Rapid Fire Oral Presentation Clinical Oncology Society of Australia Annual Scientific Meeting 2026

Predicting hospitalisation in patients with lung cancer - a model for personalised care (144384)

Sheng Xiang Franklin SXF Chen 1 , Nicole Knox 2 , Joseph Descallar 3 , Shalini Vinod 4
  1. Illawarra Cancer Care Centre, Wollongong, NSW, Australia
  2. University of Wollongong, Wollongong, NSW, Australia
  3. Ingham Institute for Applied Medical Research, Liverpool , NSW, Australia
  4. Radiation Oncology, Liverpool Cancer Therapy Centre, Liverpool, NSW, Australia

Introduction:

Patients with lung cancer have a high prevalence of comorbidity, with 67% having at least one comorbidity and 45% classified as frail. These characteristics are independently associated with hospitalisation. Currently, there are no published Australian data on hospitalisation rates for patients with lung cancer or contributing factors.

We aimed to identify factors associated with hospitalisation within 90 days of starting treatment among patients with lung cancer, characterise the nature of hospitalisations, and develop a nomogram to predict hospitalisation.

Methods:

A retrospective audit of electronic medical records was conducted for patients with lung cancer who received at least systemic treatment or radiotherapy between January 2023 and December 2024 within South Western Sydney Local Health District, New South Wales, Australia. Patient demographics, tumour characteristics, treatment types, and hospitalisation outcomes were collated and analysed. Univariate and multivariable logistic regression models were used to analyse hospitalisation. Based on the final model, a nomogram was constructed.

Results:

In the cohort of 600 patients, 213 (35.5%) were hospitalised, and 60 (10.0%) died within 90 days of starting treatment. Univariate analysis revealed that a higher Charlson Comorbidity Index (CCI), more advanced cancer stage, higher neutrophil count at diagnosis, lower albumin at diagnosis, and palliative intent were associated with higher odds of hospitalisation (P < 0.05).

Multivariable analysis showed that CCI, cancer stage, and neutrophil count at diagnosis were independently associated with hospitalisation in patients with lung cancer (P < 0.05). The nomogram, based on the multivariable analyses, achieved a Harrell’s C-index of 0.632 and a calibration curve with a mean absolute error of 0.021.

Conclusion:

The developed nomogram can modestly predict hospitalisation in patients with lung cancer, personalising patient care, enabling close monitoring and intervention in hopes of reducing hospitalisation. Further research is needed to externally validate the nomogram and its effectiveness in reducing hospitalisation rates.