Criteria for evaluating diagnostic methods

Diagnostic matrix and its elements

To assess how well an instrumental or laboratory test performs, physicians form a special statistical group. It includes both people with a confirmed diagnosis and completely healthy individuals. The results of any study are distributed into four categories that form a four-fold table:

• True positive (a). The patient actually has the disease, and the test successfully showed it.
• False positive (b). The person is completely healthy, but the test result turned out to be positive. A "false alarm" situation occurs.
• False negative. The pathology is present in the body, but the method did not detect it. A dangerous "miss" of the disease occurs.
• True negative (g). The examinee is healthy, and the test correctly yielded a negative result.

Method sensitivity

Sensitivity reflects the ability of a specific diagnostic tool to detect pathology in patients who actually suffer from it.

The calculation logic is based on the ratio of the number of true positive results to the total number of sick people. It is calculated by the formula: true positives (a) divided by the sum of true positives and false negatives, after which the result is multiplied by 100%.

From a clinical standpoint, this criterion shows what percentage of real patients will receive a positive test result. If sensitivity is high, the method minimizes the number of false-negative responses — meaning it practically does not miss the disease.

For example, in diagnosing myocardial ischemia, the sensitivity of ECG during bicycle ergometry ranges from 50 to 80%. This means that out of a hundred patients who actually have ischemia, the method will find the problem in only 50–80 people. In the remaining 20–50 examinees, the test will give a false-negative result (pathology is present, but it is not visible on the ECG). If this term is applied to an individual clinical symptom, it will mean the frequency of occurrence of this sign in a specific nosology.

Method specificity

Specificity demonstrates the ability of a method to unmistakably confirm the absence of disease in healthy individuals.

In this case, the ratio of true negative results to the total number of healthy people in the sample is calculated. Formula: true negatives (g) divided by the sum of false positives (b) and true negatives (g), multiplied by 100%.

The clinical meaning of specificity lies in the percentage of healthy people in whom the test will turn out to be negative. A high indicator suggests that the method rarely makes mistakes in healthy individuals and does not provoke a "false alarm".

Returning to the example of myocardial ischemia: the specificity of bicycle ergometry is 80–95%. Consequently, out of a hundred completely healthy patients, the test will correctly show the norm in 80–95 people. The remaining 5–20 people will receive a false-positive result indicating non-existent ischemia. For symptoms, high specificity means that the sign is extremely rarely observed in healthy people or in completely different pathologies.

Informativeness and approaches to diagnosis verification

Unlike the previous criteria, informativeness is not a quantitative indicator. It characterizes the volume of useful information that a method can provide for a specific pathology and in a strictly defined clinical situation. For example, chest plain radiography has high informativeness in diagnosing pneumonia, but is completely uninformative for detecting exertional angina.

Since there are few absolutely specific (pathognomonic) signs in medicine, and sensitivity can suffer due to the absence of symptoms in some patients, physicians use comprehensive approaches to diagnosis:

• Symptom combinations. The combination of clinical signs always possesses greater specificity than each symptom individually.
• Diagnostic criteria. Developed by specialized expert communities to increase confidence in the diagnosis. They include a combination of data with a pre-known accuracy. For example, the American College of Rheumatology criteria allow diagnosing systemic lupus erythematosus with a sensitivity and specificity of about 96%. Nevertheless, such criteria are periodically revised and are not absolute due to the individual course of the disease.
• "Gold standard". This is a method that provides the most accurate confirmation of pathology (e.g., retrograde pancreatocholangiography in primary sclerosing cholangitis).
• High-sensitivity tests. Studies like C-reactive protein or rheumatoid factor are great for screening and reliably ruling out diseases.

Anamnestic factors (gender, age, ethnicity) can become a crucial "final touch" to the diagnosis. For example, familial Mediterranean fever can be suspected in a combination of fever, abdominal pain, and Armenian or Arab ethnicity. Pulmonary hypertension should be suspected in a young woman with characteristic symptoms who took anorectics. Ultimately, the diagnosis is formulated by synthesizing all data, where the main nosological form is always placed first.

Briefly: answers to topical questions

Class and exam questions

What else to study on this topic

• The problem of specificity: why the same sign occurs in different diseases.
• Limitations of diagnostic criteria of specialized associations.
• The role of demographic and anamnestic factors as a key to diagnosis.
• Rules for data synthesis and formulation of the main nosological form.
• The use of high-sensitivity markers for screening and ruling out a diagnosis.

Other topics in the 'General diagnostic principles' section

Prepared by WhiteCoat Pro faculty in accordance with the medical school curriculum. For educational purposes only.