Evaluation of the national surveillance of Legionnaires' disease in Norway, 2008-2017.

BACKGROUND: In Norway, Legionnaires' disease is reportable upon clinical suspicion to public health authorities and mandatorily notifiable through the Norwegian surveillance system for communicable diseases (MSIS) for both clinicians and laboratories. In the summer of 2017, several European countries reported high notification rates for Legionnaires' disease, which was not observed in Norway. We evaluated MSIS to assess if it meets its objectives of detecting cases and trends in incidence of Legionnaires' disease. METHODS: We retrieved MSIS data from 2008 to 2017 and calculated timeliness as days from sampling to notification, and internal completeness for key variables as the proportion of observations with a value. Where possible, we assessed internal validity on the presence of a plausible value. To estimate external completeness and validity we linked MSIS with hospital reimbursement claims in the Norwegian Patient Registry. To assess acceptability and representativeness, we surveyed doctors in 39 hospitals on their units' diagnostic and notification procedures, and their use of MSIS. RESULTS: There were 438 notified cases. Internal completeness and internal validity were high for key variables (≥95%). The median delay from sampling to notification was 4 days. There were 73 patients in MSIS only, 70 in the Norwegian Patient Registry only, and 351 in both registers. The external completeness of MSIS was 83% (95% CI 80-86%). For external validity, the positive predictive value of MSIS was 83% (95% CI 79-86%). Forty-seven respondents from 28 hospitals described testing procedures. These were inconsistent: 29 (62%) reported no systematic application of criteria for requesting legionella testing. Eighteen (38%) reported testing all patients with suspected pneumonia and a travel history. Thirty-one (66%) found the notification criteria clear. CONCLUSIONS: Our results suggest that the surveillance in MSIS can detect incidence changes for Legionnaires' disease over time, by place and person, but likely does not detect every case diagnosed in Norway. We recommend wider investigation of diagnostic procedures in order to improve representativeness and awareness of MSIS notification criteria among clinicians in order to improve acceptability of the surveillance. We also recommend a more comprehensive assessment of whether patients only registered in the Norwegian Patient Registry were true Legionnaires' disease cases.