Neurological Complications after Tuberculous Meningitis in a Multi-State Cohort in the United States

Abstract

Background and Purpose

To assess the rate of neurological complications and mortality after tuberculous meningitis in the United States.

Methods

The authors performed a retrospective cohort study of all patients 18 years or older hospitalized for tuberculous meningitis in California between 2005–2010, New York between 2006–2012, and Florida between 2005–2012. Outcomes of interest were mortality and the following neurological complications: stroke, seizure, hydrocephalus requiring a ventriculoperitoneal shunt, vision impairment, and hearing impairment. Kaplan-Meier survival statistics were used to assess the cumulative rate of neurological complications and death. Cox proportional hazards regression was used to compare rates of complications in patients with and without human immunodeficiency virus (HIV) after adjustment for comorbidities.

Results

806 patients with tuberculous meningitis were identified, among whom the cumulative rate of any complication or death was 55.4% (95% CI, 51.5–59.3%). More than two-thirds of complications occurred during the initial hospitalization for tuberculous meningitis. Individual neurological complications were not uncommon: the cumulative rate of stroke was 16.8% (95% CI, 14.0–20.0%), the rate of seizure was 18.8% (95% CI, 15.4–22.8%), and the rate of ventriculoperitoneal shunting was 8.4% (95% CI, 6.4–10.9%). Vision impairment occurred in 21.6% (95% CI, 18.5–25.1%) of patients and hearing impairment occurred in 6.8% (95% CI, 4.9–9.4%). The mortality rate was 21.5% (95% CI, 18.4–24.9%). Patients with HIV infection were not at increased risk of complications compared to patients without HIV (hazard ratio, 1.2; 95% CI, 0.9–1.6).

Conclusions

Tuberculous meningitis is associated with significant risk of neurological complications and death even in the United States.

Introduction

Tuberculosis (TB) is a global disease epidemic, with an estimated 9.6 million new cases occurring in 2014 [1]. Tuberculous meningitis (TBM) occurs in 1–5% of patients with TB and is characterized by a progressive granulomatous inflammation of the basal meninges. TBM is associated with a mortality of 19–28% in HIV-uninfected patients and 40–58% in HIV-infected patients [2–5]. TBM also frequently leads to neurological complications including hydrocephalus [6–11], stroke [9, 10, 12–24], seizure [5, 6, 9, 24], vision loss [5, 25, 26], and hearing loss [27–29]. However, the majority of studies on TBM have been performed in resource-limited settings where there may be suboptimal prevention, diagnosis, and treatment of the disease [2–5]. Furthermore, the average length of follow-up in previous studies was relatively short and therefore may not have accurately captured delayed neurological complications [2–5]. Our aim was to evaluate the mortality and long-term risk of neurological complications after TBM in a high-resource setting such as the United States using a large heterogeneous cohort of patients.

Methods

Design

The authors performed a retrospective cohort study using administrative claims data from all emergency department (ED) visits and hospitalizations at nonfederal acute care hospitals in California from 2005 to 2011, New York from 2006 to 2013, and Florida from 2005 to 2013. Trained analysts used standardized methods to collect data regarding discharges and reported these to state health agencies for regulatory purposes. After quality checking, these data were provided in a de-identified format to the Agency for Healthcare Research and Quality for its Healthcare Cost and Utilization Project [30]. Each patient was assigned an anonymous, unique linkage number that allowed for longitudinal tracking of ED encounters and hospitalizations [30, 31]. The Weill Cornell Medicine institutional review board approved our analysis of these data.

Patient Population

Our cohort consisted of all patients 18 years or older discharged at the time of their first recorded hospitalization for TBM, as defined by the International Classification of Diseases, 9^th Revision, Clinical Modification (ICD-9-CM) code 013.x in any discharge diagnosis position. Non-residents of New York, Florida, and California were excluded to maximize follow-up. In order to ensure at least 1 year of follow-up, patients whose index visit for TBM occurred in 2013 in New York or Florida or 2011 in California were excluded.

Measurements

Our outcomes of interest were death and the following neurological complications: stroke, seizure, hydrocephalus requiring ventriculoperitoneal shunt (VPS), vision impairment, and hearing impairment. Stroke was a composite of both ischemic and hemorrhagic stroke; ischemic stroke was defined using ICD-9-CM codes 433.x1, 434.x1, or 436 in any diagnosis code position in the absence of a primary discharge code for rehabilitation (V57) or any codes for subarachnoid hemorrhage (430), intracerebral hemorrhage (431) or trauma (800–804 and 850–854). Hemorrhagic stroke was defined by the presence of a discharge code for intracerebral hemorrhage (431) or subarachnoid hemorrhage (430) in the absence of a primary discharge code for rehabilitation (V57) or trauma (800–804 and 850–854). This algorithm has been validated to have a sensitivity of ≥82% and a specificity of ≥93% for both ischemic and hemorrhagic stroke subtypes [32]. Seizure was defined using ICD-9-CM codes 345.x in any discharge diagnosis position; this schema has a positive predictive value ranging from 84–98% in adult patients [33, 34]. Hydrocephalus was considered as only those cases that required a permanent VPS, defined by the presence of an ICD-9-CM procedure code 0.23 and 0.24 in any discharge diagnosis position; these codes have a 95% sensitivity and 100% specificity based on medical record review [35]. Using the methods of prior studies [36], vision impairment was defined as ICD-9-CM codes 360–379 in any discharge diagnosis position and hearing impairment as ICD-9-CM codes 380–390 in any discharge diagnosis position. Finally, subgroup analyses were performed stratified by the presence or absence of HIV, and compared the risks of complications associated with HIV after adjustment for the following covariates: age, sex, race, insurance status, HIV status, and the Elixhauser comorbidity index [37].

Statistical Analyses

Kaplan-Meier survival statistics were used to calculate cumulative rates of individual complications as well as the composite of any complication or death. Patients were censored at the time of death or at the end of the follow-up period. Cox proportional hazards regression was used to compare the likelihood of complications in patients with and without HIV after adjustment for demographics and the Elixhauser comorbidity index [38]. All analyses were performed using Stata/MP, version 13 (StataCorp, College Station, TX). The threshold of statistical significance allowed for an alpha error of 0.05.

Results

806 patients with TBM were identified. Their mean age was 50.7 (±17.1) years, 62.9% were male, and 14.5% had HIV. During a mean follow-up of 2.9 (±2.4) years, the cumulative rate of any complication or death was 55.4% (95% CI, 51.5–59.3%) (Figure 1). Patients with complications were slightly older and had a higher prevalence of medical comorbidities compared to TBM patients without complications (Table 1).

Figure 1. The Cumulative Rate of Complications or Death after Tuberculous Meningitis.

Kaplan-Meier curve showing the cumulative rate of neurological complications or death after tuberculous meningitis.

Table 1.

Characteristics of Patients, Stratified by Presence of any Complication, including Death

Characteristica	Any Complication (N = 414)	No Complication (N = 392)
Age, mean (SD), y	56.2 (15.3)	46.6 (20.1)
Female	155 (37.4)	144 (36.7)
Raceb
White	88 (21.9)	75 (20.0)
Black	97 (24.1)	82 (21.8)
Hispanic	111 (27.6)	96 (25.5)
Asian	75 (18.7)	81 (21.5)
Other	31 (7.7)	42 (11.2)
Payment source
Medicare	115 (27.8)	92 (23.5)
Medicaid	145 (35.0)	139 (35.5)
Private	112 (27.1)	98 (25.0)
Self-pay	19 (4.6)	32 (8.2)
Other	23 (5.6)	31 (7.9)
State
California	209 (52.5)	189 (48.5)
New York	107 (48.4)	114 (51.6)
Florida	98 (52.4)	89 (47.6)
HIV	63 (15.2)	54 (13.8)
Elixhauser comorbiditiesc, mean (SD)	2.9 (1.9)	2.2 (1.7)

^aData are presented as number (%) unless otherwise specified.

^bSelf-reported by patients or their surrogates. Numbers do not sum to group totals because of missing race/ethnicity data in 3.5% of patients.

^cThe Elixhauser comorbidities represent a comprehensive set of 28 comorbidity measures for use with large administrative datasets.

Abbreviations: SD, standard deviation

By 8 years, 21.5% (95% CI, 18.4–24.9%) of patients had died. Individual neurological complications were not uncommon. The cumulative rate of stroke was 16.8% (95% CI, 14.0–20.0%), of which the majority were ischemic (12.8% (95% CI, 10.3–15.7%)). The cumulative rate of seizure was 18.8% (95% CI, 15.4–22.8%) and that of VPS surgery was 8.4% (95% CI, 6.4–10.9%). Vision impairment occurred in 21.6% (95% CI, 18.5–25.1%) of patients and hearing impairment occurred in 6.8% (95% CI, 4.9–9.4%) of patients. The majority of complications occurred during the index hospitalization for TBM (Table 2), although when individually assessed, the cumulative rate of seizures, hydrocephalus requiring ventriculoperitoneal shunt, hearing impairment and death more than doubled throughout the follow-up period. Finally, the results of our study were similar when each state was analyzed separately.

Table 2.

Rates of Complications over Time

Complication	During Initial Hospitalization	6 Months	1 Year	8 Years
All Strokea	11.2% (9.0–13.3%)	13.4% (11.0–16.2%)	15.0% (12.5–18.0%)	16.8% (14.0–20.0%)
Ischemic Stroke	8.4% (6.5–10.4%)	10.5% (8.4–13.1%)	11.8% (8.6–14.5%)	12.8% (10.3–15.7%)
Seizure	7.7% (5.8–9.5%)	10.2% (8.1–12.8%)	11.9% (9.6–14.7%)	18.8% (15.4–22.8%)
VPS	3.4% (2.1–4.6%)	6.6% (4.9–8.8%)	7.2% (5.5–9.6%)	8.4% (6.4–10.9%)
Vision Impairment	13.3 % (10.9–15.6%)	17.5% (14.8–20.6%)	18.7% (15.9–21.9%)	21.6% (18.5–25.1%)
Hearing Impairment	3.0% (1.8–4.2%)	4.9% (3.5–6.8%)	5.0% (3.6–7.0%)	6.8% (4.9–9.4%)
Death	10.5% (8.4–12.7%)	16.8% (14.2–19.7%)	18.3% (15.7–21.4%)	21.5% (18.4–24.9%)
Any Complication	36.8% (33.5–40.2%)	45.7% (42.2–49.5%)	48.9% (45.3–52.6%)	55.4% (51.5–59.3%)

^aIncludes both ischemic and hemorrhagic stroke subtypes

Abbreviations: VPS, ventriculoperitoneal shunt

The cumulative rate of complications in HIV-infected individuals was 62.7% (95% CI, 51.5–74.0%) as compared to 54.1% (95% CI 50.1–58.3%) in HIV-uninfected patients. After adjusting for demographics and the Elixhauser comorbidity index, however, HIV infection was not associated with a higher risk of neurological complications or death (hazard ratio, 1.2; 95% CI, 0.9–1.6) (Table 3). Only hearing impairment was statistically more common in patients with HIV co-infection.

Table 3.

Rates of Complications, Stratified by HIV Status

Complication	HIV Positive	HIV Negative	Hazard Ratioa
All Strokeb	17.7% (11.4–26.9%)	16.6% (13.7–20.1%)	1.3 (0.7–2.3)
Ischemic Stroke	16.0% (9.9–25.1%)	12.2% (9.7–15.4%)	1.5 (0.8–2.8)
Seizure	21.7% (12.8–35.3%)	18.2% (14.7–22.3%)	0.6 (0.3–1.2)
VPS	4.9 % (1.8–13.0%)	9.0% (6.8–11.8%)	0.4 (0.1–1.2)
Vision Impairment	24.4% (16.8–34.6%)	21.1% (17.8–25.0%)	1.1 (0.7–1.7)
Hearing Impairment	11.2% (5.7–21.3%)	6.0% (4.1–8.6%)	2.3 (1.0–5.2)
Death	21.1% (14.1–30.8%)	21.5% (18.2–25.3%)	1.4 (0.9–2.5)
Any Complication	62.7% (51.5–74.0%)	54.1% (50.1–58.3%)	1.2 (0.9–1.6)

^aAdjusted for demographics and the Elixhauser comorbidity index

^bIncludes both ischemic and hemorrhagic stroke subtypes

Abbreviations: VPS, ventriculoperitoneal shunt

Discussion

In a large cohort of patients with TBM in the United States, a neurological complication or death occurred in more than half of all patients. The majority of neurological complications occurred during the index hospitalization for TBM, suggesting that the highest at–risk period occurs soon after diagnosis. Concomitant HIV infection was not associated with a higher risk of mortality or neurological complications from TBM.

Comparison with other studies

Our results should be considered in light of several important prior international studies on TBM. Previously reported rates of individual neurological complications vary significantly, potentially in part due to small study populations and differences in patterns of care that may not apply universally. Hearing loss, for example, has previously been reported in 11–54% of patients with TBM [27–29]; however, streptomycin was part of the standard TBM treatment regimen in these studies, and as a known ototoxic agent [27], may partly explain the higher rates of hearing loss. Furthermore, complication rates differ depending on access to medical resources including antimicrobial therapy and advanced diagnostic techniques. The rate of stroke after TBM varies between 15 and 57% [9, 10, 12–24], with lower numbers reported in studies where stroke was identified clinically or by CT [15, 17] and higher numbers in studies in which serial MRIs [10, 14, 16, 21, 22] or autopsies were performed [12, 13]. The previously documented high rates of neurological complications may also reflect the fact that the majority of prior studies were performed in endemic areas, which are typically resource-limited countries with major gaps in healthcare infrastructure; however, in this study more than half of TBM patients died or had a neurological complication even in the United States, a country with general availability of critical care units, specialized healthcare providers including neurosurgeons, advanced neuroimaging, modern laboratory techniques to identify TBM, access to antimicrobial therapy, and rehabilitation centers. The high rate of complications seen in TBM is likely due to several factors including the already advanced severity of disease at presentation in the majority of patients, lack of accurate diagnostic measures to identify TBM, increased prevalence of drug resistant TBM, poor central nervous system (CNS) penetration of antimicrobials, and lack of knowledge regarding the appropriate timing and dose of corticosteroids [1, 2, 39, 40]. Furthermore, as the majority of complications occurred during the index hospitalization for TBM, further investigation into the pathophysiology of neurological injury due to TBM may lead to improved methods of both risk stratification and prevention.

Impact of HIV on risk of death or neurological complications after TBM

Contrary to the majority of prior literature performed in endemic regions, we found that patients with HIV infection were not at increased risk for death or neurological complications. Prior studies, however, recruited patients who were severely immunocompromised and who had poor access to antiretroviral therapy [2, 5]. In the study by Thwaites et al., the median CD4 count was 66 cells per cubic millimeter and none of the patients were receiving antiretroviral therapy [2]. Similarly, in the study by Heemskerk et al., the median CD4 count was 38 cells per cubic millimeter and antiretroviral therapy was not initiated until eight weeks after antituberculosis therapy was started [5]. Our results are, however, consistent with prior research from high-resource countries showing no difference in mortality between patients with TBM and HIV infection compared to HIV-uninfected patients [6, 41].

Limitations

Our study has several important limitations. First, this was a retrospective study that relied on administrative data to identify complications of TBM. Although well-validated ICD-9-CM codes were used to identify neurological complications, the authors could not assess certain clinical and microbiological data such as clinical severity of TBM at presentation, presence of multidrug resistance bacteria, and type or efficacy of treatment regimen, all of which may have affected rates of complications. Similarly, there was a lack of granularity regarding clinical characteristics of outcomes and thus, it is uncertain whether complications were directly related to TBM or a consequence of concomitant medical conditions. Second, the authors lacked data on the degree of immunosuppression in patients with HIV such as CD4 count and viral load. Third, patients with TBM who presented with a complication outside of New York, Florida and California were not captured as outcomes. The authors attempted to minimize this risk by only including residents of the three states studied. Lastly, the authors could not account for out-of-hospital death, which may have led to an underestimation in the mortality rate.

Conclusions

TBM is the most devastating manifestation of TB. Neurological complications after TBM were common even in a high-resource country such as the United States. Furthermore, although incident neurological complications most often occurred during the initial hospitalization for TBM, a substantial proportion occurred after hospital discharge, emphasizing the need for close follow-up of these patients.

Highlights.

• Tuberculous meningitis is associated with a significant risk of neurological complications and death in the United States.

• A neurological complications or death occurred in over half of all patients with tuberculous meningitis.

• The majority of neurological complications or death occurred within the hospitalization for tuberculous meningitis.

• Concomitant HIV infection does not appear to be associated with a higher risk of mortality or neurological complications in patients with tuberculous meningitis.