improve the care of older adults [7–9]. Additionally, etiologies and clinical management may be unique to older adults and warrant additional considerations in their medical care [7, 10].

By comparing older and younger adults with encephalitis, we hoped to capture meaningful differences in their presentations and outcomes. The objective of our study was to identify unique features in the older adult population that may suggest associated risk factors, clinical findings, and underlying etiologies that should be studied further and be considered in clinical practice.

METHODS

Study Population

This retrospective study was of a cohort of encephalitis patients, looking at the impact of age and other clinical features associated with adverse clinical outcomes. Encephalitis cases recruited into the study were derived from 3 major health systems—1 in New Orleans, Louisiana, and 2 in Houston, Texas—comprising 19 hospitals in total. The study was approved by the Tulane University Institutional Review Board, the University of Texas Health Committee for the Protection of Human Subjects, the Memorial Hermann Research Review Committee, and the Harris Health System Research and Sponsored Programs. Inpatient electronic clinical records from 2000 to 2017 with a discharge diagnosis code for encephalitis were reviewed. Due to the low positive predictive value of diagnosis codes for encephalitis [11, 12], cases were then further screened independently by 2 physicians to confirm their diagnoses. Cases that met the 2013 International Encephalitis Consortium (IEC) criteria [13] were retained and further reviewed retrospectively. Information regarding demographics, clinical presentation, medical management, and final disposition were all recorded. The outcome of interest was determined per the review of 2 physicians, based on the rehabilitation documentation and final discharge note. As the main exposure of interest, the population was dichotomized into those less than 65 years of age (yoa) and those greater than or equal to 65 years of age.

Definitions

Per the 2013 IEC guidelines, an encephalitis case was determined by the presence of the major criterion (presentation of altered mental status without alternative cause) and at least 2 of the following minor criterion: fever 72 hours before or after presentation; new-onset seizures; new-onset focal neurological findings; white blood cell count >5 mm³ in the cerebrospinal fluid (CSF); new imaging findings; and/or abnormal electroencephalogram consistent with encephalitis [13]. While we recognize that the IEC guidelines are not formal diagnostic criteria, restricting our case definition to its standardization was used to assist with generalizability and in assuring the best capture of true encephalitis cases possible. The inclusion, exclusion, and distribution of cases are described in Figure 1, and further descriptions of excluded cases from a subsample of this study have been described in a prior publication [14]. Case outcomes were defined using the Glasgow Outcome Scale, where 5 is considered a good recovery, 4 is consistent with moderate disability, 3 is consistent with severe disability, 2 is consistent with a vegetative state, and 1 indicates death. Anything that was scored as a 4 or less was defined as an adverse clinical outcome (ACO) [15].

Statistical Analysis

A bivariate analysis was performed between adults <65 yoa and those ≥65 yoa using the Fisher exact or Chi-square test. The relationships between clinical risk factors, age, and ACOs were assessed by using a modified Poisson regression model to determine the crude risk ratios. A multivariable modified Poisson regression model with a robust error variance was used to compute adjusted risk ratios. Model selection was based on the high prevalence of the outcome of interest (>10%) and the overestimation that would occur in a traditional logistic regression [16]. Variables included in the model consisted of risk factors that were determined to be clinically relevant and found to have a statistically significant crude association (P value <.05). All statistical analyses were conducted using RStudio (version 1.0.153) [17], using the “geepack” package [18].

RESULTS

Demographics and Clinical Presentation

After reviewing all possible cases with an encephalitis-related diagnostic code, 340 were found to have evidence of a diagnosis consistent with the 2013 IEC guidelines. Of the 340 cases, the mean age was found to be 48, with a range between 18 and 95 yoa (Figure 2). Overall, 45.9% of participants were female, 41.3% were Caucasian, 34.4% were African American, 21.1% were Hispanic, and 3.2% were Asian.

After the cases were dichotomized, there were 278 individuals <65 yoa and 62 individuals ≥65 yoa. Comparisons of features related to their demographics and clinical presentations are shown in Table 1. Between the 2 age groups, there were no significant differences between gender, race, and ethnicity. Older adults were more likely to have a Charlson Comorbidity Index (CCI) score >1 at presentation (67.7% vs 37.1% in younger adults; P < .001); however, younger adults were more likely to have a diagnosis of human immunodeficiency virus (HIV; 25.9% vs 3.4% in older adults; P = .004). Clinical symptoms at presentation showed no differences in prevalences of reported headaches, nausea, subjective fever, stiff neck, or malaise. However, older adults were less likely to report photophobia (P = .05). Clinical signs at presentation showed no differences in prevalences of fever, nuchal rigidity, Glasgow Coma Scale (GCS) scores <13, focal neurological deficits, or seizures. However, the presence of a vesicular or petechial rash was found more commonly in older adults (P = .06).

Laboratory Results and Clinical Management

Laboratory results, imaging, clinical management, and rates of adverse clinical outcomes are summarized in Table 2. Serum and CSF serologies were similar between each group, with the only difference being that older adults had more cases where CSF protein values were >120 mg/dL (P = .08). Viral etiologies were much more common in older adults: specifically, herpes simplex virus (HSV; P = .02) and arboviruses (P = .005). Of note, the arboviral panel consisted of serology testing for California encephalitis virus, Eastern and Western equine encephalitis viruses, St. Louis encephalitis virus, and West Nile virus. Anti-N-methyl-D-aspartate (NMDA) receptor antibodies were more commonly found in younger adults (0.07), but were only assessed in 5 of the 62 older adults. Abnormal findings on computer tomography (CT) were more common in older adults (P = .008).

Empiric antibiotic and antiviral medication administrations were similar between each group. However, empirical steroids were not given as frequently in older adults, as compared to younger adults (P = .003). Rates of admission to intensive care and of intubation were similar in both groups. Finally, ACOs were found more commonly in older adults, as compared to those <65 yoa (P = .04; Table 2).

More than half of all cases in both age groups had no known etiology (58.3% in adults <65 yoa and 51.6% in adults ≥65 yoa). Of the cases with a known etiology, viral causes (including arboviruses) were more frequent than bacterial, fungal, parasitic, and autoimmune diagnoses (Table 3). This viral predominance was significantly greater in older adults. The frequency of ACOs was higher in older adults and consistent across each etiological group that was captured in that population (unknown = 62.5%, bacterial = 60%, viral = 61.5%, arboviral = 66.7%, and total = 62.9%). In comparison, younger adults were found to have an overall lower prevalence of ACOs (total = 47.7%) but had greater variability between etiological groups (unknown = 46.3%, bacterial = 36.8%, viral = 46.2%, arboviral = 45.5%, fungal = 72.7%, parasitic = 83.3%, and autoimmune = 50%; Table 3).

Clinical Outcomes and Risk Factors

Of the 340 cases, 172 (50.6%) of them had confirmed ACOs. In the older adult population, when compared to the younger cohort, this proportion was increased by 14.4% (from 50.6% to 65%, respectively) and was found to be statistically significant (P = .04; Table 2). An exploration of meaningful risk factors was performed by testing crude associations of clinical features and ACOs in each age group, as demonstrated in Table 4. In younger adults, the number of comorbidities, fevers, GCS scores <13, seizures, and abnormal imaging results were all associated with increased risks of ACOs (all features, P <0.05). After adjustment, only fevers, GCS scores <13, and abnormal CT findings were found to be meaningfully associated with increased risks in ACOs in younger adults (risk ratio [RR] 1.47 [P = .03], RR 1.50 [P = .003], and RR 1.32 [P = .05], respectively). In older adults, the presence of an HIV/acquired immunodeficiency syndrome diagnosis was associated with an increases risk of an ACO (RR 1.56; P = .002). However, having either a CCI score >1 or a CSF white blood cell count <100 cells/µL was, conversely, found to be protective (RR = 0.60 [P < .001] and RR 0.62 [P = .01], respectively). After adjustment, we found that HIV remained a meaningful risk factor for ACOs (RR 1.82; P = .003); however, it is important to note that this is based on 1 single case in the older adult population. A CCI score > 1 was associated with a decline in the associated risk (RR 0.55; P = .003).

Looking at the entire cohort population, the unadjusted RR associated with age ≥65 and having an ACO was 1.29 (95% confidence interval [CI] of 1.02–3.30; P = .03). Additionally, symptoms of fever, GCS scores <13, seizures, abnormal imaging results, and abnormal Electroencephalogram (EEG) findings at presentation all had positive correlations with ACOs (Table 5). After adjustment, abnormal findings on CTs, magnetic resonance imaging, or EEG results were no longer significantly associated with ACOs. However, being ≥65 yoa or presenting with either fever, a GCS score <13, or seizures independently increased the risk of ACOs (RR = 1.33, 95% CI 1.06–1.56 [P = .01]; RR = 1.42, 95% CI 1.07–1.89 [P = .02]; RR = 1.45, 95% CI 1.15–1.81 [P = .001]; and RR = 1.31, 95% CI 1.02–1.68 [P = .04], respectively).

DISCUSSION

Serious infections and associated sequelae disproportionally afflict those less than 18 or greater than 65 years of age. Older individuals are also associated with more severe manifestations of infectious syndromes and typically have worse outcomes [19]. However, the literature regarding CNS infections in older adults demonstrates that age alone is often complicated by other risk factors, and it is difficult to accurately correlate its contribution to morbidity and mortality outcomes [9]. Closer examinations of other risk factors would seem to suggest that confounding variables may be contributing to the poorer outcomes seen in older adults. Varying clinical presentations, multiple comorbidities, and clinical preconceptions of older patients often prevent accurate medical care [20]. Encephalitis, in particular, is a notorious disease that is difficult to diagnose and treat in individuals of any age.

In our study, we found that demographics and clinical presentations were largely the same between young and older individuals with encephalitis. There does appear to be a distinct difference in comorbidities between age groups. Older individuals as a whole were, unsurprisingly, found to have a greater burden of comorbidities. But younger patients were distinctly more likely to have HIV. Most recent studies have shown that the highest burden of HIV in the United States currently afflicts individuals between the ages of 40 and 60. Older individuals only account for 7.3% of all individuals with HIV in the United States [21], and this uneven distribution was reflected in our population.

Of all 57 HIV patients, 32 (56.1%) had a confirmed etiology (cryptococcus = 9, syphilis = 3, cytomegalovirus = 5, John Cunningham [JC] virus = 1, HSV = 1, varicella zoster virus = 7, toxoplasma = 5, and NMDA = 1). In both age groups, we were surprised to find a limited association between HIV and immunosuppression on ACOs. We also assessed CD4 cell count values of all people living with HIV as an independent variable, and found no significant relationship with ACOs. This nonintuitive finding has been demonstrated in other studies looking at neurological infections [6, 8, 9]. We find the consistency of this phenomenon highly intriguing and a feature we hope to study in future analyses.

Looking globally at the clinical features that were statistically trending and significant (P values < .10 and <.05, respectively) for older adults, we found that there were higher prevalences of comorbidities and of development of a rash that is consistent with viral encephalitis. Elevated protein levels in CSF were also noted and can be associated with viral encephalitis, but are not very specific. This is consistent with the microbiological evidence showing significantly higher prevalences of HSV and arboviral-related encephalopathies in older individuals. This would be in line with most literature, particularly with mosquito- and tick-borne encephalitis, which are more likely to develop in older adults and in individuals with a greater number of comorbidities [22–24].

Yet, despite the high prevalence of viral encephalitis, less than half (48%) of all older individuals were tested for any arbovirus at all. The testing frequencies for enterovirus and varicella zoster virus were 33.9% and 21%, respectively. HSV was slightly better, with 66.1% of older adults tested. Yet, we found that NMDA receptor antibodies were only assessed in 5 of the 62 older adults. We fear that these findings for global undertesting are possibly due to perceptions that alternative diagnoses, such as delirium or dementia, in older individuals may lead to delayed or limited screening for encephalitic etiologies.

According to the 2013 IEC guidelines, older adults in the United States are particularly susceptible to Eastern equine encephalitis, St. Louis encephalitis, and West Nile virus encephalitis [13]. Evidence also strongly encourages the use of empiric antibiotic, antiviral, and steroid therapies while an etiology is being determined. While antibiotics and antivirals were given in similar frequencies to both groups, rates of steroid administration were found to be quite lower in older adults. While the efficacy of corticosteroid use is not ubiquitous and is often case-dependent, there is still no clear explanation for why older adults as an entire population should be less likely to receive steroid treatment. We suspect that unfamiliarity with recent empiric treatment recommendations and hesitancy to administer steroids to older patients coincide to prevent optimal treatment in this population. Continued efforts to standardize best practices for all age groups and educate physicians are important and need to be emphasized.

In our study, we also found that rates of ACOs were similar to those in prior research of encephalitis and other CNS-related infections [8, 9]. A crude correlation between age groups showed that older adults had an increased likelihood of an ACO. Similarly, we found that individuals presenting with fever, altered mentation, and seizures had increased likelihoods of developing ACOs, particularly in individuals <65 yoa. After adjusting for these factors, as well discrepancies in imaging findings and medical management, we found that being ≥65 yoa remained an independent risk factor for ACOs, as compared to being a younger addult. Additionally, individuals with fever, altered mentation (GCS score <13), and/or seizures at presentation were more likely to experience ACOs.

Looking at each age group independently, worse outcomes in adults <65 yoa were found to be distinctly associated with having a fever, altered mentation, or abnormal CT findings at presentation. However, individuals ≥65 yoa were found to have a different set of clinical risk factors associated with ACOs after adjustment. Surprisingly, there was a greater percentage of abnormal CT findings in older individuals. This could possibly be due to the types of etiologies affecting older patients, physiological changes that may have impacted interpretations [25], or increased numbers of artifacts [26]. Conversely, individuals with a greater number of comorbidities (CCI > 1) were found to be associated with a lower risk of ACOs. We suspect that this finding is due to individuals with diagnosed comorbidities likely already being established in the health-care system, indicating that they have the means and access to health care. Those without access are more likely to have comorbidities that are undiagnosed and poorly controlled. Additionally, those with limited access to health care would delay their presentation when encephalitis symptoms develop.

The strengths of this study are featured in its large sample size, volume of details collected during review, and its strict adherence to IEC guidelines during the case vetting procedure. The nature of our case definition and the limitations of the retrospective study design likely contribute to an under-ascertainment of several variables, particularly rare or nonroutine diagnoses. Additionally, our screening for autoimmune assays was restricted to NMDA-related inflammation, and we did not seek out a more robust profile of etiologies, thus underestimating our overall interpretation of autoimmune encephalitis as a whole. The duration of our study spans across 17 years, where major changes in encephalitis etiologies, standards of testing, and guidelines for treatment could have changed drastically and repeatedly. Lastly, it is also important to note the geography of our study, which was located in the southern United States, where there may be a unique prevalence of mosquito-borne encephalitis that may not be generalizable to other areas that are not endemic or have higher prevalences of other forms of arboviruses.

CONCLUSION

After adjusting for other risk factors, there was a greater risk of ACOs in older adults, compared to younger adults. When etiology was determined, older adults were more likely to develop encephalitis secondary to a viral agent. However, testing and treatment remain poor in all age groups, and particularly in older adults, in regards to the frequency of testing for causative agents and empiric steroid administration. In addition to age, confirmed fever, altered mentation, and seizures at presentation were all independent risk factors for ACOs.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Financial support. This work was supported by the Grant A Starr Foundation.

Potential conflicts of interest. R. H. has received grants and personal fees from, and is a speaker for, Biofire. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References