Variation, precision and validity of 1-year survival estimates for lung, breast, colon and prostate cancer in South East England primary care trusts

Abstract

Background: English cancer policy has encouraged primary care trusts (PCTs) to consider their 1-year cancer survival estimates. This study quantifies variation in these estimates across 39 PCTs in the London and South East Coast strategic health authorities, and explores their precision, possible confounding by age and bias due to death certificate only (DCO) registrations.

Study design: Retrospective observational study.

Methods: One-year relative survival estimates and data on DCO registrations for patients diagnosed with lung, colorectal, breast and prostate cancers between 2002 and 2006 were extracted from the UK Cancer Information Service. Direct age standardization was performed with weightings derived from the standard cancer patient population for Europe. Pearson correlation coefficients between survival estimates and DCO proportions were calculated.

Results: Mean 1-year PCT survival estimates ranged from 6.9 to 19.4 percentage points, and the precision of individual estimates ranged from ±0.9 to ±6.5 percentage points (at 95% confidence level). Age standardization significantly changed the estimates of nine PCTs for breast cancer, five PCTs for lung cancer and three PCTs for colorectal cancer. None of the prostate cancer estimates were affected significantly. DCO proportions were positively associated with lung cancer survival and negatively associated with colorectal and breast cancer survival.

Conclusions: PCT 1-year cancer survival estimates may be informative, but caveats relating to data quality and hence the validity of the estimates means that they require careful investigation before naïve use, as random variation, confounding due to age and bias due to DCO registrations may be significant.