Off-hours cases, performed during weekends or holidays, or started during the evening or late afternoon, have greater risk-adjusted mortality than those performed during regular hours (eg, Monday through Friday, from 7:00 am to 2:59 pm).1–3 Furthermore, each additional handoff during a case (eg, in the evening) may increase the risk of patient mortality.4,5 Thus, anesthesiologists providing care during off hours, including relieving cases in evenings, are more likely to care for patients at greater risk of sustaining a major adverse event than when the anesthesiologists work during regular hours.
Many hospitals use subspecialty teams during regular hours. By subspecialty teams, we refer to groups of anesthesiologists, surgeons, operating room (OR) nurses, and other providers who commonly work together to care for patients undergoing similar types of surgical procedures (eg, hip and knee arthroplasty). Given that the risk of mortality can be greater when a case is performed during off hours (ie, on weekends or nights),1–3 or includes one or more handoffs of care between successive anesthesia providers,4,5 our broad objective was to explore the logical inconsistency of using subspecialty teams during regular hours but not during weekends or evenings.
One reason that subspecialty teams might not be necessary during off hours is because the number of different types of procedures (diversity) performed during off hours is much less than during regular hours.6,7 Thus, effectively, anesthesiologists who provide care during off hours are participating in an acute care team, which is, in essence, a subspecialty team for a relatively limited number of procedures. To explore this possible reason for an absence of subspecialty call teams during off hours, we analyzed data8 from the Anesthesia Quality Institute National Anesthesia Clinical Outcomes Registry (NACOR). Among the hospitals in the United States (US), we estimated the average number of common types of anesthesia procedures (diversity), and the average difference in the number of common procedures between 2 off-hours periods (regular hours versus weekends and regular hours versus evenings). (For convenience, we use the word “procedure” to refer to any surgical or medical procedure during which anesthesiologists provide care.)
A second reason that subspecialty teams might not be necessary during off hours is because, at many facilities, procedures performed during off-hours are very similar to those performed during regular hours. For example, suppose that, at a given hospital, there are 20 procedures commonly performed (see Methods). In addition, suppose the relative distribution of these and all other procedures is very similar between regular hours and weekends (eg, similarity coefficient ≅ 0.90). At that hospital, the range of an anesthesiologist’s clinical activities during off hours would be no different from regular hours. To explore this possible reason, we also used NACOR data to estimate the average similarity in the distributions of procedures between regular hours and weekends, and between regular hours and evenings in US facilities.
Anesthesiologists specializing during regular hours can compensate, when working weekends and evenings, for relative unfamiliarity with procedures and the occurrence of uncommon events by asking colleagues for advice. Recently, for example, we compared NACOR data for case starting and ending dates and times with call logs of the Malignant Hyperthermia Association of the United States (MHAUS) Malignant Hyperthermia (MH) Hotline.9,a Calls received on weekends accounted for 15.3% ± 1.5% (SE) of all MH Hotline calls, which was significantly greater (P < .0001) than the 5.2% ± .1% of anesthetic minutes during weekends and the 4.3% ± .1% of general anesthetic minutes during weekends.8,9 Help may have been requested more commonly during off hours than regular hours because there were fewer anesthesia colleagues locally present and available for consultation. To explore this possible reason for our previous results,8,9 we investigated whether differences in the diversity of procedures between regular hours, evenings, and weekends, as would be seen nationally (ie, pooled distribution), would match the differences observed for individual facilities. If they were substantively different, then pooled national data (eg, listing what procedures occur commonly) would not provide relevant insight regarding anesthesia practice in individual facilities.
The University of Iowa institutional review board determined that the analysis of NACOR data did not meet the regulatory definition of human subjects research.
The data analyzed were all from US anesthesia groups that submitted cases to NACOR for all 12 months of 2013.8 The cases were transmitted monthly by participating anesthesia practices, as described previously (Figure 1).10 For each of the cases, data extracted were the date and time of the start of billed anesthesia care, duration of anesthesia care, whether or not general anesthesia was administered, facility, and primary surgical procedure based on the Current Procedural Terminology (CPT®) code. From the surgical procedure code, the anesthesia CPT code was determined using the American Society of Anesthesiologists 2014 Crosswalk® (ie, “procedure”). The case time data were predominantly anesthesia times from electronic billing records, values that are generally reliable because they are a substantial component of professional fees and thus are audited heavily by primary payers (eg, the Centers for Medicare and Medicaid Services).
Days with atypically low workloads (eg, holidays) were identified by using the counts of anesthetics nationwide on each Monday through Friday.11 All dates up to the fifth percentile of workdays were US federal holidays. Thus, dates with caseloads up to the fifth percentile were treated as holidays.11
The statistical measure of diversity we used was reviewed recently in an Anesthesia & Analgesia Statistical Grand Rounds.7 For each hospital and period (eg, weekends), we calculated the proportion of all anesthetic cases attributable to each procedure. The sum of the square of the proportions is the Herfindahl index. The inverse of the Herfindahl represents the effective number of common procedures.7,b For the example hospital shown in Table 1, the summary measure (inverse Herfindahl) was equal to the number of procedures that each accounted for ≤1% of anesthetics. Tables 1 and 2 highlight that “the effective number of common procedures” does not refer to the same set of procedures among facilities. For example, “anesthesia for electroconvulsive therapy” (CPT 00104) is a common procedure nationally and a common procedure at the hospital represented in Table 2, but not at the hospital represented in Table 1.
Estimating the total number of unique procedures at each facility would generally be of interest. However, because of the many rare procedures, it would be misleading (ie, biased) to take the observed number of unique procedures as an estimate of the diversity of a facility.7 The probability distribution of numbers of procedures follows a log-normal distribution.12 The consequence is that each increase in the number of procedures results in an increase in the inverse of the Herfindahl.7,13,14 Because the inverse of the Herfindahl (ie, the number of common procedures) is more robust than the observed number of procedures, we used it as our statistic.7
The similarity index was also reviewed in the recent Statistical Grand Rounds.7 The similarity index for a given hospital and between 2 periods (eg, regular hours [from 7:00 am to 2:59 pm] and evenings [from 3:00 pm to 10:59 pm]) is a pairwise assessment by procedure, like a correlation coefficient.6,7,15–17 To calculate this index, first, one selects a procedure at random from among all the procedures performed at the hospital during one period (eg, regular hours). Then one selects a procedure from another period (eg, weekends).7 The probability that the 2 selected procedures are the same equals the numerator of the similarity index. The denominator normalizes the range to be from .0 (when there is no overlap) to 1.0 (when the distribution of cases among the procedures is identical for the 2 periods).7 Relatively large values of the similarity index are ≥.80, indicating high levels of similarity.7,16 Values that are small are <.30, indicating low levels of similarity.7,16 Similarity index values between .30 and .80 are considered “moderate.”7,16
We report results as mean ± standard error of the mean among facilities, specifically the mean of differences in the number of common procedures, of numbers of common procedures, and of similarity indices. The standard error of the estimated number of common procedures at each facility was estimated using the first-order and second-order Taylor series expansions (ie, Delta method; see study by Dexter et al7 [equations 10-12 and footnote m]).7,18 The standard error of the similarity index at each facility was calculated using equation 20 in the study by Dexter et al7. The means and standard errors of the means among facilities were calculated using the Dersimonian and Laird random-effects model (ie, where the effects were estimated using weighted means like that of a meta-analysis but with facilities rather than studies). The corresponding P values for significant pairwise differences were 2 sided. All these calculations were performed using N = 399 facilities (ie, the facilities performing >1 procedure, and with ≥12 cases [average 1 per month] on evenings and weekends; Figure 1). The 2 percentages of the 399 facilities having substantive differences (≥5) in numbers of common procedures between periods were each compared with 50% (ie, “most”) using 1-sided exact binomial tests. The percentage of the 399 facilities having a large similarity index (≥.80)7,16 was also compared with 50%.
Finally, we also performed these calculations using pooled national data to approximate the perspective of a national report or call center (Figure 1).9 The times of all 2,064,984 cases were converted to Central Time based on the zip code of the location of the case and the date of the start of the case.8,c
Analysis by Facility
For most facilities, the number of common procedures (ie, inverse of the Herfindahl index) differed by <5 procedures: between regular hours and evenings (74.4% of facilities, P < .0001), and between regular hours and weekends (64.7% of facilities, P < .0001). The average number of common procedures was 13.59 ± .12 for regular hours, 13.12 ± .13 for evenings, and 9.43 ± .13 for weekends. The pairwise differences by facility were .13 ± .07 procedures between regular hours and evenings and 3.37 ± .12 procedures between regular hours and weekends (P = .090 and P < .0001 vs 0, respectively). Weekends generally had fewer numbers of common procedures, but not substantively less (Figure 2; median differenced 1.90 ± .32 procedures).
The distributions of common procedures were moderately similar between regular hours and evenings (average similarity index, .59 ± .01) and between regular hours and weekends (average similarity, .55 ± .02). Finally, most (> 1/2) facilities had small to moderate (<.8) similarities (62.9% and 73.4% for evenings and weekends, respectively; both P < .0001; Table 2).
Analysis Using Pooled National Data
A centralized (national) call center would see cases from the perspective of the time zone of the call center (see footnote c in Methods). Using all 654 facilities combined (Figure 1) and the Central Time zone, the differences in diversity of procedures (ie, numbers of common procedures) were: −8.06 ± .12 for regular hours versus evenings and 6.35 ± .13 for regular hours versus weekends.e (The negative sign means there were greater numbers of common procedures during evenings versus regular hours; the positive sign indicates that many procedures are performed infrequently on weekends.) These results differed from those of the preceding section. Using all facilities, each with its local time zone, the differences in numbers of common procedures still differed from the preceding section: −6.58 ± .11 for regular hours versus evenings and 6.44 ± .13 for regular hours versus weekends. Finally, limiting consideration to the 399 facilities with >1 case performed during evenings and weekends (Figure 1), the differences in the numbers of common procedures remained unlike that of the preceding section: −5.18 ± .12 for regular hours versus evenings and 7.59 ± .13 for regular hours versus weekends.
We examined why the preceding Results section showed that most individual facilities had comparable diversity between regular and off hours, but that was not so using nationally pooled data. Among all 654 facilities in the Central Time zone, the number of common procedures was 30.93 ± .05, 38.99 ± .11, and 24.58 ± .12 for regular hours, evenings, and weekends. The estimates were essentially the same using the local time zones: 31.20 ± .05, 37.77 ± .11, and 24.76 ± .12. The estimates also were essentially the same among the 399 facilities pooled: 32.23 ± .05, 37.41 ± .11, and 24.64 ± .12, respectively. However, these pooled estimates were >2x greater than when the number of common procedures was calculated for each facility and then averaged: 13.59 ± .12, 13.12 ± .13, and 9.43 ± .13. Equally weighting the facilities, only 3.51%, 1.50%, and 2.76% of the facilities had estimated common procedures at least as great as those nationally. Thus, individual facilities specialize, which results in national reports for common anesthesia procedures, the number of such procedures, and the distribution of those procedures being severely biased estimates for the vast majority of individual facilities in the US (Figure 2).
Our results show that, for most individual facilities, the numbers of different procedures commonly performed during regular hours and off hours (diversity) are essentially the same, but there is only moderate similarity in the procedures between these periods. Thus, anesthesiologists who work principally within a single specialty during regular work hours will likely not have substantial contemporary experience with many procedures performed during off hours. The importance of this finding, from an outcomes perspective, likely depends on the specific procedures of each facility (Tables 1 and 2). In addition, we found that the procedures commonly performed at most individual facilities are fewer in number than those that are commonly performed nationally (ie, because facilities specialize, the relative distribution of procedures nationally is not indicative of most anesthesia practices). Thus, analysis of anesthesia specialization patterns should be based on local practice patterns rather than upon pooled national data.
When our results are combined with previous findings of patient outcomes,1–5 there would appear to be a logical inconsistency for a facility to choose and use subspecialty anesthesia teams to improve patient outcomes during regular hours, but not to do so during off hours. Specifically, among hospitals in Australia, the risk-adjusted odds of dying within 48 hours after surgery was greater (1.19) on weekends than during regular hours.1 Among hospitals in England, the risk-adjusted odds of death within 30 days were greater (1.82) for surgery on a weekend rather than on a Monday.2 From the AQI data, nationwide in the US, surgery starting weekdays after 4:00 pm had greater adjusted odds (1.64) for mortality within 48 hours.3 In addition, each increase in the number of handoffs was associated with greater odds (1.44) of complications5 and major in-hospital mortality/morbidity (odds 1.08).4 However, a (very large) study at a single hospital found no difference in complication rates based on the timing of start of the case during evenings.19 Thus, we limited our study to results relevant to the apparent inconsistency between the greater patient risk off hours versus regular hours and subspecialty knowledge less readily available during off hours.
At an individual facility, it might be reasonable to not use subspecialty anesthesia teams when procedures performed during evenings and weekends represent a homogeneous (ie, nondiverse) set of procedures (ie, only a few different procedures take place during off hours). However, in our study, we showed that this is not so for most facilities. It might also be reasonable for an individual facility not to use subspecialty anesthesia teams when procedures performed during evenings and weekends are similar to those performed during regular hours (ie, the distribution of off-hours procedures matches that of regular hours). Again, our results provide a clear indication that this is not so for most facilities.
We emphasize that our results do not imply that every facility should necessarily change its off-hours call practices to provide subspecialty anesthesia teams during off hours. This is because our results reflect the average nationwide, and we have shown the practices of individual facilities can differ quite significantly from the aggregate. Rather, the strongest implication of our study is that each facility should examine the logic of its staff scheduling based on its local quantitative distribution of procedures between regular and off hours. For example, the hospital in Table 2 was chosen because of its low similarity of procedures between regular hours and weekends. The example shows procedures with relatively low absolute patient risk. A reasonable conclusion for this sample hospital would be that if there are anesthesiologists who infrequently work off hours (ie, when obstetrical procedures are relatively common), those anesthesiologists should not be those who limit their practice during regular hours to the other common class of procedures (ie, gastroenterology).
We also emphasize that our study should not be interpreted as suggesting either benefit or lack of benefit of the use of anesthesia teams. What we have addressed is solely the logical inconsistency of having anesthesia specialty teams only during regular hours. First, the issues related to specialized anesthesia care can apply to surgeons and or nursing. Second, use of specialized anesthesia teams can be difficult to implement at small facilities because, for staff scheduling to be by service, generally there need to be different numbers of specially trained anesthesiologists scheduled among days of the week.20 Third, although it is logical to assume that use of anesthesia subspecialty teams results in improved patient outcomes, we are unaware of any published peer-reviewed studies demonstrating such benefit other than for the traditional subspecialties (eg, pediatric anesthesiology and the care of infants). There is a spectrum of subspecialization between the infant undergoing complex congenital heart surgery and the otherwise healthy adult undergoing craniotomy for an acute subdural hematoma. Fourth, at the present, the limited knowledge of whether there is benefit to the use of anesthesia teams is that there is small operational performance, not patient outcomes.21 For details, see our recent Special Article.21
In conclusion, we suggest that facilities use subspecialty anesthesia teams (either formally or effectively) during regular hours and analyze their own data (eg, as in Tables 1 and 2), evaluate whether changes are needed in the use of such teams during off hours, and consider potential implications related to staff scheduling.
Name: Franklin Dexter, MD, PhD.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Conflicts of Interest: The Division of Management Consulting performs some of the analyses described in this study. Dr. Dexter receives no funds personally other than his salary and allowable expense reimbursements from the University of Iowa, and has tenure with no incentive program. He and his family have no financial holdings in any company related to his work, other than indirectly through mutual funds for retirement. Income from the Division’s consulting work is used to fund Division research.
Name: Richard H. Epstein, MD.
Contribution: This author helped conduct the study and write the manuscript.
Conflicts of Interest: None.
Name: Richard P. Dutton, MD, MBA.
Contribution: This author helped design the study.
Conflicts of Interest: None.
Name: Hubert Kordylewski, PhD.
Contribution: This author helped conduct the study.
Conflicts of Interest: None.
Name: Johannes Ledolter, PhD.
Contribution: This author helped analyze the data.
Conflicts of Interest: None.
Name: Henry Rosenberg, MD.
Contribution: This author helped design the study.
Conflicts of Interest: None.
Name: Bradley J. Hindman, MD.
Contribution: This author helped design the study and write the manuscript.
Conflicts of Interest: None.
This manuscript was handled by: Nancy Borkowski, DBA, CPA, FACHE, FHFMA.
1. Singla AA, Guy GS, Field JB, Ma N, Babidge WJ, Maddern GJNo weak days? Impact of day in the week on surgical mortality.ANZ J Surg2016861520
2. Aylin P, Alexandrescu R, Jen MH, Mayer EK, Bottle ADay of week of procedure and 30 day mortality for elective surgery: retrospective analysis of hospital episode statistics.BMJ2013346f2424
3. Whitlock EL, Feiner JR, Chen LLPerioperative mortality, 2010 to 2014: a retrospective cohort study using the national anesthesia clinical outcomes registry.Anesthesiology201512313121321
4. Saager L, Hesler BD, You J, et al.Intraoperative transitions of anesthesia care and postoperative adverse outcomes.Anesthesiology2014121695706
5. Hyder JA, Bohman JK, Kor DJ, et al.Anesthesia care transitions and risk of postoperative complications.Anesth Analg2016122134144
6. Dexter F, Wachtel RE, Yue JCUse of discharge abstract databases to differentiate among pediatric hospitals based on operative procedures: surgery in infants and young children in the state of Iowa.Anesthesiology200399480487
7. Dexter F, Ledolter J, Hindman BJQuantifying the diversity and similarity of surgical procedures among hospitals and anesthesia providers.Anesth Analg2016122251263
8. Dexter F, Dutton RP, Kordylewski H, Epstein RHAnesthesia workload nationally during regular workdays and weekends.Anesth Analg201512116001603
9. Dexter F, Rosenberg H, Epstein RH, Semo JJ, Litman RSImplications of national anesthesia workload on the staffing of a call center: the malignant hyperthermia consultant hotline.A A Case Rep201554346
10. Liau A, Havidich JE, Onega T, Dutton RPThe national anesthesia clinical outcomes registry.Anesth Analg201512116041610
11. Dexter F, Epstein RHHoliday and weekend operating room on-call staffing requirements.Anesth Analg200610314941498
12. Dexter F, Traub RD, Fleisher LA, Rock PWhat sample sizes are required for pooling surgical case durations among facilities to decrease the incidence of procedures with little historical data?Anesthesiology20029612301236
13. Gosselin FAn assessment of the dependence of evenness indices on species richness.J Theor Biol2006242591597
14. Jost LEntropy and diversity.Oikos2006113363374
15. Yue JC, Clayton MKA similarity measure based on species proportions.Commun Stat Theory Methods20053421232131
16. Wachtel RE, Dexter F, Barry B, Applegeet CUse of state discharge abstract data to identify hospitals performing similar types of operative procedures.Anesth Analg201011011461154
17. Wachtel RE, Dexter EU, Dexter FApplication of a similarity index to state discharge abstract data to identify opportunities for growth of surgical and anesthesia practices.Anesth Analg200710411571170
18. Simpson EHMeasurement of diversity.Nature1949163688
19. Sessler DI, Kurz A, Saager L, Dalton JEOperation timing and 30-day mortality after elective general surgery.Anesth Analg201111314231428
20. Dexter F, Wachtel RE, Epstein RH, Ledolter J, Todd MMAnalysis of operating room allocations to optimize scheduling of specialty rotations for anesthesia trainees.Anesth Analg2010111520524
21. Dexter F, Wachtel RE, Epstein RHDecreasing the hours that anesthesiologists and nurse anesthetists work late by making decisions to reduce the hours of over-utilized operating room time.Anesth Analg2016122831842