Limits...
Chances and limitations of wild bird monitoring for the avian influenza virus H5N1--detection of pathogens highly mobile in time and space.

Wilking H, Ziller M, Staubach C, Globig A, Harder TC, Conraths FJ - PLoS ONE (2009)

Bottom Line: Although more than 25,000 birds were sampled in Germany per year, the upper 95% confidence limit did not fall below 5% in the outbreak regions for most of the time.The proposed analysis can be used to monitor water bodies and high risk areas, also as part of an early-warning system.Chances for an improved targeting of the monitoring system as part of a risk-based approach are discussed with the perspective of reducing sample sizes.

View Article: PubMed Central - PubMed

Affiliation: Institute of Epidemiology, Friedrich-Loeffler-Institut, Federal Research Institute for Animal Health, Wusterhausen, Germany.

ABSTRACT
Highly pathogenic influenza virus (HPAIV) H5N1 proved to be remarkably mobile in migratory bird populations where it has led to extensive outbreaks for which the true number of affected birds usually cannot be determined. For the evaluation of avian influenza monitoring and HPAIV early warning systems, we propose a time-series analysis that includes the estimation of confidence intervals for (i) the prevalence in outbreak situations or (ii) in the apparent absence of disease in time intervals for specified regional units. For the German outbreak regions in 2006 and 2007, the upper 95% confidence limit allowed the detection of prevalences below 1% only for certain time intervals. Although more than 25,000 birds were sampled in Germany per year, the upper 95% confidence limit did not fall below 5% in the outbreak regions for most of the time. The proposed analysis can be used to monitor water bodies and high risk areas, also as part of an early-warning system. Chances for an improved targeting of the monitoring system as part of a risk-based approach are discussed with the perspective of reducing sample sizes.

Show MeSH

Related in: MedlinePlus

Results of the HPAIV H5N1 monitoring.Regions with cases of HPAIV H5N1 infection in wild birds in Germany (A). Spatial distribution of the wild bird monitoring in German municipalities 2007 (samples per square kilometer) (B). After the initial outbreak on the Isle of Ruegen (1), only three cases were recorded in inland municipalities at the coastline of Mecklenburg-Western Pomerania (2). Most of the cases in the neighboring states of Brandenburg (3) and Schleswig–Holstein (4) were reported several weeks afterwards in March and April 2006. HPAIV H5N1 cases at Lake Constance (5) were connected to further spread in smaller wetland areas in Bavaria (6). The massive outbreak in 2007 at the Helme reservoir at Berga-Kelbra (7) was related to four cases eastward in Saxony (8).
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2722081&req=5

pone-0006639-g003: Results of the HPAIV H5N1 monitoring.Regions with cases of HPAIV H5N1 infection in wild birds in Germany (A). Spatial distribution of the wild bird monitoring in German municipalities 2007 (samples per square kilometer) (B). After the initial outbreak on the Isle of Ruegen (1), only three cases were recorded in inland municipalities at the coastline of Mecklenburg-Western Pomerania (2). Most of the cases in the neighboring states of Brandenburg (3) and Schleswig–Holstein (4) were reported several weeks afterwards in March and April 2006. HPAIV H5N1 cases at Lake Constance (5) were connected to further spread in smaller wetland areas in Bavaria (6). The massive outbreak in 2007 at the Helme reservoir at Berga-Kelbra (7) was related to four cases eastward in Saxony (8).

Mentions: Constant temporal and spatial sampling rates are required for mobile pathogens like HPAIV H5N1 to achieve a considerable degree of confidence in the results. Analysis of the temporal distribution of the sampling within the wild bird monitoring program showed, however, that the sampling rates decreased after an initial period of intensive testing upon detection of the index case and reached a more or less constant level (Figure 3). Short-term infrequent sampling produced unbalanced datasets with high autocorrelation in time, ranging between 0 samples on some days and up to 300 on others. The variation in space exhibited a similar pattern (Figure 4). The allocation of the monitoring activities to the different German federal states led to a nearly homogeneous distribution of the sample size over the entire country at the state level Measure of overall spatial autocorrelation: (Moran's I: 0.09; C-l: 0.08-0.1; expected value: −0.00008) [20]. Yet, irregular sampling at the district or municipality level produced data with considerable spatial dependencies. Sampling often concentrated around outbreak areas and on susceptible species. This prompted for instance high sample sizes at the Baltic Sea coast and at Lake Constance in reaction to outbreaks in these locations.


Chances and limitations of wild bird monitoring for the avian influenza virus H5N1--detection of pathogens highly mobile in time and space.

Wilking H, Ziller M, Staubach C, Globig A, Harder TC, Conraths FJ - PLoS ONE (2009)

Results of the HPAIV H5N1 monitoring.Regions with cases of HPAIV H5N1 infection in wild birds in Germany (A). Spatial distribution of the wild bird monitoring in German municipalities 2007 (samples per square kilometer) (B). After the initial outbreak on the Isle of Ruegen (1), only three cases were recorded in inland municipalities at the coastline of Mecklenburg-Western Pomerania (2). Most of the cases in the neighboring states of Brandenburg (3) and Schleswig–Holstein (4) were reported several weeks afterwards in March and April 2006. HPAIV H5N1 cases at Lake Constance (5) were connected to further spread in smaller wetland areas in Bavaria (6). The massive outbreak in 2007 at the Helme reservoir at Berga-Kelbra (7) was related to four cases eastward in Saxony (8).
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2722081&req=5

pone-0006639-g003: Results of the HPAIV H5N1 monitoring.Regions with cases of HPAIV H5N1 infection in wild birds in Germany (A). Spatial distribution of the wild bird monitoring in German municipalities 2007 (samples per square kilometer) (B). After the initial outbreak on the Isle of Ruegen (1), only three cases were recorded in inland municipalities at the coastline of Mecklenburg-Western Pomerania (2). Most of the cases in the neighboring states of Brandenburg (3) and Schleswig–Holstein (4) were reported several weeks afterwards in March and April 2006. HPAIV H5N1 cases at Lake Constance (5) were connected to further spread in smaller wetland areas in Bavaria (6). The massive outbreak in 2007 at the Helme reservoir at Berga-Kelbra (7) was related to four cases eastward in Saxony (8).
Mentions: Constant temporal and spatial sampling rates are required for mobile pathogens like HPAIV H5N1 to achieve a considerable degree of confidence in the results. Analysis of the temporal distribution of the sampling within the wild bird monitoring program showed, however, that the sampling rates decreased after an initial period of intensive testing upon detection of the index case and reached a more or less constant level (Figure 3). Short-term infrequent sampling produced unbalanced datasets with high autocorrelation in time, ranging between 0 samples on some days and up to 300 on others. The variation in space exhibited a similar pattern (Figure 4). The allocation of the monitoring activities to the different German federal states led to a nearly homogeneous distribution of the sample size over the entire country at the state level Measure of overall spatial autocorrelation: (Moran's I: 0.09; C-l: 0.08-0.1; expected value: −0.00008) [20]. Yet, irregular sampling at the district or municipality level produced data with considerable spatial dependencies. Sampling often concentrated around outbreak areas and on susceptible species. This prompted for instance high sample sizes at the Baltic Sea coast and at Lake Constance in reaction to outbreaks in these locations.

Bottom Line: Although more than 25,000 birds were sampled in Germany per year, the upper 95% confidence limit did not fall below 5% in the outbreak regions for most of the time.The proposed analysis can be used to monitor water bodies and high risk areas, also as part of an early-warning system.Chances for an improved targeting of the monitoring system as part of a risk-based approach are discussed with the perspective of reducing sample sizes.

View Article: PubMed Central - PubMed

Affiliation: Institute of Epidemiology, Friedrich-Loeffler-Institut, Federal Research Institute for Animal Health, Wusterhausen, Germany.

ABSTRACT
Highly pathogenic influenza virus (HPAIV) H5N1 proved to be remarkably mobile in migratory bird populations where it has led to extensive outbreaks for which the true number of affected birds usually cannot be determined. For the evaluation of avian influenza monitoring and HPAIV early warning systems, we propose a time-series analysis that includes the estimation of confidence intervals for (i) the prevalence in outbreak situations or (ii) in the apparent absence of disease in time intervals for specified regional units. For the German outbreak regions in 2006 and 2007, the upper 95% confidence limit allowed the detection of prevalences below 1% only for certain time intervals. Although more than 25,000 birds were sampled in Germany per year, the upper 95% confidence limit did not fall below 5% in the outbreak regions for most of the time. The proposed analysis can be used to monitor water bodies and high risk areas, also as part of an early-warning system. Chances for an improved targeting of the monitoring system as part of a risk-based approach are discussed with the perspective of reducing sample sizes.

Show MeSH
Related in: MedlinePlus