Measuring the mortality reductions produced by organized cancer screening: a principled approach

Seminar Series

Friday, February 1, 2019 - 12:00
MSRB1 #001 Seminar Room
James A Hanley, PhD

In cancer screening trials, or in comparisons involving regions that did/did not introduce population-based screening programs, the mortality reductions are usually summarized by an overall (single-number) mortality reduction.  But this overall mortality reduction is an average of minimal reductions in the first years, larger ones after some years, waning ones starting some years after the last screens, smaller ones among those screened a few times and larger ones among those screened several times. 

Statistical models recognizing the impacts of successive rounds of screening, proposed in the 1960s, were extended in the 1990s to plan screening trials.  However, their central principles have not been used in the analysis of the data from trials.  Prevailing study-design and data-analysis practices still use the test statistics and single-summary estimates used in 1972, and assume that a successful screening program will produce proportional hazards.
Liu et al. (IntStatRev2015) developed a model for the expected reductions in each (Age,Year) cell of a Lexis diagram.  It describes the effect of one round of screening with 3 parameters (1) when in follow-up time the reduction produced by this one round is maximal (2) how large this reduction is and (3) how dispersed in follow-up time the reductions are.  Using the screening history in each (Age,Year) cell as a design matrix, reductions from previous screens are combined.  Ultimately (if follow-up extends far enough beyond the last screen) the resulting hazard ratio curves  have bathtub shapes that are modulated by the screening histories.
I illustrate this model using data from screening trials in prostate, colon, lung, and ovarian cancer, and -- for breast cancer -- using population data (Njor, JMedScr2015).  I show how analyses of such screening data can be extended/refined to incorporate the numbers and timing of screening invitations in relation to where in the Lexis diagram the deaths do/do not occur.

Speaker:    James A. Hanley, PhD
Department of Epidemiology, Biostatistics and Occupational Health
McGill University,  
Montreal, Canada