Mortality in Cerebral Venous Thrombosis: Results from the National Inpatient Sample DatabaseNasr D.M.a · Brinjikji W.b · Cloft H.J.b · Saposnik G.c · Rabinstein A.A.a
Departments of aNeurology and bRadiology, Mayo Clinic, Rochester, Minn., USA; cDepartment of Neurology, University of Toronto, Toronto, Ont., Canada Corresponding Author
Background: Outcomes of cerebral venous thrombosis (CVT) vary from full recovery to death. Few studies have been performed examining epidemiologic and medical risk factors associated with high mortality in CVT. In this study, we examined the National Inpatient Sample (NIS) to determine the epidemiologic and medical risk factors associated with increased mortality from CVT. Materials and Methods: Using the NIS from 2001 to 2008, patients who suffered from CVT were identified using the ICD-9 codes 437.6 (nonpyogenic thrombosis of intracranial venous sinus), 325 (phlebitis and thrombophlebitis of intracranial venous sinuses) and 671.5 (peripartum phlebitis and thrombosis, cerebral venous thrombosis, thrombosis of intracranial venous sinus). We analyzed the associations of demographic factors, risk factors, comorbidities, complications of CVT, and therapeutic interventions with in-hospital mortality. We performed a multivariate logistic regression analysis to determine which variables were independently associated with in-hospital mortality. Results: 11,400 patients were hospitalized with CVT between 2001 and 2008. Two-hundred and thirty-two (2.0%) suffered in-hospital mortality. Patients 15–49 years old had the lowest mortality rate (1.5%) compared with 2.8% for patients aged 50–64 (p < 0.001) and 6.1% for patients ≥65 years old (p < 0.001). The most common condition associated with CVT was pregnancy/puerperium (24.6%), and these women had a low mortality rate (0.4%). On multivariate analysis, the comorbidity most strongly associated with increased risk of mortality was sepsis (mortality rate 15.6%, OR = 7.5, 95% CI = 4.79–11.53, p < 0.001). Malignancy, underlying autoimmune disease and substance abuse were also independently associated with mortality, but with lower mortality rates (<5%). Complications associated with increased risk of mortality included paralysis (8.0%, OR = 3.4, 95% CI = 3.17–6.96, p < 0.001), intracranial hemorrhage (8.7%, OR = 5.4, 95% CI = 4.38–7.96, p < 0.001), and hydrocephalus (15.0%, OR = 3.2, 95% CI = 5.54–15.11, p = 0.004). Demographic variables associated with decreased mortality on multivariate analysis were male gender (2.1%, OR = 0.62, 95% CI = 0.43–0.87, p = 0.006) and Asian/Pacific Islander race (OR = 0.00, 95% CI = 0–0.27, p < 001). Conclusions: CVT is associated with a low in-hospital mortality rate. Amongst patients suffering CVT, male gender and Asian/Pacific Islander race were independently associated with lower odds of in-hospital mortality when compared to their female and white counterparts, respectively. Septic patients with CVT have the greatest risk of in-hospital mortality. Hydrocephalus, intracranial hemorrhage, and motor deficits are also associated with higher risk of death. Our results build on previous evidence that serves to define a group of patients with CVT at high risk of early death.
© 2013 S. Karger AG, Basel
Cerebral venous sinus thrombosis (CVT) is relatively rare, accounting for 0.5–1% of all strokes . It most commonly afflicts young women, but can occur at any age and may be encountered in the hospital in patients with severe infections and malignancies causing hypercoagulability [1,2,3]. Although prognosis is usually favorable and mortality is relatively low, the outcome may vary [4,5,6]. There are limited data on the risk of death among hospitalized patients with CVT [4,5,6]. Factors influencing the chances of mortality in these patients have not been extensively studied.
The goals of this study were (1) to determine the risk of death in hospitalized patients with CVT using the large registry of the National Inpatient Sample (NIS), and (2) to identify the magnitude of effect for different clinical factors associated with a fatal outcome.
Materials and Methods
We analyzed the NIS hospital discharge database for the period 2001–2008 from the Healthcare Cost and Utilization Project of the Agency for Healthcare Research and Quality, Rockville, Md., USA. The NIS is a hospital discharge database that represents 20% of all inpatient admissions to nonfederal hospitals in the United States. Diagnoses, comorbidities and procedures for each admission were identified using International Classification of Diseases 9th Revision (ICD-9) codes and Clinical Classification Software (CCS) codes. CCS codes are employed by the NIS database to collect similar diagnoses that span numerous ICD-9 codes into one or several codes in order to ease statistical analysis. Further description is provided at http://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccsfactsheet.jsp. Patients who suffered from CVT were identified using the ICD-9 codes 437.6 (nonpyogenic thrombosis of intracranial venous sinus), 325 (phlebitis and thrombophlebitis of intracranial venous sinuses) and 671.5 (peripartum phlebitis and thrombosis, cerebral venous thrombosis, thrombosis of intracranial venous sinus).
Patients were stratified by age group (15–49, 50–64 and ≥65 years old), race/ethnicity (Caucasian, African-American, Hispanic, Asian-American/Pacific Islander), and gender. Comorbidities analyzed included pregnancy/puerperium, thrombophilia (ICD-9 286.9), autoimmune disease (CCS diagnosis code 210–211), sepsis (CCS diagnosis code 2), substance abuse (CCS diagnosis code 660–661), head trauma (CCS diagnosis code 233–235) and malignancy (CCS diagnosis code 11–44). We also analyzed the frequency of a number of complications and clinical sequelae of CVT including seizure (CCS diagnosis code 83), paralysis (CCS diagnosis code 82), intracranial hemorrhage (ICH) (ICD-9 diagnosis code 431) and hydrocephalus (ICD-9 diagnosis code 3313–3314). The frequency of angiography (ICD-9 procedure code 8841), fibrinolytic therapy (ICD-9 procedure code 9910), ventricular drainage (ICD-9 procedure code 231–239), and craniectomy (ICD-9 procedure code 121–201) was also determined.
Our primary outcome was in-hospital mortality. The frequency of in-hospital mortality was determined for the various demographic factors, comorbidities, complications and interventions.
All continuous variables are presented as means ± standard deviations. For univariate analyses, Student’s t tests were performed for continuous variables and χ2 analysis was performed for categorical variables. When analyzing the primary outcome of mortality, we calculated the odds ratios and 95% confidence intervals for each demographic factor, comorbidity, complication and intervention. A multivariate logistic regression analysis was performed to discriminate independent associations. Associations with a p value <0.01 were considered statistically significant. Weights were applied to each admission per recommendations from the NIS to make the data nationally representative. All statistical analyses were performed using the SAS-based statistical software package JMP 9.0 (www.jmp.com).
Between 2001 and 2008, a total of 11,400 hospitalized patients age ≥15 had a diagnosis of CVT. The mean age of patients diagnosed with CVT was 38.1 ± 34.8 years (minimum age = 15, maximum age = 94). There was a female predominance (8,938 patients, 78.3%). The racial/ethnic distribution in order of decreasing frequency was Caucasian (5,595 patients, 70.0%), African-American (1,193 patients, 14.9%), Hispanic (977 patients, 12.2%) and Asian-American/Pacific Islander patients (229 patients, 2.9%). In-hospital mortality occurred in 232 (2.0%) cases.
The results of the univariate analyses assessing the associations of demographic factors, comorbidities, complications, and interventions with in-hospital mortality are shown in table 1. Age was directly related to mortality; the mortality rate was 1.5% in patients 15–49 years, 2.8% in patients 50–64 years (p < 0.001), and 6.1% in patients 65 years and older (p < 0.001). There were no significant differences in the risk of mortality related to gender. Asian-American/Pacific Islanders had much lower mortality, but this finding was based on a much smaller number of patients of this racial extraction.
|Table 1. Demographic factors, comorbidities, complications and interventions in patients with CVT: frequency and in-hospital mortality rates|
Pregnancy/puerperium was the most common condition associated with the diagnosis of CVT (2,801, 24.6%), but the mortality rate was very low in these cases (0.4%). The most common complications were seizure (14.7%) and ICH (6.4%). The most common intervention performed for CVT patients was angiography (14.2%). The comorbidities associated with significantly higher mortality rates with CVT were sepsis, malignancy, substance abuse, and underlying connective tissue disease. The mortality rate in patients who did not suffer head trauma (2.0%) was significantly lower than the mortality rate in patients suffering head trauma (7.5%, p = 0.01). In a sensitivity analysis, the mortality rate related to sepsis was similarly high (29/161, 17.9%) when the population was restricted to patients with ICD-9 code 437.6 (nonpyogenic thrombosis of intracranial venous sinus). The symptoms/complications associated with a significantly higher mortality rate were hydrocephalus, ICH and paralysis. Patients necessitating ventricular drainage (147, 1.3%) and craniectomy (190, 1.7%) had the highest mortality rates (30.0 and 24.0%, respectively).
The results of the multivariate analysis are displayed in table 2. Male gender, Asian-American/Pacific Islander race and pregnancy/puerperium were associated with decreased mortality. Sepsis, malignancy, substance abuse, autoimmune disease, presence of paralysis, ICH and hydrocephalus were associated with increased mortality. The likelihood of mortality increased with age, but the associations between mortality and age groups did not retain significance on multivariate analysis.
|Table 2. CVT mortality: multivariate analysis|
In the present study, we analyzed the NIS registry and identified the largest number of adult patients with CVT reported to date. The overall in-hospital mortality rate was low (2.0%). Our analysis showed known and new variables associated with increased risk of fatal outcome. These variables were sepsis, malignancy, substance abuse, autoimmune disease, motor deficits, ICH and hydrocephalus. The identification of factors associated with higher mortality can be beneficial to select candidates for more aggressive treatment early after diagnosis of CVT.
The mortality rate in our cohort (2%) is comparable to 30-day mortality rates found in previous large – albeit still much smaller – series of CVT patients. In the International Study of Cerebral Vein and Dural Sinus Thrombosis (ISCVT) the 30-day mortality rate was 3.4% (21/624) , in a Dutch study 3.3% (3/90) , and in a Mexican cohort 3% (2/59) . Several of the factors associated with early mortality in this study are consistent with previous reports. Older age, underlying malignancy, motor deficits and ICH on admission were associated with death or dependency in the ISCVT cohort as they were in our study [4,7,8]. Lower mortality risk was also reported in the analysis of a Mexican registry . Motor deficits and ICH were independent predictors of unfavorable outcome at discharge in a Pakistani series . Coma or altered mental status on admission, thrombosis of the deep venous system, and posterior fossa hemorrhagic lesions were also associated with poor prognosis in the ISCVT cohort [4,7,8,10], but we could not reliably assess for the presence of these factors in our population. Central nervous system infection was a marker of poor outcome in the ISCVT study and sepsis had one of the strongest associations with early mortality in our analysis. A previously published study using the NIS demonstrated that predictors of mortality were age, ICH, hematologic diseases, malignancy and central nervous system infection . Our study provides additional data on comorbidities associated with increased mortality, such as sepsis, substance abuse, autoimmune disease, motor deficits and hydrocephalus. In addition, our study demonstrates that on multivariate analysis, age was not predictive of increased mortality and that male gender and Asian/Pacific Islander race are associated with decreased mortality rates.
The strong association of sepsis with in-hospital mortality in our study deserves to be underscored. Mortality is known to be high in cases of septic thrombosis of the cavernous and superior sagittal sinuses . Using the NIS, Haghighi et al.  found the mortality associated with pyogenic CVT was slightly higher than in the nonCVT group (4.55 vs. 3.52%). Outcomes were poorer in a Middle Eastern series with a higher rate of central nervous system infection . In our study, we could not determine how often sepsis was related to central nervous system infection. However, the mortality rate remained high when we restricted our analysis to patients with the diagnostic code for nonpyogenic CVT. Thus, based on our results, septic patients with CVT are at higher risk of death regardless of the source of infection. This frequency and implications of this association deserve attention and further investigation.
The increased risk of CVT with pregnancy and puerperium is well known [1,2,3,4,6,13]. In our multivariate analysis, odds of mortality were higher in female patients; however, the risk of mortality in the setting of pregnancy and puerperium was low. Racial/ethnic differences in CVT outcome have not been previously investigated. Higher rates of poor outcome have been reported in Middle Eastern series [9,14], but the significance of these findings is unclear in the absence of comparative studies across racial groups. In our cohort, Asian-Americans/Pacific Islanders had much lower mortality than Caucasians; however, the number of cases of CVT was much smaller in the former racial group rendering the findings difficult to interpret.
The main strength of our study lies in the size of the population analyzed, which was much larger than all cohorts previously reported. This allowed us to investigate for independent associations in multivariate analysis. However, this study also has several limitations. It is based on an administrative database and therefore it is susceptible to coding errors. Case ascertainment is also a serious limitation. Even with advanced imaging techniques, the diagnosis of CVT can still be difficult to make due to factors such as anatomic variations and radiologist experience. While our study demonstrated a mortality rate similar to other, larger studies of CVT, we also demonstrated markedly lower ICH and seizure rates when compared to some studies [4,6,9]. While the reasons behind this are unclear, explanations include the possibility that cases included in this database were less severe than those included in larger studies, underreporting or undercoding of seizures and ICH and the possibility that patients included in this database may not have been routinely screened for ICH and seizure as they often are in these studies. Various variables of interest were not available for our analysis, including the severity of clinical presentation (e.g. presence of altered level of consciousness), deep-venous system involvement, and causes of death (i.e. neurological versus nonneurological). We also lacked data on death after discharge and the functional outcome of survivors. Yet, we do not think that these limitations negate the validity of the associations reported.
Our results build on previous evidence that serves to define a group of patients with CVT at high risk of early death. Current guidelines favor the prompt initiation of anticoagulation upon diagnosis of CVT and also antibiotics in cases of septic sinus thrombosis [2,15]. However, they also advise to consider more aggressive interventions in patients with unfavorable prognostic indicators who fail to improve rapidly with the initial medical treatment. Endovascular thrombectomy and thrombolysis [16,17] and decompressive craniectomy  can be effective options in selected patients with severe complications resulting from CVT although the value of these interventions needs to be further studied.
Alejandro A. Rabinstein, MD
Mayo Clinic, Department of Neurology, W8B
200 First Street SW
Rochester, MN 55905 (USA)
Received: June 25, 2012
Accepted: September 14, 2012
Published online: February 14, 2013
Number of Print Pages : 5
Number of Figures : 0, Number of Tables : 2, Number of References : 17
Vol. 35, No. 1, Year 2013 (Cover Date: February 2013)
Journal Editor: Hennerici M.G. (Mannheim)
ISSN: 1015-9770 (Print), eISSN: 1421-9786 (Online)
For additional information: http://www.karger.com/CED