Perioperative medicine training resources: U2M1: Minimising common adverse events and impact on the perioperative period

Abstract
Purpose of review: The central question of preoperative assessment is not "What can be done?" but "What should be done and how?" Predicting a patient's risk of unwanted outcomes is vital to answering this question. This review discusses risk prediction tools currently available and anticipates future developments.

Recent findings: Simple, parsimonious risk scales and scores are being replaced by complex risk prediction models as high-capacity information systems become ubiquitous. The accuracy of risk estimation will be further increased by improved assessment of physical fitness, frailty, and incorporation of existing and novel biomarkers. However, the limitations of risk prediction for individual patient care must be recognized.

Summary: Risk prediction is transforming from clinical estimation to statistical science. Predictions should be used within the context of a patient's baseline risk (life expectancy independent of surgery), personal circumstances, quality of life, their expectations and values, and consideration of outcomes that are meaningful for the patient.

Abstract
Risk prediction models are frequently used to identify high-risk patients undergoing emergency laparotomy. The National Emergency Laparotomy Audit (NELA) developed a risk prediction model specifically for emergency laparotomy patients, which was recently updated. In this study, we validated the updated NELA model in an external population. Furthermore, we compared it with three other risk prediction models: the original NELA model, the Portsmouth Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity (P-POSSUM) model, and the American Society of Anesthesiologists Physical Status (ASA-PS). We included adult patients undergoing emergency laparotomy at Zealand University Hospital, from March 2017 to January 2019, and Herlev Hospital, from November 2017 to January 2020. Variables included in the risk prediction models were collected retrospectively from the electronic patient records. Discrimination of the risk prediction models was evaluated with area under the curve (AUC) statistics, and calibration was assessed with Cox calibration regression. The primary outcome was 30-day mortality. Out of 1226 included patients, 146 patients (11.9%) died within 30 days. AUC (95% confidence interval) for 30-day mortality was 0.85 (0.82-0.88) for the updated NELA model, 0.84 (0.81-0.87) for the original NELA model, 0.81 (0.77-0.84) for the P-POSSUM model, and 0.76 (0.72-0.79) for the ASA-PS model. Calibration showed underestimation of mortality risk for both the updated NELA, original NELA and P-POSSUM models. The updated NELA risk prediction model performs well in this external validation study and may be used in similar settings. However, the model should only be used to discriminate between low- and high-risk patients, and not for prediction of individual risk due to underestimation of mortality.

Abstract
Background: The American College of Surgeons (ACS) NSQIP Surgical Risk Calculator (SRC) plays an important role in risk prediction and decision-making. We sought to enhance the existing ACS NSQIP SRC with functionality to predict geriatric-specific outcomes and assess the predictive value of geriatric-specific risk factors by comparing performance in outcomes prediction using the traditional ACS NSQIP SRC with models that also included geriatric risk factors.

Study design: Data were collected from 21 ACS NSQIP Geriatric Surgery Pilot Project hospitals between 2014 and 2017. Hierarchical regression models predicted 4 postoperative geriatric outcomes (ie pressure ulcer, delirium, new mobility aid use, and functional decline) using the traditional 21-variable ACS NSQIP SRC models and 27-variable models that included 6 geriatric risk factors (ie living situation, fall history, mobility aid use, cognitive impairment, surrogate-signed consent, and palliative care on admission).

Results: Data from 38,048 patients 65 years or older undergoing 197 unique operations across 10 surgical subspecialties were used. Stable model discrimination and calibration between developmental and validation datasets confirmed predictive validity. Models with and without geriatric risk factors demonstrated excellent performance (C statistic >0.8) with inclusion of geriatric risk factors improving performance. Of the 21 ACS NSQIP variables, CPT code, COPD, age, functional dependence, sex, disseminated cancer, diabetes, and sepsis were the strongest risk predictors, and impaired cognition, fall history, and mobility aid use were the strongest geriatric predictors.

Conclusions: The ACS NSQIP SRC can predict 4 unique outcomes germane to geriatric surgical patients, with improvement of predictive capability after accounting for geriatric risk factors. Augmentation of ACS NSQIP SRC can enhance shared decision-making to improve the quality of surgical care in older adults.

Abstract
Background: Preoperative risk prediction is important for guiding clinical decision-making and resource allocation. Clinicians frequently rely solely on their own clinical judgement for risk prediction rather than objective measures. We aimed to compare the accuracy of freely available objective surgical risk tools with subjective clinical assessment in predicting 30-day mortality.

Methods and findings: We conducted a prospective observational study in 274 hospitals in the United Kingdom (UK), Australia, and New Zealand. For 1 week in 2017, prospective risk, surgical, and outcome data were collected on all adults aged 18 years and over undergoing surgery requiring at least a 1-night stay in hospital. Recruitment bias was avoided through an ethical waiver to patient consent; a mixture of rural, urban, district, and university hospitals participated. We compared subjective assessment with 3 previously published, open-access objective risk tools for predicting 30-day mortality: the Portsmouth-Physiology and Operative Severity Score for the enUmeration of Mortality (P-POSSUM), Surgical Risk Scale (SRS), and Surgical Outcome Risk Tool (SORT). We then developed a logistic regression model combining subjective assessment and the best objective tool and compared its performance to each constituent method alone. We included 22,631 patients in the study: 52.8% were female, median age was 62 years (interquartile range [IQR] 46 to 73 years), median postoperative length of stay was 3 days (IQR 1 to 6), and inpatient 30-day mortality was 1.4%. Clinicians used subjective assessment alone in 88.7% of cases. All methods overpredicted risk, but visual inspection of plots showed the SORT to have the best calibration. The SORT demonstrated the best discrimination of the objective tools (SORT Area Under Receiver Operating Characteristic curve [AUROC] = 0.90, 95% confidence interval [CI]: 0.88-0.92; P-POSSUM = 0.89, 95% CI 0.88-0.91; SRS = 0.85, 95% CI 0.82-0.87). Subjective assessment demonstrated good discrimination (AUROC = 0.89, 95% CI: 0.86-0.91) that was not different from the SORT (p = 0.309). Combining subjective assessment and the SORT improved discrimination (bootstrap optimism-corrected AUROC = 0.92, 95% CI: 0.90-0.94) and demonstrated continuous Net Reclassification Improvement (NRI = 0.13, 95% CI: 0.06-0.20, p < 0.001) compared with subjective assessment alone. Decision-curve analysis (DCA) confirmed the superiority of the SORT over other previously published models, and the SORT-clinical judgement model again performed best overall. Our study is limited by the low mortality rate, by the lack of blinding in the 'subjective' risk assessments, and because we only compared the performance of clinical risk scores as opposed to other prediction tools such as exercise testing or frailty assessment.

Conclusions: In this study, we observed that the combination of subjective assessment with a parsimonious risk model improved perioperative risk estimation. This may be of value in helping clinicians allocate finite resources such as critical care and to support patient involvement in clinical decision-making.