Avian Influenza Outbreaks in Chickens, Bangladesh

To determine the epidemiology of outbreaks of avian influenza A virus (subtypes H5N1, H9N2) in chickens in Bangladesh, we conducted surveys and examined virus isolates. The outbreak began in backyard chickens. Probable sources of infection included egg trays and vehicles from local live bird markets and larger live bird markets.

To determine the epidemiology of outbreaks of avian infl uenza A virus (subtypes H5N1, H9N2) in chickens in Bangladesh, we conducted surveys and examined virus isolates. The outbreak began in backyard chickens. Probable sources of infection included egg trays and vehicles from local live bird markets and larger live bird markets.

T he threat that highly pathogenic avian infl uenza (HPAI)
A virus subtype H5N1 poses to poultry and public health has intensifi ed (1). As the virus becomes established in poultry in developing countries, the number of human cases increases (1,2). Countries in the Asian Association for Regional Co-operation are especially vulnerable to virus perpetuation because of insuffi cient biosecurity, rearing of chickens and ducks together, selling of live birds, and defi cient disease surveillance. To prevent human infection with avian infl uenza (H5N1), knowledge of avian infl uenza epidemiology is needed. We therefore describe the epidemiology of HPAI outbreaks in chickens in Bangladesh.

The Study
Through July 10, 2007, we investigated 52 outbreaks caused by HPAI virus (H5N1) and 3 outbreaks caused by low-pathogenicity avian infl uenza (LPAI) virus (H9N2) in chickens in Bangladesh. After a high number of chicken deaths on a farm was reported to an upazila (a lower administrative unit of Bangladesh) veterinarian, the sick chickens on the farm were examined. From each of 55 outbreaks, 2 dead chickens were sent to a fi eld disease investigation laboratory or to the Central Disease Investigation Laboratory, where oropharyngeal swabs were tested for avian infl uenza A virus antigen. From cases with positive results, tracheal samples were referred to the National Reference Laboratory for Avian Infl uenza (NRL-AI) for viral RNA extraction and purifi cation (3), reverse transcription-PCR that used a primer set of hemagglutinin (H) genes (4), and end-product visualization. When NRL-AI confi rmed H5, the farm was considered HPAI affected and was reported to the Department of Livestock Services. Tracheal samples from chickens involved in 37 outbreaks, including those that were A-antigen positive but H5 negative, were sent to the Veterinary Laboratory Agency in the United Kingdom for confi rmation. A farm on which infl uenza subtype H9N2 was found was considered LPAI affected. All farms affected with HPAI or LPAI virus were called avian infl uenzaaffected farms. A district or upazila with at least 1 avian infl uenza-affected farm was considered an infected district or infected upazila.
To collect information about the farms, we used a pretested questionnaire administered by 2 veterinarians. The form had space where veterinarians could add additional comments on the probable virus sources for infections by backward tracing (window <21 days of onset of clinical signs) and sources of spread by forward tracing (window between onset of clinical signs and culling), which they obtained by interviewing the affected farmers and allied personnel. Farm geographic coordinates were recorded. Through another questionnaire, we collected data on commercial and backyard farms and outbreaks from the upazila livestock offi ces. All avian infl uenza data stored at the Department of Livestock Services head offi ce, fi eld disease investigation laboratories, the Central Disease Investigation Laboratory, and NRL-AI were also collected.
Summary statistics were computed and plotted by using Excel (Microsoft, Redmond, WA, USA), Arc View 9.1 (Environmental Systems Research Institute, Redlands, CA, USA), and STATA 7 (Stata Corp., College Station, TX, USA). A 7-day rolling mean of avian infl uenza-affected farms, according to date of clinical onset of disease, was calculated from January 12, 2007, and plotted as a bar chart on day 4 for each value. Attack rates of farms were calculated separately for upazilas of every infected district (Table 1).
We found that the H cleavage site of the selected infl uenza subtype H5N1 isolates (determined from the Veterinary Laboratory Agency) contained polybasic amino acids, which are characteristic of HPAI A viruses ( Table  2). According to categories established by the Food and Agriculture Organization (5), 5 breeder and 2 layer farms had production system 2, 28 layer farms and 1 broiler farm had system 3, and 20 backyard farms had system 4. The 7-day rolling means of the numbers of avian infl uenza-affected farms are shown in Figure 1; the temporal and spatial spreads, in Figure 2. The index farm was recorded on January 15, 2007, at a local live bird market in Sarishabari upazila in Jamalpur district. We hypothesized that the infection probably came from chickens in nearby backyard farms because high numbers of deaths in this population went uninvestigated.
The outbreaks peaked on March 26, 2007, when 11 affected farms in 3 districts-Dhaka, Gazipur, and Narayangogj-formed a cluster, indicating a common source. The source may have been larger live bird markets, which probably infected chickens of the 15 districts. The fi rst avian infl uenza-affected backyard farm was reported on March 22, 2007, the date when avian infl uenza was confi rmed in Bangladesh. Of the 20 backyard farms, 14 were in 7 northern districts.
The overall attack rates for the upazilas of the infected districts were 6/1,000 commercial farms and 1/100,000 backyard farms (Table 1). Uninvestigated deaths of back-yard chickens could result in underestimation of the attack rate.
Among the 9 probable sources of infection, egg trays and contaminated vehicles from larger live bird markets and local live bird markets accounted for 47% of probable virus sources, eggs for 48%, and apparently healthy chickens for 5%. One avian infl uenza-affected farm disposed of ≈1,000 dead chickens in an open fi eld before diagnosis was confi rmed. For the backyard chickens, sources of spread were selling chickens (5%), giving chickens to relatives or neighbors (15%), moving birds through local poultry vendors, and hiding birds during culling operations (10%).
On the index farm, chickens in 1 shed were infected, but chickens in 2 other sheds <40 yards away remained clinically unaffected through the time of culling (71 days after clinical onset). Although the media reported that a corporate-run poultry farm, Biman Poultry Complex, was the fi rst avian infl uenza-affected farm in Bangladesh, our investigation found it to be the third. Of the 5 breeder farms, 2 had imported chicks from the United States and 1 from  (6,7), genetic variation in receptors (8)(9)(10), poor surveillance of disease in humans, or using antiviral drugs during culling of birds.

Conclusions
Our investigation showed that the epicenter of the HPAI outbreaks in Bangladesh was the Sarishabari upazila of Jamalpur district and that the primary source of infection was backyard chickens. Phylogenetic analysis on 1 infl uenza virus (H5N1) isolate showed that it belongs to the subclade 2.2 of the Qinghai lineage (11), most closely related to viruses isolated from Afghanistan, Mongolia, and Russia (11). Therefore, the virus might have entered Bangladesh through migratory birds (12)(13)(14). The presence of infl uenza virus subtype H9N2 in chickens on 3 farms, however, raises the question of when this virus was introduced to Bangladesh. An earlier introduction or emergence of LPAI virus (H9N2) in backyard chickens cannot be ruled out because ≈18% of backyard chickens tested during 2000-2003 were seropositive for avian infl uenza virus (15).
This study illustrates the progression of HPAI in Bangladesh. Further study is needed to provided more evidence for the sources we have identifi ed.