Open Access Research article

Lessons from a one-year hospital-based surveillance of acute respiratory infections in Berlin- comparing case definitions to monitor influenza

Matthias Nachtnebel123*, Benedikt Greutelaers123, Gerhard Falkenhorst4, Pernille Jorgensen4, Manuel Dehnert1, Brunhilde Schweiger5, Christian Träder6, Silke Buda7, Tim Eckmanns8, Ole Wichmann4 and Wiebke Hellenbrand4

Author affiliations

1 Department of Infectious Disease Epidemiology, Robert Koch Institute, DGZ-Ring 1, Berlin 13086, Germany

2 Post Graduate Training in Applied Epidemiology, Robert Koch Institute, Berlin, Germany

3 European Programme for Intervention Epidemiology Training (EPIET), European Centre for Disease Prevention and Control (ECDC), Stockholm, Sweden

4 Immunization Unit, Robert Koch Institute, Berlin, Germany

5 National Reference Centre for Influenza, Robert Koch Institute, Berlin, Germany

6 Vivantes Clinic, Berlin, Germany

7 Respiratory Disease Unit, Robert Koch Institute, Berlin, Germany

8 Surveillance Unit, Robert Koch Institute, Berlin, Germany

For all author emails, please log on.

Citation and License

BMC Public Health 2012, 12:245  doi:10.1186/1471-2458-12-245

Published: 27 March 2012



Surveillance of severe acute respiratory infections (SARI) in sentinel hospitals is recommended to estimate the burden of severe influenza-cases. Therefore, we monitored patients admitted with respiratory infections (RI) in 9 Berlin hospitals from 7.12.2009 to 12.12.2010 according to different case definitions (CD) and determined the proportion of cases with influenza A(H1N1)pdm09 (pH1N1). We compared the sensitivity and specificity of CD for capturing pandemic pH1N1 cases.


We established an RI-surveillance restricted to adults aged ≤ 65 years within the framework of a pH1N1 vaccine effectiveness study, which required active identification of RI-cases. The hospital information-system was screened daily for newly admitted RI-patients. Nasopharyngeal swabs from consenting patients were tested by PCR for influenza-virus subtypes. Four clinical CD were compared in terms of capturing pH1N1-positives among hospitalized RI-patients by applying sensitivity and specificity analyses. The broadest case definition (CD1) was used for inclusion of RI-cases; the narrowest case definition (CD4) was identical to the SARI case definition recommended by ECDC/WHO.


Over the study period, we identified 1,025 RI-cases, of which 283 (28%) met the ECDC/WHO SARI case definition. The percentage of SARI-cases among internal medicine admissions decreased from 3.2% (calendar-week 50-2009) to 0.2% (week 25-2010). Of 354 patients tested by PCR, 20 (6%) were pH1N1-positive. Two case definitions narrower than CD1 but -in contrast to SARI- not requiring shortness of breath yielded the largest areas under the Receiver-Operator-Curve. Heterogeneity of proportions of patients admitted with RI between hospitals was significant.


Comprehensive surveillance of RI cases was feasible in a network of community hospitals. In most settings, several hospitals should be included to ensure representativeness. Although misclassification resulting from failure to obtain symptoms in the hospital information-system cannot be ruled out, a high proportion of hospitalized PCR-positive pH1N1-patients (45%) did not fulfil the SARI case-definition that included shortness of breath or difficulty breathing. Thus, to assess influenza-related disease burden in hospitals, broader, alternative case definitions should be considered.