It has been an objective of healthcare legislation in most countries to increase the number of patients in day-case surgery and to improve outpatient care. Day-case surgery becomes increasingly important because of budgeting of costs in the health care system. Therefore, it is of interest to the daily organization of a surgical day-care unit to find the reasons that may lead to a prolonged stay of patients in the postanaesthetic care unit (PACU) or even their unanticipated admission to the ward. Evaluation of these factors is important as they may have a direct impact on the budget .
Earlier studies suggested that postoperative pain, and postoperative nausea and vomiting (PONV) are important factors for a prolonged stay after ambulatory surgery [2–4]. However, both pain and PONV occur in the postoperative period when prophylactic strategies cannot be applied any more; and often they are unpredictable. The aim of our study was to identify factors that may predict prolonged stay and admission at a much earlier stage (i.e. that are already obvious before or during surgery) in a cohort of patients undergoing day-case surgery at a University Hospital. Our intention was to improve the organizational procedures in our day-case unit. We studied potentially important factors that are related to the patients themselves, to surgery, to anaesthesia, and to the logistics of a day-care unit.
The anaesthesia day-care unit of the Justus-Liebig University Hospital consists of eight beds (two beds for children older than 3 months and six beds for adults). The unit, staffed by two anaesthesiologists and eight nurses, is managed in two shifts from 06:00 to 20:30 hours. From 1997 onwards, all data collected during the stay of day-case patients were recorded with the online record keeping system NarkoData (Imeso GmbH, Hüttenberg, Germany) . This program collects all relevant data including vital signs and drugs as well as the dataset for quality assurance according to the German Society of Anaesthesiology and Intensive Care Medicine . Pre-and postoperative monitoring is recorded in the PACU and the day-care unit and intraoperative data are recorded in the operating room. At the end of anaesthesia care, the complete protocol file is stored in a database (Oracle 7, Oracle Corporation, 500 Oracle Parkway, Redwood Shores, CA 94065, USA) after numerous checks of plausibility and integrity. Manually recorded anaesthetic records (n = 21) were excluded from this evaluation, as they did not provide the same information as the computerized patient records. Patients were discharged from the day-care unit when their physical status was equivalent to a score > 7 according to the modified Post-Anaesthesia-Discharge-Scoring-System (PADSS) of Chung and colleagues .
The length of postoperative stay was arbitrarily classified as ≤ 3 h, 3–6 h, ≥ 6 h, and unanticipated hospital admission. Factors that may have an impact on the length of postoperative stay were categorized as being related to: the patient [gender, age, body mass index (BMI), ASA physical status, diagnosis according to the International Classification of Diseases (ICD-9)]; to anaesthesia [intubation, type of anaesthesia (regional or balanced anaesthesia, combination of both, intravenous anaesthesia (IVA), IVA plus regional anaesthesia], drugs (inhaled anaesthetics, induction hypnotics, opioids, muscle relaxants); to surgery [duration, speciality (general, traumatology, urology and maxillofacial surgery, others)], type of surgery according to the International Classification of Procedures in Medicine (ICPM), blood loss, pre- and intraoperative haemoglobin concentration, volume of infused colloids and crystalloids, blood products, PONV, postoperative pain; and to logistics and organization of the day-care unit (time of admission, preoperative waiting time).
All data were exported from the database into the statistics program SPSS® (SPSS Software GmbH, München, Germany). We performed an univariate analysis. Gradual multiple logistic regression analysis of nominal and metric predictors was not possible as such analysis requires the complete data of all variables that may be considered as predictors of prolonged stay or admission. Thus, for all cases to be included in the analysis, a complete set of data is needed. For the purpose of this analysis, we defined some factors as independent predictors although they are most probably dependent (for instance, blood loss, volume of infusions, and intraoperative haemoglobin concentration).
Nominal and ordinal parameters were tested with the χ2-analysis for their association with the length of postoperative stay as the variable. Pearson's phi-coefficient (CC) was used to measure the association of nominal parameters. This estimate becomes 0 in case of total independence, and is close to 1 in case of total dependence. The maximum value of CC depends on the number of cells in the cross-table. Gamma was used to measure the association of ordinal variables.
When there was evidence of an association of general parameters (for instance, intravenous hypnotics), further differentiation was carried out. In case of a statistically significant correlation between the factor ‘intravenous hypnotics’ and length of stay, the impact of specific drugs (for instance, thiopental, etomidate, propofol) was tested. For metric parameters of the variable length of postoperative stay, Spearman's ranking correlation coefficient was determined and checked if it differed significantly from zero. Metric parameters were additionally checked with an analysis of variance for similarity of group mean values in relation to length of stay . Hospital charts of all 169 patients who had been admitted to in-patient care were reviewed and the definite cause for admission identified.
From January 1st 1997 to June 30th 1999, complete data sets of 3152 day-case patients were collected. Of those, 415 (13.2%) stayed ≤ 3 h in the day-care unit, 1743 (55.3%) 3–6 h, 825 (26.2%) ≥ 6 h and 169 (5.4%) patients were admitted. Tables 1 and 2 show the distribution of nominal and ordinal parameters in relation to the length of postoperative stay. Table 3 presents data on the degree of association between nominal and ordinal predictors and the length of postoperative stay. Table 4 shows standard deviations and results of the analysis of variance of metric parameters in relation to the length of postoperative stay. Table 5 shows the degree of association between metric parameters and the length of postoperative stay.
Factors related to patients
Female gender (Table 2) and an older age (Table 4) had a significant impact on admission. Body mass index (Table 4) and ASA classification (Table 2) did not show a relevant or a significant correlation with length of postoperative stay.
Factors related to anaesthesia
The type of anaesthesia (Table 1), drugs used for anaesthesia (Table 1) and PONV (Table 2) had a significant impact on the length of postoperative stay. Intubation, IVA and spinal anaesthesia prolonged stay and increased the likelihood of Hospital admission. Balanced anaesthesia, peripheral nerve blocks and the combination of balanced anaesthesia and regional anaesthesia shortened the length of stay. The most commonly used inhalation anaesthetics were halothane and isoflurane. With halothane, the length of postoperative stay was significantly shortened. With propofol, the most frequently used intravenous hypnotic, there was an extension of the length postoperative stay (P < 0.05, n = 1495). Thiopental did not show a significant effect. Of the 80 patients who had received etomidate for induction, a highly significant percentage had to be admitted to hospital. Fentanyl (n = 1718) was associated with an increased length of stay. Remifentanil (n = 62) had no significant impact on the duration of stay or on admittance. The use of vecuronium (n = 1864) correlated with a longer postoperative stay.
Factors related to surgery
Intraoperative blood loss, intraoperative haemoglobin concentration and the volume of infused colloids and crystalloids (Table 4) were the strongest predictors for a prolonged length of stay. Patients who required unanticipated admission had a fourfold increased blood loss (227 vs. 55 mL) compared with patients who were discharged ≤ 3 h (Table 4). Duration of surgery (Table 4), diagnosis and the surgical speciality (Table 2) were also significantly correlated with the length of stay.
Trauma patients had the longest postoperative stay. Surgery of skin and subcutaneous tissue, most often performed by the Department of traumatology, was associated with a higher rate of unanticipated admissions. Patients undergoing surgery according to ICPM ‘endoscopic diagnostics by incision or intraoperative’ or ‘maxillofacial’ often stayed longer than 6 h.
Factors related to logistics and organization
Preoperative waiting time and the time of day of admission to the day-care unit (Table 4) had a significant impact on the postoperative length of stay in the PACU.
Reasons that led to unanticipated admission of 169 patients were mainly surgical: 62 (36.7%) patients were bleeding, 33 (19.5%) had high pain scores and 53 (32.6%) had other surgery-related reasons. Other reasons for unanticipated admission were lack of vigilance (9 patients, 5.3%), PONV (7 patients, 4.1%) and the patient's wish to stay overnight (4 patients, 2.4%).
The objective of this study was to identify factors which were associated with a prolonged postoperative stay of patients in our day-care unit. An improved organization of a busy day-case unit may be achieved by an improved selection of patients, type of surgery and co-ordination of procedures. In our study, 169 of 3152 patients (5.4%) required unanticipated admission. This is a high rate compared with other reports . The study by Fortier and his colleagues  was performed in North America: their long experience in day-case surgery, and the high percentage of procedures which are carried out on an ambulatory basis, may explain their low admission rate of only 1.4%.
The most important factors
The most reliable predictors with the highest correlation for a prolonged stay in the recovery room were increased intraoperative blood loss associated with low intraoperative haemoglobin values and an increased volume of infused colloids. The duration of surgery was identified as a further important factor. Patients admitted to the hospital had the longest surgical procedures. These data are supported by findings of others [2,3,9–11]. In one study, a duration of surgery of more than 60 min was a crucial predictor for the unanticipated admission of a patient . Chung and Mezei  estimated that for each prolongation of the surgical procedure of 30 min, there was a 9% increase in the length of postoperative monitoring. In the study by Osborne and Rudkin , investigating 6000 ambulatory patients, 0.95% of all admissions were due to surgical reasons compared with 0.13% only which were due to anaesthesiological reasons. In our study, we found similar results for the relation between surgery- and anaesthesia-related factors; 87.6% of unanticipated admissions were related to surgical problems, and only 10% of admissions were due to anaesthesiological complications.
Factors related to the patient
Female gender increased the risk of a prolonged stay, although the absolute risk for unanticipated admission was only slightly higher for women than for men (5.7% vs. 5.2%). We also identified advanced age as a risk factor. Fortier and his colleagues  identified advanced age and male gender as predictive factors in their study. Body mass index had no impact. This may be due to the exclusion of patients with significant obesity from ambulatory surgery. The ASA physical status had no significant influence on the postoperative length of stay. This may have been due to the limited number of patients with ASA status III (n = 125). In patients with a higher ASA status, a higher admission rate would be expected. Other studies have suggested a significant influence of higher ASA status or of concomitant diseases (e.g. congestive heart failure) on the rate of unanticipated admission [2,3,9].
Factors related to anaesthesia
In contrast to other studies [10,12], we did not find an increased probability of admission in patients who had received a general anaesthetic compared with patients who had received regional anaesthesia. For spinal anaesthesia, we even found a prolonged postoperative stay. This may be due to spinal anaesthesia-related intraoperative adverse events (for instance, decrease in blood pressure, or bradycardia) or postoperative complaints (for instance, atonic bladder) . In one study, patients receiving spinal anaesthesia or a peripheral nerve block with mainly short-acting local anaesthetics had a shorter length of stay compared with patients undergoing general anaesthesia . In our cohort, most peripheral nerve blocks were performed with short acting local anaesthetics for surgical procedures of short duration. This group of patients had a short duration of stay in the day-care unit.
Patients who have received an intravenous anaesthetic were at higher risk for a prolonged stay. This is likely to be a secondary effect, as this anaesthesia technique has been used mainly with fentanyl in orthopaedic surgery: each of these factors, however, already implied a higher risk. The use of fentanyl, for instance, increased the probability for prolonged monitoring of the patient and hospital admission: 6.3% of all day-case patients receiving fentanyl were admitted. In a previous study, the newer opioid remifentanil  did not have a negative influence on the admission rate. The use of remifentanil – until now rarely used in our Institution because of cost – has to be reconsidered for day-case surgery in our clinic. Vecuronium was used almost exclusively for intubation. Thus, it is not surprising that additional to intubation, neuromuscular blocking agents were associated with a significant influence on postoperative length of stay.
With the exception of etomidate, none of the used intravenous hypnotics (thiopental and ketamine) showed any significant association with the hospital admission rate. We cannot exclude a secondary effect with etomidate, as it was often used in patients with cardiac or pulmonary diseases and in elderly patients, due to its favourable haemodynamic characteristics. We were unable to confirm a positive effect of propofol on the length of stay as described by other authors .
Surprisingly, we found that with halothane there was a lower risk of admission compared with isoflurane. In our clinic, halothane is used for short surgical procedures in children only. Thus, we are probably again dealing with a secondary effect. Isoflurane, which is the standard inhalation anaesthetic, had no significant influence on postoperative stay.
Factors related to the surgical speciality
Reports on the impact of the surgical speciality and the type of the procedure on the length of stay in day-case surgery are inconsistent. Mingus and his colleagues found no influence of the type of surgery on the admission rate in 8549 patients . Two studies reported an increased probability of admission among patients undergoing ear nose and throat (ENT) surgery [3,12]. In the study by Gold and his colleagues , surgery of the lower abdomen and urological procedures were associated with a higher admission rate. In our cohort, patients undergoing orthopaedic surgery were monitored for a longer period in the day-care unit and had a higher admission rate than other groups of patients. This variability of results puts into question the use of general risk scores for surgical procedures. An assessment of the patient should always be carried out individually in each clinic.
Factors related to logistics and organization
The time of admission to the day-care unit had a variable influence. Patients who were admitted early in the morning had the longest duration of postoperative monitoring. Patients that were admitted later during the day had an increased rate of unanticipated admission. This corresponds to the findings of Fortier and his colleagues , who observed a higher admission rate among patients in whom surgery was completed after 15:00 hours. Furthermore, we found that patients with the longest preoperative waiting times had the shortest postoperative day-case monitoring times. These data suggest that patients who are scheduled late, and in whom the remaining postoperative time until closure of the day-care unit is insufficient for discharge in an adequate physical status, are likely to be admitted to Hospital. These problems may be addressed by an improved scheduling of day-case patients.
The analysis confirms that there are multiple reasons for a prolonged stay of patients in a day-case surgery unit. Surgical reasons (for instance, intra- and postoperative bleeding, type and duration of surgery) play a major role compared with factors that are related to the patient or the anaesthetic technique. Such analyses may help to further reduce the number of patients who stay too long in the day-care unit after ambulatory surgery, and to limit the number of unanticipated admissions.
We would like to thank the Büro für Statistik GbR, Gießen for the help in data management and statistical evaluation.
Part of this work has already been published (the same cohort of patients described by Junger, A, Benson M, Klasen J, Sciuk G, Fuchs C, Sticher J, Hempelmann G. Einflussfaktoren und Prädiktoren für die ungplante stationäre Aufnahme tageschirurgischer Patienten [Influences and predictors of unanticipated admission after ambulatory surgery]Anästhesist 2000;49: 875–880) and permission for reproduction has been granted by the copyright holders Springer-Verlag GmbH & Co.
1 Dexter F, Tinker JH. Analysis of strategies to decrease postanesthesia care unit costs. Anesthesiology
1995; 82: 94–101.
2 Chung F, Mezei G. Factors contributing to a prolonged stay after ambulatory surgery. Anesth Analg
1999; 89: 1352–1359.
3 Fortier J, Chung F, Su J. Unanticipated admission after ambulatory surgery – a prospective study. Can J Anaesth
1998; 45: 612–619.
4 Green G, Jonsson L. Nausea: the most important factor determining length of stay after ambulatory anaesthesia. A comparative study of isoflurane and/or propofol techniques. Acta Anaesthesiol Scand
1993; 37: 742–746.
5 Benson M, Junger A, Quinzio L, Fuchs C, Sciuk G, Hempelmann G. Clinical and practical requirements of online software for anesthesia documentation an experience report. Int J Med Inf
2000; 57: 155–164.
6 DGAI-Kommission ‘Qualitätssicherung und Datenverarbeitung in der Anästhesie’. Kerndatensatz Qualitätssicherung in der Anästhesie. Anästh Intensivmed
1993; 34: 331–335.
7 Chung F, Chan VW, Ong D. A post-anesthetic discharge scoring system for home readiness after ambulatory surgery. J Clin Anesth
1995; 7: 500–506.
8 Sachs L. Angewandte Statistik.
Berlin, Heidelberg: Springer-Verlag, 1997.
9 Mingus ML, Bodian CA, Bradford CN, Eisenkraft JB. Prolonged surgery increases the likelihood of admission of scheduled ambulatory surgery patients. J Clin Anesth
1997; 9: 446–450.
10 Osborne GA, Rudkin GE. Outcome after day-care surgery in a major teaching hospital. Anaesth Intensive Care
1993; 21: 822–827.
11 Gold BS, Kitz DS, Lecky JH, Neuhaus JM. Unanticipated admission to the hospital following ambulatory surgery. JAMA
1989; 262: 3008–3010.
12 Pavlin DJ, Rapp SE, Polissar NL, Malmgren JA, Koerschgen M, Keyes H. Factors affecting discharge time in adult outpatients. Anesth Analg
1998; 87: 816–826.
13 Nolte H. Praxis der Regionalanästhesie. In: Doenicke A, Kettler D, List WF, Radke J, Tarnow J, eds. Anästhesiologie
, 7th edn. New York: Springer, 1998: 486–526.
14 Scholz J, Steinfath M. Ist Remifentanil ein ideales Opioid für das anästhesiologische Management im 21 Jahrhundert? [Is remifentanil an ideal opioid for anaesthesiologic management in the 21st century?]. Anasthesiol Intensivmed Notfallmed Schmerzther
1996; 31: 592–607.