Predictive Model of Failure of Outpatient Laparoscopic Cholecystectomy Jose Bueno Lledo ´, MD,* Manuel Planells, MD, PhD,* Alejandro Espı´, MD, PhD,w Alfonso Serralta, MD, PhD,* Rafael Garcı´a, MD,* and Angel Sanahuja, MD, PhD* Introduction: The aim of our study was to review our experience and to determine a predictive model of factors for unanticipated admissions after ambulatory laparoscopic cholecystectomy (LC). Materials and Methods: Between January 1999 and June 2003, 410 consecutive LCs were performed as outpatient procedures. We performed univariate analysis and logistic regression models of preoperative and intraoperative variables. The scoring system developed allowed calculating the ambulatorization probability of LC in each patient. Validation and calibration of the model were realized by means of Hosmer-Lemeshow test. Results: Three hundred sixty-three patients were strictly ambulatory (86.8%). Forty-two patients required overnight admission (10.2%), most of them because of social factors, and 5 patients were admitted. Predictive factors related to overnight stay or admission were: age of patient over 65 years [P = 0.021; odds ratio (OR) = 2.225; 95% confidence interval (CI), 1.130- 4.381], operation duration superior to 60 minutes (P = 0.046; OR = 2.403; 95% CI, 1.106-5.685), and ‘‘dissection difficulty’’ intraoperative score superior to 6 (P = 0.034; OR = 3.063; 95% CI, 1.086-8.649). The right classification index of the predictive system was 91.7%, reaching a sensibility of 99.7% and specificity of 31.9%. Conclusions: Outpatient LC is safe and feasible. Age of the patient, operation duration, and complexity of surgical dissec- tion during LC are independent factors influencing ambulator- ization rate. Key Words: predictive factors, cholelithiasis, laparoscopic cholecystectomy, ambulatory surgery (Surg Laparosc Endosc Percutan Tech 2008;18:248–253) L aparoscopic cholecystectomy (LC) is accepted as the ‘‘gold standard’’ in the treatment of symptomatic cholelithiasis. However, LC on an outpatient basis remains controversial in some countries. Rates of ambulatory LC have increased in Europe, although they are low in comparison with the development that it has obtained in the United States. 1 It still contrasts with the reticence of many surgeons in Spain to establish the ambulatory LC like a sure and feasible procedure. Unanticipated admission percentage represents an index of quality that measures the success of this surgery, and it oscillates among 1% to 39% due mainly to postoperative symptoms (vomits and abdominal pain), conversion to open surgery, and a patient’s insecurity. 2,3 A high unanticipated admission rate may reflect deficien- cies in patient selection criteria, which for its character- istics, antecedents, or preoperative findings were not candidates to outpatient LC. The aim of our study was to analyze preoperative and intraoperative variables related to ambulatorization and to develop a global scoring model to predict the individual probability of patients to unsuccessful out- patient LC. MATERIALS AND METHODS We prospectively analyzed 410 consecutive patients undergoing elective LC for symptomatic gallbladder disease during the period between January 1999 and June 2003. Patients with suspicion of choledocholithiasis, cholecystitis, unstable American Society of Anesthesio- logists (ASA) III or IV classification, unavailability of a competent adult to accompany the patient home and look after the patient for 24 hours, or no initial acceptation of the ambulatory discharge were excluded from the study. Patient education regarding the outpatient nature of the procedure was initiated by the surgeon during the first visit. Informed consent was obtained. All the procedures were scheduled in the morning. Patients were admitted an hour before surgery on the day of operation. Anesthesia was performed in all patients by the same anesthetic equipment and was protocolized. Pro- phylaxis against postoperative pain and nausea was achieved with metoclopramide endovenous and ketopro- fen intramuscularly. Ondansetron was required only in 12 cases to avoid postoperative nausea or vomiting. Anesthesia was initiated by fentanyl 0.1 to 0.10 mg and propofol 0.05 to 0.10 mg/kg. During induction, muscle relaxant (rocuronium) was given (0.6 to 1 mg/kg). Anesthesia was maintained by propofol, fentanyl, and Copyright r 2008 by Lippincott Williams & Wilkins Received for publication March 4, 2007; accepted January 14, 2008. From the *Instituto de Cirugı´a General y del Aparato Digestivo (ICAD), Clı´nica Quiro´ n; and wServicio de Cirugı´a General y Aparato Digestivo, Hospital Clı´nico Universitario, Avda Blasco Iba´n˜ ez, Valencia, Spain. Reprints: Dr Jose Bueno Lledo´ , MD, C/ Padre Rico 7, puerta 6 CP46008 Valencia, Spain (e-mail: buenolledo@hotmail.com). ORIGINAL ARTICLE 248 Surg Laparosc Endosc Percutan Tech  Volume 18, Number 3, June 2008