Acta Medica Colombiana
Print version ISSN 0120-2448
GOMEZ ZULETA, Martín Alonso et al. "UNAL" predictions scale to identify patients with upper gastrointestinal tract bleeding who need emergency endoscopy. Acta Med Colomb [online]. 2006, vol.31, n.4, pp. 389-399. ISSN 0120-2448.
Determination of clinical variables that allow to predict which patients with upper gastrointestinal tract bleeding (UGTB) present active bleeding is very important because it would help to identify those who need en emergency endoscopy (within the first 12 hours after admission to the hospital) or if the procedure may be delayed to do it within the next 24 hours. Besides the above, a predictive scale would be very useful for the hospitals' and health centers' emergency services because it facilitates the optimization of resources, and further more in Colombia where there are not enough or very few hospitals that provide endoscopy services 24 hours a day. The purpose of this work was to assess the usefulness of the clinical and laboratory findings to identify those patients with UGTB with active bleeding at admission and who need emergency endoscopy, to build an original scale that later on was validated in a prospective way in a different group of patients. Objective: to define clinical and laboratory variables in patients with UGTB, that can be used to make a scale that allows the identification of those with active bleeding at admission and who need emergency endoscopy. Patients and methods: it is an analytical observational cross over study taking the information from those patients that were admitted in the emergency service of a third level hospital. The clinical findings at admission were taken as prediction variables and the presence or absence of active bleeding during the endoscopy within the first 12 hours ware taken as outcome variables. The sample is made up of two prospective patients' cohorts who were consecutively selected. The first cohort was recruited from January 2004 until August 2005 for internal validation of the scale, and the second cohort, another independent group, was recruited from September 2005 to June 2006. The inclusion criteria were patients older than 18 years with a clinical diagnosis of UGTB. The 97 scale was built using the variables that were identified as predictors of active bleeding in the multivariance analysis. Results: in the first cohort we included 109 patients with UGTB (age X = 59.3 +/-19.1, men 69.7%). The main symptom for presentation was melena in 73.4% and the main endoscopic finding was peptic ulcer (58.7%). Six variables (lypotimia, hemodinamic instability hemoglobine less than 8, white cells higher than 12000, blood in the SNG, and age >65) were identified as independent predictors of active bleeding in emergency UGTE and were used to do the following scale: Bleeding scale = 5 x (Blood in the SNG) +4 (hemoglobin <8) + 4x (lipotimia) + 2x (white cells> 12.000) + 2x (age > 65) + 1x (homodynamic instability. The scale values ranged between 0 and 18 and the operative characteristics curve was used (ROC) in the determination of the cut off point where the highest sensitivity and specificity are shown and the predictions capacity of the scale was assessed, finding that the point was 7 in such a way that if the patient has less than 7 points it means absence of active bleeding and does not need an emergency UGTE. However, if the patient has 7 or more points there is active bleeding and needs emergency UGTE. This scale was externally validated in the other independent group with 98 patients (age X = 56.2 ± 16, men 62.2%), of which 15.6% had active bleeding. Using the scale with the established cut off point, a sensitivity of 100% was found, specificity of 83.1%; predictive value of 64.3%; negative PV of 100%. Conclusions: the findings suggest that a simple scale, easy to remember, built with the clinical and laboratory parameters available in any hospital in our country, may identify patients with UGTB that need an emergency UGTE as well as those that do not need it. This scale with solid mathematical basis may be used in those hospitals that do not have gastroenterology services 24 hours a day or as a tool to establish priorities in terms of patients' remission from the centers that do not have UGTE services available. The results of this study allow us to recommend the clinical application of this original scale because it had en excellent performance in the external validation, especially to rule out active bleeding and was consistent with high discriminative value when assessed in two independent patients' groups.
Keywords : upper gastrointestinal tract bleeding; endoscopy; prediction scale.