Ep25-335-23 It’s not TB but what could it be? Abnormalities on chest X-rays from the 2016 Kenya National Tuberculosis Prevalence Survey

Mungai, Brenda, Joekes, Elizabeth, Mansini, E, Obasi, Angela ORCID: https://orcid.org/0000-0001-6801-8889, Manduku, V, Mugi, B, Kirathe, D, Kiplimo, R, Sitienei, J, Oronje, Rose, Morton, Ben ORCID: https://orcid.org/0000-0002-6164-2854, Squire, Bertie ORCID: https://orcid.org/0000-0001-7173-9038 and MacPherson, Peter ORCID: https://orcid.org/0000-0002-0329-9613 (2020) 'Ep25-335-23 It’s not TB but what could it be? Abnormalities on chest X-rays from the 2016 Kenya National Tuberculosis Prevalence Survey' in 51st World Conference on Lung Health of the International Union Against Tuberculosis and Lung Disease, Online, 21st-24th October 2020.

Text Output 12. FINAL CXR Abstract for the 51st Union Conference_11thMay2020 3pm.docx Restricted to Repository staff only Download (130kB)

Abstract

Background: The prevalence of diseases other than tuberculosis(TB) detected on chest-Xray(CXR) during TB screening in Kenya is unknown. Our study aimed to characterise and quantify non-TB abnormalities on CXR and to compare radiologist interpretation with Computer-Aided Detection for Tuberculosis (CAD4TB). We hypothesized that non-TB abnormalities requiring further clinical input are prevalent and may be missed using CAD4TB.
Design/Methods: We undertook a cross-sectional study from May 2019-February 2020, analyzing CXRs from the 2016 Kenya National TB Prevalence Survey, sam- pling films classified either as “abnormal, suggestive of TB” or “abnormal other”. We developed a reporting tool which comprised four anatomical categories and a list of common diagnoses. Readers were blinded, films double reported and discordant results resolved by a third reader. We used CAD4TB 6.0. and R v3.6.2. for analysis.
Results: Of 1123 films sampled, 600(53.4%) were ab- normal (Figure-1). Prevalence of abnormalities in major categories: 26.3% (95% CI 23.7%-28.9%) heart and/ or great vessels, 26.1% (95% CI23.5%-28.8%) lung parenchyma, 7.6% (95% CI 6.1%-9.3%) pleura and 3% (95% CI 2.1%-4.2%) mediastinum. Prevalence of active-TB 4% (95% CI 2%-4%), severe post TB lung changes (bronchiectasis/destroyed lung) 2% (95% CI 0-2%). Non-TB related diagnoses: cardiomegaly 23.1% (95% CI 20.6%-25.6%), suspected cardiac failure 1.9% (95% CI1.2-2.8%), non-specific airspace opacification/ interstitial disease 6% (95% CI 4%-8%), suspected emphysema 2% (95% CI 2%-4%) and mediastinal masses 0.8% (95% CI 0.4%-1.5%). Median CAD4TB scores: Severe post TB lung changes 76 (IQR 71-81), active-TB 66 (IQR 55-72), suspected emphysema 57 (IQR 54-59), non-specific airspace opacification/interstitial disease 56(IQR 50-61), mediastinal mass 52 (IQR 47-59) and cardiomegaly 50(IQR 46-56).
Conclusions: Abnormalities unrelated to TB were prev- alent, most notably cardiomegaly. These non-TB ab- normalities will go undetected using CAD stratification based on threshold scores alone. Further refinement of CAD algorithms to include non-TB diagnoses could attenuate this risk. Incorporation of blood pressure monitoring and spirometry should be considered in TB screening activities.

Item Type:	Conference or Workshop Item (Poster)
Additional Information:	Abstract Only: Published in The International Journal of Tuberculosis and Lung Disease, 'The Abstract Book', Vol 24 No 10 October 2020 Supplement 2
Subjects:	WA Public Health > Health Problems of Special Population Groups > WA 395 Health in developing countries WA Public Health > Health Administration and Organization > WA 540 National and state health administration WF Respiratory System > WF 140 Diseases of the respiratory system (General) WF Respiratory System > Tuberculosis > WF 200 Tuberculosis (General) WF Respiratory System > Tuberculosis > WF 225 Mass chest X-ray
Faculty: Department:	Clinical Sciences & International Health > Clinical Sciences Department Clinical Sciences & International Health > International Public Health Department Education
Related URLs:	https://conf2020.theunion.org/wp-content...
Depositing User:	Elly Wallis
Date Deposited:	06 Nov 2020 14:44
Last Modified:	11 Nov 2020 12:14
URI:	https://archive.lstmed.ac.uk/id/eprint/15949

Statistics

View details

Actions (login required)

Edit Item