4.2 Materials and methods
4.2.2 Study period
The study was conducted over a period of 17 months, from 1 December 2008 to 3 April 2010 (inclusive). Flocks were visited on 14 occasions termed ‘sampling rounds’ in the re- mainder of this paper. Sampling rounds were conducted during the two high risk periods (rounds 1 to 5 between December 2008 and March 2009 and rounds 10 to 14 between De-
4.2 Materials and methods 41 cember 2009 and April 2010) and one low risk period (rounds 6 to 9 between March 2009 and November 2009). The distinction between low and high risk periods was based on a frequency analysis of previous AI outbreaks (Pfeiffer et al. 2007, Minh, Morris, Schauer, Stevenson, Benschop, Nam & Jackson 2009) and the prevalence of HPAI H5N1 circu- lation under the national surveillance programmes (Taylor & Dung 2007, Long 2008). Sampling was carried out at 28-day intervals during the two high risk periods and at 70- day intervals during the low risk period.
4.2.3
Collection of samples
Oropharyngeal and cloacal swab samples were collected from all study birds during each sampling round. The two swabs from each bird were mixed together in a single 15 mL tube containing 2.5 mL viral transport media. Viral transport media was validated by a sub-experiment to evaluate factors influencing the survival of AI viruses. Samples were kept in cool boxes at 10–12◦C. These were either transported to provincial offices where they were refrigerated at 4 ◦C for a maximum of two days and then transported to the laboratory, or were transported directly to the regional animal health laboratory (RAHO7) on the day of sampling. Blood samples were collected from birds that were greater than four weeks of age at the time of sampling.
4.2.4
Laboratory procedures
Individual swab samples were mixed by vortexing. The mixed material was then cen- trifuged at 3,500 rpm for 5 minutes. The supernatant solution was transferred into two tubes of 1.5 mL (one aliquot was used for virus isolation and the other used for RRT- PCR M gene detection). The aliquot used for virus isolation was stored at -80◦C, while the aliquot used for RRT-PCR was stored at -20 ◦C. Five of the individual RRT-PCR aliquots were then pooled and tested using M gene RRT-PCR. Four pools per flock were constructed at each sampling round (Figure 4.2).
Serum was treated at 56◦C for 30 minutes to inactivate non-specific inhibitors of haemag- glutination. Treated serum samples were stored at 4◦C prior to further testing or at -20 ◦C for long-term storage.
The USDA-validated M gene RRT-PCR (Spackman et al. 2002, NVSL 2008), customised by the Vietnamese Department of Animal Health (MARD 2005a) was used to detect the M gene of AI viruses. The RRT-PCR was firstly used as a screening test for pooled samples. Following screening all individual samples from RRT-PCR positive and suspect pools were further tested for M gene using the same RRT-PCR technique.
RNA was extracted using the MagMAX-96 AI/ND RNA Isolation Kit (Ambion). The Quiagen one-step RT-PCR kit (QIAGEN 2008) was used to prepare the master mix, and amplification and fluorescence detection were carried out using a BioRad real-time PCR cycler. After an initial reverse transcription step at 50 ◦C for 30 minutes and an initial denaturation step at 95 ◦C for 15 minutes, 45 cycles (94 ◦C for 15 seconds followed by 60◦C for one minute) were performed with fluorescence detection at the end of the annealing-extension step. Samples were declared positive when the cycle threshold (Ct) for the sample was less than 40.
The H5 Multiplex assay used in this study was developed and validated by the Australian Animal Health Laboratory (Australian Animal Health Laboratory 2010). The Multiplex TaqMan assay consists of two sets of primers and probes targeting two different regions (C-terminus and N-terminus) of the HA gene of H5 viruses.
The Quiagen one-step RT-PCR kit was used to prepare the master mix and amplification and fluorescence detection were carried out on a BioRad real-time PCR cycler. After an initial reverse transcription step at 45◦C for 10 minutes and an initial denaturation step at 95◦C for 10 minutes, 45 cycles (95◦C for 15 seconds followed by 60◦C for 45 seconds) were performed with fluorescence detection at the end of the annealing-extension step. Positive controls consisted of the H5 antigens that had been designed from Vietnamese H5 viruses. Negative controls were RNase-free water. Samples were declared positive when the cycle threshold for the sample was less than 40.
The HI test was used to detect the amount of antibody against H5 viruses. The protocol for this test is described in the OIE Manual of Diagnostic Tests and Vaccines for Terres- trial Animals (OIE 2009) and has previously been customised for use under Vietnamese laboratory conditions (MARD 2005a). Ten dilutions were performed to titrate the quan- tity of the specific antiserum in each serum sample. Both positive and negative control antigens and antisera were used in each HI testing round. HI titres were measured based on the inhibition serum dilutions, which were positive if the dilutions were at least 1/16.
4.2 Materials and methods 43
4.2.5
Data management
Test results were recorded for each study bird following each sampling round. All test results were entered into a commercial spreadsheet package by laboratory staff. These data were then imported into a relational database (subsequently referred to as the labora- tory database), which also incorporated individual bird-level details recorded at the time of sampling.
4.2.6
Statistical analyses
The period prevalence and incidence rate of influenza Type A and H5 virus infection were estimated at both the bird and flock level. Period prevalence for a given time interval (e.g. the low and high risk periods or the entire study period) was defined as the number of birds (or flocks) positive at the start of the interval plus the number of birds (or flocks) identified as positive throughout the interval divided by the total number of birds (or flocks) that were tested during the interval. A flock was declared positive if it had at least one bird that was positive to the M gene RRT-PCR or H5 Multiplex assay. Incidence rate was defined as the total number of newly infected birds (or flocks) divided by the total number of bird (or flock)-months at risk for that interval.
Standard errors for the period prevalence and incidence rate estimates respecting the mul- tistage sampling design were calculated using the survey package (Lumley 2004) imple- mented within R (R Development Core Team 2012). Sampling weights (Lohr 2009) were calculated as the inverse of the probability of selection. At the flock level the probability of selection was equal to (the number of selected villages divided by the total number of villages within the selected districts)×(the number of selected flocks divided by the total number of flocks within the selected villages). At the individual bird level the probability of selection was equal to the flock level probability of selection multiplied by (the total number of study birds divided by an estimate of the total number of birds within the study flocks). Our estimates of the standard errors of period prevalence and incidence rate did not account for the repeated sampling of individual birds over time. Our reasons for this were firstly because of the complexity of having to deal with birds joining and leaving the study population throughout the study period and secondly because accounting for levels beyond the first stage strata (i.e. district) and first stage cluster (i.e. village) has minimal
effect on the sample variance estimates (Wolter 2007). Using this approach sampling was assumed to have occurred with replacement at the final sampling stage (i.e. flock, for the flock-level analyses and bird for the individual bird-level analyses).
A scatter plot was constructed to represent changes in Ct value estimates as a function of calendar date, stratified by village. Superimposed on each scatterplot was a line of best fit and its confidence interval calculated using a local polynomial regression method.