9. Haematological characteristics of Australian mammals

INTRODUCTION

ESTABLISHING REFERENCE VALUES

The calculation of a meaningful reference interval relies on the selection of ‘healthy’ individuals; however, ‘health’ is more difficult to define than disease because it depends on how rigorously it is investigated and therefore, a reference interval may be affected by the methods used to determine if individuals are ‘healthy’. Using only the results of a clinical examination to select animals may not detect subclinical disease; in some cases, endemic disease may exist in a colony or free-living population and a decision must be made whether to screen animals for a particular disease and subsequently exclude from the selection pool those that are positive or to accept the subclinical disease as part of the characteristics of the population.

The use of ancillary tests (such as radiology and laboratory assays) may allow detection of subclinical disease but will significantly add to the cost of the exercise. Furthermore, an inherent problem arises when using laboratory assays in this role, namely, deciding what level for a given analyte should be used to exclude an individual from the sample pool. The presence of ‘outliers’ in the data may be identified mathematically (Lumsden and Mullen, 1978) but their detection is not routinely recommended because values from skewed populations may be misconstrued as outliers (Horn et al., 1998). However, an animal with a clearly defined pathological process (such as anaemia, as evidenced by decreased haematocrit, erythrocyte concentration and haemoglobin concentration) should be removed from the selection pool.

In many cases, the selection of individuals for the composition of reference intervals is simply based on availability, rather than selecting the best animals for the purpose. The number of animals assessed is an important factor influencing the statistical methods used to assess the data, as well as the ‘width’ of the interval.

The International Federation of Clinical Chemists (IFCC) guidelines recommend that samples be obtained from 120 individuals if the data has parametric distribution and 200 individuals if the data has non-parametric distribution (Solberg, 1993). The difficulty in obtaining samples from that number of animals of any species of Australian mammal, as well as the cost associated with the analysis of the samples, would almost certainly preclude adhering to these recommendations.

Commonly in veterinary medicine, the central 95 per cent of values from a subset (sample) of individuals from a population is used as an estimate of that population, which is typically described by the 2.5 and 97.5 percentiles. Meaningful parametric or non-parametric estimation of the percentiles requires that the minimum sample size is 100/P where P is the lower percentile (e.g. 100/2.5 = 40) (Lumsden and Mullen, 1978). When there are less than 40 observations from clinically healthy individuals with no obvious outliers, then the observed lowest and highest values are best estimates of the 95% reference limits (Lumsden, 1998). When the data has Gaussian distribution, 95 per cent of healthy (‘normal’) individuals fall within the interval defined by the mean plus 1.96 standard deviations (upper limit) and the mean minus 1.96 standard deviations (lower limit) and this calculation is commonly used to establish a clinical reference interval (Lumsden and Mullen, 1978).

As the sample size (i.e. the number of observations) increases, there is a decrease in the width of the interval and when there are more than 60 observations (for parametric data) or 120 observations (for non-parametric data) there is a decreasing effect on the reduction of the interval width (Lumsden, 1998).

Many other statistical methods have been proposed to allow representative reference intervals to be calculated from smaller samples sizes than recommended by the IFCC and may be applicable to the small populations of most species of wildlife that are encountered in practice (Horn et al., 1998, 1999; Virtanen et al., 1998; Wright & Royston, 1999; Linnet, 2000).

The ultimate aim is to produce a reference interval that the user is confident reflects the characteristics of the studied population and consequently, any divergence reflects processes that require clinical attention. To achieve this aim the selection of appropriate animals and appropriate statistical methods are both crucial.

GUIDE TO INTERPRETING THE TABLES

For each species the results of individual studies are listed and referenced. The number of animals in the study is shown in parentheses and if it differs between analytes in the same study it is reported after the value of the analyte. Multiple samples from the same individual are noted in the column heading. Single sex populations are stated, otherwise combined male and female data are presented. Where significant physiological differences were reported in the same study, these are presented in a separate column. Additional relevant information is given as footnotes. The data in the body of the tables is presented either as an interval or as mean ± standard deviation (SD) or standard error of the mean (SEM).

The data is presented in the standard units used for haematology in Australia, which are a practical application of metric units, rather than strictly conforming to the Système International d’Unités (SI units) (Ogrim and Vaughan, 1977). For example, concentrations are reported as per litre (cubic decilitre), rather than per cubic metre and ‘amount’ is reported in grammes rather than moles. Where necessary, data published in nonstandard units has been converted (see Appendix 2).

In addition to the values from Australian mammals, the haematological values of some ‘non-Australian’ species have been included; typically, these are closely related species to those found in Australia (e.g. tree-kangaroos from New Guinea).

Abbreviations and symbols

Stay updated, free articles. Join our Telegram channel

Join