Cost containment and reduction are major goals in health care today. To decrease costs, hospital managers need to know the principal determinants of cost. These determinants are not necessarily obvious.
There is a widespread perception that anesthesiologists can decrease operating room (OR) costs by working more quickly. However, once a hospital staff has been paid for a day of work, virtually no money will be saved if they happen to finish early. Costs can greatly decrease if the same personnel can do more operations each day, without overtime. The goal of this study was to mathematically test whether anesthesiologists can significantly decrease OR costs by working more quickly (i.e., by letting more operations be done each day). We know of no previous study that has quantitatively examined this question.
We defined anesthesia-controlled time (ACT) as the sum of 1) the time starting when the patient enters an OR to when positioning or skin preparation can begin plus 2) the time starting when the surgical dressing is completed to when the patient leaves the OR. Using statistical analysis applied to a real patient database, we tested whether eliminating ACT would allow surgeons to do extra scheduled cases, or see more scheduled clinic patients, during an 8-h workday. We consider, in the discussion section, applications of our analysis to OR management and pharmacoeconomics.
Patient and Procedure Selection
Eleven surgical procedures were selected to represent a wide range of surgical specialties, case length and complexity, and postoperative lengths of hospital stay. These operations included neuroplasty of median nerve at carpal tunnel (carpal tunnel release), coronary artery bypass using venous grafts (CABG), gastroplasty for morbid obesity (gastroplasty), unilateral reducible nonrecurrent inguinal herniorrhaphy at 5 yr of age or older (herniorrhaphy), laparoscopic cholecystectomy, reconstruction of mandibular ramus with internal rigid fixation (mandibular osteotomy), inguinal approach orchiopexy (orchiopexy), reduction mammoplasty, unilateral acetabular and proximal femoral replacement (total hip replacement), tympanostomy with ventilating tube insertion under general anesthesia (tympanostomy), and vasovasostomy. Identification codes for these procedures were obtained from the American Medical Association Physician's Current Procedural Terminology (CPT) 1993 publication. The CPT codes were used to search the University of Iowa Hospitals' Information System database to identify patients.
Two exclusion criteria were applied to the identified patient records, to limit our database to relatively "routine" operations. The criteria were intended to increase the likelihood that our results would apply not only to other tertiary care centers, but to community hospitals as well. First, we included patients only if they 1) were admitted to the hospital 1 day before or on the day of surgery or 2) had ambulatory surgery. Second, we excluded patients if they underwent "unrelated tests or procedures" pre- or postoperatively. For the analysis, "unrelated tests or procedures" were those that generated separate CPT codes. For example, a patient would be excluded who underwent cardiac catheterization before total hip replacement, during the same hospital admission.
The most recent 70 consecutive patients operated on or before September 30, 1993 but after July 1, 1987, and who were not excluded, were selected for each procedure. These dates were chosen arbitrarily. For some operations fewer than 70 cases were found.
Time Collection and Determination
ACT, surgeon-controlled time (SCT), case time, and between-case time (BCT) were explicitly defined. At our institution, the patient enters and leaves the OR with the anesthesia provider. SCT was defined as the time starting from when patient positioning and/or skin preparation can begin to when surgical dressing is completed. BCT was defined as the time starting when the patient leaves the OR to when the next patient enters the OR, assuming the next patient is scheduled to immediately follow. Case time was defined as the sum of ACT, SCT, and BCT.
The times recorded for each patient that we used in this study were ACT (minutes) and SCT (minutes). These times are recorded by OR nurses for all surgical operations at our institution. Our study was conceived after the data reported here were collected. All physicians caring for the patients were compensated by salary alone, with no bonuses or incentives for clinical productivity or workload. We limited our study to ACT, BCT, and SCT, because our preliminary observational study (not reported) of recorded times showed that these three were the most accurate. We calculated means and standard deviations of ACT and SCT for each of the 11 operations.
Actual measurements of BCT between specific procedures of the same type were not obtainable from our database for all patients, for two reasons. First, only infrequently would one of these 11 procedures happen to follow an identical procedure. Second, if a particular case was the last case of the day in an OR, then it would not have a corresponding BCT. Therefore, we devised a method to estimate valid BCTs for the operations included in our study.
BCTs were available for all 2415 consecutive cases done in our ambulatory surgery center and all 4275 consecutive cases done in our main operating suite between 7:00 AM and 3:00 PM from March 1, 1993 to March 1, 1994. We chose these dates because they were a recent 1-yr interval for which data were available. From these BCTs, we constructed reasonable estimates of BCTs for the operations included in our study. Procedures done almost exclusively in the ambulatory surgery center were carpal tunnel release, herniorrhaphy, tympanostomy, and vasovasostomy. We assigned these operations the mean BCT for the entire ambulatory surgery center. Procedures performed exclusively in the main operating suite included CABG, gastroplasty, laparoscopic cholecystectomy, mandibular osteotomy, orchiopexy, reduction mammoplasty, and total hip replacement. We assigned these operations the mean BCT for the main operating suite.
Statistical Analysis of Times
We defined total time as the time theoretically required to sequentially complete a specified number of the same type of case in 1 day in one OR. We estimated the putative percentage decrease in ACT needed to reliably detect a 30-min decrease in total time (i.e., for 95% of the days). We chose 30 min, as this might be enough time to do an additional short, scheduled case.
We compared total times between two hypothetical ORs: one with the actual ACT for the specific operation under consideration, and the other with the putative decrease in ACT. Each of the two hypothetical ORs had the same specified numbers of one specific type of case in it each day. Therefore, we were able to compare total times between the two rooms by comparing mean case times between rooms. Thus, although the statistical analysis compared mean case times, the results were identical to a comparison of total times. We compared mean case times, in each of the two hypothetical ORs, by using two-group one-sided t-tests. We calculated the specific number of cases per day by taking the maximum integer of the following fraction. The numerator was the sum of mean ACT, mean SCT, and mean BCT. The denominator was 60 times 8 = 480 min per 8-h day.
t-tests require that acceptable Type 1 and Type 2 error rates be specified. By definition, hypothetical improvements will decrease ACT. Therefore, the Type 1 error rate can be 100% (i.e., alpha = 1.0). To pick a Type 2 error rate, we considered that there is a high cost associated with having to cancel scheduled cases. If total time with the additional case was longer than expected, overtime would have to be paid. Therefore, we wanted the chance of this Type 2 error to occur to be less than 5% (i.e., beta = 0.05). We thus calculated the percentage decreases in mean ACT needed to detect (using a two-group one-sided t-test, alpha = 1.0) a predictable (beta = 0.05), short (30-min), decrease in mean total time (number of cases per day times mean case time) occurring at the end of the day.
The actual standard deviations of SCT for each case type were included in the calculations. Variation in ACT was assumed to equal zero. Therefore, variations in total time used in the calculations were deliberately too small. Decreasing variation in total time had the effect of decreasing the number of minutes by which ACT would have to be decreased to predictably detect a 30-min decrease in total time.
Description of Patient Groups and Operations
(Table 1) gives information and sample sizes for each of the 11 surgical procedures included in the study. Similar data have been previously reported by other groups; the original aspect of our work is the analysis .
The statistical method used the measured standard deviation of SCT to give the decrease in individual case times needed to achieve a predictable 30-min decrease in total time. This putative decrease in case time was compared to the measured mean ACT to give the corresponding putative percentage decrease in ACT. Therefore, to interpret the statistical results, we were particularly interested in comparing mean ACT to the standard deviation of SCT. In our database, actual mean ACT was less than 1 SD of SCT for all operations other than tympanostomy Table 1.
Effect of Decreasing ACT on Personnel Time
We calculated deliberate underestimates of the percentage decreases in ACT needed to predictably (95% of days) finish the cases in an OR 30 min earlier each day Table 2. For tympanostomy, mean ACT exceeded 1 SD of SCT (9 min vs 8 min) Table 1. Decreasing ACT by 86% would allow one additional scheduled 30-min case be performed in an OR during a an 8-h workday. For all other operations studied, mean ACT was less than the standard deviation of SCT. Mean case time would have to be decreased by an amount greater than mean ACT to finish 30 min earlier on almost all days. Therefore, eliminating ACT still would not allow for one additional scheduled, 30-min case Table 2. Actual ORs usually accommodate a mixture of cases. An OR that contains both tympanostomies and other longer cases would require a decrease in ACT that exceeds the 86% required for an OR with only tympanostomies.
Our results depend, in part, on the number of cases performed per day. The more cases done in an OR each day, the greater the chance of detecting a decrease in total time for a given percentage decrease in ACT. At teaching hospitals fewer cases may be performed in an OR each day than in community hospitals. We were concerned that our results may be biased by this fact. For example, for several of these types of operations, we typically do three cases in each OR during each 8-h day Table 2. But anesthesiologists in California are reported to do four cases per day . Therefore, we repeated our analysis after arbitrarily doubling the number of cases that could be performed during an 8-h day, by using 16 h in the analysis. With this admittedly absurd scenario, for tympanostomy ACT now could be decreased by less than 60% to reliably finish earlier each day Table 3. For all other operations, ACT would still have had to be decreased by more than 100% to have additional scheduled time at the end of the day Table 3. Therefore, we doubt our results Table 2 have been substantively biased by our longer case times.
Application of Our Results to OR Management
Reasonably achievable decreases in ACT cannot decrease scheduled OR time sufficiently to permit even one extra 30-min scheduled case each day Table 1 and Table 2. These findings have several applications.
Example 1. A hospital has an OR suite that does not operate to capacity. Despite this, the surgeons want to have ACT decreased. They contend that this would permit them to see more clinic patients in the afternoons. Our results predict that decreases in ACT of any achievable magnitude would not help Table 2. If ACT were decreased, ORs would, on average, finish somewhat sooner. But, the decreases in total daily OR time would be so variable that scheduling additional patients would be unwise, if avoidance of overtime was also important. SCT and variation in SCT controls the OR schedule. ACT does not. If the surgeons were to schedule the time "gained," they would have to frequently cancel these cases or appointments, or incur overtime costs.
Example 2. Assume that nurses and other OR personnel rarely receive overtime. The only financially advantageous reason to decrease ACT would be to be able to schedule more cases. Our results predict that under no reasonable circumstance would decreasing ACT save money Table 2.
Example 3. OR personnel working in a large suite are often being paid overtime. Surgeons operate in select rooms in an OR suite during nonprime hours (e.g., from 3:00 PM to 7:00 AM). Add-on cases are only put into prime-time operating rooms if these cases are expected to be finished during peak time. The only potential advantage to decreasing ACT in the prime-time rooms would be to permit additional scheduled cases in those rooms. Practically achievable decreases in ACT will permit no additional scheduled cases in those rooms Table 2. Therefore, there is no financial advantage to decreasing ACT in these rooms.
Example 4. An OR suite is operating beyond scheduled capacity. Nurses and other OR personnel are paid overtime every day. Decreasing ACT might make financial sense. Extra cases cannot be reliably scheduled Table 2. However, overtime payments may be decreased, although unreliably.
Example 5. Nurses and other OR personnel work shifts. A manager contends that by decreasing ACT an OR could finish sooner, so fewer personnel could be scheduled to work during off-peak hours. Our results predict this plan will not work Table 2. Personnel who work shifts must be scheduled to work on given days. Therefore, for decreasing ACT to be helpful, it would have to ensure that the OR can be reliably scheduled to finish earlier. Decreasing ACT cannot do this, because the variability in SCT is large Table 1 and Table 2.
Example 6. An OR is available from 7:00 AM to 12:00 PM (i.e., for 5 h). A surgeon wants to schedule two 3.0-h cases. Anesthesiologists are asked whether decreasing ACT would reliably (>95% chance) get the two cases done by noon. Our analysis shows that this cannot be done Table 2. The effects of decreasing ACT are too small compared with the major determinants of the OR schedule: SCT and its variability.
Example 7. A manager suggests that a hospital fund a preoperative intravenous catheter team to decrease ACT and so permit more scheduled cases. At our institution, most intravenous catheters are inserted in the OR, and so become part of ACT. Lowering ACT will not permit additional scheduled cases Table 2. Therefore, the cost of a preoperative intravenous catheter team cannot be justified by an expected increase in scheduled cases.
Example 8. Another manager recommends that a hospital open a preoperative procedure room. At our institution regional blocks, arterial catheters, central venous catheters, and pulmonary artery catheters usually are placed in the OR, and so become part of ACT. Because these could be done in a preoperative procedure room, our results can be used to assess the financial impact. We found that achievable reductions in ACT cannot result in reliably scheduled extra cases Table 2. Therefore, even if a preoperative procedure room results in decreased ACT, there still cannot be more scheduled cases during a regular workday. If this facility required extra personnel for monitoring, then costs may even increase.
Example 9. An anesthesiology group is considering using some new medications with more rapid onset and shorter durations of action. Their motivation is the contention that using them will decrease ACT and thus reduce cost. Again, unfortunately, this alone will not predictably shorten the OR day Table 2. If the newer drugs also cost more, then the drugs will actually increase costs.
Example 10. Some radiologists complain that patients needing general anesthesia take a disproportionate amount of time because of inductions and awakening within their scanner or fluoroscopy suites. We found decreasing ACT would not permit more scheduled cases for almost all operations Table 2. This condition exists because mean ACT was small compared to mean SCT and, importantly, the standard deviation of SCT. Radiology procedures (e.g., magnetic resonance imaging, computerized tomography, radiation therapy) probably have less variability in operator-controlled time. Decreasing ACT may permit more scheduled cases. If the need for such additional cases can be consistently predicted, then productivity can be increased by decreasing ACT.
As explained in Results, we can predict our ACT results by comparing the standard deviation of SCT to mean ACT. We can apply the same analysis to BCT. Mean BCT was less than 1 SD of corresponding SCT for all operations other than tympanostomy and carpal tunnel release Table 1. Therefore, we would expect that achievable (<100%) decreases in BCT would permit one additional, short, scheduled case to be performed Table 2 only after these two operations. Tympanostomy and carpal tunnel release had case times less than 76 min. Therefore, our time data suggest that decreasing BCT might be helpful in an OR if it is operated to capacity with cases lasting 75 min or less. Otherwise, decreasing BCT probably would have little or no effect on the ability to reliably schedule additional cases.
Application of Our Results to Other Hospitals
A limitation of our study is that the data come from only one hospital. We believe that our methodology ensures that our results will apply generally, for the following reasons. First, we designed our statistics to underestimate the percentage decrease in ACT needed to do an extra, scheduled, short case. We did so largely by assuming that case time only varied because of variation in SCT. Second, we repeated our analysis after doubling the number of cases we could hope to do in an 8-h day Table 3. Increasing the number of cases had the effect of decreasing the percentage decrease in ACT needed to get extra scheduled time each day. However, the results did not change substantively. This lack of sensitivity of our results to large changes in ACT, SCT, and BCT ensures that random errors that may have been made in recording these times had no effect on our conclusions. Third, and most important, our hospital's practices are likely biased toward relatively large ACTs compared to practices at community hospitals. Nevertheless, we found that decreasing ACT would not allow extra cases and therefore would not reduce costs at our hospital Table 2. Decreasing the presumably shorter ACT in community hospitals would be even less likely to allow extra scheduled cases in an 8-h day.
Summary and Conclusions
We analyzed the impact of reducing ACT on the possibility of reliably scheduling additional cases (or clinic visits). Even an unachievable decrease in ACT to zero would not allow an OR to predictably finish earlier each day, unless all case times in that room were less than 45 min. Anesthesiologists alone cannot significantly generate more scheduled OR time by working more quickly. Methods to decrease ACT (e.g., using preoperative intravenous catheter teams, procedure rooms, and/or shorter acting drugs) may simply increase costs. Anesthesiologists, surgeons, and nurses must work collectively to achieve cost savings in the OR.