Management of non-variceal upper gastrointestinal tract hemorrhage: Controversies and areas of uncertainty

Table 1 Sensitivity, specificity, positive predictive value and negative predictive value for death and rebleeding using 6 common upper gastrointestinal tract hemorrhage scoring systems

Scoring system	Death	Rebleeding
Pre-endoscopy Rockall risk score
Sensitivity	100 (43.8-100)	69.6 (49.1-84.4)
Specificity	18.5 (14.8-22.9)	17.5 (13.8-22.0)
PPV	1.0 (0.4-3)	5.5 (3.4-8.8)
NPV	100 (94.4-100)	89.2 (79.4-94.7)
Post-endoscopy Rockall risk score
Sensitivity	33.3 (6.1-79.2)	87.0 (67.9-95.5)
Specificity	29.6 (25.1-34.6)	31.1 (26.4-36.3)
PPV	0.4 (0.1-2.2)	8.1 (5.3-12.1)
NPV	98.1 (93.4-99.5)	97.2 (92.0-99.0)
Blatchford scoring system
Sensitivity	100 (83.89-100)	94.29 (81.40-98.42)
Specificity	1.83 (0.71-4.61)	0.98 (0.27-3.50)
PPV	8.51 (5.58-12.79)	14.04 (10.17-19.06)
NPV	100 (51.01-100)	50.00 (15.00-85.00)
Forest classification
Sensitivity	85.00 (63.96-94.76)	71.43 (54.95-83.67)
Specificity	50.23 (43.66-56.79)	50.49 (43.68-57.28)
PPV	13.49 (8.6-20.54)	19.84 (13.81-27.65)
NPV	97.35 (92.49-99.10)	91.15 (84.77-95.12)
Cedars-Sinai Medical Center predictive index
Sensitivity	95.00 (76.39-99.11)	80.00 (64.11-89.96)
Specificity	41.55 (35.22-48.17)	41.67 (35.12-48.53)
PPV	12.93 (8.44-19.31)	19.05 (13.52-26.15)
NPV	98.91 (94.09-99.81)	92.39 (85.12-96.26)
Baylor College scoring system
Sensitivity	87.50 (52.91-97.76)	30.77 (12.68-57.63)
Specificity	58.49 (45.09-76.74)	47.92 (34.47-61.67)
PPV	24.14 (12.22-42.11)	13.79 (5.50-30.56)
NPV	96.88 (84.26-99.45)	71.88 (54.63-84.44)

95% confidence intervals are recorded in parentheses. PPV: Positive predictive value; NPV: Negative predictive value.

Table 2 Areas requiring further investigation

Pharmacologic therapy prior to endoscopy	Determine optimal route and dosage of PPI in UGIH (continuous infusion vs intermittent IV bolus vs oral dosing)
Timing of endoscopic evaluation	Defining optimal timing of initial endoscopy, implementation of early discharge in low risk patients
Prokinetic agents as endoscopic adjuncts	Clearly define the role for prokinetics in UGIH with randomized controlled trials, specifically define the optimal agent, dose and timing prior to endoscopy
Long-term PPI management	Clarification of potential long-term sequelae of PPI including: osteoporosis, c. difficile infection and community-acquired pneumonia
PPI and thienopyridines	Consensus on the clinical importance of this interaction, a complete randomized controlled trial to support the truncated COGENT trial data
H. pylori and UGIH	H. pylori testing that allows accurate test results in the setting of acute UGIH

PPI: Proton pump inhibitor; H. pylori: Helicobacter pylori; UGIH: Upper gastrointestinal tract hemorrhage; IV: Intravenous; COGENT: Clopidogrel and the optimization of gastrointestinal events trial.