To the Editor: The highly pathogenic strain of avian influenza virus subtype H5N1 presents a major challenge to global public health systems. Currently, influenza (H5N1) infection is a zoonosis with a 60% case-fatality rate for affected persons over 3 continents; the virus could mutate to become directly transmissible among humans (1). This potential for pandemic transmission must be reduced through early detection of transmission foci, followed by rapid implementation of control measures (2). In the following analysis, we demonstrate that single carcasses of birds, mostly found by members of the public, were the primary indicators for avian influenza virus activity in Sweden and Denmark in 2006.

Influenza virus (H5N1) is amplified by commercial and backyard poultry and free-ranging birds. Whether captive birds (e.g., poultry) or wild birds are responsible for the spread of the virus remains a matter of debate (3). Initial spread from Southeast Asia before 2005 was likely the result of transport of infected poultry because the spread was not easily explained by natural bird movements (4,5). However, its spread to Western Europe in late 2005 could be explained by weather-induced migration of waterfowl after a freeze in Eastern Europe (6,7). Since spreading to Sweden and Denmark in early 2006, the virus has been detected there in dead birds of numerous species (Table). Detections in carcasses of primarily free-ranging birds have become the principal means of tracking spread of the virus in Europe.

To better understand how avian mortality surveillance could be refined for monitoring the spread of influenza virus (H5N1), we analyzed the weekly official reports of such detections in Sweden and Denmark in 2006 (8). Virus surveillance in both countries includes both active cloacal swabbing of free-ranging wild birds and passive collection of tracheal swabs from bird carcasses. For the analysis, all carcasses of a single species collected on 1 day within a single locality constituted 1 record. For each record, we evaluated whether the carcass(es) were reported by a member of the public versus a civil servant, the number of carcasses tested, and the number of positive detections.

Our analysis evaluated 44 records; a total of 70 birds, of 14 species, tested positive for the virus in 22 localities of Sweden and Denmark. Almost all of these records (n>40, 91%) referred to dead birds found by members of the public rather than civil servants. A smaller portion than expected were Anseriformes (i.e., ducks, geese, or swans; n = 32, 73%). Other orders of birds represented were Falconiformes (hawks, falcons; n = 8, 18%), Strigiformes (owls; n = 2, 5%), Podicepidiformes (grebes; n = 1, 2%), and Charadriiformes (gulls, shorebirds; n = 1, 2%). In addition, birds of other orders tested positive in Denmark but were excluded from the analysis for lack of supporting data. Most (75%) of the records referred to singleton carcasses; the remaining 25% represented multiple detections, ranging from 2 to 9 individual birds of a single species. A majority (73%) of influenza virus (H5N1)–positive localities hosted solely singleton carcasses, whereas the other 27% hosted multiple dead birds. No virus activity was detected through active free-ranging bird surveillance, even though 9,260 live birds were captured and sampled during 2006 in Sweden and Denmark.

The pattern of virus activity observed in Sweden and Denmark was unexpected. Rather than die-offs of large numbers of waterfowl during winter when they congregate, small numbers (mainly singleton birds) were affected late in winter, just before spring migration. During the spring breeding season, less transmission was observed. The predictive power of detecting the virus in free-ranging migratory birds for forecasting poultry outbreaks or human disease remains undetermined. Some of these birds may have been infected in areas remote from the site of detection. However, several of the affected birds in this report were either resident nonmigratory species (eagle owl, Eurasian magpie) or captive domesticated species (muscovy, peafowl, chicken), which indicates local transmission. Health authorities will be better prepared to prospectively minimize transmission in new regions with early warning provided by singleton carcass surveillance.

Surveillance results from Sweden and Denmark highlight the importance of public participation in avian mortality surveillance for influenza virus (H5N1); the preponderance of detections from singleton carcasses; and the broad spectrum of affected species, particularly raptors. A raptor was the index case in Denmark (7). Current surveillance efforts in regions free from the virus favor investigation of significant death events of waterfowl and active sampling of healthy waterfowl as the means for early detection (e.g., 9). Many national surveillance programs are heavily influenced by the influenza virus (H5N1) outbreak in 2005 at Qinghai Lake in China, where hundreds of geese, gulls, and cormorants died during the breeding season (10). However, large die-offs may be anomalous or restricted to communal breeding sites of waterfowl where juvenile birds amplify and spread the virus within the breeding colony. Testing of public-reported singleton carcasses provides a more sensitive and robust means of early detection of this virus.