3. Results
3.1. Vaccination compliance
Between 2016 to 2019 we followed the annual vaccination compliance among 566 cattle herd owners insured by ‘Hachaklait‘ (the veterinary cooperation providing private veterinary services to the farmers). As depicted in Figure 3, we observed a reduction in vaccination against LSD from 61% in 2016 to 27% in 2019. We also found that farmers who withdrew from vaccination were less likely to vaccinate in the following years. Farmers who vaccinated their herd only in 2017 were more than 8times more likely to vaccinate their herd in 2019 as compared to the farmers who didn’t vaccinate their herd in 2017 and 2018. The difference between farmers who didn’t vaccinate in both years to farmers who vaccinate in both years was even more striking, where the latter was 120 times more likely to vaccinate in 2019 (Table 1 ). A strong association was also found between vaccination and the geographical location of the herd; farmers whose herd was located in the centre of Israel were 20 times less likely, to vaccinate than farmers whose herd was located in the north (Table 1 ).
3.2. Analysis of intention to vaccinate
To find the causes for this reduction in vaccination compliance we distributed 90 questionnaires (Figure 1 ), based on the TPB model, among Israeli dairy farmers (see detailed description in the methods section). These questionnaires were analysed in a two-stage approach. We first wished to characterize the factors determining the farmers’ intention to vaccinate as a proxy for vaccination behaviour. At the next stage, we aimed to explore which of these factors was both likely to change with time and influence vaccination compliance.
Before the characterization of the factors influencing intention, we wished to test the assumption that in our data the intention to vaccinate represents vaccination behaviour. For this purpose, we used the 56 questionnaires for which we had data on voluntary vaccination (Figure 1, Data-intersect 1 ). As expected, analysis of these questionnaires showed that vaccination among farmers who claimed they are likely to vaccinate was 4.5 higher than among other farmers (p-value=0.007). Analysis of 33 of these farmers for which we had data on future behaviour showed a similar result though only marginally significant (p=0.067). Surprisingly, despite this strong association, only 19 out of the 38 (50%) farmers who claimed they are likely to vaccinate indeed vaccinated their herd (Figure 1 ). The vaccinating farmers showed a stronger belief in the ability of the vaccine to reduce clinical signs of LSD (Table 2 ). However, the most prominent difference between vaccinating and non-vaccinating farmers was recorded in the indirect measurement of the PBC. Specifically, these farmers were less concerned about the vaccine’s cost and the time and manpower required for vaccination (Table 2 ).
We next analyzed the entire set of 90 questionnaires to identify the factors that determine vaccination intention. Among the background factors, farmers whose herds were located in the centre of Israel and felt lower perceived risk regarding the frequency and consequences of an LSD epidemic showed less intention to vaccinate their herds (Table S2a,b ). Furthermore, as expected, higher vaccination intention was associated with both direct and indirect measurements of the three RAA constructs: attitude, subjective norms and perceived behavioural control (Table S3 ). Among the attitude factors, the highest influence was recorded for the perceptions regarding the benefit from vaccination, reduction of the severity of clinical signs, receiving money from the insurance company, peace of mind and reducing the number of disease cases. Farmers’ intention to vaccinate was significantly influenced by all the nine investigated subjective norm factors. Finally, among the PBC factors, the most influential were the vaccine’s cost and information on the vaccine’s effectiveness (Table S3 ).
At the following stage, we attempted to determine which of these factors have changed as time elapsed from the LSD epidemic in 2012 and might have caused the reduction in vaccination during 2016-2019. As explained in the ‘Materials and Methods’, since we could not measure the actual change of beliefs during this period we used the comparison between farmers’ answers before and after the 2019 epidemic as a proxy for the farmer’s change of beliefs shortly after the epidemic and long after an epidemic.When comparing these two groups, we found that after the 2019 epidemic, the farmers showed higher intention to vaccinate (Table 3). Furthermore, vaccination norms among farmers who answered the questionnaires before the epidemic in 2019 were significantly lower than the norms among farmers answering the questionnaires after the epidemic (Table 3 ). Specifically, after the epidemic the farmers perceived a stronger recommendation for vaccination by the ’Hachaklait‘ and the veterinary services (Table 3 ). In addition, after the epidemic, the farmers were more concerned about the vaccine’s adverse reactions although the overall attitude was not significantly different between the two groups (Table 3 ). These findings suggest a possible connection between a post epidemic change in these specific norms and a change in vaccination intention.
The multivariable linear regression revealed four statistically significant factors associated with farmers intention to vaccinate: the time of questionnaire filling (higher for after the 2019 epidemic), the farm location (lower in the centre of Israel) and two beliefs: The farmer’s perceived recommendation to vaccinate by the private vet, and the farmer belief that the vaccine will be economically worthwhile. (Table 4 ).