Equis ISSN 2398-2977

Blood: biochemistry - overview

Contributor(s): Ruth Morgan, Vetstream Ltd

Overview

Sampling

This article is available in full to registered subscribers

Sign up now to purchase a 30 day trial, or Login

Tests

Methodologies

  • A number of different methodologies may be available for any given test.

It is important to be aware of which method is used so as to correctly interpret the results.

Availability

  • Many tests are routinely available - if in doubt, contact the laboratory to confirm availability of a specific test.

Validity

Sensitivity

  • The ability of a test to produce a positive result in all animals that do have disease.

A high sensitivity may be associated with a reduction in speicificity as false positive reactions may occur.

Specificity

  • The ability of a test to produce a positive result only in animals that do have disease.

A high specificity may be associated with a lower sensitivity as some false negative may occur.

Predictive value

  • Predictive value = how well the test performs in a given population of animals.
  • Influenced by the prevalence of the disease in the population being tested.
  • Positive predictive value is probability of an abnormal test result indicating presence of disease.
  • Negative predictive value is probability of a test result that is within normal limits corresponding to the absence of disease.
  • Predictive value of positive test = p x sensitivity/p x sensitivity + (1-p) x (1-specificity).
  • Predictive value of a negative test = (1-p) x specificity/(1-p) x specificity + P X (1-sensitivity).
  • p = prevalence of disease.
  • 1.Sensitivity = TPx100/TP+FN = 235x100/235+15 = 235/250x100 = 94%.
  • 2.Specificity = TN/TN+FPx100 = 200/200+16x100 = 200/216x100 = 93%.

It is estimated that x-disease occurs in about 20% of the population that will be tested.

  • 3.Predictive value of a positive test = p(sens)/p(sens) +(1-p)(1-spec) = 0.20(0.94)/20(94)+(0.80)(0.07) = 0.1880/0.1880+0.0560 = 0.1880/0.2440 = 0.77. So you would expect disease in 77 out of every 100 with a positive test.
  • If prevalence is only 2%: PVPT = 0.02(0.94)/0.02(0.94)+(0.80)(0.07) = 0.0188/0.0188+0.0560 = 0.0188/0.0748 = 0.25. So you would expect to have x-disease in only 25 animals out of every 100 with a positive test.
  • 4.Predictive value of a negative test (PVNT) = (1-p)(spec)/(1-p)(spec)+p(1-sens).
  • If prevalence is 20%: (0.80)(0.93)/(0.80)(0.93)+0.20(0.06) = 0.7444/0.744+0.012 = 0.744/0.756 = 0.98. So would expect 98 out of every 100 animals with a negative test to be truly free of x-disease.
  • If prevalence is 2%: (0.98)(0.93)/(0.98)(0.93)+(0.02)(0.06) = 0.9114/0.9114+0.0012 = 0.9114/0.9126 = 0.998. So would expect 99.8 out of every 100 dogs with a negative test to be truly free of x-disease.

Technique (intrinsic) limitations

  • In general, results of any given test are most significant when interpreted in conjunction with other laboratory and clinical findings.

Result Data

This article is available in full to registered subscribers

Sign up now to purchase a 30 day trial, or Login

Further Reading

Publications

Refereed papers

  • Recent references fromPubMed.


ADDED