key: cord-318871-ffyavhm0
authors: Sy, K. T. L.; Haw, N. J. L.; Uy, J.
title: Previous and active tuberculosis in COVID-19 patients increases risk of death and prolongs recovery
date: 2020-07-26
journal: nan
DOI: 10.1101/2020.07.22.20154575
sha: 
doc_id: 318871
cord_uid: ffyavhm0

Background: There is growing literature on the association of SARS-CoV-2 and other chronic respiratory conditions, such as COPD and asthma. However, little is known about the relationship between coinfection with tuberculosis (TB) and COVID-19 outcomes. We aimed to compare the risk and survival time of death and recovery among COVID-19 patients with and without TB. Methods: We created a 4:1 propensity score matched sample of COVID-19 patients without and with TB, using COVID-19 surveillance data in the Philippines. We conducted a longitudinal cohort analysis of matched COVID-19 patients as of May 17, 2020, following them until June 15, 2020. The primary analysis estimated the risk ratios of death and recovery comparing COVID-19 patients with and without TB. Kaplan-Meier curves described time-to-death and time-to-recovery stratified by TB status, and differences in survival were assessed using the Wilcoxon test. Results: The risk of death in COVID-19 patients among those with tuberculosis was 2.17 times greater compared to those without tuberculosis (95% CI: 1.40-3.37). The risk of recovery in COVID-19 patients among those with tuberculosis was 25% less than the risk among those without tuberculosis (RR=0.75, 0.63-0.91). Similarly, time-to-death among COVID-19 patients with TB was significantly shorter (p=0.0031) and time-to-recovery among those with TB was significantly longer than patients without TB (p=0.0046). Conclusions: Our findings show that coinfection with tuberculosis increases morbidity and mortality in COVID-19 patients. Our findings reiterate the need to prioritize routine and testing services for tuberculosis, even with increased disruptions to health systems during the SARS-CoV-2 pandemic.

As of July 2020, the global burden of the COVID-19 pandemic has reached 8 million cases, and has caused large-scale outbreaks in many countries (1). The COVID-19 pandemic has caused substantial strain to healthcare systems worldwide, particularly in resource-limited settings with high prevalence of comorbid conditions such as TB. TB is also major global cause of morbidity and mortality, and was the cause of 1.4 million deaths in 2018, the largest single contribution by any one infectious disease (2) . TB disproportionately affects low-and-middle income countries (LMICs), where TB epidemics are fueled by coinfection with HIV/AIDS (3) and multidrug resistance (4, 5) . There have been significant TB treatment and program disruptions due to the pandemic (6, 7) , which have disproportionate impacts on socially disadvantaged communities with TB (8) . In particular, there has been decreases in care seeking for TB treatment, lessened enrollment in TB research studies, interruptions in availability of TB medicine and other products, and issues with obtaining food and nutritional support during the pandemic (9).

The COVID-19 outbreak in the Philippines is a continuing public health crisis, with 12, 513 reported cases as of May 17, 2020. TB is also a major public health problem in the Philippines as it has the third highest prevalence of TB globally, with one million individuals with active TB disease (10). Most research on the SARS-CoV-2 pathogen have been conducted in highincome countries, and studies suggest worse outcomes among COVID-19 patients with other respiratory diseases in these countries, such as COPD (11, 12) and asthma (13) . However, little is known about the relationship between coinfection with tuberculosis (TB) and COVID -19 outcomes. Pulmonary TB involves active destruction of lung parenchyma (14), which may predispose COVID-19 patients with TB to more aggravated disease and greater mortality. We hypothesize that COVID-19 patients with previous or active TB may have more severe clinical outcomes than those without TB. This study compares the risk of and time-to-death and . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted July 26, 2020. . recovery among COVID-19 patients who have and do not have TB coinfection in the Philippines.

Data were obtained from the Philippine national COVID-19 surveillance managed and implemented by the Philippine Department of Health Epidemiology Bureau (DOH-EB). One of the authors (NH) is involved in COVID-19 surveillance as a consultant. Data on Philippine COVID-19 cases have been de-identified and most of the data are publicly available through the DOH DataDrop initiative (https://www.doh.gov.ph/covid19tracker), thus institutional review board approval was deemed unnecessary.

We included all reported COVID-19 cases in the Philippines as of May 17, 2020, following them until June 15, 2020. COVID-19 cases were confirmed cases with positive real-time reverse transcription polymerase chain reaction (RT-PCR) from laboratories accredited by the DOH and Research Institute for Tropical Medicine (RITM) (15). Information on confirmed cases were collected using structured interviews upon initial consultation by the physician or nurse on duty as part of routine surveillance using a case investigation form (CIF), filled out manually or electronically. We used propensity score matching to create similar populations of COVID-19 patients with and without previous and active TB. To create the propensity scores, we predicted the risk of having TB using logistic regression with the potential confounders age, sex, and other comorbid conditions (chronic pulmonary obstructive disorder (COPD), asthma, diabetes, hypertension, diabetes, cancer, renal disease, cardiac disease, and autoimmune disorders). We matched 1 COVID-19 patient with TB to 4 without TB, using nearest neighbor matching of propensity scores, a caliper of 0.05, and with no replacement. As a secondary analysis, we . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted July 26, 2020. . https://doi.org/10.1101/2020.07.22.20154575 doi: medRxiv preprint created a propensity score matched subsample of only admitted COVID-19 patients, and conducted the same analysis.

Our main exposure of interest was a history or current diagnosis of TB. Comorbidity data and death certificate data encoded as part of the CIF were used as data sources to determine the history of TB infection.

The two main outcomes of interest were (1) death and (2) recovery. We modelled this as a dichotomous (yes/no) variable, as well as a time-to-event variable. The time-to-event variables for both primary outcomes were the time from symptom onset to outcome (i.e., time from symptom onset to death, time from symptom onset to recovery). Death was defined as deaths happening during active COVID-19 infection and declared as a death by the DOH-EB. Recovery was defined as those who had been declared as recovered by the DOH-EB based on rapidly evolving surveillance criteria on testing, clinical improvement, and additional days of quarantine.

We also assessed the time from hospital admission to discharge among those admitted to the hospital; hospitalization dates were validated from medical records. As a secondary outcome, we also assessed the relative risk of admission among the propensity score matched sample.

Descriptive statistics were generated to characterize the sample; t-tests were used to assess differences between age between those with and without TB. We conducted Pearson chisquare tests to examine differences in gender, comorbidities, health status and admission status. The primary analysis estimated the relative risks (risk ratios) of death and recovery with a modified Poisson regression with a robust variance estimator and a log link.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 26, 2020. . https://doi.org/10.1101/2020.07.22.20154575 doi: medRxiv preprint Kaplan-Meier curves were used to plot survival curves of the time-to-event variables stratified by TB status. We conducted nonparametric analysis for differences in survival in those with and without TB using the Wilcoxon test. Individuals were censored at their last date of follow-up if they did not have the outcome of interest. For the analysis of time-to-recovery, we censored those who had died on last day of administrative follow-up for the whole sample. This was done as to not upwardly the bias estimate of recovery by removing those who died as censored observations. All statistical analysis was conducted in R 4.0.0 (16).

The initial unmatched sample had 12,513 COVID- 19 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 26, 2020. Kaplan-Meier survival analysis demonstrated similar findings: the time-to-death among those with TB was significantly less than those without TB for both the propensity matched sample (p=0.0031) and in the subsample of those admitted (p=0.0052). Additionally, the time-torecovery among COVID-19 patients with TB was significantly greater than those who did have TB (full sample p=0.0046; admitted only p=0.02) (Figure 1) . However, the time-to-admission among those with and without TB were not significantly different (p=0.17).

The ongoing SARS-CoV-2 pandemic poses a challenge for TB prevention and treatment worldwide. Our findings show COVID-19 patients with TB have a two-fold increased risk of death, and are less likely to recover. Moreover, they died faster and recovered more slowly than those without TB, contributing to substantial morbidity and mortality. Our findings are consistent with preliminary findings on the Western Cape province of South Africa, where a current diagnosis of TB increased risk of death by 2.5 times, and a previous diagnosis of TB increased risk of death by 50% (17). To our knowledge, this is the first longitudinal cohort study examining the associations between COVID-19 and TB in a high-burden TB setting. As COVID-19 cases . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 26, 2020. . https://doi.org/10.1101/2020.07.22.20154575 doi: medRxiv preprint are continuing to increase in many low and middle income countries (LMICs) with high TB prevalence, such as Southern Africa, Latin America, and Asia (1), mitigation strategies need to target individuals with TB in order to reduce the burden of COVID-19 globally. Our results demonstrate the direct relationship of COVID-19 and TB, but SARS-CoV-2 also indirectly contributes additional TB-related morbidity and mortality. COVID-19 has and will continue to cause substantial strain to healthcare systems worldwide, and predictions expect increased mortality among TB patients due to significant TB program disruptions, diagnostic delays, TB treatment interruptions, and lack of access to drugs (6, 7, 9) .

There are a number of limitations of the study. We utilized propensity score matching to adjust for sources of measured confounding; however, residual confounding could still be an issue on confounders we did not match on. Specifically, we did not have data on HIV/AIDs in our cohort, and this might have biased our estimates since those with TB are more likely to have HIV.

However, propensity score matching on all other measured comorbid conditions would reduce confounding substantially, and any other potential residual confounding would be limited.

Moreover, as this data was obtained from the Philippine national surveillance effort on COVID-19, there are some missing dates for the time-to-event outcomes. In survival analyses, we excluded individuals with missing dates, and conducted statistical tests to ensure that the confounder distribution of the included sample for each time-to-event analysis between those with and without TB were not significantly different (Supplementary Table 2 ) to ensure limited confounding. In addition, the main outcomes death and recovery could potentially have missing data due to loss-to-follow up; survival analysis accounted for the missing data by censoring on the last day of follow-up for patients without the outcome.

Our findings reiterate the need to continue prioritizing routine and testing services for TB, even with overarching increased disruption to health and social systems during this pandemic.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 26, 2020. . https://doi.org/10.1101/2020.07.22.20154575 doi: medRxiv preprint Moreover, since a large proportion of patients with TB are also coinfected with HIV, further research on the co-occurrence of these three epidemics, as well as other endemic communicable and noncommunicable diseases in high TB-burden countries, is imperative in order to accurately assess the burden of SARS-CoV-2 globally. Additional research needs to focus on the interrelationship between TB and COVID-19 for appropriate planning and resource allocation, as SARS-CoV-2 continues to spread worldwide.

K. Sy contributed to conceptualization. K. Sy, N. Haw, and J. Uy contributed to data acquisition.

K. Sy contributed to data analysis. All authors contributed to interpretation of results and manuscript writing.

This work was not supported by any funding sources.

The authors have declared no conflicts of interest.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 26, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 26, 2020. . is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 26, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted July 26, 2020. . https://doi.org/10.1101/2020.07.22.20154575 doi: medRxiv preprint

World Health Organization (WHO) 2019 Global Tuberculosis Report

HIV and tuberculosis: a deadly human syndemic

MDR/XDR-TB management of patients and contacts: Challenges facing the new decade. The 2020 clinical update by the Global Tuberculosis Network

Recent updates on drug resistance in Mycobacterium tuberculosis

Contacts or care? Impact of COVID-19-related disruption on tuberculosis burden. Centre for Mathematical Modeling of Infectious Diseases Repository

Tuberculosis and coronavirus: what do we know? Epidemiologia e servicos de saude : revista do Sistema Unico de Saude do Brasil

Imperial College, Avenir Health

World Health Organization. It's time to end TB in the Philippines

The impact of COPD and smoking history on the severity of COVID-19: A systemic review and meta-analysis

ACE-2 expression in the small airway epithelia of smokers and COPD patients: implications for COVID

The European respiratory journal

Mycobacterium tuberculosis subverts negative regulatory pathways in human macrophages to drive immunopathology