OBJECTIVES:

Peer-reviewed cost-effectiveness analysis (CEA) publications provide vital information for health economic value-based decisions, therefore timely publication of CEA is essential. A literature analysis of CEA publications in non-small cell lung cancer (NSCLC) treatment in North America (NA) and Europe (EU) was performed to assess the timing of CEA publication date (CDP), relative to drug approval date (DD) and clinical/real-world data (for CEA model input) publication date (CRD).

METHODS:

An EMBASE literature analysis from January 2005-March 2017 was conducted with the following key criteria: “original articles”, “English”, “human”, “non-small cell lung cancer”, “cost-effectiveness analyses”. Information extracted from the eligible publications included: CPD, CRD, country, drug, and type of economic evaluation. DDs were retrieved from regulatory bodies’ websites and news reports. To determine publication timing CPD – CRD and CPD – DD were calculated, and descriptive statistics used to evaluate the timings.

RESULTS:

Fifty-two eligible publications were identified: 22 from NA (Canada: 18%, America: 68.2%) and 30 from EU, which consisted of publications from 7 countries, with France having the most (36.7%) publications. Across all publications, mean CPD – DD delay was 3.75 years (NA: 3.7 years, EU: 3.8 years) and mean CPD – CRD delay was 2.1 years (NA and EU: 2.1 years), and only 25% used real-world data. Most common types of economic evaluations were cost-utility analysis (44.2%), CEA (11%), direct cost (13.5%), and budget impact model (7.7%). Fourteen of the 24 NSCLC treatment drugs had at least 1 CEA publication; erlotinib had the most (48%) publications.

CONCLUSIONS:

As peer-reviewed CEA publications are slow in timing, have an incomplete coverage of drugs, and limited coverage of EU countries, publication timing and availability could be improved to support a faster and better informed appraisal for NSCLC medication. Furthermore, CEA publications could benefit from using more (local) real-world data as input for CEA models.