Does Preendoscopy Rockall Score Safely Identify Low Risk Patients following Upper Gastrointestinal Haemorrhage?
Table 4
Sensitivity and specificity of various Glasgow Blatchford scores for primary and secondary outcomes.
Requirement for intervention
Rebleeding within 14 days
Death within 30 days
(95% CI)
(95% CI)
(95% CI)
All ages
GBS = 0
Sensitivity
100% (97.0–100)
100% (78.9–100)
100% (74.0–100)
Specificity
2.4% (0.8–5.5)
1.4% (0.5–3.2)
1.4% (0.4–3.1)
Positive predictive value
47.3% (42.2–52.4)
6.1% (3.8–8.9)
4.7% (2.8–7.3)
Negative predictive value
100% (35.9–100)
100% (35.9–100)
100% (35.9–100)
All ages
GBS ≤ 1
Sensitivity
100% (97.0–100)
100% (78.9–100)
100.0% (74.0–100)
Specificity
6.8% (3.8–11.1)
3.8% (2.1–6.4)
3.8% (2.1–6.3)
Positive predictive value
48.4% (43.2–53.6)
6.2% (3.9–9.1)
4.8% (2.9–7.5)
Negative predictive value
100% (68.1–100)
100% (68.1–100)
100.0% (68.1–100)
All ages
GBS ≤ 2
Sensitivity
97.8% (94.4–99.4)
100% (78.9–100)
94.4% (72.7–99.9)
Specificity
11.1% (7.2–16.2)
7.4% (4.9–10.6)
7.0% (4.6–10.1)
Positive predictive value
49.0% (43.8–54.3)
6.4% (4.1–9.4)
4.7% (2.8–7.4)
Negative predictive value
85.2% (66.3–95.8)
100% (81.7–100)
96.3% (81.0–99.9)
<70 years old
GBS ≤ 2
Sensitivity
98.9% (96.1–99.9)
100.0% (78.9–100)
94.4% (72.7–99.9)
Specificity
9.7% (6.0–14.5)
6.0% (3.8–9.0)
5.7% (3.5–8.6)
Positive predictive value
48.9% (43.7–54.1)
6.3% (4.0–9.3)
4.6% (2.7–7.3)
Negative predictive value
90.9% (70.8–98.9)
100.0% (78.1–100)
95.5% (77.2–99.9)
Data are from 388 patients admitted with upper gastrointestinal haemorrhage over 5 years (2006–2011). Glasgow Blatchford score (GBS) was determined retrospectively. Specificity and sensitivity for various scores, previously shown to delineate low likelihood of requiring intervention, were defined. Primary outcomes are endoscopic or surgical therapy or blood transfusion and secondary outcomes are rebleed within 14 days or death within 30 days.