A retrospective analysis was conducted on US adult inpatient surgical procedures occurring between March 1, 2017, and December 31, 2021, within the PINC AI Healthcare Database (PHD) [28]. The PHD is a large, US hospital–based, service-level, all-payer database that contains information on inpatient discharges, primarily from geographically diverse nonprofit, nongovernmental, and community and teaching hospitals and health systems from rural and urban areas. Hospitals or health care systems submit administrative, health care use, and financial data from patient encounters. Inpatient admissions include over 108 million visits with more than 8 million per year since 2012, representing approximately 25% of annual US inpatient admissions.

This study used preexisting data with no identifiable information and therefore does not require institutional review board review per Federal Regulations for the Protection of Human Research Subjects (45CFR 46.102(e)) or patient consent [29]. All patient-related study data (eg, demographics, disease state, and information on billed services such as medications, laboratory tests, diagnostics, and therapeutic services) were accessed in compliance with the Health Insurance Portability and Accountability Act of 1996. This analysis was conducted and reported per the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.

US adults aged ≥18 years and who received rocuronium or vecuronium during an inpatient surgical procedure were included. Exclusion criteria were a diagnosis of myasthenia gravis or renal failure, receiving pyridostigmine therapy, NMB reversal with both sugammadex and neostigmine, pregnancy (proxied by women undergoing obstetrical procedures), or those diagnosed with COVID-19 (for encounters occurring in 2020 and 2021). For any hospitalized patient undergoing multiple surgeries during a calendar 30-day period or a given inpatient stay, only the first surgery was included in the analysis.

For each eligible patient, information on demographics, clinical characteristics (eg, age, gender, anthropometrics, and comorbidities), insurance status, admission status (eg, elective, emergency, or trauma), site characteristics (eg, hospital size and geographic region), and anatomic location of the surgery were collected. Additionally, the type of NMB agent administered (rocuronium or vecuronium) as well as the reversal strategy (eg, neostigmine, sugammadex, or no active pharmacologic reversal) were recorded. The use of rocuronium or vecuronium for NMB was the inclusion criteria for this study due to the aim of quantifying NMB reversal practice changes in the sugammadex-eligible population. Data were categorized by time period in the following manner: baseline period (BP, March 1, 2017, to February 29, 2019); before COVID-19 era (BC, March 1, 2019, to February 29, 2020); early COVID-19 era (EC, April 1, 2020, to July 31, 2020); middle COVID-19 era (MC, August 1, 2020, to December 31, 2020); and late COVID-19 era (LC, January 1, 2021, to December 31, 2021). For this study, the post–COVID-19 period encompasses EC, MC, and LC time periods. The month of March 2020 was omitted in these analyses to account for a transition period and due to the unavailability of COVID-19 diagnostic information. The EC period was predominated by the early part of the breakout, the lack of information beyond testing for COVID-19 and implementing strict measures to reduce viral exposure within the hospital setting; the MC period correlated with the initial availability of COVID-19 vaccination for health care workers, thus (theoretically) lessening the impact of the pandemic on health care decisions; the LC period reflects when vaccines were available to the general public and restrictive infection control measures were loosened.

NMB use was summarized by characteristics using descriptive statistics. Similarly, NMB reversal strategies (ie, sugammadex, neostigmine, or no reversal) were summarized by time period, patient, site, and procedural characteristics using descriptive statistics.

To identify factors related to NMB reversal choice during the COVID-19 and pre–COVID-19 eras, 2 multivariable logistic regression models were developed similarly to previous studies that modeled NMB reversal choices using PHD through June 2019 [8,9]. The first logistic regression models (model 1a and model 1b) aimed to test the effect of the COVID-19 time period on reversal choices by accounting for patient, hospital, and procedural characteristics. Encounters spanning both time periods (pre– and post–COVID-19) were included in these models to test for the overall effect of the COVID-19 era on reversal patterns, after accounting for all the covariates. Model 1a evaluated the effect of active (pharmacological) versus no (nonpharmacological) NMB reversal while model 1b evaluated sugammadex versus neostigmine reversal. Model 1 takes into account the trend in reversal over time by modeling the changes in NMB reversal as a linear trend over the period of January 1, 2018, to December 31, 2021. The effect of EC, MC, and LC eras are modeled as an interaction of the corresponding time period flag with the time-trend variable. For example, a positive interaction term coefficient between a post–COVID-19 period and time-trend variable in model 1b indicates a more rapidly increasing likelihood of sugammadex being used in the post–COVID-19 era as compared to the trend in the pre–COVID-19 time periods. The results from model 1 provide an overall estimate of the effect of the COVID-19 time period but do not provide insight into the effects for various population subtypes.

Model 2 was constructed and used in a counterfactual analysis to address model 1’s inability to evaluate changes in NMB reversal over time within patient subgroups. Models 2a (active vs no NMB reversal) and 2b (sugammadex vs neostigmine) were constructed using pre–COVID-19 data (January 1, 2018, to February 29, 2020) to be able to predict how a patient would have their NMB reversed (or not) based on their encounter characteristics. These models also include a continuous time variable that accounts for a linear trend to the log likelihood of the NMB reversal choice, to extrapolate this trend to the COVID-19 eras. Model accuracy, such as the receiver operating characteristic curve, is reported to help gauge the utility of these models for counterfactual analysis.

Counterfactual analysis was conducted to predict how COVID-19–era patients would have been reversed had they been treated during the pre–COVID-19 era based on their demographic, clinical, and institutional characteristics. The differences between the observed sugammadex reversal in the COVID-19 eras (actual) and the hypothetical or predicted reversal had each of those patients been seen pre–COVID-19 based on model 2 (counterfactual) were calculated. The differences in actual versus counterfactual reversal choices were then compared for each of the patient demographic, clinical, and institutional characteristics (eg, age group, comorbidities, surgery type, or hospital size). The counterfactual model was calibrated by adjusting the cutoff probability threshold to result in the same number of predicted classes (eg, sugammadex and neostigmine) as were actually observed in the combined COVID-19 eras. The odds ratios (OR) were also normalized such that the total patient-weighted OR was 1, which removed any residual time-dependent drift from the counterfactual model. The NMB reversal was compared between actual and counterfactual for each demographic, clinical, and institutional characteristic by obtaining ORs and CIs based on contingency tables for each covariate.

All statistical analyses were performed using SAS software (version 9.4; SAS Institute Inc).

Among the nearly 39.4 million inpatient encounters evaluated between March 1, 2017, and December 31, 2021, in the PHD, a total of 3,289,747 patients and 3,602,887 procedures involved the use of rocuronium or vecuronium and met all inclusion and exclusion criteria. The number of encounters included 1,644,370 during BP, 820,078 during BC, and 1,138,439 during the 3 post–COVID-19 periods (Table S1 in Multimedia Appendix 1). Patient demographics and characteristics were generally similar across the time periods despite attaining statistical significance driven by the large sample size (Table 1). Mean age (SD) ranged from 58.5 (16.76) years in the EC period to 59.0 (16.35) years during BC. A slightly higher percentage of patients were women (range 108,541/209,451, 51.8% in EC to 890,910/1,644,370, 54.2% in BP), and most patients identified as White (range 477,774/628,197, 76.1% in LC to 1,287,545/1,644,370, 78.3% in BP) throughout the study.

The percentage of patients with at least 1 comorbidity trended higher during this study’s period, increasing from 80.4% (1,321,911/1,644,370) during BP to 85.2% (535,076/628,197) by LC. The largest increases in comorbidity rates (>2% increase from BP to LC) were observed in cardiac arrhythmias, fluid or electrolyte disorders, and obesity or overweight conditions. The percentage of admissions due to elective procedures decreased between BC and EC (from 451,190/820,078, 55%, to 98,637/209,451, 47.1%) and there was a corresponding rise in the percentage of emergency or urgent admissions during these time periods (from 348,840/820,078, 42.5%, during BC to 104,123/209,451, 49.7% during EC).

Among the 3.6 million patient encounters included in this analysis, a majority involved teaching hospitals (range 885,068/1,644,370, 53.8%, in BP to 358,262/628,197, 57% in LC) and approximately 90% occurred in urban institutions (range 188,028/209,451, 89.8% in EC to 571,412/828,197, 91% in LC, Table S2 in Multimedia Appendix 1). The largest proportion of encounters (1,516,497/3,602,887, 42.1%) involved hospitals with 500 or more beds, while institutions with fewer than 200 beds accounted for approximately 15.5% (559,884/3,602,887) of encounters. Nearly half (1,736,173/3,602,887, 48.2%) of the encounters involved institutions in the South, 23% (828,275/3,602,887) from the Midwest, 14.6% (525,401/3,602,887) from the West, and the remaining 14.2% (513,038/3,602,887) from the Northeast.

During the total study period, the vast majority of encounters involved rocuronium use with or without succinylcholine (3,229,651 encounters, 89.6%; Table S3 in Multimedia Appendix 1). A general trend of increasing rates of rocuronium only (with or without succinylcholine) was observed during this study’s period, increasing from 87.1% during BP to 93% during LC. The use of succinylcholine with rocuronium or vecuronium was used in 5.3% of patient encounters overall. This rate increased from 4.8% during BP to a peak of 6.9% during the EC period, before falling to 5.4% during LC.

Before the COVID-19 outbreak, the use of sugammadex for NMB reversal steadily increased following its approval in 2016, with approximately 1 in 4 encounters using this agent for reversal during BP (Table 2; Figure 1) [9]. This trend continued through the post–COVID-19 eras, reaching 51.1% (321,268/628,197) of encounters during LC. Consequently, reversal with neostigmine decreased from 47.1% from BP to 26.6% during LC. Overall, the rate of active NMB reversal with either sugammadex or neostigmine gradually increased over time. Spontaneous reversal steadily decreased from 27.5% (451,838/1,644,370) of encounters during BP to 22.3% (139,854/628,197) during LC. The trends in sugammadex, neostigmine, and active reversal were approximately linear from 2018 until the end of this study’s period (R²>0.99, for sugammadex and neostigmine, R²=0.95 for active reversal).

When comparing patient characteristics by reversal type (ie, spontaneous, sugammadex, or neostigmine), the distribution by age, race, and ethnicity was similar, though statistical significance was achieved due to the large sample size (Tables S4-S6 in Multimedia Appendix 1). Encounters involving reversal with neostigmine or sugammadex tended to involve younger patients (mean 57.4, SD 17.15 to 58.2, SD 16.56 years for neostigmine and 58.7, SD 16.83 to 59.2, SD 16.43 years for sugammadex) compared to spontaneous reversal (mean 59.2, SD 16.28 to 59.9, SD 15.83 years). Women comprised a higher proportion of those reversed with sugammadex (49,534/92,709, 53.4%, to 231,852/417,266, 55.6%) or neostigmine (36,704/67,321, 54.5%, to 441,284/775,266, 56.9%) and a lower proportion of spontaneous reversal (22,303/49,421, 45.1%, to 217,774/451,838, 48.2%) compared to men. Those who underwent spontaneous reversal were more likely to have ≥1 comorbidity (379,136/451,838, 83.9% to 124,367/139,854, 88.9%) compared to those reversed with sugammadex (334,071/417,266, 80.1%, to 273,297/321,268, 85.1%) or neostigmine (608,704/775,266, 78.5%, to 137,412/167,075, 82.2%).

The use of NMB reversal agents was similar based on institution type. During BP, sugammadex was used in 24.8% (219,463/885,068) of encounters in teaching hospitals and 26.1% (197,803/759,302) in nonteaching hospitals. The use of sugammadex increased to 50.2% (179,721/358,262) among teaching hospitals and 52.4% (141,547/269,935) in nonteaching hospitals during the LC era.

Multivariable regression analyses were used to identify time trends and their interaction terms with post–COVID-19 time periods for pharmacologic (active) versus no pharmacologic (spontaneous) reversal (model 1a), and for reversal with sugammadex versus neostigmine (model 1b; Table 3; Figures 2 and 3). The overall yearly time-trend throughout this study’s period demonstrated an increase in the use of active reversal (using either sugammadex or neostigmine) compared to no pharmacologic reversal (OR 1.129; P<.001). However, there were variations in the interaction term coefficient when analyzing specific post–COVID-19 time periods (refer to the Multivariable Analysis section of the Methods section for details on the analysis approach). Active reversal did not show a significant association, or change in trend, during EC (OR 1.002, 95% CI 0.997-1.008; P=.44) or MC (OR 0.996, 95% CI 0.991-1.001; P=.12). A slight but statistically significant decrease in the active reversal trend (ie, there was a slowing of the trend toward increased use of active reversal) was observed in LC (OR 0.987, 95% CI 0.983-0.992; P<.001). Based on these observations, the counterfactual analysis (model 2a) was not evaluated.

Significant associations in the use of active reversal were also observed based on patient, procedure, and institutional factors (Table 3; Figure 2). Except for those aged 18-30 years, fewer older adults (aged 41 to 70 years) tended to show a decreased odds of active reversal (OR range 0.941-0.983; reference those aged 31-40 years), while older adults (aged >70 years) were more likely to use active reversal (OR range 1.094-1.349). Compared to elective surgical procedures (reference), emergency, trauma, and urgent admissions revealed significantly decreased use of active pharmacologic reversal (OR 0.641, 95% CI 0.637-0.645; 0.612, 95% CI 0.602-0.622; and 0.668, 95% CI 0.662-0.675; respectively; P<.001 for each).

When comparing the use of sugammadex versus neostigmine in model 1b (Table 3, Figure 3), the yearly time-trend from January 2018 onwards demonstrated a steady increase in the use of sugammadex over neostigmine (OR 1.388, 95% CI 1.381-1.396; P<.001). Analysis of the specific post–COVID-19 time periods revealed a small but statistically significant interaction with the time trend in NMB reversal (Table 3). A slight but statistically significant increase in sugammadex reversal was observed during EC (OR 1.008, 95% CI 1.003-1.014; P=.003), followed by negation of that trend during MC (OR 0.992, 95% CI 0.987-0.997; P<.001). There was no significant interaction identified in the LC period (OR 1.001, 95% CI 0.996-1.005; P=.81).

Other covariates in model 1b that were significantly associated with sugammadex reversal included older age categories, urgent or emergent and trauma admissions, cardiac comorbidities (including arrhythmias, peripheral vascular disorders, congestive heart failure), obesity, chronic pulmonary disease, and diabetes. Most surgical types were associated with higher rates of sugammadex reversal as compared to the reference (musculoskeletal surgeries or surgeries involving the nervous system) with the exception of female genitalia. Hospitals with fewer beds (0-199 or 200-399 vs 400+) were associated with a lower likelihood of sugammadex reversal. Hospitals in the Northeast and South geographic regions of the United States also had significantly lower odds of sugammadex reversal as compared to the West and Midwest.

When comparing EC, MC, and LC time periods within patient subgroups, only a few differences were observed in actual NMB reversal compared to expected use. Most differences were observed in LC among institution and multimodal NMB characteristics (Figure 4). After normalization, only a few patient and institutional characteristics showed a significant deviation from the expected trend of the sugammadex reversal rate. Of the patient characteristics that had an observable counterfactual difference, patients with Hispanic ethnicity were reversed with sugammadex less frequently in the LC era as compared to how they would have been reversed before COVID-19. Institutions with 400 or more beds or classified as teaching institutions also had a relative decrease in sugammadex reversal rates as compared to expected trends from before COVID-19 data. On the other hand, small hospitals (0-199 beds), mid-sized hospitals (200-399), and those located in the south of the United States had higher rates of sugammadex reversal than expected from pre–COVID-19 trends, having relative actual or counterfactual ratios greater than 1.

This study used a large health care database comprising 931 sites across the United States to identify changes in use trends for NMB and NMB reversal agents for inpatients before and after health care systems experienced the COVID-19 outbreak. Through multivariable regression analysis, we identified factors related to the patient, procedure, and institution that were associated with NMB reversal choices.

We originally hypothesized that the use of sugammadex for NMB reversal in the post–COVID-19 period would increase given the evidence demonstrating decreased PACU time and, thus, diminished potential exposure to COVID-19. When analyzing changes in the sugammadex use trend for NMB reversal, a slight, transient effect was observed during the post–COVID-19 time points. During EC, a small but statistically significant increase in sugammadex use (compared to neostigmine) was revealed, though this trend was negated by an equivalent decrease during the MC period. However, the association with sugammadex use was small and short-lived, thus arguing that other factors in the complex process of NMB and NMB reversal selection are influencing these decisions. Logistic regression analysis showed that sugammadex use was favored in emergency and urgent admissions compared to elective admissions, and the number of emergency and urgent admissions increased from 42.5% (348,840/820,078) in the BC period to 49.7% (104,123/209,451) in the EC period (Table 1). However, this did not translate to an increase in sugammadex reversal in the counterfactual analysis (model 2b, Figure 4), which largely showed no significant difference in the sugammadex reversal rates of patients being treated in the COVID-19 eras relative to how they would have been treated before COVID-19. It is also important to point out that this study is attempting to identify an association or change in trend, beyond the currently established time-trend, which has observed a strong, steady increase in sugammadex before COVID-19, likely due to increasing evidence of the benefit of sugammadex in avoiding rNMB and in quicker time to reach a train-of-four (TOF) ratio of >0.9 [17,30]. To help account for this, the analysis used data starting in January 2018, which showed a more linear and predictable increase in sugammadex use leading to the start of the COVID-19 pandemic in March 2020. Given the strong association of sugammadex in the yearly time-trend (OR 1.388, 95% CI 1.381-1.396), it is possible that small but significant changes to the NMB reversal trend following COVID-19 are being masked by the existing time-trend.

The counterfactual analysis in this study was intended to identify trends in patient and institutional characteristics that deviated from overall sugammadex reversal patterns captured in the multivariate analysis, which assumes constant effects for all of the covariates. Most of the changes that deviated from the prepandemic trend, as observed from the significantly higher or lower actual or counterfactual ratio in Figure 4, were a reversal of the characteristics that were found to be associated with the early adoption of sugammadex [9]. For example, a lower-than-predicted use of sugammadex in trauma center patients and large hospitals in the peri–COVID-19 time periods may be explained as a renormalization caused by a higher-than-expected adoption of sugammadex in these settings in the initial years after sugammadex approval.

Though this study used robust methodology and a large US database of over 3.5 million inpatient encounters, several limitations must be addressed. The PHD includes patients covered by all payer types from both teaching and nonteaching institutions of various bed sizes. However, it is not representative of geographic location, with the South region more heavily weighted, which could limit the generalizability of the results. The PHD did not include information on the depth of NMB block (moderate vs deep), use of quantitative neuromuscular monitoring, or detection of postoperative rNMB, which can impact NMB reversal selection. The PHD also did not provide individual hospital data on census or capacity limitations. Methods were proposed to account for controlling for the variability in experience between regions, sites, and time periods relative to this capacity. However, there was still a possibility that the impact of a burden on each hospital was either captured incorrectly in general or relative to the BC experience, though the burden was likely to be just as well or just as poorly, captured from one site to another. This study does not take into account reporting sites that were not continuously enrolled throughout this study’s period. During this study’s period extending over 4.5 years, practice and policy changes in surgery and anesthesiology were likely to occur that could have influenced the selection of NMB and NMB reversal agents. Additionally, variations in hospital protocols as well as access to reversal agents were not accounted for in this study. These could include external factors on anesthesia practice among institutions, such as adherence to enhanced recovery protocols, quantitative neuromuscular monitoring following surgery, budgetary constraints, and quality initiatives. Certain patient characteristics (eg, American Society of Anesthesiology physical status classification or smoking status) and procedure data (eg, drug dose or duration of procedure) were not available that could impact NMB reversal choice. By the nature of this study and data collection, there is a potential for recording errors. Lastly, the determination of early, mid, and late COVID-19 time periods was largely subjective. In the absence of nationwide, standardized protocols to guide surgical and anesthesia practices in the wake of the COVID-19 outbreak, each institution adapted independently to the pandemic based on available resources and local impacts of the pandemic, which can vary widely over time and location. Despite the limitations of the data source and our limited ability to identify certain details, our study provides an aggregate observation on the effect of the COVID-19 pandemic on NMB reversal practices in non–COVID-19 patients in the United States.

Before COVID-19, rocuronium (with or without succinylcholine) was the predominant NMB used among US inpatient procedures, accounting for approximately 87% (1,432,947/1,644,370) of patient encounters during the BP. Preference for rocuronium over vecuronium continued through the post–COVID-19 time periods, with 93% (583,815/628,197) of encounters using rocuronium (with or without succinylcholine) in LC. These findings were consistent with prior studies on NMB use among US inpatients. Bash et al [9] demonstrated a trend in preference for rocuronium over vecuronium (with or without succinylcholine) from 2014 to 2019 among US inpatients, increasing from 84.3% in 2014 to 90.7% by the first half of 2019. This trend was even more pronounced among US outpatients, with rocuronium (with or without succinylcholine) accounting for over 96% of NMB use during the first half of 2019 [8].

Database studies revealed trends in NMB reversal favoring active over spontaneous (or no pharmacologic agent) reversal. Among US inpatients, the percentage of encounters using spontaneous reversal gradually decreased from 36.5% in 2014 to 34.3% in 2016 [9]. This decreasing trend accelerated in 2016 (with the approval of sugammadex) and reached 27.6% by the first half of 2019. This current study demonstrated that a decreasing trend in the use of spontaneous reversal continued through the COVID-19 time periods. Logistic regression estimates did not reveal any significant association, or change in the trend, between active versus spontaneous reversal during the EC and MC time periods. During LC, a small but significant association was observed showing a decrease in the rate of active reversal change (effective change in OR of time trend from 1.129 to 1.115). Analyses revealed several patient, procedural, and institutional factors with significant associations with the choice of reversal approach. The most pronounced association identified was related to admission type, with emergency, trauma center, and urgent admissions strongly favoring the use of spontaneous reversal compared to elective procedures. The percentage of elective admissions decreased substantially from 55% (451,190/820,078) in BC to 47.1% (98,637/209,451) in EC, and only partially returned during the MC and LC periods. This was likely the result of a nationwide response to postpone nonessential, elective surgeries as a means to limit COVID-19 exposure in hospitals and focus health care resources on managing the pandemic [14,15].

Contrary to our hypothesis, this study suggests that the impact of COVID-19 on NMB reversal selection was generally limited during the post–COVID-19 era throughout the United States. Though the change in trend for sugammadex use was small and transient in the post–COVID-19 era, the steady trend of increasing sugammadex use over neostigmine that started before COVID-19 and continued in the post–COVID-19 era eclipsed the small transient effects of the pandemic. This trend may be attributed to evidence demonstrating certain benefits of sugammadex over neostigmine, including diminished reversal time, more rapid discharge from the PACU, and a lower incidence of rNMB [17,30-32]. However, the lack of an acceleration of sugammadex use during the post–COVID-19 periods may be attributed to several factors, including clinical inertia or a lack of evidence related to the potential reduction in viral exposure associated with quicker NMB reversal time (and earlier extubation in the operating room). Educational programs can help to ensure current standards of care are attained and maintained in the postoperative setting.

Nonetheless, neostigmine remains a reasonable alternative for NMB reversal in certain patients with minimal blockade. Recent guidelines from the American Society of Anesthesiologists and the European Society of Anesthesiology and Intensive Care (both guidelines released after the date of final data collection of this study) confirm neostigmine’s place in therapy and offer recommendations on the appropriate use of this agent in NMB reversal when accompanied with quantitative neuromuscular monitoring [33,34]. American Society of Anesthesiologists recommends quantitative neuromuscular monitoring for all patients and prefers sugammadex over neostigmine at deep, moderate, and shallow depths (ie, TOF ratio <0.4) of NMB induced by rocuronium or vecuronium [33]. Yet, neostigmine is indicated as a reasonable alternative for patients with minimal blockade (ie, TOF ratio = 0.4 to <0.9) [31]. Similarly, the European Society of Anesthesiology and Intensive Care recommends sugammadex for deep, moderate, and shallow NMB (TOF ratio <0.4) induced by rocuronium or vecuronium, while neostigmine can be considered following advanced spontaneous recovery (ie, TOF ratio >0.2) [34]. Future research using databases that collect TOF information would be instrumental in understanding the impact of these guidelines on current trends and outcomes in NMB reversal selection.

This large, retrospective database study analyzed over 3.5 million inpatient encounters throughout the United States to identify changes in NMB use and reversal trends during the COVID-19 pandemic. We hypothesized that sugammadex use for NMB reversal would accelerate during the post–COVID-19 eras as a means to reduce PACU or operating room time and, subsequently, the risk of COVID-19 exposure. However, our findings demonstrated only a slight, transient acceleration of sugammadex use during the EC that was largely negated with time. This study did not attempt to investigate the reasons for the lack of change in the existing trend in the use of NMB reversal agents. Additional research is needed to better understand how pandemic-related practice changes have affected long-term NMB and reversal selection based on patient, procedural, or institutional factors, and potentially recognize patient subpopulations that experienced greater changes in anesthesia practice during this period.