Uniform and blinded cause of death verification of the NELSON lung cancer screening participants.

Primary outcome of the Dutch-Belgian lung cancer screening trial (NELSON) is lung cancer-specific mortality. Accurate assessment of the cause of death (CoD) is crucial. As death certificates regarding the CoD can be inaccurate, a clinical expert committee (CEC) was formed to assign the CoD. In this study, the medical files of deceased lung cancer patients were reviewed and the outcomes were compared with official death certificates. The first 266 completed medical files of Dutch deceased participants who were diagnosed with lung cancer during the study or of those with lung cancer on the death certificate were selected and blinded towards arms and patients identity. The end product of the review process consisted of six possible categories which defined the graduation of certainty that lung cancer was the primary CoD. The percentage agreement and the Cohen's kappa statistics between the two CEC-memberswere calculated. The sensitivity and specificity of the official death certificates were determined. The results indicated that, the overall concordance and the Cohen's kappa between the CEC-memberswere 86.1% and 0.57(0.45-0.69, p<0.001), respectively. This level increased with the numbers of cases evaluated. The sensitivity and the specificity of the official death certificate were 92.6% and 98.8%; 6.5% cases were reclassified to lung cancer specific death, which is lower than in the National Lung Screening trial(22.0%). Concluding, each death should be reviewed by at least two members. So far, in the NELSON trial, possible biases related to lung cancer death seem relatively small.