# Preoperative Hill's classification as a predictor of postoperative gastroesophageal reflux disease following laparoscopic sleeve gastrectomy: a prospective multicenter study from Egypt

## TL;DR

This study shows that preoperative Hill classification can predict the risk of developing GERD after sleeve gastrectomy, helping surgeons make better decisions for high-risk patients.

## Contribution

The first multicenter Egyptian study to use Hill grading with machine learning to predict postoperative GERD after sleeve gastrectomy.

## Key findings

- Preoperative Hill Grades III–IV independently predicted postoperative GERD with high statistical significance.
- 36.8% of patients developed objective GERD at 1 year, including 26% with de novo GERD.
- A Random Forest classifier using Hill grade, hiatal hernia, and HbA1c achieved an AUC of 0.84 for GERD prediction.

## Abstract

Laparoscopic sleeve gastrectomy (LSG) is the most widely used bariatric procedure, but its influence on gastroesophageal reflux disease (GERD), whether inducing, aggravating, or improving it, remains a subject of ongoing debate. This study evaluated whether preoperative Hill’s classification can predict postoperative GERD and identified other endoscopic factors related to this outcome in an Egyptian multicenter cohort.

In this prospective multicenter cohort, 106 consecutive morbidly obese patients undergoing primary LSG (January 2023–June 2024) were evaluated. All underwent standardized preoperative esophagogastroduodenoscopy (EGD) with Hill, Los Angeles (LA), and Z-line appearance (ZAP) grading; postoperative assessments (GerdQ and repeat EGD) were performed at 6 months and 1 year. GERD was defined as erosive esophagitis (LA ≥ B) on follow-up EGD, LA Grade A esophagitis with concurrent pathological acid exposure (DeMeester score > 14.7), or, in symptomatic patients with normal endoscopy, pathological acid exposure alone. Ambulatory 24-h pH monitoring (off PPI therapy) was performed preoperatively in all patients and postoperatively only in symptomatic patients with a normal endoscopy. Follow-up was complete for all 106 patients at 1 year.

At 1 year, 39/106 patients (36.8%) had objective GERD, including a 26.0% incidence of de novo GERD among those without preoperative evidence of the condition. Preoperative Hill Grades III–IV were strongly associated with postoperative GERD in the primary analysis and remained independently predictive after adjustment for age, sex, BMI, and %EWL (aOR 3.92; 95% CI 1.98–7.76; p < 0.001). This association remained robust in a sensitivity analysis applying a stricter, contemporary GERD definition (aOR 3.71; 95% CI 1.82–7.55; p < 0.001). Time-to-event analysis showed reduced GERD-free survival in patients with Hill III–IV (1-year GERD-free survival 41.9% vs 71.4% for Hill I/II; Cox HR 3.60, 95% CI 1.85–7.00; log-rank p < 0.001). Preoperative esophagitis (LA ≥ A) and distorted Z-line (ZAP ≥ II) were associated with GERD in univariate analyses, but Hill grade was the only robust independent predictor. A Random Forest classifier incorporating Hill grade, hiatal hernia, and HbA1c achieved an internal AUC of 0.84 (95% CI 0.76–0.91), outperforming logistic regression (AUC 0.76; p = 0.03).

Preoperative Hill’s grading of the gastroesophageal flap valve (GEFV) independently predicts GERD after LSG. Its prognostic value persists after adjusting for hiatal hernia and is consistent under contemporary diagnostic criteria. Concurrent cruroplasty does not completely offset the risk of a defective valve. Routine preoperative EGD with Hill assessment improves risk stratification and may guide individualized surgical planning, including consideration of RYGB in high-risk cases. This is the first multicenter Egyptian study to apply Hill grading together with machine-learning methods for reflux prediction.

Not applicable.