Effect of ACE2 Expression Inhibiting Drugs on COVID-19 Disease Severity, Outcome and Length of Admission in Ethiopian Patients: A Causal Inference Using Marginal Structural Model with Inverse Probability Weight

The role of drugs that inhibit ACE2 expression on COVID-19 disease severity, progression and outcome has been debatable with studies reporting contradictory ndings. So far, there is no such study conducted in Africa. Having clarity on this issue is relevant as these drugs are the commonly prescribed medications for patients with co-morbid illnesses who are reported to be vulnerable to COVID-19 poor outcome. Therefore, the aim of this study was to assess the effect of acute or chronic ACEIs, ARBs and/or NSAIDs use on COVID-19 disease severity, outcome and length of admission among patients with COVID-19 admitted to the Millennium COVID-19 Care Center in Ethiopia. To and weighting Three outcome models; Multinomial Logistic Regression, Log Binomial Regression and Negative Binomial Regression models were tted to assess the effect of ACEIs, ARBs and/or NSAIDs use on disease severity, outcome and length of admission respectively. To predict all the three outcomes, the treatment variable alone was tted as an explanatory variable after adjusting for inverse probability weights. The study was conducted after obtaining ethical clearance from St. Paul’s Hospital Millennium Medical College Institutional Review Board. Written informed consent was obtained from the participants. The study had no risk/negative consequence on those who participated in the study. Medical record numbers were used for data collection and personal identiers were not used in the research report. Access to the collected information was limited to the principal investigator and condentiality was maintained throughout the project.


Results
Among the 945 patients studied, 115 (12.2%) had a history of ACEIs, ARBs and/or NSAIDs use. At admission, the majority (39.6%) had mild disease and 272 (28.8%) had severe disease. Among the study participants, 900 (95.2%) were discharged improved and the rest 45 (4.8%) died. The median length of admission was 14.0 days (IQR, [13][14][15][16]. Multinomial Logistic Regression, Log Binomial Regression and Negative Binomial Regression models were tted to assess the effect of ACEIs, ARBs and/or NSAIDs use on disease severity, outcome and length of admission respectively. In all the three outcome models, ACEIs, ARBs and/or NSAIDs use didn't show a statistically signi cant association with the outcomes.

Conclusions
Acute or chronic use of ACEIs, ARBs and/or NSAIDs showed no effect on COVID-19 disease severity, outcome and length of admission and therefore should not be withdrawn from patients who need these therapies.

Background
The Coronavirus pandemic has affected the entire world causing a tremendous loss to human life and also caused a burden to the existing health care system making it di cult to provide the best care possible for a better outcome especially in the developing countries. Because of this, the patient admission, treatment and discharge criteria has been continually improved by the World Health Organization (WHO) to accommodate the most in need cases to the health care provision system. To that end, risk strati cation based on disease presentation, severity, patient characteristics, existing medical conditions and drug intake history has been given great importance. Therefore, providing preventive services and strict observation for high risk groups should be strictly applied to prevent deterioration and complication at which point the care provided might not bring favorable results. With this aim, different researches were conducted with results showing geographical disparity and also inconsistency even in similar setups calling for the need for more research to be conducted especially in African setup with limited research reports on COVID-19 so far.
Among the proposed important predictors of COVID-19 disease progression and outcome is a history of taking drugs that inhibit the expression of Angiotensin-converting enzyme 2 (ACE2). This is proposed because of the pathological process of the SARS-CoV-2 virus entry into the body using ACE2. Therefore, taking these type of drugs (ACE inhibitors (ACEIs), angiotensin receptor blockers (ARBs) and nonsteroidal anti-in ammatory drugs (NSAIDs)) increase the level of ACE2 and in turn increasing the possibility of the virus to enter the body. Therefore, to better understand the effect of these drugs on disease progression and outcome (hospital/ ICU admission, length of hospital stay, complications, need for mechanical ventilation and mortality), different studies in different countries were conducted reporting contradicting results.
Systematic review and meta-analysis reports demonstrated that there is no increased risk of any of the disease related complication or outcome in those with ACEIs and ARBs intake history [1][2][3]. In addition, NSAIDs were reported to have no effect on disease progression and outcome in another systematic review [4]. Similarly, studies conducted in Italy, China, Korea, Spain and the United States show that the use of ACEIs, ARBs and NSAIDs did not affect disease progression, complication and outcome [5][6][7][8][9][10].
On the contrary, a study conducted in Turkey and Saudi Arabia reported that ACEIs and ARBs therapy were associated with higher risk of severe or critical COVID-19 disease, need of ICU care and higher incidence of in-hospital death [11,12]. As opposed to this, a study conducted in Kuwait showed that use of ACEIs and ARBs is inversely associated with ICU admission and mortality implying that these drugs have a protective effect from adverse disease outcome [13].
ACEIs, ARB s and NSAIDs are widely used drugs for the treatment of chronic conditions which are a problem of the developing countries, as much as it is a developed countries problem, showing an increasing trend in recent years with a larger proportion (77%) of deaths reported from low-and middleincome countries [14]. With such varying and contradicting reports and in the face of lack of evidence generated on the effect of these drugs on the disease in the African population, clinical judgement to continue or discontinue these life-saving medications should solely rely on evidence generated from the local population. Therefore, the aim of this study was to assess the effect of acute or chronic ACEIs, ARBs and/or NSAIDs use on COVID-19 disease severity, outcome and length of admission among patients with COVID-19 admitted to the Millennium COVID-19 Care Center in Ethiopia from July 2nd to December 25th, 2020.

Study setting, design and population
An institution-based retrospective cohort study was conducted at Millennium COVID-19 Care Center (MCCC), a makeshift hospital in Addis Ababa, Ethiopia dedicated for isolating and treating COVID-19 cases only. It is the largest Center in the country with a large ow of patients. The admission pattern initially included all types of patients who tested positive for SARS-Cov-2 with the aim of controlling the pandemic.
The follow up was made from July 2nd to December 25th, 2020. The source population was all cases of COVID-19 admitted at MCCC with a con rmed diagnosis of COVID-19 using RT-PCR, as reported by a laboratory given mandate to test such patients by the Ethiopian Federal Ministry of Health and who were on follow up from July 2nd to December 25th, 2020 [15].
All consecutively admitted patients with COVID-19 during the follow up period and who consent to participate were included in the study. With these criteria, a total of 945 patients with COVID-19 were included in the nal analysis.

Eligibility criteria
All patients with COVID-19 who were on treatment and follow up at the center from July 2nd to December 25th, 2020 and who consent to participate were included.

Data Collection Procedures and Quality Assurance
Data was extracted from patients' admission, follow up and discharge medical records using a pretested electronic data abstraction tool that is adopted from the WHO CRF form by trained data collectors. Appropriate infection prevention and control measures were followed during the data collection process.
Data quality was further assured through double data entry, and data cleaning through checking for inconsistencies, numerical errors and missing parameters. Where discrepancies are observed, data entered was veri ed with the primary data source. Once data cleaning was complete, data was exported to STATA software version 14 (College Station, TX) for analysis.

Statistical Analysis
Data was summarized using frequency tables and percentages. To compare the socio-demographic and clinical characteristics between the two groups (ACEIs, ARBs and/or NSAIDs users Vs Non users), Chisquare test, Fischer's exact test and independent t-test were used.
To identify the effect of ACEIs, ARBs and/or NSAIDs use on COVID-19 disease severity (mild vs moderate vs severe), disease outcome (alive vs dead) and length of admission (in days), Marginal Structural Model (MSM) with inverse probability weighting (IPW) approach was used.

Treatment model
The treatment model that uses binary logistic regression was tted to estimate the probability of exposure given the covariates (propensity score). The estimated probability of exposure was used to compute the inverse probability weights for each individual. The inverse of the probability of exposure was then used to weight each individual in the estimation of the marginal odds ratio. Variables to be included in the nal treatment model were selected by univariate analysis at 25% level of signi cance and also based on the existing literature reviewed.

Outcome models
There are three outcome variables in this study; disease severity (mild vs moderate vs severe), disease outcome (alive vs dead) and length of admission (in days). All the three outcomes were predicted by including the treatment variable alone in the respective models after adjusting for inverse probability weights.
To identify the effect of treatment on COVID-19 disease severity, Multinomial Logistic Regression model was used where adjusted relative risk (ARR), P-value and 95% CI for ARR were used to test the presence of statistically signi cant relationship.
To identify the effect of treatment on COVID-19 disease outcome, Log Binomial Regression model was used where adjusted relative risk (ARR), P-value and 95% CI for ARR were used to test the presence of statistically signi cant relationship.
To identify the effect of treatment on length of admission, Negative Binomial Regression model was used where adjusted relative risk (ARR), P-value and 95% CI for ARR were used to test the presence of statistically signi cant relationship. Negative binomial Poisson regression model was used instead of Standard Poisson regression model because the assumption of Standard Poisson regression model (mean equals variance) was checked and there was over dispersion depicted by comparison of mean and variance of the outcome variable and con rmed by the signi cance of dispersion parameter. And nally Model tness was checked for the Negative binomial Poisson regression model using Pearson chi square and deviance tests and the model ts the data well.
In all the three models, with a p-value of ≤ 0.05, the treatment was considered as a signi cant predictor of disease severity.
One hundred fteen (12.2%) had a history of ACEIs, ARBs and/ or NSAIDs use. At admission, the majority (39.6%) had mild disease and 272 (28.8%) had severe disease. Among the study participants, 900 (95.2%) were discharged improved and the rest 45 (4.8%) died. The median length of admission was 14.0 days (IQR, 13-16). (Table 1) Comparison of socio-demographic and clinical characteristics based on drug use history Based on the chi-square or Fischer's exact test and independent t-test result, a signi cant difference between those who has a history of ACEIs, ARBs and/or NSAIDs use and those who don't showed that the two groups showed a signi cant difference in age category, the presence of shortness of breath, disease severity, outcome and length of admission.
Accordingly, a signi cantly higher proportion of patients who has a history of ACEIs, ARBs and/or NSAIDs use history are in the age range of 50-59 years (27.0 % Vs 11.3%, p-value<0.0001) and 60 years and older (47.0 % Vs 20.4%, p-value<0.0001) compared to those with no drug use history.
A signi cantly higher proportion of patients with a history of ACEIs, ARBs and/or NSAIDs use has a shortness of breath at presentation compared to those with no drug use history (41.7% % Vs 25.1%, p-value<0.0001).
A signi cantly higher proportion of patients with a history of ACEIs, ARBs and/or NSAIDs use had severe COVID-19 disease at presentation compared to those with no drug use history (47.0% % Vs 26.3%, p-value<0.0001). Similarly, a signi cantly smaller proportion of patients with a history of ACEIs, ARBs and/or NSAIDs use had mild and moderate COVID-19 disease at presentation compared to those with no drug use history (26.1% % Vs 41.4%, p-value<0.0001 for mild disease and 27.0% Vs 32.3%, p-value<0.0001).
A signi cantly higher proportion of patients with a history of ACEIs, ARBs and/or NSAIDs use died of COVID-19 compared to those with no drug use history (9.6% % Vs 4.1%, p-value=0.010).
A statistically signi cant difference was observed in the length of admission where having a history of ACEIs, ARBs and/or NSAIDs use was associated with a delayed recovery compared to those with no drug use history (14.5 days Vs 14.4 days, p-value=0.002). But, this difference might not have a signi cant clinical implication. (Table 2) Treatment model: Logistic regression of factors affecting use of ACEIs, ARBs and/or NSAIDs The treatment model using a binary logistic regression model was run by including variables that were signi cant on univariate analysis at 25% level of signi cance and also from variables selected to be useful based on literature review. Accordingly, age category, sex, cardiac illness, hypertension, TIIDM, asthma, fever, cough, sore throat, runny nose, chest pain, myalgia, arthralgia, fatigue, shortness of breath and headache were included in the nal treatment model. By tting the nal treatment model, propensity score was estimated and it was used to compute the inverse probability weights for each individual. The inverse of the probability of exposure was then used to weight each individual in the estimation of the marginal odds ratio. (Table 3)  Outcome Model: Effect of ACEIs, ARBs and/or NSAIDs use on disease severity, outcome and length of admission Three outcome models; Multinomial Logistic Regression, Log Binomial Regression and Negative Binomial Regression models were tted to assess the effect of ACEIs, ARBs and/or NSAIDs use on disease severity, outcome and length of admission respectively. To predict all the three outcomes, the treatment variable alone was tted as an explanatory variable after adjusting for inverse probability weights.
Accordingly, on the three outcome models, ACEIs, ARBs and/or NSAIDs use didn't show a statistically signi cant association with all the three outcomes at 5% level of signi cance. (Table 4, 5 and 6)

Discussion
In this study, we assessed the effect of acute or chronic ACEIs, ARBs and/or NSAIDs use on COVID-19 disease severity, outcome and length of admission among patients with COVID-19 admitted to the Millennium COVID-19 Care Center in Ethiopia from July 2nd to December 25th, 2020. To our knowledge, this is the rst study conducted in the African set up. Understanding this helps (provides an input) in modifying the risk strati cation, prevention and admission practices so that better patient outcome can be achieved.
Based on the chi-square/ Fischer's exact test and independent t-test result, a signi cant difference between those who has a history of ACEIs, ARBs and/or NSAIDs use and those who don't showed that the two groups showed a signi cant difference in age category, the presence of shortness of breath, disease severity, length of admission and outcome. Accordingly, a signi cantly higher proportion of patients with a history of ACEIs, ARBs and/or NSAIDs were 50 years and older, had shortness of breath at admission, severe disease at presentation, had delayed recovery and died. But on further regression analysis using MSM model with IPW approach, use of ACEIs, ARBs and/or NSAIDs did not show a signi cant effect on disease severity, outcome and length of admission. Although there are few contradictory reports showing that these drugs have a signi cant effect on both directions affecting the disease outcome both negatively and positively [11][12][13], this nding is supported by a number of other institution and community based studies including systematic reviews conducted in non-African setup [1][2][3][4][5][6][7][8][9][10]. In addition, a WHO review based on studies conducted outside Africa also showed that there is no wellestablished evidence that patients on these drugs are at higher risk of poor outcome [16].
ACEIs, ARBs and/or NSAIDs are drugs which are widely used for the treatment of chronic medical conditions. In the current study, chronic medical conditions were reported in 413 (43.7%) of the participants among which hypertension and cardiac disease, which rely mainly on these drugs for treatment and control, constitutes 313 (75.8%) of the co-morbid illnesses. In addition, at the global and national level, these conditions are found in a considerable proportion of the general population, implying that the issue of continuing or discontinuing these drugs in patients with COVID-19 will continue to be raised. These medical conditions are also found to be signi cant determinants of disease severity and outcome among patients with COVID-19 [17][18][19][20][21]. Part of COVID-19 management is stabilizing existing co-morbid conditions so that the body can be at its best immunity for ghting the virus. Therefore, taking these medications is crucial to control the co-morbid conditions which otherwise could exacerbate and lead to progression of the disease ending up in complication and death.
Therefore, the use of these drugs is crucial as part of the management of patients with COVID-19 with comorbid conditions without affecting the COVID-19 disease severity, outcome and length of admission.

Conclusion
Based on the nding of this study, acute or chronic use of ACEIs, ARBs and/or NSAIDs showed no effect on COVID-19 disease severity, outcome and length of admission. Therefore, we recommend continuation of these drugs for the greater bene t of controlling the co-morbid conditions of patients of any COVID-19 severity. were used for data collection and personal identi ers were not used in the research report. Access to the collected information was limited to the principal investigator and con dentiality was maintained throughout the project.

Consent for publication
Not applicable Availability of data and materials: All relevant data are available upon reasonable request.