Detection of swIAV by RT-qPCR in the 25 enrolled farms
Detection of Influenza A virus by RT-qPCR in at least one sample was possible in 20 (80%) out of the 25 enrolled farms. In the remaining five farms swIAV could not be detected by RT-qPCR, however, antibodies against swIAV were measurable by hemagglutination inhibition test and ELISA in all five farms (data not shown).
Detection of swIAV in different specimen in the 20 RT-qPCR positive farms
Subsequently, results are restricted to the 20 RT-qPCR positive farms. A total of 26.1% (218/834) of all available samples, independent of the sampling material, revealed positive RT-qPCR results. Details on positive sampling materials and age groups in each of the 20 RT-qPCR positive farms can be found in supplementary table 1. The detection rate of swIAV RNA in the different sampling materials is shown in figure 1A for sows and suckling piglets and in 1B for nursery pigs. Cochran’s Q test revealed significant differences in the detection rate between the distinct sampling materials in suckling piglets (p= 0.034) and in nursery pigs (p<0.001), although the post-hoc test suggested no pair-wise differences between sampling materials (p>0.05) in suckling piglets. In the nursery, pair-wise comparisons have shown significant differences between nasal swabs and dust wipes (p=0.001), nasal swabs and oral fluids (p<0.001), contact wipes and dust wipes (p=0.047), contact wipes and tracheobronchial swabs (p=0.028), contact wipes and oral fluids (p<0.001), dust wipes and tracheobronchial swabs (p=0.001) and tracheobronchial swabs and oral fluids (p<0.001), respectively. No statistically significant differences were observed between nasal swabs and contact wipes (p=0.89), nasal swabs and tracheobronchial swabs (p=1.00) and dust wipes and oral fluids (p=0.23). In sows, no significant differences in the detection rate between the sampling materials could be shown according to Cochran’s Q test, however pairwise comparison revealed significant differences between contact wipes and dust wipes (p=0.043).
Probability of swIAV detection varies with sample matrix
The probability of sample materials testing positive for swIAV by RT-qPCR was investigated through a multivariable analysis, within each age group, using equally numbered three-level mixed effect logistic regression models. Differences in RT-qPCR detection rate between the different sampling materials are presented in table 2. In suckling piglets, nasal swabs (p=0.007) and udder skin wipes (p=0.036) showed a significant higher odds of detecting swIAV by RT-qPCR compared to contact wipes, whereas in the nursery the odds to detect swIAV by RT-qPCR were significantly higher in group samples (dust wipes (p<0.001; p<0.001), contact wipes (p=0.002), oral fluids (p<0.001; p<0.001) compared to individual samples (nasal swabs, tracheobronchial swabs) with the exception of the comparison between contact wipes and nasal swabs (p=0.181).
Table 2. Odds ratio (OR) of finding a positive result by sample type using a multivariable analysis, accounting for the within pooled sample and farm-level dependence of observations compared to the respective sampling materials as baseline.
Age group
|
Sample type
|
Reference
|
OR with 95%CI
|
p-value
|
Sows
|
CW NS
|
DW
|
2.76 (0.85; 8.98) 2.76 (0.85; 8.98)
|
p=0.091 p=0.091
|
|
NS
|
CW
|
1.00 (0.29; 3.43)
|
p=1.000
|
Suckling piglets
|
NS
|
CW
|
9.44 (1.86; 48)
|
p=0.007
|
USW DW
|
5.13 (1.12; 23.53) 2.34 (0.52; 10.51)
|
p=0.036 p=0.266
|
DW USW
|
NS
|
0.25 (0.05; 1.15) 0.57 (0.13; 2.51)
|
p=0.075 p=0.461
|
DW
|
USW
|
0.43 (0.097; 1.92)
|
p=0.270
|
Nursery
|
DW CW OF
|
NS
|
5.92 (2.67; 13.09) 1.68 (0.78; 3.59) 13.52 (5.70; 32.09)
|
p<0.001 p=0.181 p<0.001
|
TBS
|
|
0.48 (0.21; 1.07)
|
p=0.073
|
CW
|
TBS
|
3.50 (1.56; 7.83)
|
p=0.002
|
DW
|
12.32 (5.25; 28.13)
|
p<0.001
|
OF
|
28.17 (11.03; 71.92)
|
p<0.001
|
DW
|
CW
|
3.52 (1.65; 7.55)
|
p=0.001
|
OF
|
8.05 (3.54;18.35)
|
p<0.001
|
OF
|
DW
|
2.28 (1.06; 4.91)
|
p=0.034
|
Level of agreement between different specimen
The agreement concerning the detection of swIAV RNA between the different sampling materials resulted fair for nursery pigs (Scott/Fleiss' κ = 0.392; p<0.001) and moderate for suckling piglets (Scott/Fleiss' κ = 0.588; p<0.001) and sows (Scott/Fleiss' κ = 0.483; p<0.001). The highest agreement in both sows and suckling piglets was observed between contact wipes and dust wipes, whereas in the nursery nasal swabs and contact wipes showed the highest agreement. The agreement between the different sampling materials within each age group is shown in table 3.
Table 3. Scott/Fleiss’s kappa coefficient (κ) for the agreement of different sampling materials at sample level for the binary variable (positive or negative result of diagnosis of Influenza A) within age groups. Whenever a statistically significant association was detected (p<0.05) all pairwise comparisons were evaluated following Bonferroni's correction (pcorrected=0.05/n), where n represents the number of tests or comparisons performed.
|
|
|
Sows
|
Suckling piglets
|
Nursery
|
Material
|
|
CW
|
DW
|
CW
|
DW
|
USW
|
CW
|
DW
|
OF
|
TBS
|
NS
|
kappa (p-value)
|
0.298 (<0.001)
|
0.377 (<0.001)
|
0.330 (0.003)
|
0.504 (<0.001)
|
0.744 (<0.001)
|
0.647 (<0.001)
|
0.391 (<0.001)
|
0.360 (<0.001)
|
0.434 (<0.001)
|
CW
|
kappa (p-value)
|
-
|
0.733 (<0.001)
|
-
|
0.827 (<0.001)
|
0.487 (<0.001)
|
-
|
0.438 (<0.001)
|
0.423 (<0.001)
|
0.533 (<0.001)
|
DW
|
kappa (p-value)
|
-
|
-
|
-
|
-
|
0.666 (<0.001)
|
-
|
-
|
0.457 (<0.001)
|
0.181 (0.008)
|
OF
|
kappa (p-value)
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
0.128 (0.023)
|
Likelihood of subtyping swIAV strains by multiplex RT-qPCR depends on sample matrix
Based on previous publications and due to lower sensitivity of subtype- and lineage-specific RT-qPCRs compared to the generic IAV RT-qPCR used for initial diagnosis (19, 29), samples with Ct-values of <32 were selected for subtyping. In total 48.6% (n= 106/218) of the positive samples had Ct-values <32. Of these, subtyping was successful in 97.2% (n=103/106). Referring to the sample type, 66.7% (n=42/63) of the positive nasal swab pools, 52% (n=13/25) of positive contact wipes, 2.8% (n=1/36) of positive dust wipes, 20% (n=3/15) of positive udder skin wipes, 52.9% (n= 36/68) of positive oral fluids and 28.6% (n=4/14) of positive tracheobronchial samples had Ct-values32, respectively. The percentage of samples with Ct-values 32 in the different specimen and age groups are depicted in Figure 2.
According to multivariable analysis, nasal swabs showed a significantly higher odds of having Ct-values 32 compared to dust wipes (p<0.001), contact wipes (p=0.026), udder skin wipes (p=0.001) and tracheobronchial swabs (p<0.001) (table 4). Interestingly, dust samples showed the lowest probability of detecting samples with Ct-value results 32.
Table 4. Odds ratios (OR) of finding a sample with Ct-value 32 by sample type using a multivariable analysis, accounting for sample type, age group and within farm-level dependence of observations compared to the respective sampling materials as baseline.
Sample type
|
Reference
|
OR with 95%CI
|
p-value
|
NS
CW
|
DW
|
144.03 (26.93; 770.23)
53.77 (10.71; 269.85)
|
p<0.001
p<0.001
|
OF
|
71.41 (13.27; 384.36)
|
p<0.001
|
TBS
|
|
13.42 (2.38; 75.58)
|
p=0.003
|
NS
CW
|
USW
|
18.73 (3.27; 107.34)
7 (1.22; 40.18)
|
p=0.001
p=0.029
|
OF DW TBS
|
9.28 (1.42; 60.52) 0.13 (0.02 1.09) 1.74 (0.25; 12.34)
|
p=0.020 p=0.060 p=0.577
|
NS
OF
CW
|
TBS
|
10.73 (3.05; 37.70)
5.32 (1.57; 17.97)
4 (1.26; 12.73)
|
p<0.001
p=0.007
p=0.019
|
CW OF
|
NS
|
0.05 (0.01; 0.03) 0.49 (0.21;1.20)
|
p=0.026 p=0.119
|
OF
|
CW
|
1.33 (0.57;3.10)
|
p=0.512
|
In all 20 swIAV positive farms, one or more subtypes could be identified. The subtypes detected in this study were H1avN1, H1avN2, H1pdmN1, H1pdmN2, H1huN1 and H1huN2. The lineage differentiation in the H1 subtype indicates the original sources of the hemagglutinin (HA) as avian (av), human (hu) or human pandemic (pdm). They correspond to the phylogenetically defined clades 1C (av), 1B (hu) and 1A (pdm) (30). A stratified overview of the detection of distinct subtypes in the different specimen is shown in table 5.
Table 5. Number of subtypes (n) detected in the different specimen.
|
H1avN1
|
H1avN2
|
H1pdmN1
|
H1pdmN2
|
H1huN1
|
H1huN2
|
Total
|
NS
|
26
|
7
|
5
|
4
|
4
|
1
|
47
|
CW
|
10
|
1
|
0
|
0
|
1
|
0
|
12
|
DW
|
0
|
0
|
0
|
1
|
0
|
0
|
1
|
USW
|
2
|
0
|
1
|
0
|
0
|
0
|
3
|
OF
|
26
|
4
|
0
|
6
|
0
|
0
|
36
|
TBS
|
4
|
0
|
0
|
0
|
0
|
0
|
4
|
Total
|
68
|
12
|
6
|
11
|
5
|
1
|
103
|
In 70% (n=14/20) of the swIAV RT-qPCR positive farms only a single subtype and in 30% (n=6/20) two or more different subtypes could be detected, respectively. In 4 out of the 6 farms with multiple subtypes, distinct subtypes were found in different age groups. In addition, in 3 out of the 6 farms, different subtypes were detected in different sampling materials (supplementary table 2). Interestingly, in farms with multiple subtypes the pandemic subtype H1pdm was only found in nasal swabs.
Likelihood of detection of swIAV and swIAV subtype characterization varies with age group
Suckling piglets were 7.9 times (95% CI: 3.24; 19.28, p<0.001) more likely to have a positive result in nasal swabs compared to sows. Nasal swabs from nursery pigs were almost 9 times (95% CI: 4.07; 19.77, p<0.001) more likely to have a positive result compared to sows. No difference was observed in the odds of positive results from nasal swabs between suckling and nursery piglets (p=0.724). Accounting for all sampling materials, piglets at the beginning and at the mid of the nursery were more likely to have a positive RT-qPCR result compared to piglets at the end of the nursery, by 4.45 times (95%CI: 1.58; 4.50, p<0.001) and 2.67 times (95% CI: 1.36; 7.32, p<0.001), respectively. Suckling piglets were 152.87 times (95% CI: 6.16; 3792.21, p=0.002) and nursery pigs 19.12 times (95%CI: 1.61; 226.85, p=0.019) more likely to yield nasal swabs with Ct-values <32 compared to sows. However, no statistically significant difference was observed in the odds of a Ct-value results <32 between suckling piglets and nursery pigs (95% CI: 0.014; 1.13, p=0.065). In addition, no significant difference in the odds of a subtypeable RT-qPCR result was found between beginning and mid nursery piglets (p=0.287), beginning and end nursery piglets (p=0.897) and between mid and end nursery piglets (p=0.406).