Literature selections curated by Lewis Flint, MD, FACS, and reviewed by the Bulletin Brief editorial board.
Dyas AR, Carmichael H, Bronsert MR, et al. Does adding a measure of social vulnerability to a surgical risk calculator improve its performance? J Am Coll Surg. 2022;11(233).
This study sought to determine whether adding an assessment of socioeconomic and demographic factors—the Social Vulnerability Index (SVI)—to an accepted surgical risk calculator—the Surgical Risk Preoperative Assessment System (SURPAS)—would improve the accuracy of risk prediction. The SURPAS system used the ACS National Surgical Quality Improvement Program (NSQIP®) database to identify eight predictor variables that were then applied to predict 12 important postoperative outcomes. The SVI was added as the ninth variable, and patient addresses were used to estimate SVI status.
Data from 31,222 patients were included in the analysis, which showed that adding the SVI variable did not improve risk prediction. After further investigation, the authors determined that the eight SURPAS risk variables captured elements of social vulnerability, which probably explained the finding that risk prediction did not improve. Factors considered in the SURPAS system related to social vulnerability that could potentially influence surgical risk included higher rates of comorbidities and increased frequency of urgent and emergency procedures. The authors concluded that the SURPAS risk calculator functioned well without SVI.
Petro CC, Thomas JD, Tu C, et al. Robotic versus laparoscopic ventral hernia repair with intraperitoneal mesh: 1-year exploratory outcomes of the PROVE-IT randomized clinical trial. J Am Coll Surg. 2022;3(234).
This article reported 1-year outcomes in patients enrolled in the PROVE-IT randomized trial that compared results in patients who had robotic or laparoscopic repair of ventral hernias using intraperitoneal mesh placement. Previously reported short-term (30-day) follow-up data from the trial had shown equivalent results for both approaches in the areas of quality of life, pain, and hernia recurrence. One-year follow-up data were available for 95% (n=71) of the originally enrolled subjects.
The data analysis showed no difference in pain scores when the two groups were compared. Quality of life scores were higher for patients who had undergone robotic hernia repair, but patient-reported hernia recurrences were significantly more frequent in the robotic repair group. Reoperation rates for hernia recurrence were similar in both cohorts. The authors noted that hernia recurrence might be overestimated because of patients reporting bulges in the incision area that were not hernias. Because the small sample size in this study limited the generalizability of the conclusions, additional research will be needed to confirm factors contributing to improved quality of life scores as well as the actual hernia recurrence rate for each repair method.
James CA, Wachter RM, Woolliscroft JO. Preparing Clinicians for a Clinical World Influenced by Artificial Intelligence. JAMA. Published Mar 21, 2022.
In this opinion piece, the authors summarized available evidence and offered their views on the necessary steps that government agencies, healthcare organizations, and clinicians should take to ensure safe and effective adoption of artificial intelligence (AI) into clinical practice.
AI is the use of computers to simulate intelligent tasks humans normally perform, whereas machine learning is the process of a computer learning from data without prior programming. In the foreseeable future, AI likely will influence all aspects of clinical practice and completely change some disciplines. One potentially hazardous outcome would be for patients to rely on AI algorithms to make decisions about their care without input by clinicians. Oversight will be necessary as the influence of AI increases, and clinicians will need to commit critically reviewing data for genuine trust in AI to develop.
The authors emphasized the lessons learned from adopting electronic health records (EHR) as a source of knowledge that could guide incorporation of AI into clinical practice. EHR adoption improved patient safety but simultaneously negatively affected clinician well-being, professional satisfaction, and the physician-patient relationship.
James and colleagues used the recent implementation of the Epic Sepsis Model as an example of errors that can occur when technology is embedded into the EHR without appropriate oversight and testing. For the benefits of AI to be realized while avoiding negative effects and harms, a group of well-trained surgeons who understand the potential uses and limitations of the technology will be needed to coordinate the deployment of AI.
Because clinicians will bear much of the responsibility for protecting patients as AI is adopted, gaining the necessary knowledge will be an essential element of surgical practice.