# Standardized Methodology for Target Surveillance against African Swine Fever

^{1}

^{2}

^{3}

^{4}

^{*}

^{†}

## Abstract

**:**

^{2}grids (a total of 3953 grids). Variables related to WB density, ASF cases during the last three years, sex and age of animals, and the type of land were associated with each grid. Epidemiological models were used to identify the areas with both a lack of information and an high risk of hidden ASFV persistence. The results led to the creation of a graphic tool providing specific indications about areas where surveillance should be a priority.

## 1. Introduction

## 2. Materials and Methods

#### 2.1. Sardinian Epidemiological Landscape of ASF in 2019–2020

^{2}). As previously demonstrated, both the viral prevalence and seroprevalence have drastically decreased over the last five years across the wild boar infected zone (ZI) [22]. Otherwise, the main surveillance for ASF is limited to hunting activities during November–January, while the active searching of carcasses is almost absent, and passive surveillance is limited to wild boar roadkill [22]. More precisely, from the last PCR-positive detection until October 2020, a total of 10,869 wild boar were tested for ASF: 95% (10,360) were observed through hunting, while 5% (509) were found via passive surveillance. Most of the passive surveillance samples (56%) were concentrated during the same months as the hunting season (November–January). As shown in Figure 1, the current epidemiological context indicates the complete absence of ASFV and the presence of seropositive animals limited to two main areas: Goceano–Baronia (3953 km

^{2}) and Barbagia–Ogliastra (1896 km

^{2}).

#### 2.2. Data Collection and Management

^{2}.

^{2}for the radius [31]. In addition, the altitude (mamsl), road (km), amount of forest (km

^{2}), and protected forest (km

^{2}) were collected by Corine Land Cover 2012 (https://land.copernicus.eu/pan-european/corine-land-cover/clc-2012/view). This procedure was realized using ArcGIS

^{®}(ArcMap software by Esri, version 10.4, Environmental Systems Research Institute: Redlands, CA, USA) geoprocessing tools to join all layers built for each variable with different options. A geodesic buffer was created with the standard planar method to consider the joined attributes related to the wild boar’s home range, and all buffers were dissolved together into a single feature, thus removing any overlap. In addition, the geometric intersection of the input features was used for the water, streets, forest, and protected areas, and all attributes from the input features were transferred to the output feature class. This provides the sum of the variable values in each point or buffer. Figure 2 illustrates the flow-chart of the study design.

#### 2.3. Statistical Analysis

^{2}grid). Supposing that 45% of the total wild boar population is hunted every year, considering the WBh as the number of wild boars hunted (and consequently tested) for ASF, and WBd the estimated wild boar population density [13], the compliance C is given as follows:

_{0}and the summitry vector of m covariates (X) measured at each grid j and their respective coefficients β

_{k}represent the fixed proportion of the model; α

_{j}is the random effect representing the dispersion among grids; and the n × 1 vector of errors ε

_{ij}is assumed to be multivariate and normal, with a mean of 0 and a variance matrix of ${\sigma}_{\epsilon}^{2}\mathbf{\Re}$.

^{2}values and the Bayesian and Akaike Information Criteria (BIC and AIC). All interactions with a supposedly biologically valid foundation were tested. The intraclass correlation coefficient (ICC, i.e., the correlation between the latent linear responses conditional on the fixed-effects covariates) was calculated for each MELR model to evaluate how strongly values in the same grids resembled each other. The two models results are presented as the adjusted odds ratio (OR

_{adj}) calculated by the Lemeshow and Hosmer method [34].

#### 2.4. Map of Priority Surveillance Areas

## 3. Results

^{2}, median = 0, IQR = 0–1). The Goceano–Baronia and Barbagia–Ogliastra areas differed for the amount of forest (km

^{2}, training dataset median = 1.1, IQR = 0.6–1.9; test dataset median = 2.2, IQR = 1.3–2.5), wild boar density estimation (training dataset median = 1, IQR = 0–5; test dataset median = 3, IQR = 0–6) [13], and the absence of free-ranging pigs in the first but present in the second area. Table 1 summarizes the features related to ASF in wild boar in both the Goceano–Baronia and the Barbagia–Ogliastra areas: number of wild boar tested, PCR-positive, seropositive, and the compliance value. Data from 6488 wild boar tested within the Goceano–Baronia were used to fit the two MELR models.

^{2}for the radius [31]. The descriptive statistics in Table 2a,b show the baseline variable distribution in the training dataset (Gogeano–Baronia) based on both outcomes. Given the values imputation in the neighboring grids, over the 3953 grids of the Goceano–Baronia dataset, 178 presented the outcome 1 (presence of PCR-positive), and 229 the outcome 2 (presence of young seropositive).

#### 3.1. Mixed-Effects Logistic Regression Model Results

_{adj}= 18.71, 95% IC = 11.55–30.29, p-value < 0.0001). Each adult seropositive animal detected during the same hunting season increased the probability of finding a PCR-positive (OR

_{adj}= 1.83, 95% IC = 1.66–2.01, p-value < 0.0001). A not statistically significant coefficient was estimated for young seropositive wild boar. The grids located at altitude over 500 mamsl showed about seven times more the probability to find virus respect to those at altitude ≤ 500 mamsl (OR

_{adj}= 7.691, 95% CI = 4.978–11.881, p-value < 0.0001).

^{2}value by 0.12 points, decreasing the AIC by 5.5 points, and decreasing the BIC by 3.2 points, compared to animal density. The probability of observing a PCR-positive animal increased by about 1.5 times for every 1 km

^{2}of forest (OR = 1.53, 95% CI = 1.18–2.52, p-value = 0.001). The final AIC and BIC values were 401.6 and 487.2, respectively. Random intercepts in the output exhibited significant variation based on a likelihood-ratio test versus a one-level binomial regression model (coeff. = 3.71, p-value = 0.025) and the SD of random intercepts (3.128) was greater than twice the standard error (SE, 0.905). This result favors the random-intercept model, indicating that there is significant variation in the number of PCR-positive results between grids. Conditional on virus-positive and seropositive results, altitude, and animal density, we estimated that the latent responses within the same grids had a large correlation of approximately 0.78. Thus, 78% of the variance of a latent response was explained by the between-grid variability. The predicted values generated by this model constituted the “virus layer”.

_{adj}= 2.07, 95% CI = 1.53–2.80, p-value < 0.0001). The probability to observe the outcome was increased by 4% (OR

_{adj}= 1.04, 95% CI = 1.01–1.07, p-value = 0.01) by the increase of the animal density in the grid by one wild boar. In grids located over 500 mamsl, the probability to detect young seropositive animal was about 1.5 times more with respect to those located under 500 mamsl (OR

_{adj}= 1.68, 95% CI = 1.27–2.22, p-value < 0.0001). The probability of observing a PCR-positive animal increased by about two times for every 1 km

^{2}of forest (OR = 2.13, 95% CI = 1.77–2.54, p-value < 0.0001). The final AIC and BIC values were 462.1 and 499.6, respectively. The better fit of random-intercept compared to the simple logistic model was demonstrated by the likelihood-ratio test (coeff. 19.01, p-value = 0.002) and the random intercept’s SD and SE values (1.145, 0.699), considering that 90% of the variance of a latent response could be explained by the between-grid variability (ICC = 0.90). The predicted values generated by this model constitute the “young seropositive layer”.

^{2}). This variable had an apparently protective role in the probability of finding a PCR-positive animal (OR = 0.97, 95% CI = 0.96–0.99, p-value = 0.001) but increased the probability of detecting a young seropositive animal (OR = 1.01, 95% CI = 1.005–1.01, p-value < 0.0001) in the second model.

#### 3.2. Model Validation

## 4. Discussion

## 5. Conclusions

## Author Contributions

## Funding

## Conflicts of Interest

## References

- European Food Safety Authority (EFSA); Miteva, A.; Papanikolaou, A.; Gogin, A.; Boklund, A.; Bøtner, A.; Linden, A.; Viltrop, A.; Schmidt, C.G.; Ivanciu, C. European Food Safety Authority (EFSA); Miteva, A.; Papanikolaou, A.; Gogin, A.; Boklund, A.; Bøtner, A.; Linden, A.; Viltrop, A.; Schmidt, C.G.; Ivanciu, C.; et al. Epidemiological analyses of African swine fever in the European Union (November 2018 to October 2019). EFSA J.
**2020**. [Google Scholar] [CrossRef] [Green Version] - Dixon, L.K.; Escribano, J.M.; Martins, C.; Rock, D.I.; Salas, M.I.; Wilkinson, P.J.; Fauquet, C.M.; Mayo, M.A.; Maniloff, J.; Desselberger, U.; et al. Asfaviridae Virus Taxonomy, 8th ed.; Elsevier Academic Press: London, UK, 2005. [Google Scholar]
- Franzoni, G.; Dei Giudici, S.; Oggiano, A. Infection, modulation and responses of antigen-presenting cells to African swine fever viruses. Virus Res.
**2018**, 258, 73–80. [Google Scholar] [CrossRef] [PubMed] - World Organization for Animal Health (OIE). Standard Operating Procedures on Suspension, Recovery or Withdrawal of Officially Recognised Disease Status and Withdrawal of the Endorsement of Official Control Programmes of Members; OIE: Paris, France, 2020. Available online: https://www.oie.int/fileadmin/Home/eng/Animal_Health_in_the_World/docs/pdf/SOP/EN_SOP_Susp_Recovery.pdf (accessed on 3 July 2020).
- Blome, S.; Franzke, K.; Beer, M. African swine fever – A review of current knowledge. Virus Res.
**2020**, 287, 1–15. [Google Scholar] [CrossRef] - Penrith, M.L. Current status of African swine fever. CABI Agric. Biosci.
**2020**, 1, 11. [Google Scholar] [CrossRef] - World Organization for Animal Health (OIE). African Swine Fever (Infection with African Swine Fever Virus). In Manual of Diagnostic Tests and Vaccines for Terrestrial Animals 2019; OIE: Paris, France, 2019. [Google Scholar]
- Beltran-Alcrudo, D.; Falco, J.R.; Raizman, E.; Dietze, K. Transboundary spread of pig diseases: The role of international trade and travel. BMC Vet. Res.
**2019**, 15, 64. [Google Scholar] [CrossRef] [PubMed] [Green Version] - Halasa, T.; Bøtner, A.; Mortensen, S.; Christensen, H.; Toft, N.; Boklund, A. Simulating the epidemiological and economic effects of an African swine fever epidemic in industrialized swine populations. Vet. Microbiol.
**2016**, 193, 7–16. [Google Scholar] [CrossRef] - Montgomery, R.E. On a form of swine fever occurring in British East Africa (Kenya Colony). J. Comp. Pathol. Ther.
**1921**, 34, 159–191. [Google Scholar] [CrossRef] [Green Version] - Penrith, M.L.; Bastos, A.D.; Etter, E.M.C.; Beltran-Alcrudo, D. Epidemiology of African swine fever in Africa today: Sylvatic cycle versus socio-economic imperatives. Transbound. Emerg. Dis.
**2019**, 66, 672–686. [Google Scholar] [CrossRef] - Blome, S.; Gabriel, C.; Beer, M. Pathogenesis of African swine fever in domestic pigs and European wild boar. Virus Res.
**2013**, 173, 122–130. [Google Scholar] [CrossRef] - Loi, F.; Cappai, S.; Coccollone, A.; Rolesu, S. Standardized Risk Analysis Approach Aimed to Evaluate the Last African swine fever Eradication Program Performance, in Sardinia. Front. Vet. Sci.
**2019**, 6, 299. [Google Scholar] [CrossRef] - Kolbasov, D.; Titov, I.; Tsybanov, S.; Gogin, A.; Malogolovkin, A. African swine fever virus, Siberia, Russia, 2017. Emerg. Infect. Dis.
**2018**, 24, 796–798. [Google Scholar] [CrossRef] [PubMed] - Iglesias, I.; Martínez, M.; Montes, F.; de la Torre, A. Velocity of ASF spread in wild boar in the European Union (2014–2017). Int. J. Inf. Dis.
**2019**, 79, 69. [Google Scholar] [CrossRef] [Green Version] - Jung-Hyang, S. How far can African swine fever spread? J. Vet. Sci.
**2019**, 20, e41. [Google Scholar] - World Organization for Animal Health (OIE). WAHIS Interface—Disease Timelines. 2020. Available online: https://www.oie.int/wahis_2/public/wahid.php/Diseaseinformation/Diseasetimelines (accessed on 2 December 2020).
- Penrith, M.L.; Vosloo, W. Review of African swine fever: Transmission, spread and control. J. S. Afr. Vet. Assoc.
**2009**, 80, 58–62. [Google Scholar] [CrossRef] [Green Version] - Nurmoja, I.; Mõtus, K.; Kristian, M.; Niine, T.; Schulz, K.; Depner, K.; Viltrop, A. Epidemiological analysis of the 2015–2017 African swine fever outbreaks in Estonia. Prev. Vet. Med.
**2018**. [Google Scholar] [CrossRef] [PubMed] - Schulz, K.; Staubach, C.; Blome, S.; Viltrop, A.; Nurmoja, I.; Conraths, F.J.; Sauter-Louis, C. Analysis of Estonian surveillance in wild boar suggests a decline in the incidence of African swine fever. Sci. Rep.
**2019**, 9, 8490. [Google Scholar] [CrossRef] [PubMed] - Oļševskis, E.; Schulz, K.; Staubach, C.; Seržants, M.; Lamberga, K.; Pūle, D.; Ozoliņš, J.; Conraths, F.J.; Sauter-Louis, C. African swine fever in Latvian wild boar—A step closer to elimination. Transbound. Emerg. Dis.
**2020**. [Google Scholar] [CrossRef] [PubMed] - Loi, F.; Cappai, S.; Laddomada, A.; Feliziani, F.; Oggiano, A.; Franzoni, G.; Rolesu, S.; Guberti, V. Mathematical approach to estimating the main epidemiological parameters of African swine fever in wild boar. Vaccines
**2020**, 8, 521. [Google Scholar] [CrossRef] - European Commission. Strategic Approach to the Management of African Swine Fever for the EU; Working Document SANTE/7113/2015; European Commission: Brussels, Belgium, 2020. [Google Scholar]
- Schulz, K.; Staubach, C.; Blome, S.; Nurmoja, I.; Viltrop, A.; Conraths, F.J.; Kristian, M.; Sauter-Louis, C. How to Demonstrate Freedom from African Swine Fever in Wild Boar—Estonia as an Example. Vaccines
**2020**, 8, 336. [Google Scholar] [CrossRef] - European Food Safety Authority (EFSA); Boklund, A.; Cay, B.; Depner, K.; Foldi, Z.; Guberti, V.; Masilius, M.; Miteva, A.; More, S.; Olsevskis, E.; et al. Scientific report on the epidemiological analyses of African swine fever in the European Union (November 2017 until November 2018). EFSA J.
**2018**, 16, 5494. [Google Scholar] - Nurmoja, I.; Schulz, K.; Staubach, C.; Sauter-Louis, C.; Depner, K.; Conraths, F.J.; Viltrop, A. Development of African swine fever epidemic among wild boar in Estonia—Two different areas in the epidemiological focus. Sci. Rep.
**2017**, 7, 12562. [Google Scholar] [CrossRef] [PubMed] [Green Version] - Gervasi, V.; Marcon, A.; Bellini, S.; Guberti, V. Evaluation of the Efficiency of Active and Passive Surveillance in the Detection of African Swine Fever in Wild Boar. Vet. Sci.
**2019**, 7, 5. [Google Scholar] [CrossRef] [PubMed] [Green Version] - Schulz, K.; Conraths, F.J.; Staubach, C.; Viltrop, A.; Oļševskis, E.; Nurmoja, I.; Lamberga, K.; Sauter-Louis, C. To sample or not to sample? Detection of African swine fever in wild boar killed in road traffic accidents. Transbound. Emerg. Dis.
**2020**, 1–4. [Google Scholar] [CrossRef] - Danzetta, M.L.; Marenzoni, M.L.; Iannetti, S.; Tizzani, P.; Calistri, P.; Feliziani, F. African swine fever: Lessons to learn from past eradication experiences. A systematic review. Front. Vet. Sci.
**2020**, 7, 296. [Google Scholar] [CrossRef] - World Organization for Animal Health (OIE). African Swine Fever (ASF) Report N° 53: 4 September to 17 September. 2020. Available online: https://www.oie.int/fileadmin/Home/eng/Animal_Health_in_the_World/docs/pdf/Disease_cards/ASF/Report_53_Current_situation_of_ASF.pdf (accessed on 10 September 2020).
- Bosch, J.; Barasona, J.A.; Cadenas-Fernandez, E.; Jurado, C.; Pintore, A.; Denurra, D.; Cherchi, M.; Vicente, J.; Sanchez-Vizcaıno, J.M. Retrospective spatial analysis for African swine fever in endemic areas to assess interactions between susceptible host populations. PLoS ONE
**2020**, 15, e0233473. [Google Scholar] [CrossRef] - Cappai, S.; Rolesu, S.; Coccollone, A.; Laddomada, A.; Loi, F. Evaluation of biological and socio-economic factors related to persistence of African swine fever in Sardinia. Prev. Vet. Med.
**2018**, 152, 1–11. [Google Scholar] [CrossRef] - Piccolo, D. Modelli di regressione multipla. In Statistica; Mulino, I.L., Ed.; Bologna, Italy, 1998; pp. 872–900. Available online: https://local.disia.unifi.it/rampichini/regressione_dottorato_08.pdf (accessed on 2 December 2020).
- Lemeshow, S.A.; Hosmer, J.R. Logistic regression. In Encyclopedia of Biostatistics; Armitage, P., Colton, T., Eds.; Wiley: Chichester, UK, 2005; Volume 2, pp. 2870–2888. [Google Scholar]
- Diamond, G.A. What price perfection? Calibration and discrimination of clinical prediction models. J. Clin. Epidemiol.
**1992**, 45, 85–89. [Google Scholar] [CrossRef] - Hanley, J.A.; McNeil, B.J. The measure and use of the area under a receiver operating characteristic (ROC) curve. Radiology
**1982**, 143, 29–36. [Google Scholar] [CrossRef] [Green Version] - Silverman, B.W. Density estimation for statistics and data analysis. In Monographs on Statistics and Applied Probability; Chapman & Hall: London, UK, 1986; Volume 26. [Google Scholar]
- Diggle, P.J.; Zheng, P.; Durr, P.A. Nonparametric estimation of spatial segregation in a multivariate point process: Bovine tuberculosis in Cornwall, UK. J. R. Stat. Soc.
**2005**, 54, 645–658. [Google Scholar] [CrossRef] - Cappai, S.; Loi, F.; Rolesu, S.; Coccollone, A.; Laddomada, A.; Sgarangella, F.; Masala, S.; Bitti, G.; Floris, V.; Desini, P. Evaluation of the cost-effectiveness of ASF detection with or without the use of on-field tests in different scenarios, in Sardinia. J. Vet. Sci.
**2020**, 21, e14. [Google Scholar] [CrossRef]

**Figure 1.**Epidemiological context of African swine fever in Sardinia from April 2019, and updated on October 2020. The violet star indicates the last two PCR-positive wild boar detected in April 2019, in the protected forest of Bultei (Sassari). The blue dots indicate the seropositive wild boar (WB). The proportion of animals tested via passive surveillance in each municipality is presented as a cloropletic map. The bold black lines define the limits of the two areas of Goceano–Baronia (the largest) and Barbagia–Ogliastra (the smallest).

**Figure 2.**Flow-chart representing the study design methodology. Wild boar data were collected from the Internal database system (SIGLA) of Istituto Zooprofilattico Sperimentale della Sardegna. Based on latitude and longitude (green font), variables were associated with each grid of 1 km

^{2}by year, and the final dataset (green square) was implemented with the calculated compliance and the land features (altitude, road, forest, protected forest) collected by Corine Land Cover 2012 database.

**Figure 3.**African swine fever PCR-positive animals detected in 2018–2019 (outcome 2—red dots) and seropositive young animals in 2019–2020 (outcome 2—blue dots) in the two areas of Goceano–Baronia and Barbagia–Ogliastra (bold black lines). Points around the red and blue dots identify that the neighboring grids were positive values were imputed, considering the 5 km

^{2}for the radius.

**Figure 4.**Final passive surveillance map. The grey areas are areas where the probability of finding a pocket of infection is higher (areas 1) given the mixed-effect logistic regression (MELR) model results and the lack of information. The areas indicated as number 2 are areas where the lack of information is higher but not indicated by the MELR models. The blue lines indicate the limits of the protected forest where hunting is forbidden.

**Figure 5.**Discrimination of the models: fitting values versus validation distribution for the four MELR models. Distributions are represented as box-plots: the line splitting of each box in two represents the median value; the bottom and top edges of each box represent the lower and upper quartiles; the values at which the vertical lines stop are the upper and lower values of the data.

**Table 1.**Descriptive analyses of the Goceano–Baronia and Barbagia–Ogliastra areas on the data included in the models. Data are presented as the frequency (prevalence expressed as %) and median (interquartile range).

Area | Goceano–Baronia (3953 km^{2}) | Barbagia–Ogliastra (1896 km^{2}) | ||||||
---|---|---|---|---|---|---|---|---|

Hunting Season | Wild Boar Tested | PCR-Positive | Seropositive | Compliance | Wild Boar Tested | PCR-Positive | Seropositive | Compliance |

2017–2018 | 2104 | 6 (0.28%) | 106 (5.03%) | 30.7 (18.4–42.7) | 497 | 10 (2.01%) | 61 (12.27%) | 14.5 (7.0–27.6) |

2018–2019 | 2175 | 4 (0.18%) ^{1} | 39 (1.79%) | 28.6 (15.9–53.9) | 654 | 2 (0.31%) | 48 (7.34%) | 17.9 (10.4–36.0) |

2019–2020 | 2209 | 0 (0%) | 32 (1.45%) | 32.5 (19.7–48.2) | 702 | 0 (0%) | 43 (6.12%) | 18.6 (11.0–25.5) |

Total | 6488 | 8 (0.12%) | 177 (2.72%) | 31.8 (18.3–50.6) | 1853 | 12 (0.65%) | 152 (8.20%) | 17.6 (10.1–28.5) |

^{1}Two of these samples arise from passive surveillance activities in April 2019 in Bultei municipality.

**Table 2.**Descriptive statistics of variables in the training dataset (Goceano–Baronia) for (a) outcome 1 (PCR-positive wild boar detected in 2018–2019) and (b) for outcome 2 (young seropositive animals detected in 2019–2020). Data are presented as the frequency (%) or mean (standard deviation (SD)), median (I–III interquartile), and p-value.

(a) Variables | Outcome 1 = 1 PCR-Positive Detected in 2018–2019 (178 Grids) | Outcome 1 = 0 PCR-Positive Not Detected in 2018–2019 (3775 Grids) | p-Value |
---|---|---|---|

PCR-positive ^{1} | |||

Hunting season 2017–2018 | 1 (0–2) | 0 (0–0) | <0.0001 |

Adult seropositive ^{1} | |||

Hunting season 2017–2018 | 6 (4–8) | 0 (0–2) | <0.0001 |

Hunting season 2018–2019 | 3 (2–9) | 0 (0–1) | <0.0001 |

Young seropositive ^{1} | |||

Hunting season 2017–2018 | 2 (1–3) | 0 (0–0) | <0.0001 |

Hunting season 2018–2019 | 0 (0–0) | 0 (0–0) | NS |

Altitude (mamsl) | 700 (600–800) | 500 (300–700) | <0.0001 |

Road (km) | 1.7 (0–2.6) | 1.0 (0–2.4) | 0.007 |

Forest (km^{2}) | 1.82 (0.81) | 1.24 (0.82) | <0.0001 |

Wild boar density ^{1,2} | 5 (0–7) | 0 (0–5) | <0.0001 |

Amount of protected forest (km^{2}) | 0 (0–1.3) | 0 (0–0.7) | <0.0001 |

(b) Variables | Outcome 2 = 1Young Seropositive Animal Detected in 2019–2020(229 Grids) | Outcome 2 = 0Young Seropositive AnimalNot Detected in 2019–2020(3724 Grids) | p-Value |

PCR-positive ^{1} | |||

Hunting season 2017–2018 | 0 (0–0) | 0 (0–0) | NS |

Hunting season 2018–2019 | 0 (0–0) | 0 (0–0) | NS |

Adult seropositive ^{1} | |||

Hunting season 2017–2018 | 1 (0–4) | 0 (0–2) | <0.0001 |

Hunting season 2018–2019 | 0 (0–1) | 0 (0–1) | NS |

Hunting season 2019–2020 | 0 (0–2) | 0 (0–1) | <0.0001 |

Young seropositive ^{1} | |||

Hunting season 2017–2018 | 0 (0–0) | 0 (0–1) | NS |

Hunting season 2018–2019 | 0 (0–0) | 0 (0–0) | NS |

Altitude (mamsl) | 550 (450–700) | 500 (300–700) | 0.0026 |

Road (km) | 0 (0–2.18) | 1.1 (0–2.41) | <0.0001 |

Forest (km^{2}) | 1.78 (0.69) | 1.24 (0.82) | <0.0001 |

Wild boar density ^{1,2} | 3 (0–5) | 0 (0–5) | <0.0001 |

Amount of protected forest (km^{2}) | 1.58 (0–4.2) | 0 (0–0.05) | <0.0001 |

^{1}Considering the features of wild boar imputed in the neighboring grids, the number of positive animals correspond to the number of grids covered by a 5 km

^{2}, for which the ASF-positive cases were imputed.

^{2}Wild boar density is based on the wild boar management plan [13].

**Table 3.**Results of the mixed-effect logistic regression model with (

**a**) PCR-positive cases (2018–2019) considering outcome 1 and (

**b**) young seropositive cases (2019–2020) considering outcome 2. Data are reported as the adjusted odds ratio (OR

_{adj}), 95% confidence intervals (95% CI), and p-values.

Outcome 1: PCR-Positive 2018–2019 | Variables | OR_{adj} | 95% CI | p-Value |
---|---|---|---|---|

Presence of PCR-positive 2017–2018 | 18.71 | 11.55–30.29 | <0.0001 | |

Adult Seropositive 2018–2019 | 1.83 | 1.66–2.01 | <0.0001 | |

Altitude >500 mamsl | 7.69 | 4.98–11.88 | <0.0001 | |

Forest (by 1 km^{2}) | 1.53 | 1.18–2.52 | 0.001 | |

Sd | SE | 95% CI | ||

Random-effect | grid | 3.128 | 0.905 | 1.730–5.652 |

LR test vs. logistic regression: 3.71, p-value = 0.025 | ||||

ICC | SE | 95% CI | ||

Residual intraclass correlation | grid | 0.782 | 0.127 | 0.588–0.919 |

Outcome 2:Young Seropositive2019–2020 | Variables | OR_{adj} | 95% CI | p-Value |

Adult Seropositive 2019–2020 | 2.07 | 1.53–2.80 | <0.0001 | |

Wild boar density | 1.04 | 1.01–1.07 | 0.028 | |

Altitude >500 mamsl | 1.68 | 1.27–2.22 | <0.0001 | |

Forest (by 1 km^{2}) | 2.13 | 1.77–2.54 | <0.0001 | |

Sd | SE | 95% CI | ||

Random-effect | grid | 1.145 | 0.699 | 0.338–3.717 |

LR test vs. logistic regression: 19.01, p-value = 0.002 | ||||

ICC | SE | 95% CI | ||

Residual intraclass correlation | grid | 0.906 | 0.002 | 0.893–0.998 |

**Table 4.**Contingency tables for the discriminatory tests applied for internal validation (outcome 1 and outcome 2 on the training dataset), and eternal validation (outcome 1 and outcome 2 on the test dataset).

Model Outcome | Outcome 1 | Outcome 2 | ||||||
---|---|---|---|---|---|---|---|---|

Dataset | Observed | |||||||

Training dataset | 1 | 0 | tot | 1 | 0 | tot | ||

Predicted | 1 | 168 | 6 | 174 | 222 | 28 | 250 | |

0 | 10 | 3769 | 3779 | 7 | 3696 | 3703 | ||

tot | 178 | 3775 | 3953 | 229 | 3724 | 3953 | ||

Sensitivity | 94.4% | 96.9% | ||||||

Specificity | 99.8% | 99.2% | ||||||

Test dataset | 1 | 0 | tot | 1 | 0 | tot | ||

Predicted | 1 | 143 | 89 | 232 | 242 | 50 | 292 | |

0 | 11 | 1653 | 1664 | 9 | 1595 | 1604 | ||

tot | 154 | 1742 | 1896 | 251 | 1645 | 1896 | ||

Sensitivity | 92.9% | 96.4% | ||||||

Specificity | 94.9% | 97.0% |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Cappai, S.; Rolesu, S.; Feliziani, F.; Desini, P.; Guberti, V.; Loi, F.
Standardized Methodology for Target Surveillance against African Swine Fever. *Vaccines* **2020**, *8*, 723.
https://doi.org/10.3390/vaccines8040723

**AMA Style**

Cappai S, Rolesu S, Feliziani F, Desini P, Guberti V, Loi F.
Standardized Methodology for Target Surveillance against African Swine Fever. *Vaccines*. 2020; 8(4):723.
https://doi.org/10.3390/vaccines8040723

**Chicago/Turabian Style**

Cappai, Stefano, Sandro Rolesu, Francesco Feliziani, Pietro Desini, Vittorio Guberti, and Federica Loi.
2020. "Standardized Methodology for Target Surveillance against African Swine Fever" *Vaccines* 8, no. 4: 723.
https://doi.org/10.3390/vaccines8040723