Estimating the contribution of transmission in primary healthcare clinics to community-wide TB disease incidence, and the impact of infection prevention and control interventions, in KwaZulu-Natal, South Africa
1TB Centre, London School of Hygiene & Tropical Medicine, United Kingdom
2The Institute for Global Health and Development, Queen Margaret University, United Kingdom, Department of Infectious Disease, Faculty of Medicine, Imperial College London
3Africa Health Research Institute, School of Laboratory Medicine & Medical Sciences, College of Health Sciences, University of KwaZulu-Natal, South Africa
4Department of Infectious Disease, Faculty of Medicine, Imperial College London
5School of Public Health, University of the Witwatersrand, South Africa
* Corresponding author; email: nicky.mccreesh@lshtm.ac.ukAbstract
Background
There is a high risk of Mycobacterium tuberculosis (Mtb) transmission in healthcare facilities in high burden settings. Recent World Health Organization guidelines on tuberculosis infection prevention and control (IPC) recommend a range of measures to reduce transmission in healthcare and institutional settings. These were evaluated primarily based on evidence for their effects on transmission to healthcare workers in hospitals. To estimate the overall impact of IPC interventions, it is necessary to also consider their impact on community-wide tuberculosis incidence and mortality.
Methods
We developed an individual-based model of Mtb transmission between household members, in primary healthcare clinics (PHCs), and in other congregate settings; drug sensitive and multidrug resistant tuberculosis disease development and resolution; and HIV and antiretroviral therapy (ART) and their effects on tuberculosis. The model was parameterised using data from a high HIV prevalence, rural/peri-urban community in KwaZulu-Natal, South Africa, including data on social contact in clinics and other settings by sex, age group, and HIV/ART status; and data on the prevalence of tuberculosis in clinic attendees and the general population. We estimated the proportion of disease in adults that resulted from transmission in PHC clinics in 2019, and the impact of a range of IPC interventions in clinics on community-wide TB incidence and mortality.
Results
We estimate that 7.6% (plausible range 3.9-13.9%) of drug sensitive and multidrug resistant tuberculosis in adults resulted from transmission in PHC clinics in the study community in 2019. The proportion is higher in HIV-positive people, at 9.3% (4.8%-16.8%), compared to 5.3% (2.7%-10.1%) in HIV-negative people. We estimate that IPC interventions could reduce the number of incident TB cases in the community in 2021-2030 by 3.4-8.0%, and the number of deaths by 3.0-7.2%.
Conclusions
A non-trivial proportion of tuberculosis results from transmission in PHC clinics in the study communities, particularly in HIV-positive people. Implementing IPC interventions could lead to moderate reductions in disease burden. We therefore recommend that IPC measures in clinics should be implemented both for their benefits to staff and patients, but also for their likely effects on TB incidence and mortality in the surrounding community.
Article notes
Competing Interest Statement
The authors have declared no competing interest.
Funding Statement
The support of the Economic and Social Research Council (IK) is gratefully acknowledged. The project is partly funded by the Antimicrobial Resistance Cross Council Initiative supported by the seven research councils in partnership with other funders including support from the GCRF. Grant reference: ES/P008011/1. NM is additionally funded the Wellcome Trust (218261/Z/19/Z). RGW is funded by the Wellcome Trust (218261/Z/19/Z), NIH (1R01AI147321-01), EDTCP (RIA208D-2505B), UK MRC (CCF17-7779 via SET Bloomsbury), ESRC (ES/P008011/1), BMGF (OPP1084276, OPP1135288 & INV-001754), and the WHO (2020/985800-0). RMGJH is funded by ERC (action number 757699). TAY receives an NIHR Academic Clinical Fellowship.
Introduction
Tuberculosis (TB) is a major global public health problem, killing an estimated 1.4 million people in 20191. There is a high risk of transmission in healthcare facilities in high TB burden settings, evidenced by the elevated rate of tuberculosis in healthcare workers2. Updated World Health Organization guidelines on TB infection prevention and control (IPC) recommend a wide range of measures to reduce transmission in healthcare and institutional settings, ranging from triaging people with TB symptoms to installing ultraviolet germicidal irradiation systems (UVGI)2. These measures were evaluated and implemented as recommendations in the guidelines primarily based on evidence on their effects on risk to healthcare workers, and in hospitals settings.
Protecting healthcare workers should be a key concern of TB control programmes. However, the motivation for, and potential benefits of, IPC interventions in clinics extend beyond the reductions in disease burden among clinic staff. While healthcare workers and other clinic staff are at the highest risk of infection in clinics, due to their longer durations of exposure, the numbers of patients and other clinic attendees are far higher than numbers of staff. It is therefore likely that a large proportion of clinic-acquired TB is in patients and other clinic attendees. As a consequence, it is imperative that the impact on TB incidence in the wider community is considered when estimating the likely impacts of IPC measures.
Estimating the contribution of transmission in clinics (or other congregate settings) to overall community-wide disease burden is challenging. Taylor et al. used data on ventilation rates and a Wells-Riley approach to estimate a 0.03% risk of infection to patients per clinic visit. This approach is heavily dependent on estimates of mean quanta production rates, however, about which there is considerable uncertainty (their sensitivity analysis gave a wide range of 0.02-0.35%). Andrews et al also used a Wells-Riley based approach to determine infection risk by location (although not clinics), but removed the dependence on an assigned value for the quanta production rate by using data on contact time in multiple types of location, and calibrating their model to the prevalence of infection by age3.
In this work, we used a social contact data-based approach similar to that adopted by Andrews et al., but used an individual-based model that includes HIV/antiretroviral therapy (ART) and TB disease development and resolution, and calibrated the model to overall disease incidence. This allowed us to determine the contribution of primary healthcare (PHC) clinics not only to the incidence of infection, but also to community-wide disease incidence and mortality. This is important for determining the true contribution of clinics-based transmission to disease burden, due to the increased rates of clinic attendance by people at increased risk of progression to disease4. We also incorporated empirical data on the increased prevalence of TB in PHC clinic attendees compared to the general population, something that acts to amplify transmission in clinics4.
The IPC interventions we simulated were identified and parameterised through a rigorous multi-disciplinary approach. This work forms part of the Umoya omuhle project, that used a whole systems approach to study IPC in primary healthcare facilities in South Africa. As part of the project, system dynamics modelling was used to identify potential IPC interventions that local policy makers and health professionals active at clinic and province levels ranked highly in terms of both feasibility of implementation and perceived likely impact on overall and MDR Mtb transmission in clinics5.
Methods
Results
Fit to data
The model fit well to all the fitting targets, in the main scenario and the sensitivity analyses scenarios (Figure 1 and supporting material Table S12).
Proportion of disease from transmission in clinics
Overall, we estimate that 2.3% (plausible range 1.2-3.4%) of contact time by adults occurred in clinics in the study communities in 2019, leading to 4.9% (2.5-9.1%) of overall and MDR infections, and 7.6% (3.9-13.9%) of overall and MDR disease. The proportion of all TB disease that resulted from transmission in clinics was higher in HIV-positive people, at 9.3% (range 4.8%-16.8%), and lower in HIV-negative people, at 5.3% (2.7%-10.1%).
Intervention impact
Opening windows and doors reduced the total number of incident TB cases in the community in 2021-2030 by 5.3% (range 1.3-12.5%), simple clinic retrofits by 4.3% (0.8-11.2%), UVGI systems by 7.4% (3.2-14.7%), surgical mask wearing by patients by 4.5% (2.1-8.8%), increased CCMDD coverage by 3.4% (0.7-8.7%), queue management systems with outdoor waiting areas by 8.0% (3.8-15.2%), and appointment systems by 5.9% (2.2-12.9%) (Figure 3). Reductions in MDR-TB cases were similar to reductions in all TB cases (Supporting figure S6).
Reductions in TB deaths were 9.5-12.6% lower than reductions in cases, reflecting the time lag between developing disease and dying from TB. The reductions in deaths ranged from 3.0% (range 0.7-10.1%) for increased CCMDD coverage, to 7.2% (range 2.7-13.8%) for queue management systems with outdoor waiting areas.
Proportion of disease from transmission in clinics that is in clinic staff
We estimate that in the study community, an average of 7.1% (95% plausible range 2.3-16.7%) of all disease in adults resulting from transmission in clinics occurs in clinic staff.
Discussion
In this paper, we estimate that 7.6% (plausible range 3.9-13.9%) of tuberculosis in adults results from transmission in PHC clinics in a high HIV prevalence, rural/peri-urban setting in KwaZulu-Natal, South Africa. The proportion is higher in HIV-positive people, at an estimated 9.3% (range 4.8%-16.8%), compared to 5.3% (2.7%-10.1%) in HIV-negative people. We estimate that IPC interventions in PHC clinics could reduce the number of incident TB cases in the community in 2021-2030 by 3.4-8.0%, and deaths by 3.0-7.2%. These findings further strengthen the case for an increased emphasis on IPC in clinics, not just as a tool for protecting clinic staff and patients, but as method for reducing community-wide TB incidence and mortality.
Our findings highlight the importance of considering contact saturation13 and the population at risk when estimating the proportion of disease that results from transmission in different types of setting. We estimate that 4.9% of infections occur in clinics, more than double the estimated 2.3% of contact time occurring in clinics. This reflects the fact that contacts between household members are repeated, reducing their overall importance to transmission, and increasing the importance of other settings. It also reflects higher rates of clinic attendance by people with potentially infectious TB. We estimate that an even higher proportion of disease results from transmission in PHC clinics, at 7.7%. This reflects both the additional effects of contact saturation and repeated infections between household members, but also the higher rates of clinic attendance by HIV-positive people, who are at increased risk of progression to disease following infection.
We parameterised our model to data from a high HIV prevalence, high TB, rural/peri-urban setting in KwaZulu-Natal, South Africa. The proportion of tuberculosis that results from transmission in clinics, and the impact of IPC interventions on community-wide TB incidence, is likely to vary by setting, depending on a range of factors. These include: the proportion of people’s contact time that occurs in clinics; the prevalence of HIV and other TB risk factors, and how clinic visiting and other contact behaviour varies between people with different risk factor profiles; and the number of clinic visits people with TB need to make before receiving a diagnosis. Social contact data from sub-Saharan Africa are limited14, however our estimate of the proportion of social contact that occurs in clinics falls with the range found by other studies15 16.
The social contact data used in our model were collected in 2019, before the start of the COVID-19 pandemic. Comparable social contact data were collected from the same study community in June-August 2020, during the pandemic, and suggested that reductions in contact time in clinics may have been smaller than reductions in other congregate locations7, possibly increasing the importance of transmission in clinics to overall disease burden during this period. The IPC interventions we simulated were also designed and parameterised before the start of the pandemic, and changing views on IPC may have changed the relative impact of the different interventions over longer time periods. For instance, the acceptability to patients of mask wearing may increase, increasing the coverage that can be achieved.
The main sources of uncertainty in our estimates come from three key inputs into the model: the proportion of contact time that occurs in clinics, the prevalence of TB in clinic attendees relative to the general population, and ventilation levels in clinics relative to other congregate settings.
Additional data collection in those three areas would be valuable, both in allowing us to reduce the uncertainty in our estimates, and in allowing similar estimates to be made for other settings.
There are a number of limitations to our work. Firstly, we do not explicitly consider infection to or from clinic staff. This may have led to us underestimating the proportion of disease that results from transmission in clinics, due to amplification of transmission in clinics by clinic staff, and to us slightly underestimated the impact of the interventions on community-wide TB incidence. The underestimates are likely to have been small however, as we estimate that only 7.1% (95% plausible range 2.3-16.7%) of all disease that results from transmission in clinics is in clinic staff, and contact time between clinic attendees and staff in clinics is much lower than contact time between clinic attendees. We also do not simulate children, as social contact data from children were not available. This will have had little effect on our estimates for adults, as the risk of Mtb transmission from children is low17, but means that we cannot estimate the proportion of disease in children that comes from transmission in clinics.
We do not consider the effects of risk factors other than HIV, such as diabetes. People with diabetes and some other risk factors are both likely to visit clinics more frequently, and are at increased risk of progression to disease following infection. By not including these risk factors, we may have underestimated the proportion of disease that results from transmission in clinics.
Finally, the representation of MDR in the model is relatively simple. We implicitly assume that with high coverage of Xpert MTB/RIF18, drug resistance is diagnosed for the majority of people at the same clinic visit as their TB is diagnosed. We therefore assumed that people with infectious MDR-TB spend no more time in clinics than people with infectious DS-TB, and so the proportion of TB from transmission in clinics does not vary by drug resistance status. We were not able to explore the effects of this assumption, due to the very low incidence of MDR-TB, and the need to use an IBM to accurately capture patterns of social contact behaviour. Future work should investigate if and if so, how, the proportion of MDR-TB that results from transmission in clinics varies from the proportion of DS-TB.
To conclude, we estimate that in the setting studied, 7.7% (4.0-14.2%) of tuberculosis in adults is acquired through transmission in PHC clinics, and that IPC interventions in clinics could reduce the total number of incident TB cases in the community in 2021-2030 by 3.4-8.0%. Given the relative ease of implementing IPC measures in clinics, compared to many other proposed TB control measures, we suggest that IPC measures in clinics should be considered to be ‘low hanging fruit’, and should be implemented both for their benefits to staff, but also for their likely effects on wider TB incidence and mortality.
Supporting information
Data Availability
The mathematical model used in this work is available from https://github.com/NickyMcC/ClinicTransmissionModel The social contact data used in this work will be made available from https://datacompass.lshtm.ac.uk/ on publication of the manuscript.
Funding/acknowledgements
The support of the Economic and Social Research Council (IK) is gratefully acknowledged. The project is partly funded by the Antimicrobial Resistance Cross Council Initiative supported by the seven research councils in partnership with other funders including support from the GCRF. Grant reference: ES/P008011/1. NM is additionally funded the Wellcome Trust (218261/Z/19/Z). RGW is funded by the Wellcome Trust (218261/Z/19/Z), NIH (1R01AI147321-01), EDTCP (RIA208D-2505B), UK MRC (CCF17-7779 via SET Bloomsbury), ESRC (ES/P008011/1), BMGF (OPP1084276, OPP1135288 & INV-001754), and the WHO (2020/985800-0). RMGJH is funded by ERC (action number 757699). TAY receives an NIHR Academic Clinical Fellowship.
We would like to thank the social contact survey respondents, and all of the Umoya omuhle project team involved in the data collection. Full details are given in the supporting material.
Data availability
The mathematical model used in this work is available from https://github.com/NickyMcC/ClinicTransmissionModel The social contact data used in this work will be made available from https://datacompass.lshtm.ac.uk/ on publication of the manuscript.