Section Editor: Peter G. Duncan.
The first postanesthesia care unit (PACU) was established in 1923, and some attribute their subsequent widespread use to reducing postoperative morbidity and mortality . By providing postsurgical patients with continuous evaluation and specialized care, PACUs can contribute to increasing health care expenses. In fact, the staffing cost for a 2-h PACU stay is roughly equivalent to the staffing costs for a 24-h stay in the hospital ward. Despite the expense, analysis of factors that contribute to appropriate and inappropriate PACU length of stay (LOS) is minimal.
There are studies that measure PACU LOS as a secondary outcome when comparing different anesthetic techniques in similar patient populations undergoing a specific procedure [2,3]. However, classifying and quantifying factors that prolong LOS is difficult because appropriate and average discharge times have not been established. There has been little to no investigation to evaluate variations in phase 1 PACU LOS or to identify the determinants of phase 1 PACU LOS. There is no agreement among institutions and no literature that describe an ideal PACU LOS based on objective patient variables. Therefore, arbitrary time limits are used at our institution to quantify excessive PACU LOS, regardless of the surgical procedure, anesthetic technique, or patient's health status.
To initiate a better understanding of the PACU LOS, this study had three aims: 1) to describe actual PACU LOS for different surgical procedures; 2) to describe the appropriate PACU LOS for different surgical procedures; and 3) to identify clinical factors that contribute to variations in the appropriate PACU LOS.
This study was conducted at Barnes-Jewish Hospital south campus, a 900-bed tertiary-care academic hospital, at which approximately 23,000 anesthetics are performed per year. Approximately 15,000 postoperative patients are admitted to the phase 1 PACU each year. The remaining patients are sent directly to the phase 2 PACU, a surgical floor, or an intensive care unit and are not recovered in a phase 1 PACU. This study was approved by the Washington University Human Studies Committee.
We performed an observational study on a convenience sample of postsurgical patients who had recovered in one of our three phase 1 PACUs over a 6-week period between February and March 1995. The convenience sample excluded a significant number of PACU patients. Because the thoracic PACU is in a separate location, as is the phase 1 PACU for otorhinolaryngology, neurosurgery, and ophthalmology, patients in these PACUs were excluded. The study required close observation of patients to determine the appropriateness of discharge; therefore, it was not possible to observe patients in multiple locations. Thus, all observations were restricted to the largest phase 1 PACU. In addition, some patients from the largest phase 1 PACU were excluded, such as those receiving monitored anesthesia care, those admitted on the weekends or between 5 PM and 8 AM, and those admitted when the principal investigator was already observing six study patients. Because of the close patient observation required, the principle investigator could not observe more than six patients at a given time.
Age, gender, and race information was obtained from the preanesthetic evaluation form. Weight, height, ASA physical status, and concurrent medical disease data were also obtained from the preanesthetic evaluation form. The procedure urgency, surgical procedure, surgical time, anesthetic technique, anesthetic time, blood loss, amount of intraoperative fluids, and amount of blood product transfusion were obtained from the anesthetic records. Actual PACU LOS was defined as the time from the patient's admission to the PACU to the time that the patient left the PACU, as recorded by the PACU nurse. The medically appropriate LOS was the time required for the patient to achieve a medically stable condition for safe PACU discharge. There is no widely accepted or validated gold standard for discharge criteria. At our institution, no routine discharge criteria are used to evaluate discharge appropriateness. Rather, the decision for discharge is at the discretion and availability of individual anesthesiologists when they evaluate the patient before discharge.
To estimate appropriate PACU LOS, prompt determination of discharge appropriateness was required. Therefore, using clinical judgment, JPW determined the medically appropriate time for discharge. Evaluations were made at 30-min intervals for each patient. For patients assessed as not ready for discharge, the medical reasons prohibiting discharge were recorded. Medical reasons for continued PACU evaluation included decreased level of consciousness (arouses to physical stimulus, not verbal stimulus), spinal sensory level above the eighth thoracic level, agitation (unable to remain still to allow routine monitors to function), hypertension or hypotension (blood pressure within 20% of preoperative level), tachycardia (heart rate >100 bpm), other arrhythmias (atrial fibrillation, atrial flutter), hypoxia (oxygen saturation <90%), oxygen dependency, inadequate pain control (pain score >6 of 10), emesis, hypothermia (temperature <34[degree sign]C), oliguria (<0.5 mL of urine per kg of body weight per hour), or endotracheal intubation.
A delayed discharge was defined as a patient leaving the PACU >30 min after the principle investigator determined that discharge was medically appropriate. Premature discharge was defined as discharge before the principle investigator determined that discharge was medically appropriate.
The Charlson Weighted Index of Comorbidity  was used to evaluate the impact of concurrent disease on the recovery process and was determined using information in the preanesthetic form. The index weights the following disease states: myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, connective tissue disease, ulcer disease, liver disease, diabetes, hemiplegia, moderate or severe renal disease, diabetes with end-organ damage, any tumor, leukemia, lymphoma, metastatic solid tumor, and acquired immunodeficiency syndrome. Higher index scores correspond to more comorbid disease states.
Each operation was classified into one of three categories: elective, urgent (requiring surgery within 24 h but not immediately), or emergent (surgery required immediately). The surgical procedure performed was obtained from the intraoperative surgical record.
The six groupings for anesthetic technique were: 1) regional, 2) general mask or laryngeal mask airway case (LMA), 3) general endotracheal anesthetic (GETA) with an inhaled anesthetic, 4) GETA with a narcotic technique, 5) GETA with "other IV" drug, or 6) other. A regional anesthetic was coded when a spinal, epidural, or other regional block was performed and not supplemented with general anesthesia. An inhaled GETA was defined as isoflurane concentration >0.7 end-tidal, an enflurane concentration >1.0 end-tidal, or a halothane concentration >0.4 end-tidal for more than half of the intraoperative anesthetic time. A narcotic GETA was coded when narcotics were used without other IV maintenance drugs and when the inhaled drugs were used at concentrations less than those previously listed. Other IV GETA was coded when continuous infusions of propofol, etomidate, or brevital were used for a general endotracheal anesthetic. Anesthetic techniques that could not be coded into these categories were listed as other (for example, anesthetics combining general and regional techniques).
A uniform data collection form expedited information extraction. Data were entered into a computer spreadsheet. Statistical analysis of the data was performed by using the Statistical Analysis Software system (SAS Institute Inc., Cary, NC). Frequency distributions for actual and medically appropriate PACU LOSs were described using means, modes, and quartile percents. Data are expressed as means +/- SD. Medically appropriate and actual PACU LOSs were not normally distributed; therefore, Spearman's rank order correlation coefficients were used to identify significant bivariate predictors of appropriate PACU LOS, with P <0.05 selected as significant. Logistic analysis, with variable dummy coding as appropriate, was used to determine independently significant predictors of the appropriate PACU LOS. This also produced an odds-adjusted ratio value for the appropriate PACU LOS.
The population included 340 patients: 198 (58%) female, 142 (42%) male patients; 247 (73%) Caucasian, 92 (27%) African American, and 1 (0.3%) Hispanic patients. The mean age was 51 +/- 18 yr. Of the patients, 260 (76%) were ASA physical status II or III. The mean surgical time (surgical incision to placement of surgical dressing) was 118 +/- 83 min, and the mean anesthetic time (preoperative evaluation in holding area to arrival time in PACU) was 181 +/- 96 min.
The mean actual PACU LOS was 95 +/- 43 minutes, and the median and mode actual PACU LOS was 90 min. Values ranged from 30 to 330 min, with a 50th percentile of 90 min and a 90th percentile of 140 min.
The mean medically appropriate PACU LOS was 71 +/- 37 min, and the median and mode medically appropriate PACU LOS was 60 min. Values ranged from 0 to 240 min, with 25th and 50th percentiles of 60 min and 90th and 95th percentiles of 120 min. Figure 1 shows the distribution of actual and medically appropriate PACU LOSs.
We evaluated the association of medically appropriate LOS with demographic, clinical, surgical, and anesthetic variables. Variables predictive of the medically appropriate PACU LOS (using bivariate analysis) are shown in Table 1. Because multiple variables were significant, logistic regression was applied to determine which of these variables remained independently significant. The patients were dichotomously classified as appropriate ("yes" or "no") for discharge at each time interval. Because the mode of the medically appropriate LOS was 60 min, which is a clinically useful goal for discharge, this level was chosen for modeling the appropriate PACU LOS. Table 2 illustrates the results of the logistic regression analysis for the variables that were independently significant: anesthetic time (P = 0.0032), regional anesthetic (P = 0.0002), and amount of intraoperative fluids (P = 0.0123). The logistic Equation thatestimates the log of the odds of being discharged in <60 min is: Equation 1
The coefficient of concordance for this model is 0.772, and the likelihood ratio chi squared is 72.3 with 3 degrees of freedom (P = 0.0001). The examples below demonstrate the clinical application of the equation.
Example 1: Carpal tunnel repair with IV regional technique: anesthesia time 1h, intraoperative fluids 0.5 L, regional technique 1. Equation 2 Therefore, the odds of being discharged in <or=to60 min are 20:1. Alternatively, the probability (odds/1-odds) of being discharged within 60 min is 0.95.
Example 2: Exploratory laparotomy with general endotracheal narcotic technique: anesthesia time 3.5 h, intraoperative fluids 6 L, regional technique 0. Equation 3 Therefore, the odds of being discharged in <or=to60 min are 1:5. Alternatively, the probability of being discharged within 60 min is 0.2.
Of the 340 patients studied, 68 (20%) were classified as having delayed discharges. The causes of delayed discharges are summarized in Table 3. Of these 68 patients, 4 experienced more than one delay. For example, the patient waited for a physician release for >30 min; when the release was obtained, the nurse was unable to transport the patient to the floor. For these four patients, only the last cause of their delay is presented in Table 3.
Forty-three total delayed hours were spent in the PACU, which is 8% of the total hours spent in the PACU by all study patients. Waiting for a physician release accounted for the most delayed hours, and more than half of all delayed hours were caused by waiting for a physician release or laboratory and electrocardiographic results.
Five patients (1.5%) were discharged before discharge was medically appropriate. These patients were classified as being "discharge inappropriate" because of a decreased level of consciousness (two patients), inadequate pain control (one patient), and a decreased level of consciousness and inadequate pain control (two patients). No negative outcomes occurred in these patients.
This study documents the actual and appropriate PACU LOSs for a large cohort of patients undergoing a variety of surgical procedures at a tertiary academic teaching hospital. The study demonstrates a significant variation in both the actual and the appropriate PACU LOS. The appropriate LOS was 30 min for some patients, whereas others required >150 min. There are no accepted standards for appropriate PACU LOSs for different surgical procedures and anesthetic techniques. For example, one might expect phase 1 recovery from a carpal tunnel repair to require less time than recovery from intraabdominal surgery. In this study, we attempted an initial determination of modeling appropriate PACU LOS to predict which patients would most likely be appropriate for discharge in <60 min.
Furthermore, we demonstrated that >20% of patients at our institution experienced a delayed discharge that accounted for 8% of total PACU time. Delayed discharge was arbitrarily defined as a patient being discharged >30 min after a discharge appropriate assessment. We thought that 30 min should be adequate for nurses to perform paper work, for transporters to assist with discharge, and for other nonmedical issues to be completed. Using this definition, we identified common causes of delays. Delays related to medical personnel accounted for >50% of the delays and over half of all the delayed hours. For the 340 study patients, 8424 min were spent in the PACU after patients were judged to be ready for discharge. The cost of PACU care after the first half hour is $55.22 per half hour. Therefore, these 340 patients' care cost >$15,000 ($44 per patient) after they were medically ready for discharge. At the west pavilion PACUs at Barnes-Jewish South Hospital, this would translate to 16,000 patients at a cost of $704,000 over 1 yr. Examining the difference between actual and appropriate PACU LOSs suggests that most of the difference is caused by administrative or nonmedical issues. Methods to decrease the time required for administrative functions could safely eliminate a significant portion of unnecessary PACU LOSs.
An important goal of this project was to identify factors that influence the appropriate PACU LOS. Independent predictors of the medically appropriate PACU LOS were anesthetic time, anesthetic technique, and amount of intraoperative fluids. Using mathematical modeling, we created a prediction method to determine the odds of a patient being ready for discharge in <60 min. Identifying these factors and using the model helps to define clinically relevant targets for efficiency of care evaluations. For example, PACU efficiency efforts could focus on reviewing the charts of patients whose LOS significantly varies from the duration predicted by the anesthetic time, anesthetic technique, and the amount of intraoperative fluids received.
The results of this study could be biased for several reasons. First, the selection of study patients could have introduced bias. The sample contained only patients admitted to one of three possible phase 1 PACUs, was only collected on weekdays between 8 AM and 5 PM, and excluded some west pavilion PACU patients because it was not logistically possible for the primary investigator to observe additional patients at that time. However, the demographic, surgical, and anesthetic data indicate that a wide cohort of subjects was studied; therefore, it is unlikely that the restrictions significantly biased the results. However, there were certain surgical procedures not evaluated; therefore, our results should be confirmed in these other surgical populations. The exclusion of some PACU patients occurred during the busiest PACU times. Exclusion of these subjects may have biased our results by minimizing the actual PACU LOS, because administrative delays are more likely to be increased when the PACU is near maximal capacity.
In addition to patient selection bias, bias could also have occurred in the determination of appropriateness of discharge. Because of the lack of a gold standard to determine the appropriate PACU LOS, readiness for discharge was assessed by an anesthesiologist in this observational study. All assessments for discharge readiness were made by JPW and influenced by individual medical judgment. Because JPW was not blinded to the subject's comorbidity status, surgical procedure, or anesthetic management, there is potential for the introduction of observer bias. Indeed, it is likely that among anesthesiologists discharging PACU patients, some would have different thresholds for safe PACU discharge. It was important in this study to identify a potential area for further evaluation and to develop more objective criteria for determining discharge appropriateness. These criteria would not only assist in improving clinical care and efficiency of care, they would also allow more meaningful comparisons of PACU discharge times at different institutions.
Finally, our patient population was selected exclusively from a large academic medical center. It is possible that this influenced both the appropriate and the actual PACU LOS. One might expect longer anesthetic times for similar surgical procedures performed at an academic institution compared with a private hospital, which this may prolong the medically appropriate PACU LOS for a given procedure. In addition, large centers may experience more administrative delays, therefore prolonging the actual PACU LOS.
In conclusion, there seems to be a significant difference between the appropriate and the actual PACU LOS at our institution. Administrative issues account for a significant amount of the overall PACU LOS. The appropriate PACU LOS can be modeled by using the variables of anesthetic time, anesthetic technique, and intraoperative fluid administration. Clearly, this is an area that merits further investigation at other institutions to confirm our results. This observational study most importantly identifies an area for further investigation.
We thank the PACU staff at Barnes-Jewish Hospital South Campus for their patience with data collection.