Development and validation of a rapid-decision pathway to diagnose malnutrition in patients with lung cancer

Objectives: Malnutrition is frequently developed and outcome-related in patients with lung cancer (LC). Making a rapid and accurate diagnosis of malnutrition is the major concern for dietitians and clinicians.

Methods: We performed a multicenter, observational cohort study including 1219 patients with LC. Malnutrition was diagnosed using the Global Leadership Initiative on Malnutrition criteria, and the study population was randomly divided into a training group (n = 914) and a validation group (n = 305). A nomogram (to diagnose malnutrition) and two decision trees (to diagnose and grade malnutrition, respectively) were independently developed and tested. A random forest algorithm was used to calculate relative variable importance.

Results: The Global Leadership Initiative on Malnutrition criteria identified 292 patients with malnutrition (24%). Sex, body mass index, weight loss within 6 mo, weight loss beyond 6 mo, calf circumference, and handgrip strength to weight ratio were screened for model development. The nomogram showed good discrimination with an area under the curve (AUC) of 0.982 (95% confidence interval, 0.969-0.995) and good calibration in the validation group. A decision curve analysis demonstrated that the nomogram was clinically useful. The diagnostic tree showed an accuracy of 0.98 (Kappa = 0.942; AUC = 0.978; 95% confidence interval, 0.964-0.992), and the classification tree showed an accuracy of 0.98 (Kappa = 0.955; AUC = 0.987) in the validation group. Weight loss within 6 mo contributed the largest importance to both trees.

Conclusions: This study presents a rapid-decision pathway, including a set of tools that can be conveniently used to facilitate the diagnosis and severity grading of malnutrition in patients with LC.