Overcoming cut-off restrictions in multimorbidity prevalence estimates