Blood Pressure Management by Anesthesia Professionals: Evaluating Clinician Skill From Electronic Medical Records

Assessment of anaesthesia trainees using performance metrics derived from electronic health records: a longitudinal cohort analysis

BACKGROUND

The development of competencies in anaesthesia residents is assessed by direct observation of their performance and written and oral examinations. Little is known about how residents' competencies are reflected by objective data in anaesthetic records. We investigated whether performance metrics derived from electronic anaesthesia records are associated with anaesthesia residents' progress of training and European written examination timing and results.

METHODS

We recruited 46 anaesthesia providers who trained at the Vienna University Hospital between September 2013 and June 2021, and we reviewed the anaesthesia records of all cases they managed during the study period. We derived six performance measures based on perioperative data routinely collected: intraoperative hypotension and hypothermia, glycaemic control, postoperative nausea and vomiting prevention, lung-protective ventilation, and postoperative kidney injury. We evaluated the association between these quality metrics with training level and written exam completion status in anaesthesia residents after adjusting for patient and surgical factors.

RESULTS

We found a statistically significant association between the level of training and most performance measures. The probability of preventing intraoperative hypotension increased (OR 1.16, 95% CI 1.12 - 1.20) with the level of training, as did the probability of preventing hypothermia (OR 1.08, 95% CI 1.05 - 1.11) and administering appropriate postoperative nausea and vomiting prophylaxis (OR 1.21, 95% CI 1.15 - 1.27). However, the odds of preventing acute kidney injury decreased (OR 0.91, 95% CI 0.83 - 0.99), as did the use of lung-protective ventilation (OR 0.94, 95% 0.92 - 0.97). All participating residents who took the written exam passed it, precluding a direct pass versus fail comparison, but the exam completion status was associated with increased odds of lung-protective ventilation (OR 1.42, 95% CI 1.03 - 1.95) and decreased odds of preventing intraoperative hypotension (OR 0.7, 95% CI 0.49 - 0.99). Glycaemic control was not associated with either of the training milestones.

CONCLUSIONS

The training level of anaesthesia residents had a significant association with several performance metrics. Passing the written exam only had a modest effect. Performance measures based on patient outcomes and intraoperative care may serve as assessment methods for anaesthesia residents' progress tracking.

Impact of Predictive Hemodynamic Monitoring on Intraoperative Hypotension and Postoperative Complications in Multi-level Spinal Fusion Surgery

STUDY DESIGN

Prospective longitudinal comparative cohort.

OBJECTIVES

To determine if the use of predictive hemodynamic monitoring (PHM) during elective multi-level posterior instrumented spine fusions decreases episodes of intraoperative hypotension (IOH) and complications.

BACKGROUND

A recent study showed an association between complications and duration of IOH in patients undergoing multi-level spine fusions. Whether the use of PHM to maintain hemodynamic stability intraoperatively decreases postoperative complications has not been evaluated.

METHODS

Adults undergoing elective multi-level posterior thoracolumbar fusion with arterial line blood pressure monitoring were identified and stratified into those in which predictive hemodynamic monitoring (PHM) was used and those in which it was not. Number of minutes of hypotension (MAP <65 mm Hg) and hypertension (MAP ≥100 mm Hg), volume of fluids, blood products and vasopressors administered intraoperatively and within the first 4 hours postoperatively as well as the number and type of postoperative complications were collected.

RESULTS

The 47 cases in the PHM group and 70 in the non-PHM group had similar demographic and operative characteristics. A shorter duration of IOH was seen in the PHM group (8.13 min) compared with the non-PHM group (13.28 min, P=0.029); and a shorter duration of intraoperative hypertension seen in the PHM group (0.46 min) compared with the non-PHM group (1.38 min, P=0.032). There was a smaller number of patients in the PHM group who had a surgical site infection (2.% vs. 13%, P=0.027), postoperative nausea and vomiting (0 vs. 14%, P=0.004) and postoperative cognitive dysfunction (6% vs. 19%, P=0.049) compared with the non-PHM group. There was also a statistically significant shorter length of hospitalization in the PHM (4.62 d) compared with the non-PHM group (5.99 d, P=0.017).

CONCLUSION

Predictive hemodynamic monitoring to manage intraoperative hemodynamic instability is associated with a shorter duration of intraoperative hypotension, a lower prevalence of complications, and a decreased hospital stay in multi-level spinal fusion surgery.

Intraoperative Hypotension Is an Important Modifiable Risk Factor for Major Complications in Spinal Fusion Surgery

STUDY DESIGN

Retrospective observational cohort.

OBJECTIVES

This study explores the impact of Intraoperative hypotension (IOH) on postoperative complications for major thoracolumbar spine fusion procedures.

SUMMARY OF BACKGROUND DATA

IOH with mean arterial pressure (MAP) <65 mm Hg is associated with postoperative acute kidney injury (AKI) in general surgery. In spinal deformity surgery, IOH is a contributing factor to MEP changes and spinal cord dysfunction with deformity correction.

METHODS

A total of 539 thoracolumbar fusion cases, more than six surgical levels and >3 hours duration, were identified. Anesthetic/surgical data included OR time, fluid volume, blood loss, blood product replacement and use of vasopressors. Arterial-line based MAP data was collected at 1-minute intervals. Cummulative duration of MAP <65 mm Hg was recorded. IOH within the first hour of surgery vs. the entire case was determined. Post-op course and complications including SSI, GI complications, pulmonary complications, MI, DVT, PE, AKI, and encephalopathy were noted. Cumulative complications were grouped as none, one to two complications, or more than three complications.

RESULTS

There was a significant association between occurrence of complications and duration of IOH within the first hour of surgery (8.2 vs . 5.6 min, P <0.001) and across the entire procedure (28.1 vs . 19.3 min, P =0.008). This association persisted for individual major complications including SSI, acute respiratory failure, PE, ileus requiring NGT, and postoperative cognitive dysfunction. Comparison of patients with zero versus one to two versus three or more complications demonstrated that patients with three or more complications had a longer duration of IOH in the first hour of the surgery and that patients who had no complications received less vasopressor than patients who had one to two or three or more complications.

CONCLUSION

This study identifies duration of IOH during the first hour of surgery as a previously unrecognized modifiable risk associated with major complications for multilevel lumbar fusion surgery.

LEVEL OF EVIDENCE

III.

Individualized Perioperative Hemodynamic Management Using Hypotension Prediction Index Software and the Dynamics of Troponin and NTproBNP Concentration Changes in Patients Undergoing Oncological Abdominal Surgery

INTRODUCTION

Abdominal oncologic surgeries pose significant risks due to the complexity of the surgery and patients' often weakened health, multiple comorbidities, and increased perioperative hazards. Hypotension is a major risk factor for perioperative cardiovascular complications, necessitating individualized management in modern anesthesiology.

AIM

This study aimed to determine the dynamics of changes in troponin and NTproBNP levels during the first two postoperative days in patients undergoing major cancer abdominal surgery with advanced hemodynamic monitoring including The AcumenTM Hypotension Prediction Index software (HPI) (Edwards Lifesciences, Irvine, CA, USA) and their association with the occurrence of postoperative cardiovascular complications.

METHODS

A prospective study was conducted, including 50 patients scheduled for abdominal cancer surgery who, due to the overall risk of perioperative complications (ASA class 3 or 4), were monitored using the HPI software. Hypotension was qualified as at least one ≥ 1 min episode of a MAP < 65 mm Hg. Preoperatively and 24 and 48 h after the procedure, the levels of NTproBNP and troponin were measured, and an ECG was performed.

RESULTS

We analyzed data from 46 patients and found that 82% experienced at least one episode of low blood pressure (MAP < 65 mmHg). However, the quality indices of hypotension were low, with a median time-weighted average MAP < 65 mmHg of 0.085 (0.03-0.19) mmHg and a median of 2 (2-1.17) minutes spent below MAP < 65 mmHg. Although the incidence of perioperative myocardial injury was 10%, there was no evidence to suggest a relationship with hypotension. Acute kidney injury was seen in 23.9% of patients, and it was significantly associated with a number of episodes of MAP < 50 mmHg. Levels of NTproBNP were significantly higher on the first postoperative day compared to preoperative values (285.8 [IQR: 679.8] vs. 183.9 [IQR: 428.1] pg/mL, p < 0.001). However, they decreased on the second day (276.65 [IQR: 609.4] pg/mL, p = 0.154). The dynamics of NTproBNP were similar for patients with and without heart failure, although those with heart failure had significantly higher preoperative concentrations (435.9 [IQR: 711.15] vs. 87 [IQR: 232.2] pg/mL, p < 0.001). Patients undergoing laparoscopic surgery showed a statistically significant increase in NTproBNP.

CONCLUSIONS

This study suggests that advanced HPI monitoring in abdominal cancer surgery effectively minimizes intraoperative hypotension with no significant NTproBNP or troponin perioperative dynamics, irrespective of preoperative heart failure.

American society of anesthesiologists physical status classification significantly affects the performances of machine learning models in intraoperative hypotension inference

STUDY OBJECTIVE

To explore how American Society of Anesthesiologists (ASA) physical status classification affects different machine learning models in hypotension prediction and whether the prediction uncertainty could be quantified.

DESIGN

Observational Studies SETTING: UofL health hospital PATIENTS: This study involved 562 hysterectomy surgeries performed on patients (≥ 18 years) between June 2020 and July 2021.

INTERVENTIONS

None MEASUREMENTS: Preoperative and intraoperative data is collected. Three parametric machine learning models, including Bayesian generalized linear model (BGLM), Bayesian neural network (BNN), a newly proposed BNN with multivariate mixed responses (BNNMR), and one nonparametric model, Gaussian Process (GP), were explored to predict patients' diastolic and systolic blood pressures (continuous responses) and patients' hypotensive event (binary response) for the next five minutes. Data was separated into American Society of Anesthesiologists (ASA) physical status class 1- 4 before being read in by four machine learning models. Statistical analysis and models' constructions are performed in Python. Sensitivity, specificity, and the confidence/credible intervals were used to evaluate the prediction performance of each model for each ASA physical status class.

MAIN RESULTS

ASA physical status classes require distinct models to accurately predict intraoperative blood pressures and hypotensive events. Overall, high sensitivity (above 0.85) and low uncertainty can be achieved by all models for ASA class 4 patients. In contrast, models trained without controlling ASA classes yielded lower sensitivity (below 0.5) and larger uncertainty. Particularly, in terms of predicting binary hypotensive event, for ASA physical status class 1, BNNMR yields the highest sensitivity of 1. For classes 2 and 3, BNN has the highest sensitivity of 0.429 and 0.415, respectively. For class 4, BNNMR and GP are tied with the highest sensitivity of 0.857. On the other hand, the sensitivity is just 0.031, 0.429, 0.165 and 0.305 for BNNMR, BNN, GBLM and GP models respectively, when training data is not divided by ASA physical status classes. In terms of predicting systolic blood pressure, the GP regression yields the lowest root mean squared errors (RMSE) of 2.072, 7.539, 9.214 and 0.295 for ASA physical status classes 1, 2, 3 and 4, respectively, but a RMSE of 126.894 if model is trained without controlling the ASA physical status class. The RMSEs for other models are far higher. RMSEs are 2.175, 13.861, 17.560 and 22.426 for classes 1, 2, 3 and 4 respectively for the BGLM. In terms of predicting diastolic blood pressure, the GP regression yields the lowest RMSEs of 2.152, 6.573, 5.371 and 0.831 for ASA physical status classes 1, 2, 3 and 4, respectively; RMSE of 8.084 if model is trained without controlling the ASA physical status class. The RMSEs for other models are far higher. Finally, in terms of the width of the 95% confidence interval of the mean prediction for systolic and diastolic blood pressures, GP regression gives narrower confidence interval with much smaller margin of error across all four ASA physical status classes.

CONCLUSIONS

Different ASA physical status classes present different data distributions, and thus calls for distinct machine learning models to improve prediction accuracy and reduce predictive uncertainty. Uncertainty quantification enabled by Bayesian inference provides valuable information for clinicians as an additional metric to evaluate performance of machine learning models for medical decision making.

Incidence, severity and detection of blood pressure and heart rate perturbations in postoperative ward patients after noncardiac surgery

STUDY OBJECTIVE

We sought to determine changes in continuous mean and systolic blood pressure and heart rate in a cohort of non-cardiac surgical patients recovering on the postoperative ward. Furthermore, we estimated the proportion of vital signs changes that would remain undetected with intermittent vital signs checks.

DESIGN

Retrospective cohort.

SETTING

Post-operative general ward.

PATIENTS

14,623 adults recovering from non-cardiac surgical procedures.

INTERVENTIONS & MEASUREMENTS

Using a wireless, noninvasive monitor, we recorded postoperative blood pressure and heart rate at 15-s intervals and encouraged nursing intervention as clinically indicated.

MAIN RESULTS

7% of our cohort of 14,623 patients spent >15 sustained minutes with a MAP <65 mmHg, and 23% had MAP <75 mmHg for 15 sustained minutes. Hypertension was more common, with 67% of patients spending at least 60 sustained minutes with MAP >110 mmHg. Systolic pressures <90 mmHg were present for 15 sustained minutes in about a fifth of all patients, and 40% of patients had pressures >160 mmHg sustained for 30 min. 40% of patients were tachycardic with heart rates >100 beats/min for at least continuous 15 min and 15% of patients were bradycardic at a threshold of <50 beats/min for 5 sustained minutes. Conventional vital sign assessments at 4-h intervals would have missed 54% of mean pressure episodes <65 mmHg sustained >15 min, 20% of episodes of mean pressures >130 mmHg sustained >30 min, 36% of episodes of heart rate > 120 beats/min sustained <10 min, and 68% of episodes of heart rate sustained <40 beats per minute for >3 min.

CONCLUSIONS

Substantial hemodynamic disturbances persisted despite implementing continuous portable ward monitoring coupled with nursing alarms and interventions. A significant proportion of these changes would have gone undetected using traditional intermittent monitoring. Better understanding of effective responses to alarms and appropriate interventions on hospital wards remains necessary.

Associations between intraoperative and post-anesthesia care unit hypotension and surgical ward hypotension

STUDY OBJECTIVE

To test whether patients who experience hypotension in the post-anesthesia care unit or during surgery are most likely to experience hypotension on surgical wards.

DESIGN

A prediction study using data from two randomized controlled trials.

SETTING

Operating room, post-anesthesia care unit, and surgical ward.

PATIENTS

550 adult patients having abdominal surgery with ASA physical status I-IV.

INTERVENTIONS

Blood pressure measurement per routine intraoperatively, and with continuous non-invasive monitoring postoperatively.

MEASUREMENTS

The primary predictors were minimum mean arterial pressure (<60, <65, <70 and < 80 mmHg) and minimum systolic blood pressure (<70, <75, <80, <85 mmHg) in the post-anesthesia care unit. The secondary predictors were intraoperative minimum blood pressures with the same thresholds as the primary ones. Our outcome was ward hypotension defined as mean pressure < 70 mmHg or systolic pressure < 85 mmHg. A threshold was considered clinically useful if both sensitivity and specificity exceeded 0.75.

MAIN RESULTS

Minimum mean and systolic pressures in the post-anesthesia care unit similarly predicted ward mean or systolic hypotension, with the areas under the curves near 0.74. The best performing threshold was mean pressure < 80 mmHg in the post-anesthesia care unit which had a sensitivity of 0.41 (95% confidence interval [CI], 0.35, 0.47) and specificity of 0.91 (95% CI, 0.87, 0.94) for ward mean pressure < 70 mmHg and a sensitivity of 0.44 (95% CI, 0.37, 0.51) and specificity of 0.88 (95% CI, 0.84, 0.91) for ward systolic pressure < 85 mmHg. The areas under the curves using intraoperative hypotension to predict ward hypotension were roughly similar at about 0.60, with correspondingly low sensitivity and specificity.

CONCLUSIONS

Intraoperative hypotension poorly predicted ward hypotension. Pressures in the post-anesthesia care unit were more predictive, but the combination of sensitivity and specificity remained poor. Unless far better predictors are identified, all surgical inpatients should be considered at risk for postoperative hypotension.