Inequalities in the impact of COVID-19-associated disruptions on tuberculosis diagnosis by age and sex in 45 high TB burden countries

Tuberculosis (TB) remains a major public health priority and, prior to the emergence of COVID-19, was the leading infectious cause of mortality worldwide [1]. Large inequalities exist in TB burden and access to care globally [2]. Older people are at higher risk of developing active TB disease [3], whilst children face high rates of mortality due to TB and have particularly low rates of case detection [1, 4]. TB prevalence is high amongst men [5, 6], who, compared to women, have more limited access to care [1], and in some settings may also be at an increased risk of drug-resistance [7]. Many of these inequalities have remained consistent over the last decade or more [2].

The COVID-19 pandemic has substantially impacted TB prevention, diagnosis, and care, resulting in major reductions in TB notifications, the number of people diagnosed with and reported as having TB, worldwide [1, 8, 9]. With a wide range of disruptions caused by the COVID-19 pandemic, we might expect that the impact of these disruptions on TB burden would vary for different population groups. For example, during the first three months of the COVID-19 pandemic in the UK, children and young people have had a much greater reduction in visits to accident and emergency departments for infections than the rest of the population [10], whilst reductions in clinic attendance for acute respiratory infections in Singapore were higher for both children and older adults [11]. Elderly people may have been more inclined to “shield” during the pandemic compared to working age adults and so be less likely to attend healthcare facilities. In Malawi, TB diagnosis in women and girls was significantly more affected than in men and boys in 2020 [12], whilst in India, more women than men were diagnosed in 2020–2021, in a reversal of previous trends [13]. In Mozambique, TB notifications amongst men were more severely impacted than amongst women [14]. In addition, the proportion of transmission attributable to contact within the household is likely to have increased as a result of COVID-19-associated non-pharmaceutical interventions [8], which may disproportionately affect children, who are at a high risk of infection when exposed to TB within the household [15]. However, efforts to estimate changes in the burden of TB as a result of COVID-19-associated disruptions going forward [1, 16, 17] have primarily focused on the overall effect of these disruptions with no consideration of potential inequalities in that effect.

Assessing disparities in the impact of the COVID-19 pandemic on TB services requires identifying any inequalities in access to TB diagnosis and then prioritising key populations for catch-up campaigns. In this analysis, we investigate the impact of COVID-19 on missed or delayed TB diagnosis in 2020 by age and sex for 45 high TB, TB/HIV and multidrug- or rifampicin-resistant (MDR) TB burden countries.

The COVID-19 pandemic has caused significant disruptions to TB services, as indicated by substantially lower case notifications than expected in 2020. Compared to trends predicted from the pre-pandemic era, our results indicate that global TB notification rates were 21.8% lower than expected for adults (with a similar rate for both men and women), whilst for children and the elderly notification rates were 37.8% and 28.5% lower than expected, respectively. These findings suggest that a large number of individuals are likely suffering from untreated TB disease directly as a result of the pandemic. These individuals are also at risk of transmitting infection to their contacts, which will have long term public health consequences.

Our results are in line with most findings that COVID-19-associated disruptions have led to widespread decreases in TB diagnosis [1] and demonstrate that major inequalities in the impact of COVID-19-associated disruptions on TB diagnosis exist in high TB burden countries. However, these inequalities are extremely setting-specific, and there is no evidence of any systematic inequality in disruptions by age or sex globally. In over a half of countries, including in the WHO Eastern Mediterranean and European regions, the risk of having diagnosis missed or delayed as a result of the pandemic was higher for children than adults. However, high levels of heterogeneity and the presence of a small number of countries where more children than expected were diagnosed mean that there was no evidence for any difference in risk globally. In more than two thirds of countries, and in the WHO European and Western Pacific regions, the risk of having diagnosis missed or delayed was higher for the elderly than adults. Again, however, high levels of heterogeneity mean that this inequality in risk was not evident globally, although in countries that did experience disruptions and those with better model fits the elderly did also appear to be at a higher risk. In nearly a half of countries, the risk of having diagnosis missed or delayed differed by sex. This varied by setting, with men in the Region of the Americas representing a particular concern. However, globally there was no evidence for any difference in risk. Given already low case detection rates for children, and high burden in men and the elderly, it is encouraging that COVID-19-associated disruptions have not worsened this inequality. However, it is concerning that in countries that did appear to experience disruptions, or those with a better model fit, children, the elderly and men are all at a higher risk as a result.

In most of our analyses, heterogeneity between countries was extremely high, which we were unable to account for due to multiple sources of confounding. Although some heterogeneity was attributable to regional differences, differences between countries in population age-structure, gender roles and responsibilities, COVID-19 non-pharmaceutical interventions, COVID-19 burden on health care services, barriers to accessing health care services due to cultural practices, individual’s personal protection measures and timing of all of the above likely all contribute. Given a lack of data, no meta-regression was conducted. It is unclear why some countries demonstrate inequalities in one direction whilst others demonstrate no inequality, or the opposite. Causal drivers are likely to be highly setting specific, where identifying these drivers requires a close examination of a country’s experience of the pandemic. Whilst our analysis is not able to suggest why men in Tajikistan may have been more likely to miss or delay a diagnosis in 2020 whilst women in Uzbekistan were more likely to do so, it does point to a clear need for further, gender-sensitive analyses when planning the response in both of those countries.

Our results have focused on the year 2020. We note that our analysis is not able to capture diagnoses made in 2020 which would have occurred earlier in the year but were delayed because of COVID-19. Similarly, catch-up campaigns in 2020 may mask delays to diagnosis from earlier in the year. Although COVID-19-associated disruptions will have affected different countries at different time points throughout 2020, the use of a risk ratio here allows for a comparison in risk within, and between, countries during those periods. Extrapolating this analysis to future years should be conducted with extreme caution. The long duration of progression from M.tb infection to TB disease [26] suggests that changes in transmission of M.tb as a result of COVID-19-associated disruptions are unlikely to have had a major influence on TB disease burden and notifications in 2020, although the effect may still be non-negligible. However, from 2021 as COVID-19-related disruptions continue, it is likely that societal changes from earlier in the pandemic, such as an increased proportion of time spent in the household and not the workplace, may affect transmission dynamics (and therefore the underlying TB disease burden) differently in different groups, so it is no longer clear how best to estimate a “missed or delayed diagnosis”.

Our results are limited by both the availability of data and the methods used to estimate missing cases. Although changes to M.tb transmission are unlikely to have affected TB burden during the time period under consideration, it is possible that an increase in risk factors such as undernutrition may have, where the extent of this may vary by population group. In addition, an observed increase or decrease in notifications does not necessarily correspond to a change in diagnosis, and could instead reflect changes in reporting, although this would probably not vary by population group. In some countries, including a number in Africa in particular, attempts to counter the effects of COVID-19-associated disruptions, such as a sustained effort to increase case detection, could also have led to more notifications than expected. However, this could also reflect other factors, such as a high uptake of non-pharmaceutical interventions with a corresponding reduction in M.tb transmission, or the importance of rapid progression as opposed to remote reactivation in that setting. In general, we are not able to account for country-specific events, such as natural disasters or political unrest, which may also have affected care seeking behaviour in 2020 of one population group over another. Our analysis is predicated on good model predictions, where the use of a simple linear predictor term in our Poisson models may be unrealistic in some countries. However, given the limited data available, any more complicated model may overfit the data.

Despite these limitations, overall, we found no evidence at a global level of any difference by age or sex in the risk of disruption to TB diagnosis as a result of the pandemic. However, in many specific settings, the consequences of disruptions have been greater for some groups than others, with children and the elderly in particular being disproportionately affected in the majority of countries. It is important to recognise these inequalities when prioritising catch-up campaigns and the cultural and social issues that have contributed to them. For example, in countries where children have been disproportionately affected, active case finding interventions to mitigate the impact could prioritise schools or TB-affected households, whilst in countries where men have been disproportionately affected, out-reach campaigns and extended opening hours for clinical services could improve access to healthcare for men. Our results highlight the importance and utility of tracking age and sex differences in healthcare reporting to explore the impact of disease across these dimensions. Such data would allow for a nuanced, equitable response not just to COVID-19, but to the mitigation of humanitarian crises in general. There is a clear need for context-specific data and evidence-informed action to achieve equity in TB care, with the ultimate aim of ending TB.