Sex-Related Differences in Patients’ Characteristics, Provided Care, and Outcomes Following Spontaneous Intracerebral Hemorrhage

All patients admitted to the neurocritical care unit of the University Hospital of Zurich between January 2009 and December 2019 with spontaneous ICH were eligible for inclusion in this retrospective study. Inclusion criteria of the study were (1) adults (aged > 18 years) and (2) presence of spontaneous ICH. Exclusion criteria were (1) patients’ written or documented oral refusal to have their data analyzed for research projects; (2) ICH secondary to cavernoma, dural arteriovenous fistula, other arteriovenous malformation, or histopathologically confirmed amyloidangiopathy (in autopsies and surgical biopsy during hematoma evacuation); (3) ICH of traumatic etiology; and (4) sole intraventricular hemorrhage with no parenchyma involvement. To rule out ICH of secondary genesis, computed tomography angiography is routinely performed in our department when the patient is admitted.

Data were obtained from the hospitals’ electronic health records (KISIM-TM; Cistec,® Zurich, Switzerland). Demographic data collected were: sex, age, height, weight, and presence of comorbidities, based on the Charlson Comorbidity Index (CCI) (i.e., history of myocardial infarction, congestive heart failure, peripheral vascular disease, history of cerebrovascular event, dementia, chronic pulmonary disease, rheumatologic disease, gastric ulcer, liver disease, diabetes without or with chronic complications, kidney disease, and history of cancer) [11]. Collected symptoms at presentation included hypertension (defined by a systolic blood pressure greater than 160 mmHg), headache, nausea and vomiting, and seizure at onset. Data concerning surgically provided care included insertion of an external ventricular drain (EVD), and surgical hematoma evacuation. At our institution the indication for EVD insertion is made by the treating neurosurgeon on the basis of imaging and their clinical expertise. Additionally, hematoma evacuation is performed by osteoplastic craniotomy only. Endoscopic evacuation is not commonly used at our institution. Moreover, primary decompressive craniectomy is not part of our institution’s treatment protocol. The following intensive care unit (ICU) complications were included: pneumonia (defined as new pulmonary infiltrate by chest X-ray, increased inflammatory blood parameters, and/or positive bacterial culture of tracheal secretion), clinical or electroencephalographic seizure, and blood- and/or platelets transfusion. Radiological findings on the initial cranial computer tomography (CT) scan included the main location of ICH, the presence of a spot sign, perifocal edema, subarachnoid hemorrhage, signs of ventriculomegaly (acute abnormal expansion of the cerebral ventricles), and hemorrhage volume. The hemorrhage volume was calculated using the formula of (abc)/2 (a = maximum length in mm, b = perpendicular width to a in mm, and c = number of slices multiplied by slice thickness in mm) from the initial CT scan. A neuroradiologist (NN) blinded to the clinical presentation and outcome evaluated all imaging. As outcome, ICU-length of stay (in days), hospital-length of stay (in days), mortality (30 days and 1 year after initial ICH), and functional outcome, as evaluated with the modified Rankin scale (mRS) at discharge were collected. Additionally, medical records and family interviews were then searched for indicators for early withdrawal of therapy (within 72 hours) and premorbid functional status (“independent” or “special care”) before the occurence of ICH and noted if it applied. “Independent” refers to autonomy in activities of daily living, life participation, social relationships and occupational performance if the patient is at pre-retirement age. “Special care” refers to any limitations in daily activities that make autonomous living at home impossible and require admission to a nursery home.

Statistical analysis was performed using SPSS V.25. Descriptive statistics are reported as counts/percentages, mean ± standard deviation, or median and interquartile range, as appropriate. For the analysis of associations, patient characteristics were dichotomized depending on either sex or the insertion of an EVD, or the outcome (mRS score 0–3 vs. mRS 4–6). All continuous data were tested for normality using Shapiro–Wilk’s test. Categorical variables were compared using Pearson’s χ² test or Fisher’s exact test. Continuous/ordinal variables compared using Student’s t-test or Mann–Whitney U-test for parametric and non-parametric data, respectively, as appropriate. Multivariate binomial logistic regression was performed to ascertain the independence of the associations found on the likelihood of insertion of EVD or on the likelihood of unfavorable outcome in the prior univariate analysis. Multivariate binomial logistic regression was performed to ascertain the independence of the associations found on the likelihood of insertion of EVD in the prior univariate analysis. Significance was defined as the probability of a two-sided type 1 error being < 5% (P < 0.05).

When we considered the hemorrhage site location, more women than men presented with lobar ICH (P = 0.007). Within lobar ICH, women were more likely to present with frontal and temporal ICH, and men presented more frequently with occipital hemorrhages. In addition, the frequency of deep ICH (basal ganglia or thalamus) did not differ among men and women (P = 0.224). Similarly, the frequency of the infratentorial ICH (e.g. brainstem or cerebellum) did not differ between sexes. Hemorrhage volume, hematoma expansion, presence of subarachnoid hemorrhage, ventriculomegaly, perifocal edema, intraventricular hemorrhage, and midline shift of parenchyma the ventricle on the first head CT-scan were not significantly different between groups (Table 2).

Even though, seizures at onset were not more prevalent in women, they received anti-epileptic drugs significantly more often than men (P = 0.007). Among the other conservative medical treatments (such as administration of blood and/or platelet conserves, frequency of intubation and days on mechanical ventilation), no differences were found among men and women (Table 3).

When we considered surgical care, an EVD was inserted more often in men than in women (P-value = 0.016), whereas no differences were found in frequency of hematoma evacuation among men and women (P = 0.433) (Table 3). On the basis of the results of the univariate analysis evaluating differences depending on EVD insertion (Table 4), a multivariate analysis was performed (Table 5). Male sex, presence of intraventricular hemorrhage, ventriculomegaly, and volume as well as evacuation of ICH were independently associated with EVD insertion. Male sex in particular increased the odds of EVD insertion by 2.82 (95% confidence interval 1.61–4.95, P < 0.001).

Length of ICU-stay, length of hospital stay, 30 day- and 1 year- mortality did not differ between groups (Table 3). When we compared the outcome as evaluated using the mRS, men were more likely to suffer from an unfavorable outcome (Fig. 2). Further predictors of an unfavorable outcome at hospital discharge are shown in Tables 6 and 7. The multivariable analysis (Table 8) showed higher age, male sex, lower initial GCS score, midline shift, insertion of EVD, and higher CCI to be predictive of unfavorable outcome independent of TC transfusion, presence of ventriculomegaly, presence of intraventricular hemorrhage volume of hemorrhage and localization of the hematoma.

From all deceased patients n = 115, n = 55 (47,0%) died after early withdrawal of life-sustaining therapy. Of them n = 25 were men (45%) (P = 0.500 compared to female). Premorbid “special care” functional status was comparable in both sexes: n = 6 in women (20%) and n = 6 in men (24%) (P = 0.617).

The strength of this study lies in the use of a database of consecutive patients with spontaneous ICH during a 10-year period with several parameters collected. However, due to the retrospective nature of our study, our study design bears limitations. Firstly, this is a single center study. Therefore, the generalizability of the results may be limited because we cannot exclude detection and referral biases. While most patients with severe clinical symptoms are admitted to the Neurocritical Care Unit in our institution, some may have also been admitted to the in-house intermediate care unit and thus, not been included in our study. Furthermore, while patients with histopathologically proven cerebral amyloid angiopathy were excluded, due to the small number of biopsies/autopsies we cannot exclude the possibility that some cases have still been included. Furthermore, MRIs were not regularly performed to screen for microbleeds. Moreover, the lack of data regarding the decision-making process of the neurosurgeon to insert an EVD only allows speculations about the reason of the sex-related discrepancy in frequency of EVD-insertion. Moreover, pre-admission status could only be recorded in a retrospective manner and not based on modified Rankin Scale. Due to the retrospective study design and unavailable data on the initial ICP or its course, the effect of EVD insertion on ICP management cannot be judged in this study design.