Gastrointestinal Hemorrhage after Spontaneous Subarachnoid Hemorrhage: A Single-Center Cohort Study

Spontaneous subarachnoid hemorrhage (SAH) is a devastating disease, and gastrointestinal hemorrhage is one of several potential complications of acute strokes. We aim to analyze its prevalence, risk factors, and association with in-hospital prognosis following SAH. A total of 1047 adult patients with a primary diagnosis of spontaneous SAH were retrospectively enrolled. We retrieved medical information from the administrative database utilizing diagnostic and procedure codes of the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). Patients with SAH included 418 men and 629 women, and their mean age was 57.2 (standard deviation 14.6) years (range, 18–93 years). Gastrointestinal hemorrhage occurred in 30 of the 1047 patients, accounting for 2.9%. In a multivariate logistic regression model, the independent risk factors for gastrointestinal hemorrhage were liver disease and hydrocephalus. The in-hospital mortality rates were 43.3% and 29.3% in patients with and without gastrointestinal hemorrhage, respectively, but the difference was not statistically significant. In conclusion, the prevalence of gastrointestinal hemorrhage was 2.9% in patients hospitalized for spontaneous SAH. Underlying liver disease and the presence of hydrocephalus were both independent risk factors for this complication, which is a reminder to clinicians to pay increased attention in such cases.

Memorial Hospital. Since the study design was retrospective and delinked, patient informed consent was not required after approval by the institutional review board. All methods were carried out in accordance with relevant guidelines and regulations. We retrieved medical records from the administrative database, which included the following patient information: gender; age; admission and discharge dates; marital status; diagnostic codes by the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM); procedure codes; condition at discharge; and related data. From 2000 to 2010, a total of 1094 hospital admissions with a primary diagnosis of SAH (ICD-9-CM code 430) were identified. Patients who were readmitted, who were < 18 years of age, or who had missing documents were excluded. Eventually, we enrolled 1047 SAH patients for further analysis.
We analyzed data with SPSS software (IBM SPSS Statistics, version 20.0). Parameters were presented as numbers (percentage) or as a mean ± standard deviation (SD). Intergroup differences were assessed using the chi-square test or Fisher's exact test for categorical variables, and the Student's t-test or Mann-Whitney U-test for continuous variables. All parameters with a P value < 0.1 were included in multivariable logistic regression to adjust for independent risk factors of gastrointestinal hemorrhage after SAH. A Kaplan-Meier survival curve was constructed and compared using the log-rank test. Statistical significance was defined as a p < 0.05.

Results
The 1047 patients, including 418 men and 629 women, were diagnosed with first-ever spontaneous SAH. The mean age was 57.2 ± 14.6 years (range, 18-93 years) at the time of diagnosis. The median of length of hospital stay was 12 days (range, 1-102 days). Of these patients, 353 (33.7%) underwent surgical treatments and 172 (16.4%) underwent endovascular interventions for cerebral aneurysms. There were 256 (24.5%) cases who had mechanical ventilation for 96 hours or longer and 61 (5.8%) received tracheostomy procedures.
The mean duration of hospital stay was 15.7 ± 10.1 and 14.5 ± 12.8 days for patients with and without gastrointestinal hemorrhage, respectively (p = 0.21). At discharge, 13 of the 30 patients with gastrointestinal hemorrhage and 298 of the 1017 patients without gastrointestinal hemorrhage had died; thus, incidences of in-hospital mortality were 43.3% and 29.3%, respectively (p = 0.10). Kaplan-Meier survival curves for the two populations with spontaneous SAH showed that, by the 30th day of hospitalization, 56.7% of the gastrointestinal hemorrhage group and 71.2% of the non-gastrointestinal-hemorrhage group had survived (Fig. 2). No remarkable difference was observed between the two study groups (log-rank test: p = 0.30).

Discussion
SAH accounts for less than 10% of all strokes, with an overall prevalence of approximately 9 per 100,000 person-years 11,12 . However, as a neurological emergency with a serious outcome impact, SAH remains an intensive topic of clinical studies, and the characteristics of the complications are generally investigated 3,4,12 . Our study showed that gastrointestinal hemorrhage occurred in 2.9% of the 1047 adults with SAH during acute hospitalization. In a study focusing on post-SAH medical morbidities, Wartenberg et al. reported a gastrointestinal bleeding rate of 4% among 576 patients, which was similar to that observed by us 3 . In comparison, the prevalence of gastrointestinal hemorrhage was 3% after acute strokes (excluding SAH) 7 , and only 1.2%-1.5% in patients with ischemic stroke 6,8,9 . Therefore, though gastrointestinal hemorrhage following SAH is uncommon, its frequency is still higher than that of ischemic cerebral events which usually require subsequent thromboprophylaxis.
The pathophysiology of gastrointestinal hemorrhage after acute strokes is not completely realized, and several theories exist. Intracranial lesions or elevated intracranial pressure affect brainstem or hypothalamic nuclei with hyperactivity of vagal tone and lead to hemorrhagic peptic ulcerations because of hypersecretion of gastric acid 13 . Second, acute stroke predisposes to gastroparesis resulting from interruption of the axis between the central nervous and digestive systems and seems to increase the risk of gastrointestinal bleeding 14 . In addition, a humoral surge in catecholamines or cortisol is usual in the acute phase of stroke 15 and possibly causes vasoconstriction or mucosal damage of the gastrointestinal tract. As we all know, SAH is a highly stressful disease, and it is reported that intracranial hypertension is common after SAH, even in 48.7% of good-grade cases 16 . This also explains why the occurrence of gastrointestinal hemorrhage is more frequent in patients with SAH than in those with ischemic stroke.
Many risk factors of gastrointestinal hemorrhage after acute ischemic stroke, such as previous history of peptic ulcer disease, liver disease, cancer, sepsis, renal insufficiency, or stroke severity, have been documented 6,[8][9][10] . In patients suffering from non-traumatic intracerebral hemorrhage, increased hematoma size, septicemia, and low Glasgow coma scale score are at risk for gastrointestinal hemorrhage 17 . Our results demonstrated that underlying liver disease and the presence of hydrocephalus were both independent risk factors for post-SAH gastrointestinal hemorrhage. In contrast to ischemic stroke, it is surprising that our patients with previous peptic ulcers were not vulnerable to subsequent gastrointestinal bleeding. The reasonable explanation is that the comorbidities were probably underestimated, as ICD-9-CM code recording is occasionally incomplete in clinical practice. Additionally, the number of patients may still be relatively small from a statistical standpoint, and this is underpowered to detect the significance of some risk factors.
A connection between liver disease and gastrointestinal hemorrhage is noted in populations with either ischemic stroke or SAH. Rumalla et al. analyzed acute ischemic stroke in the largest database of the United States, but information on the subgroup with liver disease is scanty 9 . Interestingly, a study from Taiwan reported that abnormal liver function is related to gastrointestinal hemorrhage after ischemic stroke 10 , which is compatible with our findings. Taiwan is one of the endemic areas for hepatitis B and hepatitis C viral infections, and the estimated prevalence rates are 13.70% and 1.8%-5.5%, respectively 18,19 . In this series, the overall prevalence of liver disease is merely 3.2%, and appears to be understated, perhaps owing to the absence of personal awareness or general surveillance programs for hepatitis infections. Theoretically, patients with chronic liver diseases have gastrointestinal bleeding most commonly from gastroesophageal varices or gastropathy associated with portal hypertension, but non-variceal bleeding is not infrequent 20 . Clinicians must pay attention to liver function and thrombocoagulation status after SAH, especially when patients have potential liver diseases with unpredictable sources or pathogenesis of hemorrhagic gastrointestinal tract lesions.
The occurrence of hydrocephalus is a critical complication following SAH, and there are several predictive factors, including high Hunt and Hess Scale score, high Fisher score, rehemorrhage, presence of intraventricular blood, posterior circulation aneurysm location, or older age 21,22 . Most of these predictors for hydrocephalus are correlated to initial clinical severity or subarachnoid blood volume. With a greater amount of hemorrhage, the subarachnoid space is filled with blood cells and corresponding products which block the outflow tract of cerebrospinal fluid, and thus the presence of hydrocephalus can be viewed as a sign of severe SAH. Moreover, acute cerebrospinal fluid disarrangement is likely accompanied by a rise in intracranial pressure. Hydrocephalus with subsequent intracranial pressure increase potentially induces a surge of stress hormones, and may explain why hydrocephalus is significantly relevant to post-SAH gastrointestinal hemorrhage.
In this study, gastrointestinal hemorrhage was not a remarkable factor correlated with the prognosis of SAH. Nevertheless, there seemed to be a trend in the difference between the in-hospital mortality rates in patients with and without gastrointestinal hemorrhage (43.3% and 29.3%, respectively). In a total of 6853 patients with acute ischemic stroke, O'Donnell et al. showed that gastrointestinal hemorrhage was strongly associated with in-hospital death and 6-month mortality 6 . Another population-based study for ischemic stroke similarly reported an 82% increased likelihood of death at discharge in patients with gastrointestinal bleeding 9 . Because our findings reflect the observation of a single medical center, we suppose that the current evidence may be insufficient. Further studies in a larger series of cases are required to draw a firm conclusion on the relationship between gastrointestinal hemorrhage and post-SAH outcomes. There were several limitations to this study. First, the utility of ICD-9-CM coding for research was sometimes questionable, because the specific information, such as Hunt and Hess grading, Fisher grading, location or size of cerebral aneurysms, or sequential functional status, was difficult to evaluate. Second, the analysis did not contain records regarding the use of proton pump inhibitors or histamine receptor antagonists, which have a potential impact on the occurrence or severity of gastrointestinal hemorrhage. Third, the etiology of gastrointestinal bleeding was not documented, though we considered that bleeding originating from the upper gastrointestinal tract accounted for the majority of cases. Fourth, the outpatient or longer-term prognosis of SAH was not addressed here, and this is an important consideration from a public health perspective. Fifth, because of the higher rate of hepatitis in our country, these results may not be generalized to the non-endemic area. In addition, we enrolled the patients with a primary diagnosis of SAH (ICD-9-CM code 430), and this cohort included 15% non-aneurysmal SAH 23 . Since SAH is a disease with around 30% early mortality rate 1,2 , the patients with poor grading SAH may have no chance to undergo cerebral angiography or further managements. These explained why only about 50% of our SAH-patients received definite treatments of aneurysms.
In conclusions, the prevalence of gastrointestinal hemorrhage was 2.9% in patients hospitalized for spontaneous SAH. Underlying liver disease and the presence of hydrocephalus were both independent risk factors for this complication, which reminds clinicians to pay increased attention in such cases.  Table 2. Multivariable analysis of independent risk factors for gastrointestinal hemorrhage after SAH.