Accuracy and precision

Accuracy of a laboratory test is its correspondence with the true value.Accuracy is maximized by calibrating laboratory equipment with reference material and by participating in external quality control programs.Precision of a test is its reproducibility when it is repeated on the same sample.An imprecise test yields widely varying results on repeated measurement. Precision is monitored in laboratory by using control material.

Detection and quantification 

The tests performed during the physical examination are usually designed to detect symptoms or signs, in these cases, a test that detects symptoms or signs is designated as a positive test, while a test that shows no symptoms or signs is designated as a negative test, in the following Further details in a separate section.Quantification of target substances, cell types, or other specific entities is, for example, a common output of most blood tests.This not only answers whether the target entity exists, but also the degree of existence. In blood tests, the quantification is relatively specific, eg given in mass concentration, while most other tests may be quantitative as well, although less specific, eg "very pale" rather than "slightly pale" sign. Likewise, the radiological image is a technical quantification of the radiological opacity of the tissue.Especially when taking a medical history, there is no clear line between detection or quantification tests and descriptive information about an individual.For example, questions about a person's occupation or social life may be viewed as tests for the presence of various risk factors which can be viewed as positive or negative, or they may be viewed as "merely" descriptive, although the latter clinically important.

Positive or negative

The result of a test designed to detect an entity may be positive or negative: this does not correlate with poor prognosis, but rather means whether the test is valid, and whether a certain parameter assessed is present.For example, a negative screening test for breast cancer means no signs of breast cancer are found (which is actually very positive for the patient).Classifying a test as positive or negative gives a binary classification, resulting in the ability to perform Bayesian probabilities and test performance metrics, including calculations of sensitivity and specificity.

Continuous value

Tests whose results are continuous values, such as most blood values, can be interpreted as-is, or they can be converted to binary values by defining a cutoff value, and the test result is designated as positive or negative, depending on whether the result is a value higher than or below cutoff.

Interpretation

When characteristic signs or symptoms are observed, the target condition is almost certainly present, whereas if the requisite signs or symptoms are not present, the target condition is almost certainly not present.In practice, however, the subjective probability that a condition exists is never exactly 100% or 0%, so the purpose of testing is to estimate the post-test probability of a condition or other entity.Most diagnostic tests essentially use a reference group to establish performance data such as predictive value, likelihood ratio, and relative risk, which are then used to interpret individual post-test probabilities.When monitoring an individual's testing, the test results of that individual's previous testing can be used as a reference for interpreting subsequent testing.

Risks

Some medical testing procedures have associated health risks and even require general anesthesia, such as mediastinoscopy.Other tests, such as blood tests or Pap smears, carry little immediate risk.Medical testing may also have indirect risks, such as test stress, and a higher risk test may be required as follow-up to a (potentially) false positive test result.Consult a healthcare provider (including physicians, physician assistants, and nurse practitioners) for more information on any test prescriptions.

Indications

Each test has its own indications and contraindications. An indication is a valid medical reason for testing.A contraindication is a valid medical reason not to perform a test.For example, a middle-aged person may need a basic cholesterol test (medically appropriate). However, if the same test has been recently performed on the person, the presence of the previous test is a contraindication to that test (a medically valid reason not to have the test).Information bias is a cognitive bias that causes healthcare providers to order tests that yield information they do not actually expect or intend to use to make medical decisions. Medical tests are indicated when the information they generate will be used.For example, a screening mammogram is not recommended (medically appropriate) for a dying woman because even if breast cancer is found, she will die before any cancer treatment begins.In a simplified way, how much testing is done on an individual depends largely on its net benefit to that individual. A test is chosen when the expected benefit outweighs the expected harm.The net benefit can be roughly estimated as:

Where:

• bn is the net benefit of executing the test

• Λp is the absolute difference between the pre-test and post-test probabilities of the condition (such as a disease) that the test is      expected to achieve.A major factor in this absolute difference is the power of the test itself, which can be described, for example, in terms of sensitivity and specificity or likelihood ratio.Another factor is the pretest probability, a lower pretest probability leads to a lower absolute difference, with the result that even a very powerful test will achieve a low Absolute difference without any other indicators), but on the other hand, even a low power test can make a big difference in a highly  suspicious situation. Probability in this sense may also need to be considered in situations where it is not the primary objective of the test, such as relative probabilities of profiles in differential diagnostic procedures.ri is the ratio of the difference in probability expected to result in a change in the intervention (eg change from 'no treatment' to 'low dose drug treatment').For example, if the only expected effect of a medical test is to make one disease more likely than another, but both diseases have the same treatment (or neither), then that factor is low and the test is here Aspects may have no value to an individual.

• bi is the individual benefit of changing the intervention

•  hi is the harm to the individual of a change in the intervention, such as a side effect of a drug treatment

• ht is the hazard caused by the test itself.

Some other factors that affect the decision whether a medical test should be done include: cost of the test, availability of additional tests, potential interference with subsequent tests (eg, abdominal palpation may induce bowel activity whose sounds interfere with subsequent auscultation of the abdominal examination),Time spent on testing or other practical or administrative aspects.The possible benefits of diagnostic testing can also be weighed against the cost of unnecessary testing and the resulting unnecessary follow-up and possibly even unnecessary treatment for incidental findings.In some cases, ongoing testing is not expected to benefit the individual being tested.Instead, the results may help build statistics to improve healthcare for others. Patients may give informed consent to undergo medical tests that benefit others.