The sensitivity for TB of “any abnormality” as reported on CXR in children is 84%, and the specificity is 91%. It is thus more specific than symptom screening alone. The estimates of the accuracy of CXR are not, however, disaggregated by age group, and significant differences in CXR findings between younger and older children may lead to important differences in sensitivity and specificity by age group.
Abnormalities caused by TB seen on CXR in children may differ widely from those in adults. While older children may have “adult type” disease presentation, such as cavitary disease, the changes on CXR associated with TB disease in younger children may be subtle and hard to see if the quality is not optimal. When using CXR for TB screening in children, both posteroanterior and lateral views should be done. Besides cavitary disease, the other most common abnormalities are enlarged hilar lymph nodes, enlarged hilar and paratracheal lymph nodes, enlarged lymph glands compressing the airways, pneumonic consolidation with lymph node enlargement, miliary TB and pleural effusions. It may sometimes be difficult to distinguish abnormally enlarged paratracheal and hilar lymph nodes from the normal vascular structures. These subtle findings on CXR in younger children may affect the sensitivity and specificity of CXR. The help of a practitioner experienced in interpreting paediatric chest radiography may be sought to resolve questions about interpretation. CAD software for interpreting plain CXR for TB is now recommended by WHO as an alternative to human reading (Chapter 4); however, this recommendation is limited to people aged ≥ 15 years, and more data should be collected to validate the performance of CAD for TB in children.
CXR can be used in combination with symptom screening (see 6.4 for algorithm options for screening child contacts). CXR is not, however, readily available in many locations, and travel to another location for a CXR may not be feasible for a caregiver, who may be unable to make time or to afford direct or indirect costs for travel, time, support or the radiography service. Mobile CXR units may be used to reach populations that otherwise would be unable to access a health centre with a radiography machine. These, however, require financial and logistical support, and, to be clinically useful, a mobile unit would have to have a regular schedule.
CXR emits a small amount of radiation; however, the radiation risk is very low. Chapter 3 outlines additional considerations for implementing CXR, including the benefits and drawbacks of serial and parallel screening when CXR is combined with symptom screening.