An Approach for Combining Clinical Judgment with Machine Learning to Inform Medical Decision Making: Analysis of Nonemergency Surgery Strategies for Acute Appendicitis in Patients with Multiple Long-Term Conditions.

BACKGROUND: Machine learning (ML) methods can identify complex patterns of treatment effect heterogeneity. However, before ML can help to personalize decision making, transparent approaches must be developed that draw on clinical judgment. We develop an approach that combines clinical judgment with ML to generate appropriate comparative effectiveness evidence for informing decision making. METHODS: We motivate this approach in evaluating the effectiveness of nonemergency surgery (NES) strategies, such as antibiotic therapy, for people with acute appendicitis who have multiple long-term conditions (MLTCs) compared with emergency surgery (ES). Our 4-stage approach 1) draws on clinical judgment about which patient characteristics and morbidities modify the relative effectiveness of NES; 2) selects additional covariates from a high-dimensional covariate space (P > 500) by applying an ML approach, least absolute shrinkage and selection operator (LASSO), to large-scale administrative data (N = 24,312); 3) generates estimates of comparative effectiveness for relevant subgroups; and 4) presents evidence in a suitable form for decision making. RESULTS: This approach provides useful evidence for clinically relevant subgroups. We found that overall NES strategies led to increases in the mean number of days alive and out-of-hospital compared with ES, but estimates differed across subgroups, ranging from 21.2 (95% confidence interval: 1.8 to 40.5) for patients with chronic heart failure and chronic kidney disease to -10.4 (-29.8 to 9.1) for patients with cancer and hypertension. Our interactive tool for visualizing ML output allows for findings to be customized according to the specific needs of the clinical decision maker. CONCLUSIONS: This principled approach of combining clinical judgment with an ML approach can improve trust, relevance, and usefulness of the evidence generated for clinical decision making. HIGHLIGHTS: Machine learning (ML) methods have many potential applications in medical decision making, but the lack of model interpretability and usability constitutes an important barrier for the wider adoption of ML evidence in practice.We develop a 4-stage approach for integrating clinical judgment into the way an ML approach is used to estimate and report comparative effectiveness.We illustrate the approach in undertaking an evaluation of nonemergency surgery (NES) strategies for acute appendicitis in patients with multiple long-term conditions and find that NES strategies lead to better outcomes compared with emergency surgery and that the effects differ across subgroups.We develop an interactive tool for visualizing the results of this study that allows findings to be customized according to the user's preferences.

Effectiveness of emergency surgery for five common acute conditions: an instrumental variable analysis of a national routine database.

The effectiveness of emergency surgery vs. non-emergency surgery strategies for emergency admissions with acute appendicitis, gallstone disease, diverticular disease, abdominal wall hernia or intestinal obstruction is unknown. Data on emergency admissions for adult patients from 2010 to 2019 at 175 acute National Health Service hospitals in England were extracted from the Hospital Episode Statistics database. Cohort sizes were: 268,144 (appendicitis); 240,977 (gallstone disease); 138,869 (diverticular disease); 106,432 (hernia); and 133,073 (intestinal obstruction). The primary outcome was number of days alive and out of hospital at 90 days. The effectiveness of emergency surgery vs. non-emergency surgery strategies was estimated using an instrumental variable design and is reported for the cohort and pre-specified sub-groups (age, sex, number of comorbidities and frailty level). Average days alive and out of hospital at 90 days for all five cohorts were similar, with the following mean differences (95%CI) for emergency surgery minus non-emergency surgery after adjusting for confounding: -0.73 days (-2.10-0.64) for appendicitis; 0.60 (-0.10-1.30) for gallstone disease; -2.66 (-15.7-10.4) for diverticular disease; -0.07 (-2.40-2.25) for hernia; and 3.32 (-3.13-9.76) for intestinal obstruction. For patients with 'severe frailty', mean differences (95%CI) in days alive and out of hospital for emergency surgery were lower than for non-emergency surgery strategies: -21.0 (-27.4 to -14.6) for appendicitis; -5.72 (-11.3 to -0.2) for gallstone disease, -38.9 (-63.3 to -14.6) for diverticular disease; -19.5 (-26.6 to -12.3) for hernia; and - 34.5 (-46.7 to -22.4) for intestinal obstruction. For patients without frailty, the mean differences (95%CI) in days alive and out of hospital were: -0.18 (-1.56-1.20) for appendicitis; 0.93 (0.48-1.39) for gallstone disease; 5.35 (-2.56-13.28) for diverticular disease; 2.26 (0.37-4.15) for hernia; and 18.2 (14.8-22.47) for intestinal obstruction. Emergency surgery and non-emergency surgery strategies led to similar average days alive and out of hospital at 90 days for five acute conditions. The comparative effectiveness of emergency surgery and non-emergency surgery strategies for these conditions may be modified by patient factors.

Antibiotics as first-line alternative to appendicectomy in adult appendicitis: 90-day follow-up from a prospective, multicentre cohort study.

BACKGROUND: Uncomplicated acute appendicitis can be managed with non-operative (antibiotic) treatment, but laparoscopic appendicectomy remains the first-line management in the UK. During the COVID-19 pandemic the practice altered, with more patients offered antibiotics as treatment. A large-scale observational study was designed comparing operative and non-operative management of appendicitis. The aim of this study was to evaluate 90-day follow-up. METHODS: A prospective, cohort study at 97 sites in the UK and Republic of Ireland included adult patients with a clinical or radiological diagnosis of appendicitis that either had surgery or non-operative management. Propensity score matching was conducted using age, sex, BMI, frailty, co-morbidity, Adult Appendicitis Score and C-reactive protein. Outcomes were 90-day treatment failure in the non-operative group, and in the matched groups 30-day complications, length of hospital stay (LOS) and total healthcare costs associated with each treatment. RESULTS: A total of 3420 patients were recorded: 1402 (41 per cent) had initial antibiotic management and 2018 (59 per cent) had appendicectomy. At 90-day follow-up, antibiotics were successful in 80 per cent (1116) of cases. After propensity score matching (2444 patients), fewer overall complications (OR 0.36 (95 per cent c.i. 0.26 to 0.50)) and a shorter median LOS (2.5 versus 3 days, P < 0.001) were noted in the antibiotic management group. Accounting for interval appendicectomy rates, the mean total cost was €1034 lower per patient managed without surgery. CONCLUSION: This study found that antibiotics is an alternative first-line treatment for adult acute appendicitis and can lead to cost reductions.