# Predicting patient dropout: a nomogram for loss to follow-up after Helicobacter pylori eradication therapy

## TL;DR

This study creates a tool to predict which patients are likely to miss follow-up after Helicobacter pylori treatment, helping improve treatment success.

## Contribution

A novel nomogram was developed to predict patient dropout after Helicobacter pylori eradication therapy using clinical and demographic factors.

## Key findings

- Six independent risk factors for loss to follow-up were identified, including BMI > 30 and distance to hospital > 10 km.
- The nomogram achieved an AUC of 0.885 in training and 0.862 in testing, showing strong predictive accuracy.
- Decision curve analysis confirmed the model's clinical utility for predicting patient follow-up risk.

## Abstract

Helicobacter pylori (H. pylori) infection remains a global public health burden, particularly in developing countries. While its eradication is a cornerstone for gastric cancer prevention, management is challenged by high infection rates, rising antibiotic resistance, and suboptimal treatment efficacy. Compounding these issues, patient loss to follow-up (LTFU) has emerged as a critical factor directly undermining the success of eradication therapy.

This study aimed to investigate the risk factors associated with LTFU after H. pylori eradication, and to develop a predictive model for assessing the risk of LTFU.

We conducted a prospective cohort study (April 2023-September 2024) enrolling treatment-naïve patients from a tertiary gastroenterology clinic. Following data collection via questionnaires and follow-ups, a nomogram for predicting loss to follow-up (LTFU) was developed by applying LASSO regression for variable selection and logistic regression for model building. The model was evaluated by its area under the ROC curve (AUC), calibration, and decision curve analysis (DCA), with internal validation performed via 500 bootstrap resamples to confirm reliability.

A total of 145 (37.76%) patients failed to follow up. From 19 potential predictors, 6 variables were independent predictive factors. They were included in the risk score: BMI > 30 kg/m2 (OR = 3.81, 95% CI: 1.16–12.50), government employee (OR = 2.10, 95% CI: 1.21, 3.63), distance to hospital >10 km (OR = 11.27, 95%CI: 6.29–20.18), alcohol consumption (OR = 1.82, 95% CI: 1.19–2.79), outpatient waiting time (OR = 1.01, 95% CI: 1.00–1.02), and lack of awareness of follow-up (OR = 3.32, 95% CI: 1.93–5.69). In the training set, the model demonstrated an AUC of 0.885 (95% CI: 0.843–0.918), with a sensitivity of 93.58% and a specificity of 67.92%. Comparatively, in the test set, the model achieved an AUC of 0.862 (95% CI: 0.794–0.925), with a sensitivity of 83.33% and a specificity of 77.50%, effectively forecasting the risk of patient LTFU in H. pylori eradication. DCA demonstrated the favorable clinical utility of the nomogram, suggesting its potential as a valuable auxiliary tool for predicting the risk of LTFU.

The nomogram effectively assessed the risk of LTFU after H. pylori eradication, thereby contributing to improved treatment management outcomes.