Noridah, Nordin
(2008)
Predictive factors for mortality, recurrent bleeding and need
for surgery in non-variceal upper gastrointestinal (GI) bleeding
in correlation with rockall score.
Masters thesis, Universiti Sains Malaysia.
Abstract
Rockall score was widely used around the world including Malaysia. The score was useful in
predicting mortality but there were controversies regarding its use to predict recurrent bleeding
and need for surgery. The objectives of this study were to see the correlation between Rockall score and outcome of GI
bleed and to determine factors that predict the three outcomes and association with Rockall risk
group. This was a retrospective observational study of250 patients with non variceal GI bleed who
were admitted to HUSM between 2004 until2006. The mean age was 62.1 + 16 years. There were more males 57.6% (144) than females 42.4%
(106). Most patients were Malays 209 (83.6%) others were Chinese 30 (12%), Siamese 8(3.2%)
and Indians 3(1.2%). Mortality rate was 3.6% (9 patients), recurrent bleeding rate was 9.6% (24
patients) and rate for patients that need operation was 4.4%(11 patients). 85.2% (213 patients
were in the high risk group (Rockall score :55) and 14.8% (37 patients) were in the high risk
group. The high risk group was significantly associated with mortality ( p < 0.00 I), recurrent
bleeding (p value= 0.01) and the need for surgery (p = 0.013). Factors associated with mortality
in univariate analysis were older age (p=O.OOI), chronic liver disease (p = 0.008), sepsis (p< 0.001), warfarin (p=0.019), unconscious state (p<0.001), coffee ground in ryles tube
(p=0.019), low haemoglobin (p=0.030, mean± SD=6.4±2.0 gldl)) raised urea (p=0.040, mean±
SD = 22.5±13.4). In multivariate analysis significant factor associated with mortality was sepsis
(p=0.021, OR=9.9, 95%CI 1.413-69.513). Factors associated with recurrent bleeding in
univariate analysis were ischaemic heart disease (p=0.032), chronic liver disease (p=0.042),
sepsis p=(0.004), aspirin (p=0.022), symptoms of anaemia (p=0.002), unconscious state
(p=0.008), coffee ground in ryles tube (p=0.024), low haemoglobin (p=0.006, mean± SD=
6.8± 1.6), raised urea (p=0.028, mean±SD=21.0± 14. 9), raised creatinine (p=0.029 mean±SD =
290.4± 230. 7). Factor associated with recurrent bleed in multivariate analysis was creatinine.
(p=O.Ol2, 95% CI 1.0-l.O).Factors associated with the need for surgery in univariate analysis
were epigastric pain (p=0.005), symptomatic anaemia (p=O.Ol8), diastolic blood
pressure(p=0.031, mean±SD, 63.8±10.3), tender epigastrium (p=0.015), low haemoglobin
(p=0.001, mean±SD=6.4 ±1.3). Factors that were associated with the high risk group were older
age (p=0.010), ischaemic heart disease (p= 0.007), prior peptic ulcer disease (p=0.018), chronic
renal failure (p=0.004), chronic liver disease (p=0.039), sepsis (p<O.OOl), herbs (p=0.046),
unconscious state (p=0.014), low haemoglobin (p=0.022, mean±SD=7.3±2.4), APTT (p=0.026,
mean=37.9±9.4). Factors associated with high risk group in multivariate analysis were stigmata
of recent haemorrhage (p<O.OOl, OR=0.063, 95%CI =0.26-0.152), Sepsis (p=O.Ol3, OR=0.149,
CI 0.034-0.664) and warfarin (p=0.028, OR=0.182, Cl = 0.040-0.832)
The receiver Operating Characteristic (ROC) curve for mortality showed area under the curve of
0.87, for recurrent bleeding 0.72 and the need for surgery 0.65. The study showed Rockall score was useful to predict mortality but not recurrent bleeding and
need for surgery however the risk grouping analysis showed that higher Rockall score can
predict mortality, recurrent bleed and need for surgery. Most upper 01 bleeders was in the low
risk group. Sepsis was found to be an important factor to predict outcomes. A further
prospective study on the use of Rockall score in sepsis patient in upper 01 bleed should be
proposed.
Actions (login required)
 |
View Item |
