Experiments were carried out in the high containment unit at PT. Medion, Bandung, Indonesia. Four experiments were done each with 22 replicates. Each replicate consisted of one pair of broilers. Experiment 1 was carried out with unvaccinated birds, experiment 2 with birds vaccinated at day 1 of age, and experiment 3 with birds vaccinated at day 10. Experiment 4 consisted of progeny obtained from the same flock, but birds were day-old at time of challenge (see below).
Animals and housing
Approximately 200 18-day-old embryonated eggs were purchased from a commercial breeder farm. The breeders were vaccinated several times against H5N1 with Medivac® (PT Medion, Bandung, Indonesia), containing H5N1 virus strain A/chicken/Legok/2003. At time of purchase of the eggs, no clinical signs of AI in the breeders were reported, indicating the absence of HPAI H5N1 virus at that time.
After hatching at the facilities, day-old chicks were housed in one experimental unit. They were fed with a commercial ration, and had tap water ad libitum. Four groups were formed each consisting of 44 birds for each experiment one. Three groups (experiments 1-3) were of the same age at purchase and challenge and were used to determine the efficacy of vaccination. A fourth group (experiment 4) consisted of day-old chicks (DOC), and were challenged to determine whether virus could be transmitted amongst DOC. Challenge was done when birds in experiments 1-3 were 28 days old.
One week before challenge birds from experiments 1-3 were moved to the experimental units. Two rooms were available, each with two rows with cages on three levels. Birds in experiments 2 and 3 were housed together in Unit 1 and birds in experiment 1 were housed together with those in experiment 4 in Unit 2: each experimental group at opposite sides of the corridor in each house. In each cage one pair of birds from the same experiment was housed. The cages between each experimental pair were empty. Sentinel birds were placed in empty cages in the middle level below each pair in the upper level. Sentinels were SPF layers, from the SPF unit of PT Medion, and were not older than the experimental birds. Sentinels were used to monitor between-cage virus transmission. The experiments lasted four weeks after inoculation, when the surviving birds in experiments 1-3 were 56 days old.
Vaccine
An inactivated oil-emulsion vaccine was used, which contained the H5N1 strain A/Chicken/Legok/2003 (Medivac®, PT Medion, Bandung, Indonesia) [26]. The vaccine was administrated intramuscularly in one leg using 0.5 mL containing 256 HAU per dose per bird. Chickens in experiments 1 and DOC in experiment 4 remained unvaccinated; chickens in experiment 2 were vaccinated at one day of age; the birds in experiment 3 were vaccinated at 10 days of age.
Inoculation
The HPAI virus strain H5N1 A/Chicken/Legok/2003 was used for challenge. The strain was provided by PT. Medion Bandung Indonesia. The virus has been used in other transmission experiments, and was able to induce clinical signs and transmission [26, 27]. At day of challenge, when the birds in experiments 1-3 were 28 days old and in the DOC experiment (4) were 1 day old, one bird per pair was inoculated intranasally and intratracheally with 0.2 mL inoculum containing 106/mL median egg infectious dose (EID50) [26]. Before inoculation each bird that had to be inoculated was put in the empty cage near its pen mate. Eight hours after inoculation, they were placed back in their original cage.
Transmission experiments
Throughout, we refer to each experimental pair of chickens as a trial. Each experiment consisted of 22 replicate trials, and in each trial one inoculated bird (I) was placed in a cage with an uninfected contact bird (C). Both birds had received the same treatment, and were of the same age.
Transmission of virus was monitored by taking daily swab samples from the trachea and cloaca from all birds for 10 days. From birds that survived this sampling period, additional samples were taken at day 14 after challenge. The samples were stored at -70 °C until further testing. Serum blood samples were taken from surplus birds at day of hatch to determine the level of maternally derived antibodies. From the experimental birds, serum blood samples were taken two days before challenge and four weeks after challenge, at the end of the experiment. Sera were stored at -20°C until further testing. Clinical signs were recorded during four weeks after challenge.
The treatment is referred to as: unvaccinated (experiment 1), d1 vaccinated (experiment 2), d10 vaccinated (experiment 3) and DOC (experiment 4). Figure 1 gives an overview of the experimental data from days 0 to 10 after challenge. Additional samples were taken from infectious birds that were still present at day 10 of the experiment (not shown). The complete dataset is available on request from the corresponding author. All experiments were carried out in accordance to article 80 on "Research in Animal Health" of the Indonesian "Law on Livestock and Animal Health UU/18/2009".
Tests
The presence of AI virus in swabs was determined by virus isolation according to standard procedure described by OIE [28]. Briefly, three SPF embryonated chicken eggs, incubated for nine days, were inoculated with 0.2 mL swab medium per egg. After 72 h, or when the embryo had died before that time, the allantoic fluid was harvested. A haemagglutination assay (HA) was performed following standard procedure. When at least one of the eggs was positive in the HA, the swab was considered to be positive. The test results were recorded as positive for AI virus or negative [26]. A bird was considered infected if at least one sample (either tracheal or cloacal) tested positive at least once.
Serum samples were tested in a haemagglutination inhibition (HI) test according to standard procedures [28]. It is generally assumed that HI titers ≥ 32 are protective against disease, i.e. clinical signs [29]. Tests were carried out in duplo using 4 HAU of the strain A/chicken/Legok/2003(H5N1). Two-fold dilutions of the serum samples were made, and titers were expressed as the serum dilution that caused complete inhibition of agglutination [28].
Quantification of transmission
A Bayesian inferential framework based on a stochastic SEIR (susceptible-exposed-infected and infectious-removed) epidemic model was used to obtain quantitative estimates of the parameters of interest [26, 30–32]. The methods of analysis have been described in detail earlier [26]. Here we give a concise overview.
The main interest is in the transmission rate parameter, and parameters of the latent and infectious period distributions. Together, these parameters determine the basic reproduction number and the generation interval [26]. We parameterize the latent and infectious periods using gamma distributions, and assume uninformative uniform prior distributions (U(0.0001-100)) for all parameters. To be precise, we characterize gamma distributions of the latent and infectious periods by their mean and variance (and not the shape and scale parameters), and assume uniform prior distributions for the mean and variance (and not the shape and scale parameters).
In the following β denotes the transmission rate parameter, γ
E
and δ
E
the parameters determining the latent period probability distribution, and γ
I
and δ
I
the parameters of the infectious period probability distribution. Specifically, E[T
E
] = γ
E
and, Var[T
E
] = δ
E
, and E[T
I
] = γ
I
and Var[T
I
] = δ
I
represent the means and variances of these distributions. The corresponding probability densities are denoted by f
E
(x) and f
I
(x).
Further, e
k
i
k
, and r
k
are N- dimensional vectors which contain the time points of the S→E, E→I, and I→R transitions for inoculated (k = 1) and contact (k = 2) birds in the N trials. Hence, we have e1 = (0,...,0) T by definition, and all other transition times are unknown. The unknown transitions are imputed. We adopt the convention that e2jdenotes the exact time at which the contact bird in experiment j is infected, that i1jdenotes the exact time that the inoculated bird in experiment j became infectious, etc.
As in [26], the contribution of trial j to the likelihood is given by
In the above equation λ(j)(t) and S(j)(t) denote the infection hazard in trial j at time t and the probability that the contact bird in trial j remains uninfected up to time t, respectively. If we let [...] denote the indicator function, the infection hazard is given by
where the parameter t
add
represents the delay between the moment of inoculation and the moment that the inoculated birds were placed back in their cages. Hence, the function max(t
add
,i1j) marks the beginning of the at-risk period for the contact bird. In all trials and experiments, the delay is 8 h, i.e. t
add
= 0.33 (day). The probability that the contact bird in trial j remains uninfected up to time t can be expressed in terms of the infection hazard as follows
With the above preparation at hand, the likelihood function is given by the product of the contributions of the individual trials given in equation 1. The above equations are furthermore readily generalized to include differences in the epidemiological parameters of inoculated versus contact birds [26].
The epidemiological parameters and unobserved epidemiological transitions (i.e. S→E, E→I, I→R) were all updated by a random-walk Metropolis algorithm. We used Normal proposal distributions with the current value as mean. After running a number of analyses to explore the posterior distribution and optimize the proposal distributions, we used standard deviations of 0.02 for the epidemiological transitions, and 0.02-0.5 for the epidemiological parameters. The epidemiological parameters and unobserved transitions were updated in blocks, in the order (1) timing of inoculated chickens becoming infectious, (2) timing of removal of inoculated chickens, (3) timing of infection of contact chickens, (4) timing of contact chickens becoming infectious, (5) timing of removal of contact chickens, and (6) updating of the epidemiological parameters [26]. Chains were run for 350 000 cycles, of which the first 100 000 cycles were discarded as burn-in. Thinning was applied by taking output from each twentieth cycle, yielding a sample of 12 500.
Below we report not only the basic epidemiological parameters (transmissibility, duration of the latent and infectious periods), but also the generation interval and basic reproduction number. The generation interval is defined as the moment of infection of the contact bird relative to the moment of infection of the inoculated bird (i.e. it is given by e2jin trial j if the contact bird was infected), while the basic reproduction number is defined as the product of the transmission rate parameter (unit: day-1 ) and infectious period (unit: day).
Each of the Experiments 1-4 was analysed separately, assuming a common distribution of the latent period of inoculated and contact birds. Based on the results of the separate experiments, and given the observation that there may be differences between inoculated and contact birds, possibly due to differences in the inoculum size we also analysed the combined data of Experiments 1-3 while relaxing this assumption. Specifically, we allowed the mean of the latent period to differ between inoculated and contact birds, while assuming a fixed common variance (0.001) of the latent periods [26]. Furthermore, the data of Experiments 1-3 were used to explore, by means of logistic regression, whether the probability of infection could be dependent on the immune status (i.e. HI titer) of the birds just prior to the experiments.