Sensitivity of the Swedish statutory surveillance system for communicable diseases 1998-2002, assessed by the capture-recapture method.

To assess the sensitivity of the Swedish surveillance system, four notifiable communicable diseases in Sweden were examined during 1998-2002 with the two-sources capture-recapture method, based on parallel clinical and laboratory notifications. The sensitivity (proportion of diagnosed diseases actually being notified) was highest for salmonellosis (99.9%), followed by meningococcal infection (98.7%), and tularaemia (98.5%). For penicillin-resistant pneumococci, introduced as a notifiable disease in 1996, the overall sensitivity was 93.4%--increasing from 86.5% in 1998 to 98.5% in 2002. The system benefited from parallel reporting, with a sensitivity of clinical and laboratory notifications alone (all diseases combined) of 91.6% and 95.9% respectively. The sensitivity of both clinical and laboratory notifications was markedly higher in counties using the national electronic reporting system, SmiNet. Thus, sensitivity was higher for diseases with a long tradition of reporting, and there is a run-in period after a new disease becomes notifiable.