Although stress ulcer is a complication that can cause mortality and morbidity in critical patients, there is still lack of consensus about its prophylaxis. There is also few data available from Taiwan. H2 blockers are commonly used due to convenience. Some prefer sucralfate (a mucosal protective agent) for the sake of less associated nosocomial pneumonia. Recently, proton pump inhibitors were shown to have good prophylactic effects for stress ulcer. Esomeprazole, an isoform of omeprazole, has good acid suppression effect and the tablets are soluble for the use of tube feeding. Our previous study showed that there was no difference for the efficacy of stress ulcer prophylaxis between esomeprazole and sucralfate in patients admitted to medical ICU with at least one risk factor. The prevalence of nosocomial pneumonia was also similar.

We will enroll those patients that have received intracranial surgery and admitted to neurosurgical ICU. After obtaining the consent, we will give them prophylactic drugs for 7 days within 24 hours. They are randomly allocated to 2 groups. Group I: esomeprazole 40 mg qd from NG route or orally; Group II: famotidine 20 mg iv bolus q12h. We will monitor the following data: Glasgow coma scale, APACHE II score, CBC, CXR, stool character and OB test, NG aspirate. If clinical evidence of UGI bleeding occurs, endoscopy will be performed. We define the end point as overt bleeding, death or transfer out of ICU. We will compare the prevalence of UGI bleeding and nosocomial pneumonia in these 2 groups.

1. patients: the patients receiving neurosurgery and admitted to intensiv care unit within 24 hours. They are enrolled after well explanation and giving written consdent. Those are less than 18 y/o, pregnant, not suitable for NG feeding, already having GI bleeding, are excluded
2. grouping & intervention: The patients are randomized to 2 groups. 1st group:receiving esomeprazole 40 mg qd via NG; 2nd group: receiving famotidine 20 mg iv bolus q12h. These medication are used for 7 days. Estimated enrolled number is 60 for each group
3. monitoring: Glasgow coma scale , APACHE II score at baseline, CBC、CXR at basleine and qod, stool OB q3d，NG drainage、sputum、 stool character, ICU routine （TPR, BP）, ICU admitted day, 30 day mortality rate. UGI endoscopy arranged according judgement of attending doctors
4. end points: overt UGI bleeding（tarry stool,hematemesis、coffee ground substance from NG more than 60 ml, Hb decrease more than 2g/dl and endoscopic proof of bleeder）. ventilator associated pneumonia: new onset and persisted hazziness in CXR, combined with fever, leucocytosis and positive sputum smear finding.
5. statistics: the prevalence of overt bleeding and ventilator associated pneumonia is examined by Fisher's exact test, the demongraphic data and disease severity data are examined by student's t test or Chi-square test。

Inclusion Criteria:

• Within 24 hours of admission to Neurosurgical ICU after neurosurgery with ventilator support

Exclusion Criteria:

• Less than 18 y/o;
• Pregnancy;
• Not suitable for medication from NG route,
• Had GI bleeding at admission to ICU