Efficient operating room (OR) management mandates the ability to accurately predict the duration of components of care for each surgical procedure (1–3). In many institutions, the duration listed for a given surgical procedure includes an approximation of the time required for, e.g., induction of anesthesia and application of monitors. This is usually supplied by the surgeon or, alternatively, a “fixed” time period is added by the scheduling office for these tasks. However, with greater emphasis on operational efficiency, there is a need to better predict the time required by the anesthesiologist before incision. No investigation has rigorously addressed the ability of the anesthesiologist to predict the actual time required for induction of anesthesia. Therefore, we undertook an independent observer-based study to determine if attending anesthesiologists can accurately predict the length of induction and what factors influence their ability to accurately predict this time.
After Human Investigation Committee review (exemption of informed consent was granted), 1558 patients were enrolled in the study. The methodological details have been reported elsewhere in this issue (4). Of these, 1265 patients met the inclusion criteria for the analysis presented in the study. Inclusion criteria required the presence of the attending anesthesiologist before induction of anesthesia, ASA physical status (ASA PS) I–IV, and elective surgery in which no artificial airway was present before induction. This case mix reflects all adult (>16 yr old) elective cases performed in the inpatient ORs and all surgical specialties are represented. Exclusion criteria included ASA PS V, the presence of an endotracheal tube or any other artificial airway, and emergency cases. Of the 1265 cases evaluated, 1176 were performed by residents and 89 by certified registered nurse anesthetists. All data were recorded on a standardized form by trained observers who were not involved in patient care. Data regarding the identification of anesthesiology attendings who took part in this study were not recorded.
Before induction, the attending anesthesiologist was asked by the independent observer to predict the time required for induction using a scale of 5-min intervals (5–60 min). We have defined induction time as anesthesia release time (ART, the patient on the table [time zero] until the release of the patient to the surgeons for preparation and positioning). After induction, the anesthesiologist was asked to evaluate the difficulty of the induction on a scale of 1–5 (1 = very easy; 5 = very difficult). The attending anesthesiologist had access to information that is routinely available before the start of all cases: the chart, preoperative evaluation, and the patient. Clinical care was not altered as a result of a patient being enrolled in this study. Thus, decisions regarding airway management, invasive monitoring, etc. were left to the discretion of the anesthesia care team and were the attending’s responsibility. The attending also knew, before estimation of ART, which individual resident or certified registered nurse anesthetist they would be supervising. For analysis, data were examined in the aggregate (n = 1265) and then divided into 3 groups. Group I consisted of those cases in which ART was correctly predicted (i.e., within 5 min of actual ART); in Group II, the ART was under-predicted (i.e., actual time took more than 5 min of predicted ART); in Group III the ART was over-predicted (i.e., the ART duration was at least 5 minutes less than predicted ART). Data are expressed as mean ± sd. Statistical analysis included the following tests: Spearman correlation coefficients, paired Student’s t-test, and analysis of variance (P < 0.05 was considered significant).
Overall for the entire group (n = 1265), anesthesiologists were highly accurate in predicting ART (r = 0.77; P ≤ 0.001). However, for an individual case significant degrees of under-prediction (24%, 300/1265) and over-prediction (24%, 298/1265) occurred and resulted in a correctly predicted rate of only 53% (667/1265) (Table 1). Under-prediction was significant by surgical service for cardiothoracic and transplant, and over-prediction was significant for gynecologic surgery and plastic surgery (P ≤ 0.001). Under-prediction was also associated with ASA PS IV, a regional anesthetic technique, age >65 yr, and the presence of invasive hemodynamic monitoring (arterial and/or central venous catheterization) (P = 0.006). For the group in which under-prediction was <10 min (n = 142), only 2.0% of anesthesiologists rated the difficulty = 5; for the group in which under-prediction was >10 min (n = 156) 12.4% of the attendings rated the difficulty = 5. Thus, anesthesiologists were more likely to under-predict with difficult inductions. As the difficulty of the induction increased (Group 1 = very easy to Group 5 = very difficult), a lower predictive value with more difficult inductions was observed.
In the aggregate, overall mean ART was correctly predicted by the attending anesthesiologist. However, this is the result of an equal offsetting percentage of over-prediction (24%) and under-prediction (24%). In contrast, analysis by individual cases shows a correct prediction rate of only 53%. Factors such as ASA PS, age, and surgical service were predictive of both under-prediction and over-prediction of ART.
When managing an OR, rational decision-making is only possible with accurate data. Accurate prediction of surgical and anesthesia controlled time could result in decreased costs or increased revenue for the hospital (5). In addition surgical data may be used by quality assurance, regulatory, and administrative personnel. However, we found a wide variance in the ability of the anesthesiologist to predict surgical times. Accurate allotment of scheduled time for OR cases demands an understanding of many complex factors (5–6). These factors include the complexity of the surgical procedure, the individual surgeon, and the anesthesia induction time. Other factors affecting variability were the patient’s age, type of anesthesia, and ASA PS. Wright et al. (1) found that in only 24% of the cases was the actual surgical time within ±15% of the scheduled time. Strum et al. (3) found that the most important variable in surgical times was the individual surgeon.
Limitations of this study include the fact that observations were performed in a teaching setting. This may have interfered with the attending anesthesiologists’ ability to accurately and consistently predict increased ART. We previously reported that the presence of teaching increased ART and time to incision (7). This study also included cases of longer duration. Thus, in an ambulatory setting the duration of ART may be reduced with decreased use of invasive monitors and other preparation required for longer and more complex surgical procedures. This may serve to increase the anesthesiologists’ predictive capability because more complex surgical procedures were associated with a decreased ability to predict ART.
We conclude that although, in aggregate, anesthesiologists can reasonably predict ART most of the time, it is very difficult to be accurate in individual cases. We have clearly shown that induction times in elderly, high-risk patients who require invasive monitoring are difficult to estimate and almost always take longer than expected. By considering these factors, the ability to predict ART for a given case could be significantly enhanced. This can lead to more accurate scheduling and more efficient OR time management.
The authors wish to acknowledge the continuing support of Mr. Norman Roth, Senior Vice President, Yale-New Haven Hospital. In addition, we wish to recognize the assistance of our Operating Room colleagues in the Departments of Anesthesiology, Nursing and Surgery.
1. Wright IH, Kooperberg C, Bonar BA, Bashein G. Statistical modeling to predict elective surgery time: comparison with a computer scheduling system and surgeon-provided estimates. Anesthesiology 1996;85:1235–45.
2. Dexter F, Macario A. Applications of information systems to operating room scheduling. Anesthesiology 1996;85:1232–4.
3. Strum DP, Sampson AR, May JH, Vargas LG. Surgeon and type of anesthesia predict variability in surgical procedure times. Anesthesiology 2000;92:1454–66.
4. Escobar A, Davis EA, Ehrenwerth J, et al. Task analysis of the pre-incision period: an independent-observer based study of 1558 cases. Anesth Analg 2006;103:922–7.
5. Dexter F, Traub RD, Macario A. How to release allocated operating room time to increase efficiency: predicting which surgical service will have the most underutilized operating room time. Anesth Analg 2003;96:507–12.
6. Dexter F, Macario A. Decrease in case duration required to complete an additional case during regularly scheduled hours in an operating room suite: a computer simulation study. Anesth Analg 1999;88:72–6.
7. Davis EA, Escobar A, Ehrenwerth J, et al. Resident teaching versus the operating room schedule: an independent observer-based study of 1558 cases. Anesth Analg 2006;103:932–7.