Prolonged time to extubation has been defined as the occurrence of a 15-minute or longer interval from the end of surgery to removal of the tracheal tube.1–3 In routine practice, approximately 15% of extubations are prolonged.1,2 The incidences of prolonged extubations are increased by large mean extubation times, large standard deviation (SD) of extubation times, or both.1,3 The incidence of prolonged extubations contains the same statistical information as the mean and SDs of extubation times.a
Prolonged extubations matter clinically for multiple reasons:
- Prolonged extubations can be assessed retrospectively from anesthesia information management system data or prospectively from observation in operating rooms (ORs).1,2
- The incidences of prolonged extubations are modifiable (e.g., from meta-analysis of randomized trials, ≥ 95% reduction with desflurane versus isoflurane).1,3
- Cases with prolonged extubations are rated by anesthesiologists as having poor recovery from anesthesia.4
- Cases with prolonged extubations have a larger chance, than when extubations are not prolonged, that members of the OR team will be idle (i.e., unoccupied with patient-care activities) while waiting for extubation slowing workflow (P < 0.0001): 21% for <5 minutes, 42% for 5 to <10 minutes, 87% for 10 to <15 minutes, and 100% for ≥15 minutes.2
- Cases with prolonged extubations have longer times from when the patient exits the OR until the start of surgery of the surgeon’s next case in the same OR (i.e., skin incision or equivalent).1
- When surgeons score the importance of anesthesiologists’ attributes on a scale from 0, “no importance,” to 4, “a factor that would make me switch groups/hospitals,” their average score is 3.9 for “patient quick to awaken.”5
However, it is not known for how many minutes prolonged extubations increase the time in the OR and whether the increases are substantive economically. Prolonged extubations may not cause increased OR time if, for example, the additional time is shorter than the time to complete paperwork or computer charting that are mandatory before the patient can leave the OR. Prolonged extubations may also not be responsible for increasing OR time if frequently the postanesthesia care unit cannot accept another patient from the ORs due to resource constraints.6–11 Our objective was to use anesthesia information management system data to quantify the increases in the mean times from end of surgery to exit from the OR associated with prolonged extubations and to test whether the increases are economically substantive (≥ 5 minute). We performed this analysis multiple ways to try to avoid the chance of studying simply noncausal associations.
The Thomas Jefferson University IRB approved this retrospective study at its affiliated academic hospital without a requirement for written consent. The hospital is a multiple specialty, academic, tertiary hospital with an anesthesia information management system (AIMS).1,12,13 The data were collected from November 2005 through December 2012. This period included all available cases in the AIMS through the start of study design. Cases in which the patient’s trachea was not intubated or extubated while physically in the OR were excluded (e.g., when the patient arrived already intubated or when the tracheal tube was not removed at the conclusion of the case). Cases where the airway was secured via a laryngostomy or tracheostomy stoma also were excluded. The data elements used in the study were the times of patient entrance into the OR, end of surgery, extubation, and exit from the OR; positioning (prone or not prone) (Table 1); anesthesia Current Procedural Terminology® code (Table 2); a de-identified surgeon code; and the surgeon’s specialty (Table 3). There were no missing values among these data. These fields were selected from an exploratory analysis of predictors of prolonged extubation (Appendix). Readers particularly interested in the statistical methodology may want to read the Appendix before proceeding.
The historical cohort was grouped into 22 sequential quarters (i.e., 16-week periods). Because prolonged extubations were relatively uncommon, we used periods of 16 weeks rather than 4 weeks, thereby increasing the numbers of periods with an average of at least 1 prolonged extubation per week. We studied all combinations of hours from OR entrance to end of surgery and positioning (Table 1), anesthesia Current Procedural Terminology code (Table 2), or surgeon (Table 3) with at least 8 quarters where, on average, there was at least 1 prolonged extubation per week (see Appendix). Every such combination of stratification variable and quarter also had at least 1 (and usually many) prolonged extubation.
Statistics were calculated using the method of batch means, as previously described for OR management analyses.14–16 Batch means are used because the master surgical schedule influences which surgeons work on which days, and thereby causes the presence of prolonged extubations to be clustered by time of day and day of the week.17,18 Prolonged extubations can be influenced by personnel, and their availability can depend on the numbers of simultaneous ends of cases.2,19–22 The effective sample size is the number of quarters (8 ≤ N ≤ 22, see preceding paragraph), not the numbers of patients or events (prolonged extubations). For each combination of stratification variable, category of the stratification variable, and quarter, the mean time from end of surgery to OR exit was calculated separately for the extubations that were not prolonged and for those that were prolonged. The difference of the means was then taken. For each combination of stratification variable and category of the stratification variable, the mean and the SEM was calculated among quarters. For each stratification variable, Dersimonian and Laird random effects analysis was then used to obtain the pooled mean estimate among the different categories of the stratification variable.23–27 The results are reported as mean ± SEM. The lower 95% confidence limits were calculated using the Student t statistic with the sample size being the number of categories of the stratification variable and are described by the phrase “at least.”26,27 One-sided P-values testing whether the means exceeded 5 minutes were calculated.1 We selected 5 minutes as economically relevant in part because prolonged extubations have been shown to increase the mean time from OR exit until the start of surgery (incision) of the surgeon’s next case by 5 minutes.1,28,b A 5-minute delay has substantive economic value when occurring in an OR with at least 8 hours of cases.c
We performed 4 sensitivity (secondary) analyses to increase our confidence that we were studying a true relationship between prolonged extubation and increased time from end of surgery to OR exit. The first was conducted by testing whether the means exceeded 10 minutes. Another sensitivity analysis was performed by repeating the analysis with no stratification. A third approach was to test, for each stratification variable, the variability (SD) of the time from end of surgery to OR exit, instead of the mean time (Table 4). A fourth was to test, for each stratification variable, the increased percentage cases taking longer than scheduled.29–33 Finally, we examined the median extubation times. If prolonged extubation affects workflow, then once tracheal extubation occurs, the time from extubation to OR exit should be briefer.2 The times from extubation to OR exit were calculated for each studied case performed during one of the combinations of stratification variable and quarter. The median was taken for each quarter among the cases with extubation times that were not prolonged versus prolonged. The difference of the medians was calculated for the quarter. The 1-sided upper confidence limit for the increase in time was calculated using Student t test applied to the N = 22 pairs (i.e., the number of quarters). Whereas there is accurate classification of an extubation as prolonged, measurement of the time of extubation is subject to several minutes of uncertainty.13 Thus, the median of the large sample sizes within quarters was studied. A consequence of considering the medians was that this secondary analysis of median extubation times was performed without stratification by the categories of the stratification variables (Table 4).
Overall, 15.4% ± 0.4% of extubations were prolonged. Durations of cases and positioning are described in Table 1, most common anesthesia Current Procedural Terminology codes in Table 2, and types of surgeons in Table 3.
The mean times from end of surgery to OR exit were at least 12.6 minutes longer for prolonged extubations compared with extubations that were not prolonged when calculated with stratification by duration of surgery and positioning (13.0 ± 0.1 minutes). This prolongation was significantly larger (P < 0.0001) than 5 minutes, the reference time for an economically substantive difference (see Methods and its footnotes b and c). The mean times were at least 11.7 minutes longer when calculated stratified by anesthesia Current Procedural Terminology code (12.4 ± 0.4, P < 0.0001) and at least 11.3 minutes longer when calculated stratified by surgeon (12.4 ± 0.6, P < 0.0001).
We performed several sensitivity (secondary) analyses to explore the results further (see Methods). The increases in the mean times from end of surgery to OR exit associated with prolonged extubation were longer than 10 minutes: P < 0.0001, P < 0.0001, and P = 0.0024 for each of the 3 methods of stratification, respectively. The increases in the mean times from end of surgery to OR exit associated with prolonged extubation, without stratification, was at least 13.4 minutes (13.7 ± 0.2 minutes). The P < 0.0001 compared to 10 minutes.
Prolonged extubations were associated with longer SDs of the times from end of surgery to OR exit (Table 4). The increases were at least 2.4 minutes when calculated with stratification by duration of surgery and positioning (3.3 ± 0.5 minutes, P < 0.0001 for significant lengthening). The SDs were at least 1.5 minutes longer when calculated stratified by anesthesia procedure code (2.8 ± 0.7 minutes, P = 0.0008) and at least 0.9 minutes when calculated stratified by surgeon (2.6 ± 0.9 minutes, P = 0.012).
Prolonged extubations were associated with increased percentages of cases taking longer than scheduled. The increases were at least 10.1% when calculated with stratification by duration of surgery and positioning (11.0% ± 0.5%, P < 0.00001). The increases were at least 12.0% when calculated stratified by anesthesia procedure code (12.8% ± 0.5%, P < 0.00001) and at least 11.5% when calculated stratified by surgeon (12.4% ± 0.5%, P < 0.00001).
Prolonged extubations were associated with briefer median times from extubation to OR exit, all P < 0.00001 (Table 4). This finding was expected with other tasks needed for OR exit having been completed while waiting for extubation (i.e., indicates concurrent validity).2 The reduction in time was not, however, sufficiently long to compensate for the increased time associated with prolonged extubations. The 95% upper confidence limits were <1 minute for each of the 3 populations of cases (0.7, 0.6, and 0.4 minutes, respectively).
Prolonged extubations slow OR workflow and increase the time from OR exit until the start of the surgeon’s next case (see Introduction).1,2 From our current study, we can add that prolonged extubations also are associated with increases in the time from end of surgery until OR exit. All 4 analyses show the mean times exceed the economically substantive duration of 5 minutes and are very likely reproducible (since P < 0.0001). Prolonged extubations are regarded unfavorably by both anesthesiologists and surgeons,4,5 and their incidence can be reduced.1,3 We therefore recommend that anesthesia providers document the times of extubations and monitor the incidence of prolonged extubations as an economic measure. This would be especially important for providers at facilities with many ORs that have at least 8 hours of cases and turnovers (see footnote c in Methods).
Previously, Masursky et al.2 showed that the proportion of cases with at least 1 provider in the OR idle while waiting for extubation is a simple, reliable, and valid quantification of the influence of the extubation time on OR workflow (see Introduction). Among studied cases with prolonged extubation, every case had at least 1 person idle. However, that observation showed predictive validity of the measure, not that prolonged extubation caused the delayed workflow. Our current findings provide strong evidence of a causal relationship (i.e., prolonged extubations are a bottleneck to OR exit).
For example, we consider the procedures in the first and last rows of Table 2. From the last row, a prolonged extubation time after urethrocystoscopy may cause an embarrassing delay (e.g., since the procedures are brief, OR times 1.67 ± 0.02 hours). However, although this anesthesia code was one of only 14 with sufficient prolonged extubations for study (Table 2, column 1), the last row accounted for only 2.0% ± 0.2% of all prolonged extubations. In contrast, consider the first row, “anesthesia for extensive spine and spinal cord procedures,” which accounted for 17.7% ± 0.4% of all prolonged extubations. Such procedures are in prone position and are long, OR times 4.59 ± 0.03 hours. For such anesthesia codes, it was not clear before our study that prolonged extubations were causing increased OR time, as compared with other steps taking time before the extubation can occur and the patient can leave the OR. Yet for both the mean differences in times from end of surgery to OR exit between prolonged and not-prolonged extubations exceeded 5 minutes (P < 0.0001), 12.6 ± 1.4 minutes for the former and 13.4 ± 0.5 minutes for the latter anesthesia codes. More importantly, for the prone, long procedures, the mean difference exceeded 10 minutes (P < 0.0001). Thus, prolonged extubations substantively increase OR times.
Each of our 3 methods of stratification resulted in highly similar estimates of the effect of prolonged extubation on mean time in the OR. Furthermore, we have been cautious in only drawing quantitative conclusions with a substantial statistical chance of reproducibility (P < 0.0001 for the measured delay being longer than 5 minutes). However, the SEMs differ markedly among stratifications (Tables 1–3) because of limited sample sizes. Unfortunately, this cannot be rectified with more centers or even a larger center since it is influenced by surgeon (see Appendix), and more centers do not increase the number of cases of individual surgeons.34 Prolonged extubations have the highest incidence among long (e.g., ≥ 5 hour) cases and in cases where surgery is performed in the prone position (Table 1). Our prior reanalyses of randomized trial data showed that anesthetic choice substantially influences the incidence of prolonged extubations on a proportional basis.1,3 However, few of the studies included the populations with the largest incidences of prolonged extubations. Further study of such populations is recommended.
Franklin Dexter is the Statistical Editor and Section Editor for Economics, Education, and Policy for Anesthesia & Analgesia. This manuscript was handled by Steve Shafer, Editor-in-Chief, and Dr. Dexter was not involved in any way with the editorial process or decision.
The choices of stratification variables were based on exploratory analyses to predict whether a patient would have prolonged extubation. Prolonged extubations have economic consequence and the issue is whether a surgical suite should plan on a specific extubation being prolonged (i.e., at least better than a 50:50 chance that prolonged when so predicted). The variables explored for inclusion included those considered above (i.e., time from OR entrance to end of surgery, prone positioning or not (Table 1), anesthesia Current Procedural Terminology code (Table 2), surgeon (Table 3), specialty (Table 3), days from the start of study, time of day of extubation, elective versus urgent (including weekend, and/or holiday), and patient factors: ASA physical status, body mass index, and age.
Classification trees were applied to the 77,953 cases using 3 different criteria: φ coefficient, Gini index, and twoing (SYSTAT 13, Systat Software, Inc., Chicago, IL). If no node (i.e., bottom of branches) of a classification tree had a larger than 50% incidence of prolonged extubations, then the classification tree added no value, and that was what we found. The limiting factor was our requirement that there be at least 5 cases at the bottom of each branch. Substantive predictive improvement to the combination of preceding duration and positioning (Table 1) would be by surgeon (Table 3), as surgeon influences workflow. However, adding surgeon reduced the sample size resulting in too few prolonged extubations. Importantly, adding more cases, centers, etc., would have no effect, because what differs is the workflow of the individual surgeon.
As expected from the preceding results, logistic regression was equally unsatisfactory at predicting prolonged extubation. Each continuous variable was transformed individually (X−2.0, X−1.0, X−0.5, ln(X), X0.5, X1.0, and X2.0) to achieve linearity between the logit and the variable. All pairwise interaction variables were considered. Forward stepwise regression was performed, with (removal applied) both at P < 0.05 and repeated with P < 0.001 criteria, but with no resulting predictive combination having a statistically significant probability of prolonged extubation exceeding 50%. The best predictor (by F-statistic) was the interaction of (1) time from OR entrance to end of surgery and (2) prone positioning. However, when the time was binned by hour (Table 1), the residuals in the logit scale were not homogeneous, varying by surgeon. Therefore, the approach that we took for the Results was to perform 3 separate analyses using the 3 best classifiers: time and prone in combination (Table 1), anesthesia procedure (Table 2), and surgeon (Table 3).
Name: Franklin Dexter, MD, PhD.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Franklin Dexter has approved the final manuscript.
Conflicts of Interest: The Division of Management Consulting performs statistical analyses for hospitals and companies, including Merck that funded this study. Income from the Division’s consulting work is used to fund Division research. Franklin Dexter has tenure and receives no funds personally, including honoraria, other than his salary and allowable expense reimbursements from the University of Iowa. He and his family have no financial holdings in any company related to his work, other than indirectly through mutual funds for retirement.
Name: Richard H. Epstein, MD, CPHIMS.
Contribution: This author helped design and conduct the study and write the manuscript and is the archival author.
Attestation: Richard Epstein has approved the final manuscript.
Conflicts of Interest: The author has no conflicts of interest to declare.
We appreciate input from Eric M. Maiese, PhD, Merck Global Health Outcomes on study concept and design and recommendations on the paper.
a For example, Figure 3 of Reference (1) shows a surgeon’s extubation times, mean 9.27 minutes, and SD 5.68 minutes (N = 281). All extubation times combined followed a Weibull distribution (P = 0.58), but extubation times longer than the mean of 9.27 minutes followed a folded normal distribution (P = 0.45). To calculate the percentage of extubation times that were prolonged, the Z-score = (159.27)/5.68 = 1.01. Taking 100% minus the inverse of the cumulative normal distribution for that Z score gives 15.7% prolonged extubations. This model was accurate for that surgeon (actual 15.3%) and all surgeons (bias 0.7% ± 0.2%). Using 10 minutes as a threshold would have resulted in a percentage prolonged for that surgeon that was much larger, 38.8%. The statistical basis for the analysis of prolonged extubations (i.e., the folded normal) can apply provided the threshold (15 minutes) exceeds the mean, which it did for that surgeon and all 98 surgeons. Using 10 minutes as a threshold would have excluded analysis of 29.5% of the surgeons.
b The 95% confidence interval for the mean increase in time, calculated using 5% trimmed data, was 2.7 to 7.1 minutes, P < 0.0001. When this mixed effects modeling was repeated with control for other variables as fixed effects (e.g., time from OR entrance to end of surgery), estimates of mean differences > 4.0 minutes and all P < 0.0014. When analyses were repeated by Wilcoxon-Mann-Whitney stratifying by surgeon, P < 0.0001. See the second full paragraph on page 572 and the second column of page 576 of Reference (1).
c For an OR with at least 8 hours of cases, the cost of a 5-minute delay would be at least U.S.$15, where $15 < (5 minutes) × (1.1 minutes of regularly scheduled labor costs per 1.0 minute time) × ($3.35 per minute for surgical technologist, registered nurse, and anesthesiologist).28 In comparison, $15 is approximately the United States cost of 3 MAC hours of desflurane at 0.5 L/min fresh gas flow or 4 MAC hours of sevoflurane at 2 L/min. Varkey JK. Cost analysis of desflurane and sevoflurane an integrative review and implementation project introducing the volatile anesthetic cost calculator iVAC. Doctor of Nursing Practice Capstone Project, Texas Christian University 2012, page 19.
1. Dexter F, Bayman EO, Epstein RH. Statistical modeling of average and variability of time to extubation for meta-analysis comparing desflurane to sevoflurane. Anesth Analg. 2010;110:570–80
2. Masursky D, Dexter F, Kwakye MO, Smallman B. Measure to quantify the influence of time from end of surgery to tracheal extubation on operating room workflow. Anesth Analg. 2012;115:402–6
3. Agoliati A, Dexter F, Lok J, Masursky D, Sarwar MF, Stuart SB, Bayman EO, Epstein RH. Meta-analysis of average and variability of time to extubation comparing isoflurane with desflurane or isoflurane with sevoflurane. Anesth Analg. 2010;110:1433–9
4. Apfelbaum JL, Grasela TH, Hug CC Jr, McLeskey CH, Nahrwold ML, Roizen MF, Stanley TH, Thisted RA, Walawander CA, White PF. The initial clinical experience of 1819 physicians in maintaining anesthesia with propofol: characteristics associated with prolonged time to awakening. Anesth Analg. 1993;77:S10–4
5. Vitez TS, Macario A. Setting performance standards for an anesthesia department. J Clin Anesth. 1998;10:166–75
6. Dexter F, Epstein RH, Penning DH. Statistical analysis of postanesthesia care unit staffing at a surgical suite with frequent delays in admission from the operating room—a case study. Anesth Analg. 2001;92:947–9
7. Dexter F, Wachtel RE, Epstein RH. Impact of average patient acuity on staffing of the phase I PACU. J Perianesth Nurs. 2006;21:303–10
8. Epstein RH, Dexter F, Traub RD. Statistical power analysis to estimate how many months of data are required to identify post anesthesia care unit staffing to minimize delays in admission from operating rooms. J Perinesth Nurs. 2001;17:84–8
9. Dexter F, Epstein RH, Marcon E, de Matta R. Strategies to reduce delays in admission into a postanesthesia care unit from operating rooms. J Perianesth Nurs. 2005;20:92–102
10. Marcon E, Dexter F. An observational study of surgeons’ sequencing of cases and its impact on postanesthesia care unit and holding area staffing requirements at hospitals. Anesth Analg. 2007;105:119–26
11. Schoenmeyr T, Dunn PF, Gamarnik D, Levi R, Berger DL, Daily BJ, Levine WC, Sandberg WS. A model for understanding the impacts of demand and capacity on waiting time to enter a congested recovery room. Anesthesiology. 2009;110:1293–304
12. Epstein RH, Dexter F, Piotrowski E. Automated correction of room location errors in anesthesia information management systems. Anesth Analg. 2008;107:965–71
13. Epstein RH, Dexter F, Ehrenfeld JM, Sandberg WS. Implications of event entry latency on anesthesia information management decision support systems. Anesth Analg. 2009;108:941–7
14. Dexter F, Marcon E, Epstein RH, Ledolter J. Validation of statistical methods to compare cancellation rates on the day of surgery. Anesth Analg. 2005;101:465–73
15. Dexter F, Epstein RH, Marcon E, Ledolter J. Estimating the incidence of prolonged turnover times and delays by time of day. Anesthesiology. 2005;102:1242–8
16. Ledolter J, Dexter F, Epstein RH. Analysis of variance of communication latencies in anesthesia: comparing means of multiple log-normal distributions. Anesth Analg. 2011;113:888–96
17. Dexter F, Macario A, Qian F, Traub RD. Forecasting surgical groups’ total hours of elective cases for allocation of block time: application of time series analysis to operating room management. Anesthesiology. 1999;91:1501–8
18. Dexter F, Masursky D, Ledolter J, Wachtel RE, Smallman B. Monitoring changes in individual surgeon’s workloads using anesthesia data. Can J Anaesth. 2012;59:571–7
19. Dexter F, Marcon E, Aker J, Epstein RH. Numbers of simultaneous turnovers calculated from anesthesia or operating room information management system data. Anesth Analg. 2009;109:900–5
20. Epstein RH, Dexter F. Influence of supervision ratios by anesthesiologists on first-case starts and critical portions of anesthetics. Anesthesiology. 2012;116:683–91
21. Wang J, Dexter F, Yang K. Behavioral study of daily mean turnover times and first case of the day tardiness of starts. Anesth Analg. 2013;116:1333–41
22. Paoletti X, Marty J. Consequences of running more operating theatres than anaesthetists to staff them: a stochastic simulation study. Br J Anaesth. 2007;98:462–9
23. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7:177–88
24. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002;21:1539–58
25. Sidik K, Jonkman JN. A comparison of heterogeneity variance estimators in combining results of studies. Stat Med. 2007;26:1964–81
26. Jackson D, Bowden J, Baker R. How does the DerSimonian and Laird procedure for random effects meta-analysis compare with its more efficient but harder to compute counterparts? J Stat Plan Inference. 2010;140:961–70
27. Ledolter J, Dexter F. Analysis of interventions influencing or reducing patient waiting while stratifying by surgical procedure. Anesth Analg. 2011;112:950–7
28. Dexter F, Epstein RH. Typical savings from each minute reduction in tardy first case of the day starts. Anesth Analg. 2009;108:1262–7
29. Dexter F, Ledolter J. Bayesian prediction bounds and comparisons of operating room times even for procedures with few or no historic data. Anesthesiology. 2005;103:1259–167
30. Dexter F, Macario A, Ledolter J. Identification of systematic underestimation (bias) of case durations during case scheduling would not markedly reduce overutilized operating room time. J Clin Anesth. 2007;19:198–203
31. Dexter F, Epstein RH, Lee JD, Ledolter J. Automatic updating of times remaining in surgical cases using bayesian analysis of historical case duration data and “instant messaging” updates from anesthesia providers. Anesth Analg. 2009;108:929–40
32. Dexter F, Dexter EU, Ledolter J. Influence of procedure classification on process variability and parameter uncertainty of surgical case durations. Anesth Analg. 2010;110:1155–63
33. Dexter F, Ledolter J, Tiwari V, Epstein RH. Value of a scheduled duration quantified in terms of equivalent numbers of historical cases. Anesth Analg. 2013;117:205–10
34. Sulecki L, Dexter F, Zura A, Saager L, Epstein RH. Lack of value of scheduling processes to move cases from a heavily used main campus to other facilities within a health care system. Anesth Analg. 2012;115:395–401