Predictors of gastrointestinal bleeding in acute intracerebral haemorrhage

Abstract

Background: Gastrointestinal (GI) haemorrhage is an important and sometimes serious complication in critically ill neurological patients who suffered from stroke and head injury and those in intensive care. There is no study evaluating frequency, severity and risk factors of GI haemorrhage in patients with primary intracerebral haemorrhage (ICH). Aims: To evaluate the frequency, severity and predictors of GI haemorrhage in patients with ICH. Methods: In a prospective hospital-based study, consecutive CT-proven ICH patients within 10 days of the ictus were included. The patients with history of peptic ulcer, GI haemorrhage, liver and kidney disease, bleeding diathesis and those on antiplatelet, anticoagulant or nonsteroidal antiinflammatory drugs (NSAIDS) were excluded. A detailed neurological evaluation was carried out. Glasgow coma scale (GCS) was used for assessment of consciousness level and Canadian neurological scale (CNS) for severity of stroke. The haematomas were classified into small (<20 ml), medium (20-40 ml) and large (>40 ml). The occurrence of GI haemorrhage during 14 days of ictus was considered due to ICH. To evaluate the predictors of GI haemorrhage, various clinical and CT scan findings were evaluated by univariate followed by multivariate logistic regression analysis. Results: Fifty-one patients with ICH were included whose age ranged between 30 and 80 years and 14 were female. The mean GCS score was 8.9 (3-15) and CNS score was 2.2 (2-4). Haematoma was small (<20 ml) in 11 patients and medium (20-40 ml) and large (>40 ml) in 20 patients each. Evidences of septicemia were present in 20 patients. Gastric haemorrhage (GH) was noted in 15 patients which was more than 40 ml in 4 patients and one of these patients needed blood transfusion. On univariate analysis, the size of haematoma, septicemia, motor signs on the nonhemiplegic side and pupillary asymmetry were significantly related to GI haemorrhage. On multivariate analysis, the best set of predictors of gastric haemorrhage included size of haematoma, septicemia and GCS score. Conclusion: GI haemorrhage is more likely present in patients with larger haematoma having septicemia. Our study highlights the importance of septicemia, which is an important and modifiable risk factor for GI bleeding in ICH patients.