A Population-Based Study of the Incidence and Case Fatality of Intracerebral Hemorrhage of Undetermined Etiology

Abstract

Background

There is a paucity of reliable recent data regarding epidemiology of intracerebral hemorrhage (ICH) of undetermined etiology in population-based studies.

Objectives

To determine the incidence and case fatality of ICH of undetermined etiology using a population-based design.

Methods

Medical records and neuroimaging data of all patients with ICH from Stearns and Benton Counties, Minnesota, between June 1st, 2012 and June 30th, 2014 were reviewed. Patients with a first-time diagnosis of ICH were categorized as of undetermined etiology if ICH was without features typical of hypertensive etiology with normal or no magnetic resonance imaging (MRI)/angiograms. We calculated the incidences of [1] probable and possible hypertensive ICH; [2] related to arteriovenous malformation, cavernous malformation, or aneurysmal rupture (angiographic or MRI diagnoses); [3] secondary to anticoagulation; and [4] of undetermined etiology adjusted for age and sex based on the 2010 US census.

Results

Of the 50 identified ICHs among 136,654 resident populations, seven were true incident cases of ICH of undetermined etiology in this population-based study. The age- and sex-adjusted incidence of ICH of undetermined etiology was 2.6 [95% confidence interval (CI) 0.7–4.9] per 100, 000 person-years, which was lower than probable and possible hypertensive ICH incidence of 12.8 [95% CI 8.4–17.2] per 100,000 person-years. The age-adjusted case fatality rate at 1 month was 8.14 and 0.4 per 100,000 persons for probable and possible hypertensive ICHs and ICHs of undetermined etiology, respectively.

Conclusions

Our results should prompt further studies into identification of causes in ICH patients presently classified as ICH of undetermined etiology to reduce the incidence and case fatality of such ICHs.

INTRODUCTION

Intracerebral hemorrhage (ICH) can occur secondary to degenerative changes in small arteries induced by chronic hypertension or arteriovenous abnormalities, such as arteriovenous malformations or cavernous malformations [1]. Stroke of undetermined etiology has been used to designate ischemic strokes in which no etiology can be identified [2]. In approximately 10%-20% of the patients with ICH [3,4], no underlying etiology can be identified. The rate of recurrent hemorrhage and mortality is lower in patients with ICH in whom no underlying etiology is identified. Current data are available through single center studies [3,4] which can be influenced by referral patterns and diagnostic paradigms. We performed this study to determine the actual incidence of ICH of undetermined etiology in a population-based cohort using a standard definition and associated case fatality.

METHODS

The methodology of the population-based studies involving Stearns and Benton Counties situated in central Minnesota has been previously described [5]. The population is derived from a total area of 1,751 square miles or 1,120,915 acres. There were 189,093 people, 71,653 households, and 46,396 families residing within 38 cities and 46 townships in both counties according to the US 2010 census. The racial/ethnic makeup of both counties was 92.3% White, 2.9% African American, 0.3% Native American, 1.8% Asian, 0.03% Pacific Islander, and 2.7% from other categories, 2.6% of the population were Hispanic or Latino of any race. The median age was 33.7 years and 23.5% were under the age of 18 years, 14.6% from 18 to 24, 25.5% from 25 to 44, 24.40% from 45 to 64, and 12.1% who were ≥65 years of age as previously described [5]. The estimated 1.0% and 2.0% population increases between April 1, 2010 to July 1, 2013 in Stearns and Benton counties was very small to affect calculation of incidences using US 2010 census and case ascertainment in the following four years [6]. St. Cloud Hospital is the only acute care hospital and certified Primary Stroke Center with 489 licensed beds for both Stearns and Benton Counties. The hospital is the only hospital that provides neurosurgical services within Central Minnesota and all ICHs within the catchment area are admitted to St. Cloud Hospital.

Case Ascertainment

Patients admitted with ICH from June 1, 2012 to June 30, 2014 were identified through several overlapping sources. A list of patients admitted with primary or secondary diagnoses of ICH (ICD-9-CM) code 431–432.9 was retrieved from data entered in the Get With The Guidelines®-Stroke registry. The medical records of all patients with above-mentioned diagnoses codes were reviewed. Patients who had intraparenchymal hemorrhage with or without intraventricular hemorrhage were identified. Patients who had primary residence within the zip codes included in Stearns and Benton Counties were included in further analysis. One physician (MRA) classified all ICHs into the following categories described previously as follows [7] “(1) probable hypertensive ICH defined on the basis of presence of established or newly diagnosed hypertension with hematoma at typical sites for hypertensive bleeding (basal ganglia, thalamus, or pons) or at sites less likely to be associated with hypertension or to have a high frequency of underlying vascular lesions (lobar, cerebellar, or intraventricular), after other etiologies were ruled out with conventional angiography and/or magnetic resonance imaging [MRI]; (2) possible hypertensive ICH defined on the basis of presence of established or newly diagnosed hypertension with hematoma at sites less likely to be associated with hypertension or to have a high frequency of underlying vascular lesions (lobar, cerebellar, or intraventricular) without exclusion of other etiologies; (3) related to arteriovenous malformation; (4) secondary to aneurysmal rupture (categories 3 and 4 were angiographic or MRI diagnoses); (5) related to cavernous malformation diagnosed by MRI; (6) secondary to thrombolysis or anticoagulation; and (7) of undetermined etiology defined as ICH without features typical of hypertensive etiology and normal or no angiograms."

Data Collection

Patients admitted to St Cloud Hospital with the diagnosis of ICH were identified using above-mentioned methodology. The methodology of medical records review was approved by local Institutional Review Board. Within EPIC electronic medical record system, patient charts were reviewed for information regarding demographic and clinical characteristics, preexisting risk factors, Glasgow Coma Scale at admission, and location of ICH and the presence or absence of intraventricular hemorrhage based on initial CT scan findings. The location of ICH was categorized into one of the five categories: lobar, basal ganglionic, thalamic, pontine, and cerebellar. Details regarding surgical evacuation and intraventricular catheter placement, and details of medical treatment given were also extracted. Length of hospital stay and functional status defined by modified Rankin scale (mRS) at discharge was also ascertained. All vital status was ascertained by review of in-hospital and clinic records or by direct contact with patient or family by the treating physician to determine mortality at 1 and 3 months.

Sensitivity Analysis

A list of patients admitted with primary diagnoses of ICH (ICD-9-CM) code 431 and 432.9 and had primary residence within the 46 zip codes included in Stearns and Benton Counties was retrieved from Minnesota Hospital Association data from June 1, 2012 to December 31, 2013. The methodology of data collection by Minnesota Hospital Association has been previously described [8,9]. A total of 48 patients with aforementioned ICD-9-CM codes were admitted from study population of which 46 were admitted to St. Cloud Hospital during the same period of time providing a yield of 95.8%.

Data Analysis

We calculated age-adjusted incidence rates (age-adjusted to the 2010 US standard population) for all ICHs, and various subtypes of ICH as previously described [5]. Briefly, the number of ICH subtypes within each of the 14 age and gender groups were ascertained and used to calculate crude rates using population data with the counts for each of the 14 age/gender groups expressed as the number of ICH events per 100,000 population at risk [10]. The age-adjusted rate was calculated by multiplying each crude rate by the appropriate weight and summing the products.

We calculated the case fatality rate and mortality rate within 1 and 3 months of ICH onset for all ICHs, and ICH subtypes. We calculated age-specific ICH-related death rate (ASDR) per 100,000 population for each age group as follows: ASDR = deaths in age group + estimated population within of that age group × 100,000. Each ASDR was then multiplied by the proportion of the standard population within that same age group. The age-specific results were summed to get the age-adjusted death rate for ICH subtype-related deaths as follows: AADR = Summation of (ASDR × standard proportion) [11].

RESULTS

Of the 50 identified ICH patients, seven were true incident cases of ICH of undetermined etiology. The numbers of cases of other ICH subtypes were as follows: possible hypertensive (n = 7), probable hypertensive (n = 26), and secondary to thrombolysis or anticoagulation (n = 10) in this population-based study. Mean age (±SD) at diagnosis of patients with ICH of undetermined etiology was 56.9 years (±19.5 years) compared with all ICHs 70.1 (±14.9) years. The overall incidence of any ICH and ICH of undetermined etiology was 36.6 [95% CI: 26.5–46.7] and 5.12 [95% CI: 1.3–8.9] per 100,000 person-years, respectively. The age- and sex-adjusted incidence of any ICH was 18.9 [95% CI: 13.7–24.1] per 100, 000 person-years). The age- and sex-adjusted incidence of ICH of undetermined etiology was 2.6 [95% CI: 0.7–4.9] per 100, 000 person-years which was lower than probable and possible hypertensive ICH incidence of 12.8 [95% CI: 8.4–17.2 ] per 100,000 person-years.

The clinical characteristics of patients with ICH of undetermined etiology are summarized in Table 1. At admission, 2 of 7 patients with ICH of undetermined etiology and 6 of 33 patients with possible and probable hypertensive ICHs were categorized as poor grade (Glasgow Coma Scale score of <8 at admission), respectively. The location of ICH among patients with ICH of undetermined etiology and those with possible or probable hypertensive ICHs was as follows: lobar (42.8 % and 57.5 %), basal ganglia (14.2 %, and 12.1 %), intraventricular hemorrhage (0.0 % and 3.03%), cerebellar (14.2% and 12.1 %), pontine (0.0% and 3.03%), and thalamus (28.5 % and 12.1 %), Among survivors with ICH of undetermined etiology, mild (mRS 1–2) and moderate (mRS 3–5) disability was seen in 3 and 3 patients, respectively, Among survivors with possible and probable hypertensive ICHs, mild (mRS 1–2) and moderate (mRS 3–5) disability was seen in 5 and 16 patients, respectively. The unadjusted 1-month case fatality rate of ICH of undetermined etiology was significantly lower compared with possible and probable hypertensive ICHs (0.73 vs. 7.31 per 100,000 person-years, p<0.0001). The 3-month age-adjusted mortality rate was lower among patients with ICH of undetermined etiology (0.78 per 100,000 persons) compared with possible and probable hypertensive ICHs (8.96 per 100,000 persons).- The age adjusted case fatality rate at 1 month was 8.14 and 0.4 per 100,000 persons for probable and possible hypertensive ICHs and ICHs of undetermined etiology, respectively.

Table 1. Clinical and imaging characteristics of patients with intracerebral hemorrhage of undetermined etiology:

Age/gender	Cardiovascular risk factors	Initial Glasgow Coma scale score	Location	Intraventricular hemorrhage	Diagnostic tests	Surgical procedure performed/ mechanical ventilation	Days of hospitalization	mRS at discharge
84/male	History of stroke, atrial fibrillation, no anticoagulation	3	Basal ganglia	None	CT	None	3	5
49/male	None	15	Lobar	None	CT, angiogram	Mechanical ventilation	10	3
74/female	None	15	Thalamus	None	CT	None	2	2
51/male	None	15	Thalamus	None	CT	None	4	2
23/male	None	15	Lobar	None	CT, MRI, CTA, angiogram	None	2	1
56/female	None	3	Cerebellum	None	CT	Mechanical ventilation	1	6
61/female	None	15	Lobar	None	CT, transcranial Doppler, MRI	None	3	4

Abbreviations: CT, computed tomographic scan; MRI, magnetic resonance imaging; CTA, computed tomographic angiogram; mRS, modified Rankin scale.

DISCUSSION

The proportion of patients with ICH of undetermined etiology can vary from 10–35% depending upon the cohort studied and definition used [3,4,7,12]. In our study, 14.0 % of the patients with ICH were designated as ICH of undetermined etiology which is similar to the 18% rate identified in a previous study using similar definition [7]. The rate of ICH of undetermined etiology was 21% in a single-center study of a total of 1013 patients using SMASH-U criteria, structural lesion, medication, amyloid angiopathy, systemic/other disease, hypertension, undetermined [4]. The SMASH-U classifications designate lobar and subcortical ICHs in patients aged ≥55 years as related to amyloid angiopathy after exclusion of other etiologies. In another study of 3,785 ICH patients, 10% were classified as ICH of undetermined etiology based on SMASH-U classification [3]. Based on SMASH-U classification, one patient with lobar ICH would have been classified as amyloid angiopathy. The rate of utilization of angiographic studies varies among these studies but the diagnostic yield of vascular neuroimaging for structural lesions associated with ICH was low (13%) in the 80 patients who underwent conventional angiography (positive yield in 5 of 55) and/or MRI (positive yield in 6 of 43) [7]. Most patients were classified in etiologic subtypes based on location of ICH, and patients’ demographic and clinical features.

The overall crude incidence of any ICH was 36.6 per 100,000 in our study which is similar to the worldwide incidence of intracerebral hemorrhage ranging from 10 to 20 cases per 100,000 population [13]. The age- and sex-adjusted incidence of ICH of undetermined etiology was 2.6 [95% CI: 0.7–4.9] per 100, 000 person-years which was lower than probable and possible hypertensive ICH incidence of 12.8 [95% CI: 8.4–17.2] per 100,000 person-years. We found that the unadjusted 1- and 3-month case fatality rate of ICH of undetermined etiology was significantly lower compared with possible and probable hypertensive ICHs. The immediate and 5-year survival of patients with ICH of undetermined etiology was lower than that observed for patients with hypertensive ICH [3]. In another study, the immediate survival was lower among patients with ICH of undetermined etiology, but 5-year survival was similar compared with those with hypertensive etiology [4]. At admission, the percentage of poor grade patients in patients with ICH of undetermined etiology was higher than that observed in patients with possible and probable hypertensive ICHs (33.3% vs. 18.2%). The higher mortality may be related in part to the differences in proportion of poor grade patients [14, 15].

Our largest limitation was small number of events which was a consequence of restricting the denominator population to two counties and one hospital. Such a restriction resulted in high accuracy of the diagnosis of ICH of undetermined etiology and avoidance of variations in management between hospitals [5]. However, the precision of estimates is lower than that seen with a larger number of events. The relatively low precision of estimate due to large contribution of single events in the analysis can exaggerate the incidence and disability estimates associated with ICH of undetermined etiology. The small number of events also reduced our ability to detect differences in incidences between men and women and other subgroups. None-the-less, the incidence and mortality associated with ICH of undetermined etiology were highlighted by the results of our study.

We hope that our results will prompt further studies into identification of causes in ICH patients presently classified as ICH of undetermined etiology. Long-term ambulatory blood pressure monitoring [16,17], and detailed histological analysis of hematomas and perihematoma tissue in patients undergoing surgical evacuation may be potential next steps. Without such efforts, the incidence of ICH of undetermined etiology will continue to either increase or remain unchanged.

Conflict of Interest

The authors declare that they have no conflict of interest.

Financial Disclosure

None