Lead-time and overdiagnosis estimation in neuroblastoma screening

In Germany, neuroblastoma is the most frequent extracranial solid childhood tumour. Its properties made it seem an ideal candidate for screening. A German trial assessed the effect of screening at one year of age from 1995-2001 in a nationwide project. We present here the methods developed for the estimation of lead-time and overdiagnosis in this project. Follow up on 1.5 million screened children and 2.1 million control children is currently available until June 2002. Ascertainment of control cohort cases and false negative cases is complete up to this date. A method for determining an empirical lead-time distribution and overdiagnosis estimate from comparing the age specific incidences in the control group and the study group is presented. Lead-time leads to an excess of cases in the screening group at the screening age and cases missing at higher age. If more cases are observed at the screening age than can be explained by lead-time, the difference is attributed to overdiagnosis. The width of the screening age window and the empirical maximum lead-time have to be chosen from graphs. About 1.0/100000 cases (20 per cent of the possible cases) experienced lead-time while 6.8/100000 cases were overdiagnosed. The mean lead-time was estimated to be about 15 months. The number of cases who might benefit is much smaller than was expected before the study while the overdiagnosed group is much larger. The method is robust against the choices that have to be made in the estimation process.