Evaluating a parametric model to correct multiple sclerosis incidence for reporting delay.

摘要

Esbjerg et al. [1] developed a framework for correcting current calculations of multiple sclerosis (MS) incidence for two types of reporting delay. Since the publication of their report more than a decade ago, the precise incidence has become available, so that their corrected values may be compared with the actual incidence. The present letter shows that the methods of Esbjerg et al. were remarkably precise in removing the bias due to reporting delay. In Denmark, the incidence of MS is monitored in the Danish MS Registry (DMSR). However, the calculation of current incidence rates is complicated by some administrative delay in reporting cases to the registry but more importantly by the delay between onset and diagnosis of MS. The delay can be several months or years because the diagnostic criteria require that the patient has experienced two attacks separated in time [2] or, nowadays, accumulation of new MRI-lesions with time [3]. Since we only observe people who have received a diagnosis, the crude onset incidence rate will be underestimated especially for the most recent years. In statistical terms, we have to make inference about right truncated data.