Systematic review: reliability of compendia methods for off-label oncology indications.

BACKGROUND

The Centers for Medicare & Medicaid Services limit coverage of cancer drugs for off-label indications to indications listed in specified compendia.

PURPOSE

To assess whether compendia provide comprehensive, research-based, and timely information for off-label prescribing in oncology.

DATA SOURCES

6 drug compendia, English-language literature searches of MEDLINE and the Cochrane Central Register of Controlled Trials from 2006 and 2008, and American Society of Clinical Oncology annual meeting abstracts from 2004 to 2007. Data Assessment: The compendia's stated methods, literature related to off-label indications of 14 cancer drugs in 2006, updated literature related to 1 off-label indication between 2006 and 2008, and completeness of compendia content and citations were assessed.

DATA SYNTHESIS

The compendia's stated methods varied greatly from their actual practices. Compendia cited little of the available evidence, often neither the most recent nor that of highest methodological quality. Compendia differed in evidence cited, terminology, detail, presentation, and referencing. For the 14 off-label indications studied, the compendia differed in the indications included and whether and how they recommended particular agents for particular types of cancer. Update schedules varied, and documentation practices made it difficult to determine whether and when compendia content was updated. For 1 indication, compendia citations did not increase between 2006 and 2008 despite newly published articles.

LIMITATIONS

The 2006 analysis was limited to 14 off-label indications; the 2008 update examined 1 indication. Only off-label indications for cancer drugs were included, and results cannot be generalized to noncancer drugs or indications.

CONCLUSION

Oncologists rely on compendia for up-to-date access to evidence and reimbursement information for off-label indications. Current compendia lack transparency, cite little current evidence, and lack systematic methods to review or update evidence.