Comparison of New Zealand Cancer Registry data with an independent lung cancer audit.

Background: Despite the importance of New Zealand Cancer Registry (NZCR) data to research and healthcare decision-making, there has been no previous assessment of the accuracy of NZCR data since mandatory reporting commenced in 1994.

Aim: To assess the completeness and accuracy of NZCR lung cancer data.

Method: An audit of secondary care management in Auckland and Northland of lung cancer patients diagnosed in 2004 provided the opportunity to compare data from regional databases (RD) with NZCR data.

Results: Of 565 audit cases, 66 cases (12%) were not included on the NZCR listing. The NZCR listing included 9 eligible cases not identified by RD, 1 duplicate registration and 78 (13%) ineligible cases. Few differences occurred in demographic or tumour details for the 490 cases common to both listings. Tumour staging was available for 97% of cases in RD, and disease extent was recorded for 58% in the NZCR. The latter was more likely to be missing for cases with locally advanced disease (p<0.001), older age (p<0.001), or comorbidity (p<0.001).

Conclusion: Use of the NZCR alone would have reduced accrual by 12%; disease extent was absent for 42% with a systematic bias towards being unknown for cases with locally advanced disease. Use of NZCR data without recognition of this bias could lead to inappropriate conclusions. Those using NZCR data should be aware of its definitions, methodology and limitations.