INTRODUCTION

In the assessment of animal welfare the use of physiological indicators is common. Within these, blood reference values have been an important tool for practitioners in order to monitor health and welfare at individual or group level, or the immediate response of an animal towards a determined stressor (^{Waran 2007}, ^{Geffré et al 2009}, ^{Cozzi et al 2011}). For example, changes in plasma concentration of glucose, urea and proteins have been associated to significant metabolic cost by the animal, while immunosuppression can indicate a potential health risk, or a reduction in growth rate (^{Barnett and Hemsworth 1990}).

To identify those individuals that have health alterations, values from the blood analysis are usually compared with the population means or ranges of standard values (^{Herdt 2000}), which in the case of horses are usually based on Thoroughbreds, sport or pleasure horses living under good husbandry conditions (^{Pritchard et al 2009}). The use of inappropriate reference values increases the risk of erroneous conclusions by the clinician or the researcher, and may lead to further unnecessary or inappropriate analysis (^{Tsang et al 1998}).

Haemato-biochemichal values obtained abroad may not be fully applicable under local conditions due to genetic factors, differences in environmental and husbandry practices and animal's function (^{Gul et al 2007}, ^{Pritchard et al 2009}). ^{Gul et al (2007)} reported haemato-biochemical values for apparently healthy equids in Pakistan, finding increased values for biochemichal variables in horses when compared to the literature. ^{Pritchard et al (2009)} established haematological and biochemical reference intervals for working horses in Lahore (Pakistan), finding differences with reference limits from the United Kingdom (UK), for the variables studied.

In Chile working horses are still the main source of income for many families (^{Tadich and Stuardo 2014}). When the welfare of these horses is evaluated blood variables are used for assessing their health state and international reference intervals are usually applied (^{Tadich et al 2011}).

The aim of this study was to establish haematological and biochemical reference intervals for apparently healthy working horses under Chilean local husbandry conditions, and compare them with those proposed in the literature.

MATERIAL AND METHODS

RESULTS AND DISCUSSION

The RIs for common haemato-biochemical variables were calculated for apparently healthy urban-working horses in Chile (table 1) and compared with those in the literature (table 2) for horses with different functions and geographical locations.

A total of eleven haematological and fifteen biochemical variables were assessed, from a population of 320 sampled working horses. Due to damage of samples during handling, aberrant results provided by the laboratory and the establishment of outliers, not all variables have the same sample size, being 165 horses the smallest sample (fibrinogen) and 316 individuals the largest one (urea, proteins, albumin, phosphate and GGT) for calculating the reference interval (table 1). This allowed excluding atypical values that could have been the product of inadequate analysis (^{Wagemann et al 2014}) or from horses with an underlying pathology that was not detected at the moment of clinical examination. Despite elimination of aberrant data and outliers all variables included over 120 individuals, in accordance to the minimum sample size recommended by the IFCC (^{Solberg 1987}, ^{Friedrichs 2012}).

The final sample size, descriptive statistics, the RIs established, the method applied, and number of outliers are presented in table 1. Differences in the RIs established could be product of differences in laboratory analysis, but could also reflect the diversity in genetic characteristics, husbandry practices, geoclimatic conditions and adaptation to the work performed by the individuals that were evaluated (^{Satué et al 2012}, ^{Padalino et al 2014}).

Table 2 shows the comparison between RIs for haematological and biochemical variables established for Chilean urban-working horses and those established for Thoroughbred horses in Europe (^{Knottenbelt 2006}), Chilean Creole Horses in Chile (^{Wittwer 2012}) and working horses in Pakistan (^{Pritchard et al 2009}). The RIs calculated were similar to those reported for working horses in Pakistan (^{Pritchard et al 2009}), and the Chilean Creole horses studied by ^{Wittwer (2012)}; and differed more from those proposed by ^{Knottenbelt (2006)} for Thoroughbred horses, especially for RBC, Haemoglobin and enzymes (table 2).

Red blood cell count and haemoglobin RIs calculated are similar to those proposed by ^{Pritchard et al (2009)} and lower than those provided by (^{Knottenbelt 2006}) and ^{Wittwer (2012)}. The later two intervals were developed for horses dedicated mainly to sports, this could explain the difference since these horses are usually under better feeding practices; at the same time training could also influence the concentration of erythrocytes and haemoglobin. In relation to this higher RBC counts ^{Satué et al (2012)} reported that "hot-blooded breeds" have higher RBC count, Hb and PCV than draught horses, ponies or the "cold-blooded breeds", for which PCV as low as 24% can be found in healthy animals. This cold-blooded breeds category includes Chilean creole horses and the crossbred horses used for draught work in Chile. On the same line, training can affect basal levels of RBC count, Hb and PCV. For example endurance trained horses have lower resting values of these variables (^{Satué et al 2012}), type of exercise comparable to the draught work performed by urban working horses. On the other hand, a decrease in RBC and haemoglobin due to anaemia has been considered as a cause of low values, but Chile is free of equine infectious anaemia and Trypanosoma spp.¹; and iron deficiency in horses has been poorly reported (^{Reed et al 2004}), but considering the sometimes inappropriate feeding practices that working horses receive this possibility cannot be ruled out. Most working horses in Chile have been reported to have access to pasture, either in public green areas in the city or near rubbish dumps (Tadich et al 2008), being likely that they could be obtaining iron from the soil (^{Humphries et al 1983}, ^{Brommer and Sloet van Oldruitenborgh-Oosterbaan 2001}).

The N:L ratio has been reported as a more reliable indicator of stress than the use of cortisol alone (^{Stull and Rodiek 2000}), with a normal fluctuation between 1.5 and 2.5 (^{Rossdale et al 1982}). The N:L ratio could be of value when studying working horses, and increases could be expected to occur in an overworked horse. No reference intervals for this ratio were provided by the literature used for comparison (^{Knottenbelt 2006}, ^{Pritchard et al 2009}, ^{Wittwer 2012}) (table 2).

The percentage of horses that presented variables above or below the RI provided in the literature (table 3) is higher when comparing to the RIs for Thoroughbreds (^{Knottenbelt 2006}), and horses tend to adjust better to the RI for working horses in Pakistan (^{Pritchard et al 2009}) and Chilean creole horses (^{Wittwer 2012}). These results support ^{Pritchard et al (2009)} conclusions that the reference intervals established by them are more appropriate when assessing a working horse population. The major differences were found for enzymes, where the median and mean of LDH and GSH-Px were out of the reference limits established by ^{Knottenbelt 2006} and ^{Wittwer (2012)} respectively, no reference limit was provided by ^{Pritchard et al (2009)} for these enzymes. GSH-Px participates in reduction of oxidative processes and contains the major percentage of blood selenium (^{López Alonso et al 1997}, ^{Oh et al 1974}). The assessment of GsHPx in chilean horses is important since clinical cases of miodegeneration and steatosis due to selenium deficiency have been reported (^{Araya et al 2004}), together with insufficient concentrations of selenium in Chilean grasses in order to satisfy equine requirements established by the National Research Council (^{NRC 2005}, ^{Ceballos et al 1999}). Over 60% of horses in the present study presented selenium deficiency identified by a decrease in serum GSH-Px according to the reference value provided by ^{Wittwer (2012)} (table 3), but none presented clinical signs of selenium deficiency.

In relation to CK and AST, the upper reference limits established were higher than those reported in the literature for both enzymes (^{Knottenbelt 2006}) (table 2), but similar to other reports in working horses (^{Gul et al 2007}, ^{Pritchard et al 2009}, ^{Tadich et al 1997}). High serum concentrations of these enzymes could reflect a lack of adaptation to work resulting in a low-grade chronic muscular damage (^{Tadich et al 1997}). This differs with the results of ^{Vergara and Tadich (2015)} for tourism working horses in Chile, where no significant increases of these enzymes were found after work.

The upper reference limit for calcium was higher in the Chilean working horses than in the other three reference intervals. In the horse, the intestine is not a regulatory point for calcium homeostasis and intestinal calcium absorption is always turned on, thus feeding high dietary calcium increases the amount of calcium that enters the blood (^{NRC 2005}). A more detailed study on feeding practices in working horses is required to better understand these changes.

In conclusion, the reference intervals calculated differ from those found in the literature for other equines, mainly those calculated for sport horses. Working horses seem to share some similarities around different countries, which could be the result of adaptation to work and precarious husbandry conditions, especially in what refers to feeding practices that are reflected in their haemato-biochemical values.