Symptom‐ and chest‐radiography screening for active pulmonary tuberculosis in HIV‐negative adults and adults with unknown HIV status

Why is improving screening for tuberculosis of the lungs important?

Systematic screening in settings where tuberculosis is common is a recommended strategy for early detection of tuberculosis. Screening helps identify people who are more likely to have tuberculosis so they can have confirmatory testing. These are additional tests to confirm the presence of Mycobacterium tuberculosis, the bacterium that causes tuberculosis. Asking about tuberculosis symptoms (for example, cough, coughing up blood, fever, and fatigue) and doing a chest X‐ray (CXR), which shows lung abnormalities, are commonly used screening methods. Tuberculosis is treatable with antibiotics, which means that early detection may result in lower mortality and morbidity, less transmission of tuberculosis, and more equitable access to care.

Not recognizing lung tuberculosis when it is present (a false‐negative result) may result in delayed treatment and further transmission. Conversely, a screening result that is thought to be positive while it is not may result in unnecessary confirmatory tests, which burdens both the individual and the public health system.

Knowing how often screening tests lead to false‐positive and false‐negative results – this is called accuracy ‐ may help in choosing a screening method.

What is the aim of this review?

To find out how accurate asking about symptoms and CXR are as screening tests for lung tuberculosis in adults with unknown or negative HIV status.

What was studied in the review?

We studied the accuracy of three types of symptom questions: (i) cough for two or more weeks, (ii) cough of any duration, and (iii) any tuberculosis symptom. For CXR, we studied two definitions for a positive result: (i) any CXR lung abnormality and (ii) CXR lung abnormalities suggestive of tuberculosis. The results are interpreted by staff trained in radiology.

What are the main results in this review?

The review included 59 studies, of which 48 reported on one or more symptom screening questions and 37 reported on CXR.

The results below indicate a situation in which five individuals (0.5%) have lung tuberculosis among a group of 1000 screened individuals.

Cough for two weeks or more: if 1000 individuals were screened, 58 would screen positive, meaning they report cough for two weeks or more and, of these, 56 (97%) would not have lung tuberculosis. Of 1000 individuals, 942 would screen negative, meaning they do not report cough for two weeks or more and, of these, three (0.3%) would have lung tuberculosis.

Cough of any duration: of 1000 individuals, 127 would screen positive and, of these, 124 (98%) would not have lung tuberculosis. Of 1000 individuals, 873 would screen negative and, of these, two (0.2%) would have lung tuberculosis.

Any tuberculosis symptom: of 1000 individuals, 351 would screen positive and, of these, 348 (99%) would not have lung tuberculosis. Of 1000 individuals, 649 would screen negative and, of these, one (0.2%) would have lung tuberculosis.

Any CXR lung abnormality: of 1000 individuals, 113 would show lung abnormalities on CXR and, of these, 108 (96%) would not have lung tuberculosis. Of 1000 individuals, 887 would not show lung abnormalities and, of these, no one (0%) would have lung tuberculosis.

CXR lung abnormalities suggestive of tuberculosis: of 1000 individuals, 48 would screen positive and, of these, 44 (92%) would not have lung tuberculosis. Of 1000 individuals, 952 would screen negative and, of these, one (0.1%) would have lung tuberculosis.

How reliable are the results of the studies in this review?

In the included studies, the diagnosis of tuberculosis was made by assessing the study participants with confirmatory tests (the reference standard). This is the best available method for deciding whether participants really had lung tuberculosis.

However, there were problems with how the studies were conducted. In many studies, those without symptoms or CXR abnormalities were not tested with a confirmatory test. Therefore, the numbers of those without symptoms or CXR abnormalities, but nevertheless having tuberculosis (people who tested falsely negative), may have been underestimated in these studies. Consequently, screening for symptoms or CXR abnormalities might appear more accurate than it really is.

In addition, results from individual studies included in the review varied, for example, because of regional variation. Therefore, we cannot be sure that screening for symptoms and CXR abnormalities will always have the same accuracy.

What are the implications of this review?

The results of the review suggest that screening for tuberculosis with symptom questions or CXR might result in a high yield of persons with tuberculosis disease. However, this screening might also result in a high proportion of persons without the disease screening positive. Additional considerations for the best design of screening programmes include the local epidemiological situation, availability and accessibility of CXR, and the need for confirmatory tests.

How up to date is this review?

The review authors searched for and included studies published from 1 January 1992 to 10 December 2018. A repeat of the search to 2 July 2021 revealed no further studies that would inform the results of the analysis.