Global impact of COVID-19 on childhood tuberculosis: an analysis of notification data

Abstract

Background: There is concern that the COVID-19 pandemic has damaged global childhood tuberculosis management. Quantifying changes in childhood tuberculosis notifications could support more targeted interventions to restore childhood tuberculosis services. We aimed to use time-series modelling to evaluate the impact of COVID-19 on child tuberculosis notifications.

Methods: Annual tuberculosis case notification data reported to WHO by 215 countries were used to calculate annual notification counts for the years 2014-20, stratified by age groups (0-4, 5-14, and ≥15 years) and sex. We used time-series modelling to predict notification counts for 2020, and calculated differences between these predictions and observed notifications in 2020 for each of the six WHO regions and at the country level for 30 countries with high tuberculosis burden. We assessed associations between these differences and the COVID-19 stringency index, a measure of COVID-19 social impact.

Findings: From 2014 to 2019, annual tuberculosis notification counts increased across all age groups and WHO regions. More males than females in the 0-4 years age group and ≥15 years age group had notifications in all years from 2014 to 2020 and in all WHO regions. In the 5-14 years age group, more females than males were notified globally in all years, although some WHO regions had higher notifications from males than females. In 2020, global notifications were 35·4% lower than predicted (95% prediction interval -30·3 to -39·9; 142 525 observed vs 220 794 predicted notifications [95% prediction interval 204 509 to 237 078]) for children aged 0-4 years, 27·7% lower (-23·4 to -31·5; 256 398 vs 354 578 [334 724 to 374 431]) in children aged 5-14 years, and 18·8% lower (-15·4 to -21·9; 5 391 753 vs 6 639 547 [6 375 086 to 6 904 007]) for people aged 15 years or older. Among those aged 5-14 years, the reduction in observed relative to predicted notifications for 2020 was greater in males (-30·9% [-24·8 to -36·1]) than females (-24·5% [-18·1 to -29·9]). Among 28 countries with high tuberculosis burden, no association was observed between the stringency of COVID-19 restrictions and the relative difference in observed versus predicted notifications.

Interpretation: Our findings suggest that COVID-19 has substantially affected childhood tuberculosis services, with the youngest children most affected. Although children have mostly had fewer severe health consequences from COVID-19 than have adults, they have been disproportionately affected by the effects of the pandemic on tuberculosis care. Observed sex differences suggest that targeted interventions might be required. As countries rebuild health systems following the COVID-19 pandemic, it is crucial that childhood tuberculosis services are placed centrally within national strategic plans.

Funding: Medical Research Council.

Figure 1

Age-stratified notifications from 2014 to 2020 by WHO region Using data from 2014 to 2019, our model predicts notification data for 2020 with 95% prediction intervals (shaded triangles) originating from the 2019 value. Data on case numbers were square-root transformed for the purpose of visualisation.

Figure 2

Proportional difference between the number of predicted and number of observed notifications for 2020 for 29 countries with high tuberculosis burden Predicted notifications were from time-series modelling using data from 2014 to 2019. Liberia was excluded because of very high relative but low absolute changes in notifications. Countries are ordered by the total difference between observed and predicted values across all age groups. *Aggregates across the 29 included high-burden countries. †Proportion of overall notifications in the 29 reported high-burden countries in each age group accounted for by each country.

Figure 3

Age-stratified correlations between mean COVID-19 stringency index and percentage difference in tuberculosis notifications between time-series predictions and observed notifications in 2020 for 28 countries with high tuberculosis burden Liberia was excluded because of very high relative but low absolute changes in notifications, and North Korea was excluded because of the lack of a stringency index. Countries are indicated by their three-letter ISO 3166 codes. AGO=Angola. BGD=Bangladesh. BRA=Brazil. CAF=Central African Republic. CHN=China. COD=Democratic Republic of the Congo. COG=Congo. ETH=Ethiopia. GAB=Gabon. IDN=Indonesia. IND=India. KEN=Kenya. LSO=Lesotho. MMR=Myanmar. MNG=Mongolia. MOZ=Mozambique. NAM=Namibia. NGA=Nigeria. PAK=Pakistan. PHL=Philippines. PNG=Papua New Guinea. SLE=Sierra Leone. THA=Thailand. TZA=Tanzania. UGA=Uganda. VNM=Viet Nam. ZAF=South Africa. ZMB=Zambia. *Negative values indicate number of observed notifications was lower than predicted; positive values indicate number of observed notifications was higher than predicted.