Original Investigation
July 20, 2022
Association of Anesthesiologist Staffing Ratio With Surgical Patient Morbidity and Mortality
Michael L. Burns, PhD, MD1; Leif Saager, Dr med, MMM2; Ruth B. Cassidy, MA1; et alGraciela Mentz, PhD1; George A. Mashour, MD, PhD1; Sachin Kheterpal, MD, MBA1
Author Affiliations Article Information
JAMA Surg. Published online July 20, 2022. doi:10.1001/jamasurg.2022.2804
editorial comment icon Editorial
Comment
author interview icon Interviews
Audio Author Interview (16:51)
Anesthesiologist Staffing Ratio and Surgical Patient Morbidity and Mortality
Backward 15 Play Forward 15
1x
0:00 / 0:00
Subscribe to Podcast
Key Points
Question Do overlapping anesthesiologist responsibilities increase the risks of morbidity and mortality for adults undergoing surgery with anesthesia?
Findings In this cohort study that used electronic health record registry data for 578 815 adult patients from 23 institutions, anesthesia care teams in which the anesthesiologist supervised 3 to 4 overlapping operations were associated with a 14% relative increase in composite risk-adjusted surgical patient morbidity and mortality, from 5.06% to 5.75%, compared with operations with an anesthesiologist covering between 1 and 2 overlapping operations.
Meaning This study’s findings suggest that increasing overlapping clinical responsibilities of a supervising anesthesiologist is associated with increased surgical patient morbidity and mortality.
Abstract
Importance Recent studies have investigated the effect of overlapping surgeon responsibilities or nurse to patient staffing ratios on patient outcomes, but the association of overlapping anesthesiologist responsibilities with patient outcomes remains unexplored to our knowledge.
Objective To examine the association between different levels of anesthesiologist staffing ratios and surgical patient morbidity and mortality.
Design, Setting, and Participants A retrospective, matched cohort study consisting of major noncardiac inpatient surgical procedures performed from January 1, 2010, to October 31, 2017, was conducted in 23 US academic and private hospitals. A total of 866 453 adult patients (aged ≥18 years) undergoing major inpatient surgery within the Multicenter Perioperative Outcomes Group electronic health record registry were included. Anesthesiologist sign-in and sign-out times were used to calculate a continuous time-weighted average staffing ratio variable for each operation. Propensity score–matching methods were applied to create balanced sample groups with respect to patient-, operative-, and hospital-level confounders and resulted in 4 groups based on anesthesiologist staffing ratio. Groups consisted of patients receiving care from an anesthesiologist covering 1 operation (group 1), more than 1 to no more than 2 overlapping operations (group 1-2), more than 2 to no more than 3 overlapping operations (group 2-3), and more than 3 to no more than 4 overlapping operations (group 3-4). Data analysis was performed from October 2019 to October 2021.
Exposure Undergoing a major inpatient surgical operation that involved an anesthesiologist providing care for up to 4 overlapping operations.
Main Outcomes and Measures The primary composite outcome was 30-day mortality and 6 major surgical morbidities (cardiac, respiratory, gastrointestinal, urinary, bleeding, and infectious complications) derived from International Classification of Diseases, Ninth Revision and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision discharge diagnosis codes.
Results In all, 578 815 adult patients (mean [SD] age, 55.7 [16.2] years; 55.1% female) were analyzed. After matching operations according to anesthesiologist staffing ratio, 48 555 patients were in group 1; 247 057, group 1-2; 216 193, group 2-3; and 67 010, group 3-4. Increasing anesthesiologist coverage responsibilities was associated with an increase in risk-adjusted surgical patient morbidity and mortality. Compared with patients in group 1-2, those in group 2-3 had a 4% relative increase in risk-adjusted mortality and morbidity (5.06% vs 5.25%; adjusted odds ratio [AOR], 1.04; 95% CI, 1.01-1.08; P = .02) and those in group 3-4 had a 14% increase in risk-adjusted mortality and morbidity (5.06% vs 5.75%; AOR, 1.15; 95% CI, 1.09-1.21; P < .001).
Conclusions and Relevance This study’s findings suggest that increasing overlapping coverage by anesthesiologists is associated with increased surgical patient morbidity and mortality. Therefore, the potential effects of staffing ratios in perioperative team models should be considered in clinical coverage efforts.
Introduction
Overlapping responsibilities are common in anesthesiology practice, with services typically provided through clinical care teams. In this model, in-room anesthesia clinicians (certified registered nurse anesthetists [CRNAs], anesthesia assistants, or anesthesiology residents) are supervised by the anesthesiologist, who oversees multiple operations simultaneously. The ratio of anesthesiologists to the overlapping number of rooms they cover is their staffing ratio. Anesthesiology care team compositions have been studied extensively, with results suggesting no significant difference in care quality based on team composition,1,2 yet improved patient outcomes have been reported with increased anesthesiologist involvement.3 The effects of overlapping anesthesiologist clinical responsibilities on patients remain unknown. Cost reduction efforts often target high fixed-cost anesthesiology services, assuming that increased clinical responsibilities are noninferior to lower patient to anesthesiologist staffing ratios.1 Understanding the potential association with the quality of patient care is necessary to inform clinical care staffing decisions.
Although fundamental differences exist between medical specialties, and practice patterns have changed over time, research suggests that increased clinical responsibility is associated with decreased clinical care quality and poor patient outcomes. Elevated hospitalist workloads have been reported to result in admission and discharge delays, decreased care quality, and poor patient satisfaction.4,5 Surgeons have received substantial scrutiny of overlapping operative practices, raising concerns for patient safety.6 Although several studies7-9 found no adverse outcomes associated with overlapping operations, others10-12 found higher mortality, longer surgical duration, and higher postoperative complication rates in select patient subgroups. In the intensive care unit, intensivist workload requires overlapping care of critically ill patients. Greater intensivist involvement is associated with improved clinical outcomes,13 although no optimal intensivist to patient ratio has been established.14 Although 1 study15 found no association with mortality, studies consistently find that increasing clinical overlap may lead to poor patient outcomes,16 and a multidisciplinary task force has found that ratios below a critical threshold had a negative effect on education, staff well-being, and patient care.17 This study’s objective was to evaluate the association between anesthesiologist staffing ratios and surgical patient morbidity and mortality by using data from a national electronic health record registry.
Methods
Data
Data were derived from the Multicenter Perioperative Outcomes Group (MPOG) database,18 an electronic health records registry of operations during which an anesthesiologist was involved, gathered from hospitals in 18 states.19 Numerous perioperative outcome studies have used MPOG data.10,20 The study protocol—including inclusion criteria, primary outcome, and statistical analyses plan—was presented, approved, and registered by the MPOG research committee before data were accessed. This multicenter cohort study also received approval from the University of Michigan Medical School Institutional Review Board, which also waived informed consent because the study was based on the secondary use of health care data that were already collected for clinical and operational purposes. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.21
Sample Selection
From the MPOG data set, all operations performed on patients aged 18 years or older between January 1, 2010, and October 31, 2017, were extracted. Data analysis was performed from October 2019 to October 2021. All operations were used to define the staffing ratio exposure variables. However, operations that commonly occurred with fixed staffing ratios were excluded from outcome analyses; the following types of surgical procedures were considered fixed staffing ratio operations: cardiac surgery, liver transplant, cataract removal, obstetric surgery (labor epidurals, cesarean delivery), and operations with resident involvement of more than 25% (typically fixed at a maximum staffing ratio of 1:2 because of reimbursement and educational limitations). Included operations consisted of those that involved anesthesiologist supervision of CRNAs and those with less than 25% resident involvement. Anesthesia care personally performed by an attending anesthesiologist, with no in-room CRNA or anesthesiology resident, is qualitatively different and was outside the scope of this analysis. No operations in which an anesthesia assistant was involved were included in this data set. Operations that occurred overnight or during weekends and holidays, which consisted of less than 4.3% of the original data set, were omitted from the primary analysis to increase the homogeneity of the analytical data set because the staffing model may have differed in these instances. Additional exclusions are listed in Figure 1. After exclusions were applied, a propensity score–matched cohort was created as described in the Statistical Analysis subsection.
Exposure
After sign-in and sign-out adjustments were made (eMethods 1 in the Supplement), overlapping assignments were quantified using attending anesthesiologist sign-in and sign-out times to calculate a single, time-weighted average staffing ratio for each operation. Specifically, operations were grouped by anesthesiologist, date, and time. Overlaps were identified using anesthesiologist transitions (sign-in and sign-out), and time-weighted averages of the individual staffing ratios were calculated as a continuous variable for each operation. For example, if the staffing ratios for a single 120-minute operation were 12 minutes at 1:2, 60 minutes at 1:3, and 48 minutes at 1:4, then the staffing ratio was calculated as [(2 × 12) + (3 × 60) + (4 × 48)]/120 = 3.3. For analysis, this primary exposure was divided into 4 distinct groups reflecting the discrete, modifiable staffing ratios: group 1 represents a staffing ratio of 1; group 1-2, a staffing ratio greater than 1 but no more than 2; group 2-3, a staffing ratio greater than 2 but no more than 3; group 3-4, a staffing ratio greater than 3 but no more than 4. We adjusted for additional variables, including Elixhauser comorbidities (a method of categorizing comorbidities of patients based on the International Classification of Diseases, Ninth Revision [ICD-9] and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision [ICD-10] diagnosis codes found in health care and administrative data22), teaching institution status, and operative year. Race and ethnicity were not included in the adjustment because they were not observed to be different among the staffing ratio groups.
Outcome
The primary outcome was a composite of 6 major morbidities and 30-day mortality. Morbidities included cardiac, respiratory, gastrointestinal, urinary, bleeding, and infectious complications according to ICD-9 groupings, based on the US Agency for Healthcare Research and Quality’s single-level Clinical Classifications Software categories for ICD-9 diagnosis codes and manually cross-referenced to ICD-10 codes within MPOG.23 eMethods 2, eTable 1, and eTable 2 in the Supplement provide more detail.
Statistical Analysis
Preliminary exploratory data analysis techniques, such as frequency distribution, mean, median, quintile-quintile plots, and box plots, were used to assess the distribution of primary and secondary outcomes and all other variables. Conflicting data (multiple timed event documentation, mortality designation despite subsequent clinical documentation) were addressed using standard MPOG registry processes. Missing patterns and rates (<10%) were assessed, and complete case analysis was used. Two-sided statistical significance testing with a P < .05 threshold was performed. All statistical analyses were performed using SAS software, version 9.4 (SAS Institute Inc).
Propensity score matching based on an allowable absolute difference between, or a radius around, the propensity score with a tolerance of 0.01 was applied to create a sample that was more balanced with respect to patient and operative characteristics, including potential confounders. This was completed by means of a generalized SAS macro, which uses an algorithm to maximize the number of propensity score matches.24 First, continuous staffing ratios were divided into 4 groups as described in the Exposure subsection (groups 1, 1-2, 2-3, and 3-4). Variables included in the propensity score derivation model and used to calculate the likelihood of being in a particular staffing ratio group were age, sex, body mass index, American Society of Anesthesiologists (ASA) physical status (where 1 represents a healthy patient, 2 represents a patient with mild systemic disease, 3 represents a patient with severe systemic disease, and 4 represents a patient with severe systemic disease posing a constant threat to life), emergency status, anesthesia technique (including general anesthesia, peripheral block, or neuraxial block), number of Elixhauser comorbidities, type of operation according to anesthesia Current Procedural Terminology (CPT) code, surgical service category, anesthesia duration, and institution.
Before matching, surgical service categories were stratified as follows: general, gynecologic, neurologic, otolaryngologic, orthopedic, urologic, and vascular services. Surgical category contributions by anesthesiology CPT code can be found in eMethods 3 and eTable 3 in the Supplement. With the staffing ratio group indicator as the dependent variable, multivariable logistic regression models were implemented in a sequential modeling approach, in which propensity scores were estimated for 2 staffing ratio levels at a time using group 1-2 as the reference group. Operations were matched based on estimates of the propensity scores for group 1 vs group 1-2, group 1-2 vs group 2-3, and group 1-2 vs group 3-4. Once these 3 sets of matched pairs were completed, duplicates were removed and operations were divided into strata based on their corresponding match in the reference group (group 1-2). In the resulting matched data set, each operation in group 1-2 (the reference group) was matched to no more than 1 operation from each of the other staffing ratio groups (groups 1, 2-3, and 3-4). Some reference group operations could not be matched because of nonoverlapping propensity score distributions. Additional details on the list of covariates and the modeling strategy used to develop the propensity score model can be found in eMethods 4 in the Supplement.
Differences in distributions of confounders and covariates across the staffing ratio groups were analyzed via pairwise absolute standardized differences (Table 1 and Table 2). The association between staffing ratio and the collapsed composite was assessed using a multivariable conditional logistic regression model (ie, any complication vs none of 6 major morbidities or 30-day mortality). Measures with absolute standardized differences greater than 0.2 were included in the multivariable outcome model, resulting in adjustment for teaching institution status, anesthesia duration, and operative year. Additionally, age, sex, body mass index, ASA status, emergency status, anesthesia technique, number of Elixhauser comorbidities, anesthesia CPT code operative type, and surgical service category were included in the model to achieve doubly robust effect estimates. With the use of this model, adjusted odds ratios (AORs) and associated 95% CIs of the composite morbidity and mortality outcome were estimated for each staffing ratio group, with group 1-2 as the reference group.
To assess whether any single component might be associated with the results, the distribution and contribution of the separate outcome components were examined by the staffing ratio group. In addition, distribution of the outcome (eTable 2 in the Supplement) and the continuous staffing ratio were examined by institution (eMethods 5 and the eFigure in the Supplement).
Results
Study Population
From an initial data set of 3 624 399 operations, the staffing ratio was calculated for each operation. After exclusions were applied, the data set consisted of 866 453 operations during which 1960 anesthesiologists provided care in 23 distinct institutions (Figure 1). Propensity score matching then revealed similar operations for comparison among staffing ratio groups, resulting in a matched data set with 578 815 operations.
In this final sample of matched operations performed on 578 815 adult patients, 259 925 patients (44.9%) were male and 318 890 (55.1%) were female, the mean (SD) age was 55.7 (16.2) years, the mean (SD) body mass index (calculated as weight in kilograms divided by height in meters squared) was 29.6 (7.3), and the median number of Elixhauser comorbidities was 1 (IQR, 0-2) (Table 1). In all, 308 028 operations (53.2%) had an ASA status of 1 or 2, 14 354 (2.5%) were emergency status, 431 229 (74.5%) used general anesthesia, 301 687 (52.1%) occurred at teaching hospitals, and the median anesthesia duration was 106 minutes (IQR, 64-176 minutes) (Table 2).
Operation counts for the matched groups were as follows: 48 555 in group 1, 247 057 in group 1-2, 216 193 in group 2-3, and 67 010 in group 3-4. When the distributions of patient demographic characteristics and operative factors were compared among staffing ratio categories, these groups appeared well matched. When an absolute standardized difference threshold of 0.2 was used, the only factors suggesting potential imbalance were the operations occurring at a teaching institution (0.26), anesthesia duration (0.34), and operative year (0.25) (Tables 1 and 2).
Primary Outcome
Overall, morbidity and mortality occurred after 30 026 operations (5.19%; 95% CI, 5.13%-5.24%). The distribution of this composite consisted of 30-day mortality (2607 operations [0.45%]) and the following morbidities: cardiac (5133 operations [0.89%]), respiratory (6645 [1.15%]), gastrointestinal (6694 [1.16%]), urinary (5093 [0.88%]), bleeding (4457 [0.77%]), and infectious (4963 [0.86%]) complications (eTable 2 in the Supplement). The unadjusted absolute event rates were 4.88% in staffing ratio group 1, 5.06% in group 1-2, 5.23% in group 2-3, and 5.74% in group 3-4.
The observed composite morbidity and mortality rate for the reference staffing ratio group (group 1-2) was 5.06%. This observed rate was referenced to calculate estimates from the adjusted outcome model. The adjusted estimate for group 2-3 was 5.25% (95% CI, 5.09%-5.42%). Similarly, the adjusted estimate for group 3-4 was 5.75% (95% CI, 5.47%-6.04%), a 14% relative increase compared with group 1-2, and for group 1 was 5.48% (95% CI, 5.14%-5.83%) (Figure 2). The distribution and contribution of the separate outcome components were examined by the staffing ratio group.
The adjusted odds of morbidity or mortality for all staffing ratio groups were statistically significantly different from that of the reference group of 1-2 (Table 3). The AOR for group 2-3 was 1.04 (95% CI, 1.01-1.08; P = .02) times that of the reference group (group 1-2); the AOR for group 3-4 was 1.15 (95% CI, 1.09-1.21; P < .001) times that of the reference group; and the AOR for group 3-4 was 1.10 (95% CI, 1.04-1.16; P = .001) times that of group 2-3.
Discussion
In this retrospective cohort study of 578 815 matched operations performed on adult patients in 23 US hospitals, increased overlapping anesthesiologist coverage beyond 1 to 2 operations was associated with an increased risk of surgical patient morbidity and 30-day mortality. Because 313 million surgical procedures are performed worldwide each year, any small individual improvements in outcome can have major repercussions for public health.25 These results complement previous studies that have shown improved 30-day mortality and morbidity rates after complications when anesthesiologists directed anesthesia care3 and are consistent with similar studies investigating medical staff workload effects. Studies in the nursing field26-30 have shown that institutions with higher patient to nurse ratios have higher rates of overall patient death, death after complications (failure to rescue), and other adverse events. Increased physician clinical workload has led to decreased quality of care and poor clinical outcomes, with specific examples found among internal medicine hospitalists4,5 and critical care intensivists.16 One simulation study31 showed that increasing operation overlap increased the risk of anesthesiologist supervision lapse. Furthermore, a review32 revealed that patient mortality increased during times of capacity strain in 18 of 30 inpatient care studies and in 9 of 12 studies in intensive care unit settings, suggesting that hospital capacity strain is associated with increased mortality and worsened health outcomes. Collectively, these results suggest that workloads have substantial consequences for patient care quality, clinical outcomes, and individual and organizational performance.33
Models of anesthesia care delivery include anesthesiologists working alone, anesthesiologists supervising physicians in training, anesthesiologists supervising nonphysicians in a care team, nonphysicians working independently, and surgical providers directly administering or overseeing anesthesia. There has been much focus on anesthesia team composition, specifically independent practice of CRNAs and practice group compositions.34 This study focused on physician-led anesthesia care teams and overlaps in patient care, which is a common concern among anesthesia care models and medical specialties with overlapping clinical responsibilities. By removing operations with high (>25%) resident involvement, this study primarily analyzed physician-CRNA teams, the dominant practice model in US anesthesiology. There is a paucity of research in this area of anesthesia practice. In a single-center study,35 investigators observed an increased rate of critical incidents (adverse events not resulting in adverse outcomes) associated with increasing anesthesiologist workload. However, that study was limited to staffing ratios between 1.6 and 2.2 and did not adjust for variations in operative factors and patient characteristics. Because major morbidity and mortality are not common events in a broad operative population, previous studies have lacked the power to appropriately match and analyze effects of overlapping patient care with the methodological rigor applied in the present study. In addition, a robust electronic health record registry, with detailed anesthesiologist sign-in and sign-out times, is required for accurate calculation of the staffing ratios.
Staffing ratio group 1 (1 anesthesiologist with an in-room provider in a 1:1 staffing ratio for the entire operation) demonstrated a statistically significant difference compared with group 1-2, the reference group, but not when compared with group 2-3 or group 3-4 (Table 3). If 1:1 staffing occurred only with higher-acuity operations, substantial variation from the reference group (group 1-2) might be expected; however, the rationale for staffing 1:1 could include both high-risk patients (requiring more attention) and low facility caseloads (ie, idle clinicians). These findings may not be relevant to typical anesthesia practice. Anesthesiology staffing ratio groups 2-3 and 3-4 both showed statistically significant increases in composite morbidity and mortality compared with group 1-2, which suggests that increasing overlapping anesthesiologist clinical responsibilities are associated with adverse surgical patient outcomes. Although major surgical complications are uncommon, any method of further decreasing and understanding this risk is important. When 100 000 operations, which is typical annually for a major medical center, are considered, the increase in risk from 5.06% to 5.75% that we observed would translate to an additional 690 operations with adverse outcomes. These results inform the safe practice of anesthesiology and surgical care models and may also extend to physician practices outside of anesthesiology, specifically areas with elements of overlapping clinical care. Overall, these findings suggest that increasing overlapping clinical care responsibilities may increase patient risk. The results add to previous findings investigating clinical workload and safe clinical practice in other fields, an important finding for improving provider practice and patient care.
Limitations
This study has several limitations, which may restrict the broad applicability of the results. First, the cohort was limited to 23 US centers, which is not representative of all clinical environments, with potential bias to site-specific tendencies. Second, 578 815 operations were used in the final matched analysis using inclusion criteria with a relatively limited operative set, and results may differ when operations outside of those included in this study are considered. Third, although propensity score–matching criteria were used, it is challenging to address unmeasured confounders, such as intraoperative patient acuity, proximity of operating rooms, surgeons’ and other nonanesthesiologist clinicians’ levels of overload and stress, and temporal fluctuations in staffing ratios such as those that may occur during anesthesia induction and tracheal extubation. Fourth, staffing ratios were limited to between 1:1 and 1:4. Operations with a staffing ratio greater than 1:4 are considered “medical supervision” rather than “medical direction” and are relatively uncommon in the US. Our data set contained few operations with time-weighted average staffing ratios greater than 4. Therefore, this study did not include analysis of physicians with overlapping responsibilities through medical supervision. When these findings are considered, it is important to judge increased overlap balanced against potential efficiency and access benefits to assess how much overlap may be appropriate. This study is limited to physician-led anesthesiologist care teams; previous studies have shown that anesthesiology care teams improve outcomes compared with anesthesiologists working alone.36 Fifth, this study did not investigate operations with anesthesiologists supervising physicians in training by limiting resident involvement to less than 25% in each operation. Sixth, routinely collected electronic health record data have known limitations, which were mitigated by the MPOG registry’s strict data quality controls. Seventh, this analysis is limited by the perioperative outcomes available within the data set, and several common postoperative surgical outcomes may have been omitted. Furthermore, we were unable to categorize ICD-9 and ICD-10 data into a severity system such as the Clavien-Dindo system37 because the basis of these classifications depends on corrective therapies in response to a specific complication, which were unavailable to us. Eighth, this analysis used a time-weighted average staffing ratio; other definitions of staffing ratio or variations at specific critical portions of the anesthesia process may be considered in future research to elucidate the relationship further.
Conclusions
In this cohort study, increasing overlapping anesthesiologist coverage was associated with increased surgical patient morbidity and mortality, despite treatment bias for healthier patients and lower-risk operations. These findings suggest potential consequences of overlapping anesthesiologist responsibilities in perioperative team models and should be considered in clinical coverage efforts.
pubmed.ncbi.nlm.nih.gov
