Langenbecks Arch Surg (2007) 392:581–585 DOI 10.1007/s00423-007-0156-7
ORIGINAL ARTICLE
Comparison and validation of scoring systems in a cohort of patients treated for perforated peptic ulcer Mahmut Koç & Ömer Yoldaş & Yusuf Alper Kılıç & Erdal Göçmen & Tamer Ertan & Hayrettin Dizen & Mesut Tez
Received: 3 October 2006 / Accepted: 16 January 2007 / Published online: 14 February 2007 # Springer-Verlag 2007
Abstract Background and aims The aim of this study is to evaluate the predictive accuracy of different scoring systems on surgery for perforated peptic ulcer referred to an academic department of general surgery in a tertiary reference center. Patients and methods Seventy-five consecutive patients (Male/female ratio=64:11; mean age, 44 years; range, 16– 85) with perforated peptic ulcer disease were investigated. Disease severity scores and mortality predictions were calculated using the collected data during admission. Discrimination and calibration characteristics of each system, namely, the acute physiology and chronic health evaluation II and III, the simplified acute physiology score II, and the mortality probability models (MPM) II, were determined by using the area under receiver operating characteristics curve and the Hosmer–Lemeshow goodnessof-fit test, respectively. Results Among the 75 patients included, there were eight (10.6%) mortalities. All systems had a reliable power of discrimination and calibration. Among the systems tested, MPM II was the best performing as far as discrimination and calibration characteristics were considered. The parameters of MPM II system that were related to systemic
M. Koç : Ö. Yoldaş : E. Göçmen : T. Ertan : H. Dizen : M. Tez (*) Fifth Department of Surgery, Ankara Numune Education and Research Hospital, 5. Cad 10/3, 06500 Ankara, Turkey e-mail:
[email protected] Y. A. Kılıç : E. Göçmen Department of General Surgery, Hacettepe University School of Medicine, Ankara, Turkey
perfusion of the patient were significantly positive in patients who died compared to those who survived. Conclusions MPM II that predicted mortality at admission is better than the other systems in predicting mortality. Results also indicate the importance of maintenance of systemic perfusion of the patient at the early phases of peptic ulcer perforation. Keywords Peptic ulcer . Perforation . Mortality . Scoring systems
Introduction There has been a marked decrease in elective surgery for peptic ulcer disease (PUD) after the introduction of medical therapies including H2-receptor antagonists and H pylori eradication. By contrast, the number of ulcer perforation requiring emergency surgery has remained quantitatively constant. Peptic ulcer perforation (PUP) affects almost 10% of PUD patients. Overall, perforation accounts for more than 70% of mortality associated with the ulcer disease [1]. Rational planning of treatment and use of limited critical care resources for high-risk patients necessitate the stratification of patients according to disease severity and predicted mortality. Commonly used prediction systems, namely, the acute physiology and chronic health evaluation (APACHE) II or III, the simplified acute physiology score (SAPS) II, and the mortality probability models (MPM) II, have been developed using databases of general intensive care unit (ICU) populations [2]. The APACHE II system considers 76 disease categories. However, one known limitation of APACHE II is the fuzziness of this classification that may cause the specific disease entity not to be represented adequately or being
582
represented by more than one disease category. APACHE III, a refinement of APACHE II score, has additional potential uses that include the identification of factors in the ICU that contribute to the outcome and assistance in the individual patient’s decision-making [3]. The SAPS II system, which also considers laboratory and physiologic data calculation of predicted mortality, is not based on disease categories [4]. Instead of numerical values, MPM II evaluates the presence or absence of 14 specific clinical entities like cirrhosis, mechanical ventilation, cardiopulmonary resuscitation before ICU admission, or a pulse rate over 150/ min on admission. The only numerical value in this system is age. As the cutoff points of physiologic and laboratory values are more extreme compared to the other systems’ discrimination and calibration, characteristics may be different compared to other systems (Table 1) [5]. Although these models perform well in predicting the mortality of the general ICU patient population, they may well under- or overestimate mortality in selected patient subpopulations that were not well represented in the original cohort on which the model was developed [6]. The aim of this study is to evaluate the performance of the general severity-of-illness scoring systems (APACHE II and III, SAPS II, and MPM II) in surgery for perforated peptic ulcer referred to an academic department of general surgery in a tertiary reference center.
Patients and methods Patients Over a period of 1 year (2005–2006), 75 patients who underwent emergency surgery for perforated peptic ulcer were enrolled in this prospective cohort study. Excluded patients had perforated ulcers at the anastomotic site after a former ulcer surgery and others had cancer (Table 2). The diagnosis of PUP was based on clinical features, blood tests, routine laboratory tests, and radiological findings (i.e., plain abdominal X-ray in all cases and abdominal USG and CT scan). Invariably, the definitive diagnosis of PUP was obtained at surgery. Surgical procedure An open surgical approach was performed leading to a nondefinitive operation (primary suture and omentopexy) in 71 patients (94.6%) and to definitive operations (i.e.. truncal vagotomy and pyloroplasty) in four patients (5.4%). The decision to perform one or the other type of surgery depended on several known factors including the location and extent of
Langenbecks Arch Surg (2007) 392:581–585 Table 1 Variables collected by each scoring model Variables
Chronic health status AIDS Cirrhosis Lymphoma Hematologic malignancy Leukemia Hepatic failure Metastatic cancer Immunsuppression Chronic renal insufficiency Physiology Temperature Heart rate Respiratory rate Blood pressure White blood cell count Albumin Bilirubin Electrolytes Blood urea nitrogen Creatinine Urine output Blood gas Glascow coma score Acute diagnoses Acute renal failure Arrhythmias Cerebrovascular accident GI bleeding Leukemia Infection Intracranial mass effect Other Age Patient origin CPR prior to ICU admission Mechanical ventilation
MPM0 II
SAPS II
APACHE II
APACHE III
+
+ + + +
+ + +
+ + + + +
+ + + + +
+ + + + +
+ +
+ + + + + + + + + + + + +
+ +
+ +
+ +
+
+
+
+ +
+ + + + + + +
+ +
+
+ + +
+ + + + + + +
+
+ +
lesions, feasibility of a safe nondefinitive surgery, presence or absence of anaesthesiological risk factors, and surgeon’s attitude. The main outcome measured in this study was mortality. Mortality was defined as any death that occurred in the 30-day postoperative period. Disease severity scores and predicted mortality values for these patients were calculated using the prospectively collected values. Disease severity scores and mortality
Langenbecks Arch Surg (2007) 392:581–585
583
Table 2 Demographic distribution of patients Patient’s characteristics
Number
Male/female Age (65 years) Previous ulcer history (yes/no) Operation type (nondefinitive/definitive)
64:11 66:9 21:54 71:4
predictions were calculated for APACHE II and SAPS II systems. For APACHE III and MPM II systems, predicted mortality values at admission were calculated for each patient. Calculations were performed using the Muavenet Intensive Care Information System at (http://www.icu.hacettepe.edu.tr/ micis.html). ‘Discrimination’ refers to a model’s ability to distinguish survivors from nonsurvivors. Model discrimination was measured by the area under the receiver–operator characteristic (ROC) curve (AUC) to evaluate how well the model distinguished patients who experienced the event (death) from those who did not. The AUC represents the probability that the patient who died had a higher predicted probability of dying than the patient who survived. An AUC of 0.5 indicates that the model does not predict better than chance. The discrimination of a prognostic model is considered perfect if AUC=1, good if AUC>0.8, moderate if AUC is 0.6–0.8, and poor if AUC
