Search

BACKGROUND Accurate projections of procedural case durations are complex, but critical to planning of perioperative staffing, operating room resources, and patient communication. Nonlinear prediction models using machine learning methods may provide opportunities for hospitals to improve upon current estimates of procedure duration. OBJECTIVE We hypothesized a machine learning algorithm derived from a large multicenter dataset would more accurately predict surgical procedure duration when compared to a baseline linear regression approach. Using an explainable machine learning-based algorithm, results provide additional valuable insight regarding procedure duration and variability. METHODS A total of 1,177,893 procedures from 13 academic and private hospitals between 2016 and 2019 were used. Deep learning, gradient boosting, and ensemble machine learning models were generated using perioperative data available at three distinct time points: time of scheduling, time of arrival to the operating/procedure room (primary model), and time of surgical incision/procedure start. The primary outcome was procedure duration, defined by the time between arrival and departure of the patient from the procedure room. Model performance was assessed by mean absolute error, proportion of predictions within 20% of actual duration, and other standard metrics. Performance was compared to a baseline method of historical means within a linear regression model. Model features driving predictions were assessed using Shapley values and permutation feature importance. RESULTS Across all procedures, median procedure duration was 94 minutes (interquartile range of 50-167 minutes). In estimating procedure duration, the gradient boosting machine was the best performing model, demonstrating a mean absolute error of 34 minutes with 46% of predictions within 20% of actual duration in the test dataset. This represented a statistically and clinically significant improvement in predictions compared to a baseline linear regression model (43 minutes, p < 0.001; 39% of predictions within 20% of actual duration). The most important features in model training were historical procedure duration by surgeon, the word “free” within the procedure text, and time of day. CONCLUSIONS Nonlinear models using machine learning techniques may be used to generate high-performing, automatable, explainable, and scalable prediction models for procedure duration. Medi

Clinical practice guidelines recommend supervised exercise therapy (SET) as first-line therapy for patients with peripheral artery disease (PAD) and intermittent claudication (IC). The Society for Vascular Surgery Appropriate Use Criteria for IC deems excercise therapy (ET) as appropriate for all patients with IC. However, compliance with recommendations for the use of ET is often poor owing to the lack of availability, patient travel requirements, and cost. Results of home-based ET programs have been mixed with a trend toward improved results with more frequent patient engagement. The feasibility, use, and effectiveness of a 12-week app-based structured ET program using cognitive-behavioral therapy (CBT) techniques for IC was evaluated.Patients with PAD confirmed by an abnormal ankle-brachial index or toe-brachial index and IC were recruited prospectively to participate in Society for Vascular Surgery SET, a 12-week app-based ET program. Participants performed home 6-minute walk tests, completed quality-of-life surveys, received education options via micro learning courses (eg, What is PAD?, Exercise, and Nutrition), and ongoing health coaching using CBT techniques. They were instructed to record at least three 30-minute ET walks a week using their personal mobile phones. Programming also included daily doses of health education, coaching, and reminders sent via text message.One hundred thirty-nine patients (37% women; mean age, 65 years) were enrolled across 20 institutions (44% offered in-person ET programs). One hundred twenty patients (86%) completed the program. Participants recorded 201,163 minutes of walking 8,013,520 steps with the ET program, completing a total of 5049 SET walks. Nineteen enrolled participants (14%) became inactive because they either paused (14 participants: medical reasons, travel, or other) or withdrew (5 participants: security concerns, lack of motivation). Ninety-two percent of patients met their stated CBT specific, measurable, achievable, relevant, and time-bound goals. Freedom from intervention at 6 months was 92% and 69% at 12 months.Deployment of a 12-week app-based ET program for PAD incorporating CBT was feasible in achieving 86% program completion and effective at meeting guideline recommended activity goals. Ninety-two percent of participants achieved their CBT specific, measurable, achievable, relevant, and time-bound goals. The use of ET was increased by virtue of offering this program at institutions that did not offer ET.

First-line treatment of peripheral artery disease (PAD) involves medical therapy and lifestyle modification. Multiple professional organizations such as the Society for Vascular Surgery and the American Heart Association/American College of Cardiology make Class I recommendations for medical management including antiplatelet, statin, antihypertensive, and cilostazol medications, as well as lifestyle therapy including exercise and smoking cessation. Although evidence supports up-front medical and lifestyle management prior to surgical intervention, it is unclear how well this occurs in contemporary clinical practice. It is also unclear whether variability in first-line treatment prior to revascularization is associated with postoperative outcomes. This study examined the proportion of patients with claudication actively receiving evidence-based therapy prior to surgery in a statewide surgical registry.We conducted a retrospective cohort study of adult patients undergoing elective open lower extremity bypass for claudication from 2012 to 2021 within a statewide surgical quality registry. The primary exposure was optimal medical therapy, defined as an antiplatelet agent, a statin, and an angiotensin-converting enzyme inhibitor or angiotensin II receptor blocker (if the patient had hypertension) on the patient's home medication list on admission for surgery, all of which are Class I recommendations. Despite also being Class I recommendations, cilostazol was not included in the primary exposure due to its highly selective use and our inability to capture intolerance and/or contraindications that are common, and lifestyle therapies were not included as they were only recorded at the time of discharge rather than preoperatively. The primary outcomes were mortality, hospital readmission, amputation, wound complication, myocardial infarction (MI), non-patent bypass, and non-independent ambulatory status at 30 days and 1 year after surgery. Multivariable logistic regression was performed to estimate the association of receiving optimal vs non-optimal medical therapy.A total of 3829 patients with claudication underwent bypass surgery during the study period, with a mean age of 64.8 years (standard deviation, 9.8 years); 2690 (70.3%) were males, and 1873 (48.9%) were current smokers. Of the patients, 1822 (47.6%) were on optimal medical therapy prior to surgery. Additionally, at discharge, 66.5% of smokers received referral to smoking cessation therapy, and 54.1% of patients received referral to exercise therapy. In a multivariable logistic regression, compared with patients not on optimal medical therapy, patients on optimal medical therapy prior to surgery had lower 30-day odds of mortality (adjusted odds ratio [aOR], 0.45; 95% confidence interval [CI], 0.26-0.78) and MI (aOR, 0.46; 95% CI, 0.28-0.76), lower 1-year odds of mortality (aOR, 0.57; 95% CI, 0.39-0.82), MI (aOR, 0.48; 95% CI, 0.32-0.74), and lower readmission (aOR, 0.79; 95% CI, 0.64-0.96).Although medical and lifestyle management is recommended as first-line treatment for patients with PAD, only one-half of patients were on optimal medical therapy prior to surgery. Patients receiving optimal therapy had a lower risk of postoperative mortality, MI, and readmission. This suggests that not only are there significant opportunities to improve clinical utilization of evidence-based treatment of PAD, but that doing so can benefit patients postoperatively.

In 2011, the Society for Vascular Surgery (SVS) prepared a set of clinical research priorities through a survey of its membership. These priorities were developed with the goal of enhancing clinical research to improve care for vascular patients. In the subsequent decade, several of these priorities served as the focus of clinical trials and significant research efforts. It was understood from the outset that this list of priorities represented a starting point with the intention that they be reevaluated at suitable intervals. In 2021, the SVS Research Council set out to update the research priorities by surveying the SVS membership and engaged a panel of subject matter experts. This process resulted in an updated set of vascular research priorities that more clearly align with current areas of emphasis. Our priorities remain focused on basic areas including aortic disease, carotid disease, lower extremity arterial disease, venous disease, dialysis access, and medical management of vascular disease, along with the topic of health care disparities. The 10 updated priorities reported herein reflect our increasing awareness of the need to understand vascular disease pathogenesis and prevention in the context of a diverse patient population. Importantly, patient-centered outcomes and personalized vascular care are at the core of these updated priorities. Similar to the aims of the original 2011 clinical research priorities, our hope is that this updated list will help to drive large-scale investigations that will improve how we care for our vascular patients.