Association between in-transit loss, internal trailer temperature, and distance traveled by Ontario market hogs

Abstract

An observational study was conducted from July to October 2004 to determine the association between in-transit losses of swine and internal trailer temperature after controlling for loading density, trip distance, herd size, and random trip effect. A convenience sample of 3 trucking companies was used to collect temperature, relative humidity, and global positioning data for 104 trips that delivered 21 834 pigs from 371 producers to Ontario abattoirs. The association between in-transit loss and trailer temperature was determined using the 90th percentiles of internal temperature for each trip. Average loading density was 0.36 m²/100 kg pig (range 0.28 to 0.50 m²/100 kg pig). Average in-transit loss was 0.12%; however, 94% of producers experienced no losses. As the 90th percentile of internal trailer temperature increased from a range of 8.6°C to 23.3°C to a range of 23.4°C to 26.1°C, average in-transit loss ratio increased approximately 3-fold, with an additional 2-fold increase as the range increased from 26.2°C to 28.9°C to 29.0°C to 30.5°C. As the 90th percentile of temperature increased by 1°C over the full range of temperatures in this study, in-transit loss was expected to increase 1.26 times. The in-transit loss was expected to decrease 0.81 times for each 50-km increase in distance traveled between the farm and the abattoir.

Introduction

In-transit loss is a term used to describe pigs that die after leaving the farm but before being stunned at the abattoir. These deaths are a concern for the swine industry both in terms of welfare and economic loss. Over the past 35 y, estimates from observational studies have placed in-transit losses in the range of 0.07% to 0.14% in Britain, 0.085% to 0.15% in the United States, and 0.18% to 0.20% in Canada (1–6). Estimates during the past 10 y range from 0.08% to 0.22% (5,7). Factors associated with in-transit losses may include finishing-pig rearing and handling management; shipping methods (quality of home-farm shipping facilities); shipping conditions (trailer design, ride characteristics, trailer density, distance, and external temperature and relative humidity); health and genetics of the pig, and lairage conditions at the abattoir (1,3,5,6,8–19). Of these factors, higher temperature and relative humidity during the transport process contribute most to in-transit death losses (5,9).

The upper critical temperature (UCT) is the point at which an animal maintains homeothermy by actively increasing heat dissipation (8,20–22). The UCT for finishing pigs may be as low as 23°C (73°F), depending on factors such as body size, feeding, and hydration status, and conditions of transport such as load density (15,18,20). A finishing pig achieves sensible heat dissipation through radiant cooling by heat transfer to cooler surroundings, conductive cooling by direct contact with a cooler surface, and convective cooling by air movement. Since pigs do not sweat, their primary means of evaporative cooling occurs through panting, which is ineffective at high relative humidity levels (23). However, evaporative cooling is the primary means of heat dissipation at higher environmental temperatures (23). External temperature and relative humidity outside the trailer are related to conditions inside the trailer (18,19). Factors such as animal density (which affects heat and moisture production by the pigs in the trailer), trailer design, and velocity (which influence air movement in the trailer) affect internal trailer conditions (13,19,24–26). Although external temperature and relative humidity are associated with in-transit losses, it is likely that the microclimate within the trailer is equally or more strongly associated with these losses.

The purpose of this study was to determine the association between in-transit losses and internal trailer temperature after controlling for other factors known to be associated with in-transit losses, including pig density, trip distance, herd size, and the random effect of the trip.

Materials and methods

The internal temperatures in 9 trailers that transported pigs on 104 trips from the farm or assembly yard to the abattoir between July and October, 2004 were collected. Temperature and relative humidity data loggers (The HOBO; Onset Computer Corporation, Bourne, Massachusetts, USA), that recorded information once every minute, were installed in the top, bottom, and back compartments of the trailer. Global positioning systems (GPS) (Turnpike Global Technologies, Stoney Creek, Ontario) were installed in the trailers to delineate each trip’s beginning and ending. The GPS and HOBO data were merged to create a longitudinal dataset with observations at 1-minute intervals. Finally, number of pigs and pig loading density (m²/100 kg pig), calculated from internal trailer area measurements, were added to the data set. The average live market weight may range from 100 to 118 kg and as a barn is emptied, may include pigs that are substantially smaller. However, for purposes of this study, it was assumed that average live market weight was 112 kg. The loading density was adjusted to express the area per 100 kg pig.

The longitudinal data set also included information on when the truck was in motion or stopped, total stopping time in the trip for stops over 10 min, average and maximum speed between observations, and the distance traveled between observations. The total distance traveled (km), average of the maximum speeds for each period of motion (km/h), average of the average speeds for each period of motion (km/h), total stopping time during the trip (min), number of pigs on the trailer when all pigs had been loaded, and density (m²/100 kg pig) were determined for each trip. Each observation also included the trucking company, truck, trip, and destination (abattoir). The original longitudinal database was sorted by trip number and internal trailer temperature to create specific categories of temperature for each trip; the 90th percentile for temperature was determined for each trip.

The Ontario Pork Producer’s Marketing Board provided data on in-transit losses for the 104 study trips, including number of pigs shipped by each producer each day, number of pigs that died in-transit per producer; producer identification number, transport company, abattoir identification, and date. In-transit losses included pigs that died on the truck, while being loaded or unloaded from the truck, or in lairage at the abattoir. These data were merged within the 2 described longitudinal databases. Because some truckloads included pigs from more than 1 farm, a total of 371 producers’ pigs were transported on the 104 trips resulting in 654 producer-by-trip combinations. Two producers were represented on 6 trips, 10 on 5 trips, 17 on 4 trips, 34 on 3 trips, 114 on 2 trips, and 194 producers had pigs on only 1 trip. All data manipulations were performed using computer software (Microsoft Access and Excel; Miocrosoft, Redmond, Washington, USA). The final 2 cross-sectional databases based on percentiles of internal trailer temperature and the original longitudinal dataset were exported using computer software (Stata Statistical software, Version 7.0; Stata Corporation, College Station, Texas, USA), for further analyses.

The dataset that included the 90th percentile of trailer temperature was used to develop the association between in-transit loss and temperature. The number of pigs that died per producer per trip was modeled as an outcome using a negative binomial distribution with the number of pigs marketed by that producer that day multiplied by the total trip length in minutes as the time component for calculation of the rate of death. Fixed effect independent variables included in the initial model were the 90th percentile of internal trailer temperature; internal trailer relative humidity; total distance traveled; pig density (m²/100 kg pig); speed traveled (average of the average speed or average of the maximum speed: km/h); total number of stops of at least 10 min; and numbers of pigs shipped by individual producers in 2004 (as a measure of farm size). Quadratic and cubic functions were assessed against the outcome and retained for further analyses if P < 0.05. Multivariable models were built using a backward selection process. Interaction terms were tested and retained if significant. The incidence rate ratio (IRR) was used as the measure of strength of association between fixed effects and in-transit loss in these models (28). The fixed effect models were developed and tested for goodness of fit, outliers, and leverage using computer software (Stata Statistical software; Stata Corporation).

Mixed models were produced by adding the significant fixed-effect variables to trip number and truck number as random effects. Fixed effect variables with P > 0.05 were dropped from these models using a backwards elimination process. The mixed models analyses were conducted using a Glimmex macro (SAS, version 8.2; SAS Institute, Cary, North Carolina, USA) for Poisson distributions.

Results

A convenience sample of 3 trucking companies was used in this study. These companies used 3-tier, 14.5-m to 16-m (48-ft to 53-ft) trailers, and transported a minimum of 190 000 pigs to Ontario abattoirs in 2004. During this study, 9 drivers made 131 trips to the abattoirs. Complete temperature, relative humidity, and GPS data were collected for 104 of these trips between July 8 and October 12, which delivered 21 834 pigs to the abattoirs. A total of 371 producers’ pigs were transported on the 104 trips, resulting in 654 producer-by-trip combinations.

The range of internal trailer temperature values is described in Table I. The 90th percentile of internal trailer temperatures was 26.3°C. Average in-transit loss per producer was 0.12% for a total of 26 pigs. However, only 21 of the 371 producers lost pigs on these 104 trips: 94% of producers experienced no losses. The greatest number of dead pigs for 1 producer on 1 trip was 6 of the 213 pigs shipped. The trucker did not believe there was anything unusual about the pigs or the trip and did not have an explanation for these deaths. However, another producer shipped a small number of pigs on multiple trips and experienced an overall in-transit mortality of 12.5%. The average pig density was 0.36 m²/100 kg pig, with a range of 0.28 to 0.50 m²/pig.

Table I.

Range of values for factors describing 104 trips transporting finisher pigs from Ontario herds to abattoirs in July to October 2004

	Percentilesa
Factor	10th	25th	50th	75th	90th	100th
Average internal trailer temperature (°C)b	18.5	20.7	22.9	24.8	26.3	27.6
Loading density (m2/100 kg pig)	0.43	0.41	0.37	0.31	0.29	0.28
Pigs shipped by producers in 2004	165	376	693	1746	3712	11 404
Pigs shipped by producers on 1 truck	3	7	14	33	100	240
In-transit loss ratio per producerc	0.0	0.0	0.0	0.0	0.0	12.5
Number of times truck stopped for ≥ 10 mind	2	3	5	7	8	11
Total distance traveled (km)	134	181	833	1092	1113	1146

^a Average for 104 trips.

^b Based on the average value in 3 locations on a truck, calculated from 52 293 observations recorded at 1-minute intervals during 104 individual trips.

^c Number of pigs that died ÷ number shipped by the producer on these 104 trips.

^d Includes stops to load and unload pigs.

As the internal trailer temperature increased, in-transit losses also increased (Tables II and III). There was approximately a 3-fold increase in average in-transit loss ratio as the 90th percentile of internal trailer temperature increased from a range of 8.6°C to 23.3°C to a range of 23.4°C to 26.1°C, with an additional 2-fold increase as the range increased from 26.2°C to 28.9°C to a range of 29.0°C to 30.5°C (Table II).

Table II.

Average in-transit loss^a for the 90th percentiles of temperature,^b pig density, and distance travelled on 104 trips transporting pigs from Ontario herds to abattoirs in July to October 2004

Percentile	Range	Average in-transit loss (%)
90th percentile of internal truck temperature (C°)
90th	29 to ≤ 30.5	0.21
75th	26.2 to ≤ 28.9	0.17
50th	23.4 to ≤ 26.1	0.11
25th	8.6 to ≤ 23.3	0.04
Loading density (m2/100 kg pig)
90th	0.29 to ≤ 0.30	0.18
75th	0.31 to ≤ 0.36	0.12
50th	0.37 to ≤ 0.39	0.11
25th	0.41 to ≤ 0.49	0.14
Total distance travelled (km)
90th	1113.6 to ≤ 1146.5	0.00
75th	1092.4 to ≤ 1113.6	0.18
50th	181.5 to ≤ 833.2	0.13
25th	102.4 to ≤ 181.4	0.20

^a Number of pigs that died ÷ number shipped by each producer on 1 truck on 1 day.

^b Based on average temperature in 3 locations on a trailer, calculated from 52 293 observations recorded at 1-minute intervals on 104 individual trips.

Table III.

Association between in-transit losses and the 90th percentile of internal trailer temperature and distance travelled for 104 loads of Ontario market pigs in 2004

	90th Percentile of temperaturea
Random effects	Value	Sx̄	P
Trip number	4.76	0.994	< 0.0001
Error term	0.22	0.013	< 0.0001
Fixed effectsc	IRRb	Sx̄	P
Intercept	0.00	2.334	< 0.0001
Temperaturea	1.26	0.085	0.007
Distance (50-km increments)	0.81	0.045	< 0.0001

^a Average temperature in 3 areas on the trailer. The 90th percentiles of temperature are used as independent variables for the model.

^b Incidence rate ratio (IRR) indicates change in in-transit loss for every 1°C increase in temperature or 50-km increase in distance traveled, for example, in-transit loss increases 1.26 times with each 1°C increase in the 90th percentile of truck temperature.

^c A total of 371 producers’ pigs were transported on the 104 trips, resulting in 654 producer-by-trip combinations.

Incident rate ratios were used in each of the multivariable models to express the effect of each fixed variable on in-transit loss. An IRR of > 1 means that the factor contributes to in-transit losses, and an IRR of < 1 indicates a sparing effect. More specifically, as the 90th percentile of temperature increases by 1°C, in-transit loss is expected to increase by 1.26 (Table III).

Herd size, truck driver, and transport company were not associated with in-transit losses. However, trip number explained 96% to approximately 97% of the variation in in-transit loss not accounted for by internal trailer temperature and distance traveled (Table III).

Trip length affected in-transit losses. For each 50-km increase in distance, in-transit loss was expected to decrease 0.81 times (Table III). Internal trailer relative humidity was not associated with in-transit loss after controlling for trailer temperature and distance traveled (data not shown).

Analysis of outliers, leverage cases, and influential cases identified no observations that required removal from the analysis.

Discussion

The transporters in this study shipped 21 834 pigs to abattoirs, and 0.12% of these pigs (12 pigs per 10 000) died in-transit. During this same period (July 8 to October 12, 2004), 1.4 million pigs were shipped in Ontario and 0.14% (14 pigs per 10 000) died in transit (27). Of the 5.5 million pigs marketed in Ontario in all of 2004, approximately 0.10% (10 pigs per 10 000) died in transit (27). Previous observational studies investigating in-transit death and using transport companies as the basis of their sampling included 15 717 to 268 187 market pigs; losses between 0.08% and 0.27% were reported (8,12,14,29). Only 1 study (29) observed shipments that occurred at approximately the same seasonal time as those herein; however, fewer truckloads were observed in that study. To our knowledge, this is the first observational study investigating the effect of internal trailer temperature and other factors on intransit loss.

As expected, the number of in-transit losses was greater at higher internal trailer temperatures. Pigs are more sensitive to increases in dry temperature than in relative humidity (30). However, at temperatures in excess of 30°C with relative humidity in excess of 88%, both sensible and evaporative cooling mechanisms for the pig are severely compromised, and these environmental conditions may cause physiological stress in pigs (5,18). The 90th percentile of temperature for each of the 104 trips were based on the longitudinal data, that is, internal trailer temperature recorded at 3 locations in a trailer at 1-minute intervals. Pigs were exposed to the 90th percentile temperature or a higher temperature for 10% of the time during each trip. This percentile was used to reflect the duration of exposure to the highest temperatures of each trip. The 90th percentile temperatures identified in the current study identify a lower critical temperature than those reported in a previous study that also examined transport losses (31). In Europe, fans have been used in trailers to cool pigs, both in clinical trials and under actual transport conditions (26). Pigs transported in Ontario would benefit from this practice when trailer temperatures are high during the summer months.

Previous research has primarily focused on the effect of time in transit on measures of poor carcass quality and behavioral characteristics of pigs on the trailer and in lairage (5,16,27). One study found a higher in-transit mortality on trips of 50 to 150 km compared with longer trips (32). However, these European findings may not apply because pigs in Ontario travel a greater distance between the farm and the abattoir. In the current study, the shortest trip was 102 km, and 25% of the trips were less than 181 km. Under conditions in Ontario, increasing the distance by 50 km increments was associated with lower in-transit losses. The sparing effect of distance on in-transit loss may reflect that pigs had time to recover from the stress of loading (5,33,34). A pig’s heart rate increases at loading and gradually decreases during transport (5). Pigs remain standing during shorter journeys and lie down after about 3 h on the road providing there is sufficient room to lie down (35). If the pig’s heart rate is low when the pig is unloaded from the truck, then the risk of death during unloading is decreased. Other trip characteristics, such as longer highway driving, may have allowed temperatures at the 90th percentile of internal conditions to fall below external conditions (7). Highway driving was likely represented in the model by trip distance.

There were additional, significant unmeasured factors associated with the trip that explained approximately 97% of the random variation of in-transit loss beyond trailer temperature and distance traveled. The random effect of driver was not significantly associated with in-transit loss in this study, while in 2 previous studies that collected driver reports over a 2-year period, indicated that in-transit loss differed by driver (8,12). Lack of agreement among studies may be due to differences in observational periods, study populations, and managerial influence on transport procedures for drivers in the 3 transport companies in this study (8,12). In the current study, a convenience sample was used and drivers were aware that data were being collected from their loads. Therefore, drivers may have been more uniform in driving style than in other studies. The earlier studies considered trip characteristic factors such as driver opinion of journey, ease of loading, whether the journey was rough or smooth, and the effect of time of day (8,12). These factors were not specifically included herein, but may have been represented by random trip effect.

Several earlier studies have found a positive association between pig density and in-transit loss (13,16,36); however, our results did not suggest this association when data were adjusted for temperature. As loading density increases, heat production on a trailer increases (19,23,24) as does the ability for pigs to adjust body posture to facilitate heat loss. It is likely that after controlling for trailer temperature, the impact of pig density on in-transit loss is effectively explained, because pig density is an explanatory antecedent variable in the relationship between trailer temperature and in-transit loss (37). In the current study, the pig density was ≥ 0.375 m²/100 kg pig on 50% of trips, and trips were likely of sufficient duration to allow pigs to recover from the stress of loading.

Internal trailer temperatures at the 90th percentile were useful predictors of in-transit death. These values described the actual sensory experience of shipped pigs and provided an expression of time of exposure. However, as the study used a convenience sample of trucks and drivers selected by willingness to cooperate, and company size, results may not adequately represent all trips experienced by pigs in Ontario during the summer months. There is likely a greater diversity in trailer and driver style and variation in pig density. Also, although this study monitored the transport conditions for more than 21 000 pigs, only 26 pigs died in-transit. Therefore, further studies examining the impact of trailer temperature, loading density, and travel distance are warranted.

The temperature and humidity sensors used in this study are relatively inexpensive compared with the cost of in-transit loss, and could be used on trucks throughout North America. When trailer temperatures are too high, a combination of sprinklers and fans could be activated to cool pigs. These monitoring and cooling systems would be expected to reduce in-transit losses.