Co-infections of respiratory pathogens and gastrointestinal parasites in smallholder pig production systems in Uganda

A cross-sectional study was conducted to identify factors for infections of pigs with key respiratory pathogens: porcine circovirus type 2 (PCV2), porcine reproductive and respiratory syndrome virus (PPRSv), Mycoplasma hyopneumoniae (M. hyo), Actinobacillus pleuropneumoniae (App), and gastrointestinal (GI) parasites in Uganda. A structured questionnaire was used to collect data on management practices associated with infections. Ninety (90) farms and 259 pigs were sampled. Sera were screened against 4 pathogens using commercial ELISA tests. The Baerman’s method was used to identify parasite species in faecal samples. Logistic regression was done to identify risk factors for infections. Results showed individual animal seroprevalence of PCV2 was 6.9% (95% CI 3.7–11.1), PRRSv 13.8% (95% CI 8.8–19.6), M. hyo 6.4% (95% CI 3.5–10.5), and App 30.4% (95% CI 24.8–36.5). The prevalence of Ascaris spp. was 12.7% (95% CI 8.6–16.8), Strongyles spp was 16.2% (95% CI 11.7–20.7), and Eimeria spp. was 56.4% (95% CI 50.3–62.4). Pigs infested with Ascaris spp. were more likely to test positive to PCV2, odds ratio (OR) 1.86 (CI 1.31–2.60; p = 0.0002). For M. hyo, infection with Strongyles spp. was a risk factor (OR 12.9, p < 0.001). Pigs that had Strongyles and Ascaris spp. Infections (ORs 3.5 and 3.4, p < 0.001 respectively) were likely to have co-infections. The model showed that use of cement, elevated floor, and limiting contacts with outside pigs were protective while using mud and helminth infestations increased risks of co-infections. This study provided evidence that improved housing and biosecurity are critical in reducing pathogen incidence in herds. This study was done in Lira district, mid-northern Uganda, where the International Livestock Research Institute (ILRI) previously implemented a smallholder pig value chain development project (SPVCD) in 2011. In this project, a value chain assessment was conducted to select study sites using pig density, poverty levels, and market access (Ouma 2017). Our study used market access to select subcounties based on value chain domains into rural production for urban consumption (R-U) and urban production for urban (U-U) consumption (Ouma 2017). The total pig population in Lira district was estimated to be 30,000 in 2020 (district veterinary officer, personal communication). Pigs are produced under housed, tethered, and free-range systems (Kungu et al. 2019).Under these systems, pigs are housed in permanent or temporal structures made of cement, wood, or papyrus. Tethering is when pigs are tied on a rope (on a pole) to graze around the homestead, while free-range is when pigs are allowed to freely roam in the neighborhood in search of own feeds and water. Routine preventive measures such as anthelmintics are generally not practiced until pigs show visible signs of illness. In these smallholder pig production systems, biosecurity and hygiene practices are generally poor, which partly explains a high disease incidence in herds. A cross-sectional serologic study was conducted from October to December 2018. We used multistage sampling to select subcounties and villages. In the first stage, four subcounties were selected (from a total of 9): two (central division and railways) representing U-U consumption and two (Adekokwok and Ngetta) representing R-U consumption. In stage two, two (2) villages with the highest pig density were selected for the study. To determine the sample size, a formula for simple random sampling was used (Dohoo et al. 2003). A previous study in Lira district found a seroprevalence of M. hyo in pigs of 20.9% (Dione et al. 2018). Adjusting for test sensitivity and specificity, true prevalence was computed to be 24%. The required sample size of pigs was obtained from Eq. (1): where n = is the required sample size, Zα is the standard z-score from a normal distribution (1.96), p = estimated prevalence of disease (24%), q = 1 − p (76%), and d = allowable error (6%). Using this formula, an unadjusted sample size of 195 pigs was computed. To adjust for within-farm clustering, we sampled 3 pigs per herd, thus the design effect (Deff) was obtained from Eq. 2 below: where icc is the intra-cluster correlation (0.2) for respiratory disease (Dohoo et al. 2003) and n1 is the number of pigs sampled per herd (3), thus the Deff calculated is 1.4. The adjusted sample size was calculated from the equation: N = n1 × number of pigs sampled per herd (3). From this, adjusted sample size of 273 pigs was derived. In each selected village, a list of pig keeping households/farms was obtained from the district veterinary office and the area local councils. Random sampling of farms was done until the required sample size was obtained. We sampled farms regardless of health status or anthelmintic treatments to examine differences in farm husbandry practices and how these influence occurrences of specific pathogens in farms. Using a sampling frame of all pig farmers generated with field research assistants, three (3) pigs per herd were sampled until the required sample size was reached. Only pigs ≥ 2.5 months old were selected for sampling, since pigs below that age are reported to retain maternal antibodies to PCV2 and PRRSv post weaning, which could interfere with serologic tests (Opriessnig et al. 2004; Gillespie et al. 2009). App-acquired colostral antibodies were reported to decay within 2 months postpartum (Vigre et al. 2003). We sampled pigs from 2.5 months and above regardless of health status, clinical signs, or feed types given. A structured questionnaire with closed questions was designed, pre-tested by the first author in Mukono district, and revised before use. Research assistants were trained in its use before it was administered to each household head or farm manager. To ensure consistency, all questions were translated to a local language spoken in the area (Langi). The questionnaire captured data on potential risk factors for infection with respiratory pathogens. Each pig was properly restrained as described in the ILRI Standard Operating Procedures (SOPs) manual, Sect. 2, parts (c) and (d) (ILRI 2004). Smaller pigs were restrained by hand while larger ones were restrained with a metallic pig catcher (model BZ002; MG. Livestock, Shandong, China) placed behind the upper incisor teeth and the snout raised upward. Blood was then collected from the cranial vena cava or jugular vein using a 21G, 1.5″ needle into plain 5-mL BD® vacutainer tubes. The tubes were labeled with animal identification details and then placed in an ice box at 4–6 °C. After collection, samples were delivered (within 3 h) to the district veterinary laboratory for temporary storage. Blood samples were left to stand at room temperature (20 °C) overnight and serum harvested the following day into 2 mL cryotubes (Sarstedt®, Germany), labeled, and stored in a fridge at − 20 °C until testing. In the lab, sera were screened using ELISA assays according to manufacturers’ instructions for each pathogen: M. hyo and App-ApxIV (IDDEXX, Westbrook, Maine, USA) for PRRSv and PCV2 assays (Krishgen Biosystems, India). Cut-off sample to positive ratios (S/P%) for M. hyo were > 0.40 (positive) and < 0.30 (negative), App was ≥ 50% (positive) and < 40% (negative). PCV2 and PRRSv S/P cut-off ratios for positive and negative samples were ≥ 0.2 and < 0.2 respectively. Suspect samples were re-tested. Test sensitivity (Se) and specificity (Sp) for M. hyo ELISA were 85.6 and 99.6%, respectively; Se and Sp for App-ApxIV Ab ELISA test were 97.8 and 100%, respectively. Se and Sp for PRRSv were 94.0 and 94.0%, while for PCV2 Se and Sp these were 92.0 and 94.0%, respectively. Test Se and Sp were used to calculate true prevalence of respiratory infections at α = 0.05 significance level. Fecal samples (~ 3 g) were collected from the rectum of each pig using gloved hands into 5-mL plastic containers, labeled, and placed in ice box at 4 °C. Samples were taken to the district veterinary lab for temporary storage at 4 °C. Samples were transferred to the Central Diagnostic Laboratory, College of Veterinary Medicine, Animal Resources and Biosecurity (CoVAB), Makerere University for analysis 1 week after collection. Helminth species were identified and fecal egg counts per gram (EPGs) were quantified using the Baermann and McMaster methods, respectively (MAFF 1986). Data was coded and entered into Excel 16.0, and any errors in entry were corrected by cross-checking with questionnaires. RStudio was used for data analysis and presentation (R Core Team 2019). True prevalence was computed by adjusting for apparent prevalence using prevalence 0.2.0 package in R, taking into account test sensitivities and specificities (Devleesschauwer et al. 2022). Multivariable logistic regression analysis of risk factors for each pathogen was performed. The response variable was the ELISA test result with predictors pig age and husbandry practices (house type, parasites, drug use, pig mixing, hygiene score, and drainage). The model below was fitted to predict respiratory infections as a function of pig characteristics and husbandry practices (house type, parasites, drug use, pig age, pig mixing, hygiene, and drainage): where lnp^(1-p)^ is the expected log of odds of infection, β0 is the model intercept, β1, β2 are regression coefficients, and x1i, x2i, are the respective explanatory variables. Interaction terms were tested for each model, and confounding was checked by inclusion and exclusion of variables to observe a change in model coefficients. A cumulative link mixed-effects (CLMM) model was fitted to estimate the odds of co-infection (2 or more pathogens) with farm as a random effect. The CLMM model was fitted using R packages “factoextra” and “ordinal” using a backward elimination method such that all variables whose p-value > 0.2 were dropped. Residual plots and R-square statistics were used to assess the fitted models. Table ​Table11 below presents housing and husbandry practice variables used in fitting the models. Housing and husbandry practice variables used in fitting the models