Non-small cell lung cancer (NSCLC) survival is compared between patients treated in all academic (n [1289) versus all non-academic (n = 12,698) Dutch hospitals, using Kaplan-Meier estimates and a Cox proportional hazards model. For 1009 patients, treatment patterns are described. Diagnosis in an academic hospital is associated with decreased mortality. Possibilities for improvement of NSCLC care are suggested.

Background: The aims of this study are to analyze differences in survival between academic and non-academic hospitals and to provide insight into treatment patterns for non-small cell lung cancer (NSCLC). Results show the state of NSCLC survival and care in the Netherlands. 

Methods: The Netherlands Cancer Registry provided data on NSCLC survival for all Dutch hospitals. We used the Kaplan-Meier estimate to calculate median survival time by hospital type and a Cox proportional hazards model to estimate the relative risk of mortality (expressed as hazard ratios) for patients diagnosed in academic versus non-academic hospitals, with adjustment for age, gender, and tumor histology, and stratifying for disease stage. Data on treatment patterns in Dutch hospitals was obtained from 4 hospitals (2 academic, 2 non-academic). A random sample of patients diagnosed with NSCLC from January 2009 until January 2011 was identified through hospital databases. Data was obtained on patient characteristics, tumor characteristics, and treatments. 

Results: The Cox proportional hazards model shows a significantly decreased hazard ratio of mortality for patients diagnosed in academic hospitals, as opposed to patients diagnosed in non-academic hospitals. This is specifically true for primary radiotherapy patients and patients who receive systemic treatment for non-metastasized NSCLC. 

Conclusion: Patients diagnosed in academic hospitals have better median overall survival than patients diagnosed in non-academic hospitals, especially for patients treated with radiotherapy, systemic treatment, or combinations. This difference may be caused by residual confounding since the estimates were not adjusted for performance status. A wide variety of surgical, radiotherapeutic, and systemic treatments is prescribed.