Descriptive statistics.

A five-number summary (i.e., values representing the minimum and maximum, the lower and upper quartiles, and the median) was computed for the continuous variables, and frequencies were calculated for the categorical variables. Differences in the frequency of SM and WB across seasons were compared using the Kruskal-Wallis test, and Dunn’s multiple range test was used for post-hoc multiple comparisons.

Exploratory statistics.

Multivariable regression analysis was conducted to estimate the association between risk factors (independent variables) and presence/absence of SM or WB (outcome variables), considered separately (i.e., one model for each myopathy) and regardless of severity. Hence, WB was re-coded as 0 (denoting WB0) and 1 (denoting WB1 and WB2) for the logistic regression analyses. WS was not considered for regression analysis because WS1 was present in 93.8% of fillets.

Risk factors to be fitted in the final multivariable model were initially tested individually (univariable models) using binomial logistic regression due to the binary nature of the dependent variables (presence/absence), and employing a relaxed significant threshold (p < 0.02). When continuous variables were significant from univariable models, they were kept in the model as continuous variables. If they were not significant during unconditional association analysis, they were categorized into three equal quartiles (low, medium, high) using the xtile function from a statistical software package (Stata 14.0; Stata Corporation, College Station, Texas, USA), and tested again as categorical categories using univariable models. These variables included average live weight, precipitation during grow-out, precipitation during transport, temperature during grow-out, temperature during transport, mortality rate during grow-out, hold time on lairage at plant, and condemnation.

Pair-wise correlation coefficients, using Spearman’s rank test among all significant independent variables, were examined to prevent the inclusion of collinear variables in the multivariable models. When two variables were highly correlated (rho > 0.70; p < 0.05), the one with the smallest p-value was considered for inclusion in the multivariable model. All unconditionally significant variables identified during univariable screening were offered to a multivariable model, and a manual backward elimination process was performed. Variables with p > 0.05, as calculated using the likelihood ratio test, were removed for a better fit of a model. For suspected confounders, when the removal of a variable changed the coefficients of the remaining variables from the final model by more than 20%, the variable was kept in the final model, otherwise, it was dropped [42].

After the final independent variables were selected (S2 Fig), in order to evaluate and control for clustering in the occurrence of breast myopathies, 2-level random intercept logistic regression models were built [43]. Flocks were included in the models as random effects. Based on the unexplained variance components at each level of the models (breast fillets and flocks), intra-class correlation coefficients (ICC) were calculated with flocks as the level-2 variable. The ICC was computed by assuming that level 1 variance on the logit scale was: Eq 2 The formula used for the ICC calculation for the 2-level model was, Eq 3

The final results of the random intercept 2-level (flock-fillet) multivariable logistic regression models (one for SM, another for WB) were reported as the odds ratio (95% CI) of breast myopathies to occur, as an effect of the independent variable(s) included in the model (p < 0.05). An odds ratio > 1 indicates increased odds of a certain myopathy to occur, as a result of the predictor variable to increase by one unit (continuous) or in comparison to a set reference (categorical), whereas the odds ratio < 1 denotes a decreased chance.

Flock demographics variables.

The occurrence of SM was associated with the average live weight of broilers or source of chicks. When broilers were heavier than 2.45 kg at slaughter, they had higher odds of having SM compared with those smaller than 2.30 kg (OR = 1.30, 95% CI 1.01–1.69). Day-old chicks and/or hatching eggs sourced from the USA had significantly higher odds of SM occurrence than domestic-only flocks (OR = 1.50, 95% CI 1.01–2.23).

Flock health variables.

Vaccination for coccidiosis and infectious bursal disease was associated with the occurrence of SM. Flocks that received the coccidiosis vaccine had significantly higher odds of SM occurrence than flocks that did not receive the vaccine (OR = 1.46, 95% CI 1.07–2.00). Similarly, flocks that received the infectious bursal disease vaccine had significantly higher odds of SM occurrence than flocks that did not receive the vaccine (OR = 1.84, 95% CI 1.32–2.54).

Transportation variables.

When truck travel time was less than 69 min, broilers had higher odds of having SM compared with those that travelled 164 min or more (OR = 1.64, 95% CI 1.21–2.22). Likewise, when truck travel time was equal to or more than 69 min and less than 164, broilers had higher odds of having SM compared with those that travelled for 164 min or more (OR = 1.89, 95% CI 1.43–2.49).

Processing plant variables.

When hold time on lairage at plant was equal to or more than 111 min and less than 161, broilers had higher odds of having SM compared with those that were held for 161 min or more (OR = 1.35, 95% CI 1.02–1.80).

Environmental variables.

The temperature during grow-out was associated with the occurrence of SM. When temperature during grow-out was higher than 0°C and lower than 15.8 °C, broilers had higher odds of having SM compared to those flocks for which temperature was equal to or lower than 0°C (OR = 2.58, 95% CI 1.96–3.39). When temperature during grow-out was equal to or higher than 15.8°C, broilers had higher odds of having SM compared to those flocks for which temperature was equal to or lower than 0°C (OR = 1.75, 95% CI 1.31–2.34).

Flock demographics variables.

The occurrence of WB was associated with the average live weight at slaughter or source of chicks. When broilers were heavier than 2.29 kg and smaller than 2.46 kg, they had higher odds of having WB compared with those smaller than 2.30 kg (OR = 1.18, 95% CI 1.01–1.38). When broilers were heavier than 2.45 kg, they had higher odds of having WB compared with those smaller than 2.30 kg (OR = 1.23, 95% CI 1.03–1.47). Day-old chicks and/or hatching eggs sourced domestically had significantly higher odds of WB occurrence than USA-only flocks (OR = 1.32, 95% CI 1.03–1.69).

Flock health variables.

Mortality rate during grow-out and coccidiosis vaccine were associated with the occurrence of WB. When the mortality rate during grow-out was lower than 2.1%, broilers had higher odds of having WB compared with those higher than 3.0% (OR = 1.79, 95% CI 1.43–2.25). When the mortality rate during grow-out was higher than 2.0% and lower than 3.1%, broilers had higher odds of having WB compared with flock mortality higher than 3.0% (OR = 1.76, 95% CI 1.42–2.17). Flocks that did not receive the coccidiosis vaccine had a significantly higher risk of the WB occurrence than flocks that receive the vaccine (OR = 1.86, 95% CI 1.51–2.29).

Transportation variables.

Birds per crate, dead on arrival, and loading duration were associated with the occurrence of WB. When more than 10 birds were in a crate, broilers had higher odds of having WB compared with those equal to or less than 9 birds (OR = 1.46, 95% CI 1.18–1.79). For every one-unit (%) increase in DOA, a 1.04 increase in the odds of WB development was expected. When loading duration was longer than 100 min, broilers had higher odds of having WB compared with those loaded for less than 76 min (OR = 1.48, 95% CI 1.18–1.86).

Processing plant and environmental variables.

Hold time on lairage at plant was associated with the occurrence of WB. When broilers were held on lairage less than 111 min, they had higher odds of having WB compared with those held longer than 110 min or less than 161 min (OR = 1.31, 95% CI 1.06–1.61). When broilers were held on lairage longer than 160 min, they had higher odds of having WB compared with those held longer than 110 min or less than 161 min (OR = 1.52, 95% CI 1.24–1.86). The season affected the occurrence of WB. During summer, a 2.38 increase in the odds of being WB was expected compared to winter.

Flock demographics variables.

Our model shows that the odds of SM and WB increase when the average live weight at slaughter increases, i.e., the odds of SM and WB were increased when the average live weight was ≥ 2460g and 2300g, respectively. The positive association between WB and heavier birds is supported by numerous studies, and it has been proposed that the increased number, diameter, and length of muscle fibers, as seen in heavier broilers, can result in reduced blood vessel density, eventually leading to muscle fibers loss and replacement with connective tissue [48, 49]. To the authors’ knowledge, however, this is the first study documenting an association between SM and heavier birds, although the mechanism underpinning this association remains unclear. Reducing the growth rate and/or the final bodyweight of the birds would likely be able to reduce the occurrence of SM and WB; however, these are trade-offs with expected yield and performance consequences.

In Canada, approximately 78.9% of hatching eggs and chicks are domestically sourced, and 21.1% of broiler hatching eggs and chicks (hatching eggs 17.4%, chicks 3.7%) are imported from the USA under the regulations of the United States-Mexico-Canada Agreement (USMCA) and the World Trade Organization (WTO) [50, 51]. In our dataset, the domestic-only flocks accounted for 78.4%, which is similar to the proportion mandated by the USMCA and WTO agreements. In our cohort, the source of chicks affected the occurrence of SM and WB, with flocks that were entirely or in part domestically sourced showing higher odds of WB. It is unclear how the source of chicks could directly affect development of myopathies, however this parameter likely includes variability at the hatchery and parent flock level, as well as different transportation conditions of hatching eggs and chicks. Variables associated with breeder flocks can include age, strain, management practices, and lighting, all of which can affect the growth and live performance of future placement chicks, their processing yield, and ultimately the occurrence of breast myopathies [52–56]. Environmental factors during the incubation period can also affect muscular physiology. For instance, increased temperatures during late incubation, as well as mid or late hatching, have been shown to reduce the severity of myodegeneration in broilers later in life [57]. Another study documented that WB was more prevalent in broilers hatched from eggs stored longer (8 to 14 d), compared to short storage (1 to 7 d) before incubation [45]. Detailed evaluation of how factors associated with hatchery and breeder flocks may influence development of breast myopathies is warranted in future studies.

In our study, unconditional association analysis showed that the odds of having SM was 2.25 times higher in female broilers, while the odds of having WB was 1.56 times higher in males (S2 Table). Pascual et al. [58] also reported that females had a higher prevalence of SM. During our stepwise regression analyses using the likelihood ratio test, however, the sex variable was removed from the final multivariable regression models of both SM and WB. This is likely due to unknown functional relationships between sex and other independent variables, when explaining SM and WB occurrence [59]. Aguirre et al. [31] also reported that sex was not a significant variable for breast myopathies, although raw frequency indicated that male broilers had more frequently WB and WS.

It has been also suggested that nutrition and genetics affect the occurrence of breast myopathies [19, 60]. In the present study, all flocks were fed with a corn-based diet and had the same genetic background (Ross 708), and therefore, these variables were not used to build the multivariable model.

Flock health variables.

Flocks that received coccidiosis vaccine and/or infectious bursal disease vaccine had a higher risk of SM, while flocks that did not receive coccidiosis vaccine had a higher risk of WB. Vaccination might have affected the initial growth performance of the flocks, even though final weight at slaughter was similar between treated and untreated ones. Gautier et al. [61] reported that, up to 20 d, weight gain and feed intake were lower in broilers that received the coccidiosis vaccine compared to those that did not. Lee et al. [62] also showed that vaccination against coccidia had a negative impact on weight gain of broilers between 22 to 28 d. Therefore, it is possible that an initial decrease in the flock’s growth rate due to vaccination may influence development of myopathies. This is supported by Meloche et al. (2018), who reported that lighter broilers during the first stage of the rearing period had decreased prevalence of WB at slaughter [27]. Similarly, our logistic regression analysis showed an almost 2-fold increased odds of developing WB in non-vaccinated broilers compared to vaccinated ones. This effect, however, is likely not to be the same for all vaccines, as vaccination schedules and the response to different antigens varies [63]. For instance, administration of the infectious tenosynovitis vaccine was not associated with myopathy development, in our study. While growth performance could be the underlying reason for the associations between vaccination and myopathies, it remains unclear why SM and WB would present positive and negative associations, respectively, with coccidiosis vaccination.

The occurrence of WB was associated with mortality rate during grow-out, but the mechanism of the association is not clear. Mortality rate is affected by overall health, housing factors and management routines including flock size, stocking density, type of ventilation, drinking and heating systems [64]. The median mortality rate during grow-out reported in our study (i.e., 2.33%) is similar to what was reported by Torrey et al. (i.e., 2.52%) [65]. In this study, the authors reported that total mortality did not differ between conventional and slow-growing strains, even though conventional strains showed the greatest incidence of myopathies. Overall, other hidden factors, broadly associated with differing management practices, could affect the relationship with the occurrence of breast myopathies. Moreover, it should be considered that mortality rate during grow-out is a proxy variable for development of myopathies, since breast fillets were not likely assessed from birds with severe health issues, which would have been culled before slaughter. Further work is required to identify the association between mortality rate during grow-out and the occurrence of breast myopathies.

Transportation and processing plant variables.

The occurrence of SM was affected by the truck travel time and hold time on lairage at plant, whereas the occurrence of WB was affected by birds per crate, DOA, loading duration, and hold time on lairage at plant. The relationship between these variables and the occurrence of SM or WB is unclear, and not mechanistically defined. In our cohort, the reported time interval from feed withdrawal to deboning was no more than 24 hrs, and this time is too short to significantly alter the tissue structure of the pectoral muscle. While scarring (fibrosis) in the skeletal muscle develops between 1 to 3 weeks after the initial injury [66, 67], it has also been suggested that increased corticosterone levels (as seen in stressful events) shortly before slaughter may affect meat quality through increased proteolysis or altered development of rigor mortis [68, 69]. Increased proteolytic activity is mediated mainly through matrix metalloproteinases (MMPs), which include enzymes such as collagenases and gelatinases that play a role in the degradation of structural components of the connective tissue [70]. Stress can deplete glycogen and ATP reserves in the myofibers, leading to higher final pH and softer meat [71]. It is possible that pre-slaughter stressors might have affected certain visual and tactile characteristics of the fillets, affecting our classification of myopathies. For instance, altered development of rigor mortis might have affected our ability to accurately classify WB, especially in the milder phenotype (WB1).

Unconditional association showed that SM was approximately 1.5 times more likely to occur in the processing plant employing the water-chill, compared to the air-chill method. Mechanical agitation of water during water-chilling using an auger increases moisture absorption [72]. It is possible, therefore, that increased water in the interstitium may promote myofiber separation, leading to increased frequency of SM. However, the chilling method variable was eliminated during backward elimination process using the likelihood ratio test, and was not included in the final multivariable models for SM or WB. This was likely caused by unknow functional relationship between the chilling method and other variables in explaining the frequency of SM and WB.

Environmental variables.

A higher temperature during grow-out was positively correlated with the occurrence of SM, and WB had higher odds of occurring when fillets were sampled in the summer season. Environmental temperature can affect the physiology of the breast muscle. For instance, broilers experiencing high temperatures show different protein profiles in the breast muscle and variations in meat quality [69]. In turn, broilers with a rapid growth rate and high breast yields have reduced thermoregulatory capacity, making them vulnerable to thermal challenges [73]. Thus, it appears that higher temperature contributes to the development of SM and WB, although the mechanism of this association remains to be elucidated. The highest tier of temperature during grow-out category included average temperatures between 15.8°C and 22.7°C, while the overall temperature during the summer months spanned between 11.7°C (June 2019) and 22.7°C (August 2019). Considering that summer in Canada is not as hot compared to other countries with large poultry industries (such as Brazil or the southern USA), our broiler flocks were not exposed to extreme heat during the grow-out, and we did not further stratify the summer temperatures of our study.

In our model, precipitation during grow-out was removed during stepwise multivariable regression analysis. Relative humidity has been shown significantly to influence birds’ capacity to thermoregulate [74], and therefore the thermal-humidity index has been used to assess the impact of humidity and temperature on the growth performance of broilers [75]. It will be useful if future research will be done to investigate the association between the thermal-humidity index and the occurrence of breast myopathies.

The ICC values at the flock level of clustering were relatively small (< 1.8%), and the between-flock variance was relatively low compared to the within-flock variance. Low variability between the flocks could be explained by the uniformity of the broiler flocks included in our cohort, especially considering the similar target live weight, which is one of the most important predictor for SM and WB development [32]. The ICC values for SM were slightly higher than those of WB, indicating a greater variation in prevalence of SM compared to WB, especially at the single-bird level. This may indicate different mechanisms for development of WB and SM, even though they occur together macroscopically.