key: cord-0981268-s4w9bsdh
authors: Oddy, Christopher; Allignton, Jonathan; McCaul, James; Keeling, Polly; Senn, Dhanuja; Soni, Neesha; Morrison, Hannah; Mawella, Ruwani; Samuel, Thomas; Dixon, John
title: Inpatient omission of ACEi and ARBs is associated with morbidity and mortality in COVID-19
date: 2021-02-25
journal: Clin Ther
DOI: 10.1016/j.clinthera.2021.02.004
sha: 5ea1a475558e792a18b8b1c1ece15f348dc81b42
doc_id: 981268
cord_uid: s4w9bsdh

Purpose Due to the affinity of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) for the human angiotensin converting enzyme 2 (ACE2) receptor, use of angiotensin converting enzyme inhibitors (ACEi) and angiotensin receptor blockers (ARBs) has been a major concern for clinicians during the 2020 pandemic. Meta-analyses have affirmed that these agents do not worsen clinical outcomes in SARS-CoV-2 infection. To date, only a limited number of studies have directly looked at the safety of inpatient prescription of ACEi/ARBs during acute COVID-19 illness. Methods A retrospective cohort analysis was conducted to investigate the impact of inpatient provision of ACEi/ARBs on morbidity and mortality in patients admitted to hospital with COVID-19. Relationships were explored using linear and logistic regression. Findings Six hundred and twelve adult patients met our inclusion criteria of which 151 (24.7%) patients were established on ACEi/ARBs. Despite correction for known confounders, discontinuation of ACEi/ARBs was highly predictive of worsened outcomes in COVID-19. The proportion of doses omitted in hospital was significantly associated with increased mortality (p<0.001, OR=9.59 [2.55-36.09]), maximum National Early Warning Score (NEWS-2; p<0.001, OR=1.66 [1.27-2.17]), maximum oxygen requirements (p<0.001, OR=3.00 [1.83-4.91]), and maximum C-Reactive Protein concentration (CRP; p=0.030, OR=1.83 [1.06-3.17]). Implications Our data demonstrates a strong association between missed ACEi/ARB doses with increased morbidity and mortality. The available evidence supports continuation of current accepted practice surrounding ACEi/ARB therapy in acute illness – which is to limit drug omission to established acute contraindications, to actively monitor such decisions and to restart therapy as soon as it is safe to do so.

The number of confirmed severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections in the UK since the start of the pandemic exceeds two million as of the 1 st of January 2021. 1 Despite advances in both understanding and treatment, COVID-19 still has a high mortality when compared with seasonal influenza. 2 Advances in our understanding of the management of COVID-19 are clearly necessary if we are to reduce the impact of the disease. SARS-CoV-2 and its association with preexisting co-morbidities has been extensively described. 3 However, there remains a limited amount of evidence guiding clinicians on the safety of continuing, or discontinuing, patients' regular medications.

The safety of angiotensin-converting enzyme inhibitors (ACEi) and angiotensin receptor blockers (ARBs) in COVID-19 has been questioned. This concern emerged due to upregulation of the human angiotensin converting enzyme 2 (ACE2) receptor associated with ACEi/ARB therapy and the emergence of evidence that this protein acts as a functional receptor for SARS-CoV-2. 4, 5 Since then, a rapidly evolving evidence base supporting the safety of long-term ACEi/ARB use in COVID-19 has been developed. Broadly these suggest that antecedent ACEi/ARB therapy is not associated with worsened outcomes in COVID-19, and that their use might confer certain protective benefits. [6] [7] [8] [9] [10] [11] [12] [13] [14] The impact of ACEi/ARB provision during acute COVID-19 illness has been inadequately examined. To date few studies have attempted to address this question. One retrospective cohort concluded that complete suspension of therapy in hypertensive patients with COVID-19 is associated with greater morbidity and mortality. 15 Whether these findings are generalisable to all patients receiving ACEi/ARBs, and whether sporadic omission of these agents might also affect prognosis, remains unclear. The purpose of the present study is to examine, in further detail, the effects of suspending ACEi/ARBs on outcomes in patients admitted to hospital with COVID-19.

A retrospective cohort analysis was performed including patients admitted to St Helier Hospital in London (England) with COVID-19 between 10/01/2020 and 01/06/2020. Infection in all cases was confirmed by detection of SARS-CoV-2 RNA obtained by oropharyngeal/nasal swab. Participants were included if they met the following criteria:

 Their admission was coded with COVID-19 as the primary or secondary reason for admission, or COVID-19 was documented as cause 1a or 1b on deceased patients' death certificates.

 Their admission had reached a primary endpoint (deceased or discharged) by 23/07/2020.  The clinical coding of their admission was complete by 23/07/2020. Diagnosis codes are ordered according to their clinical significance, in terms of related morbidity and implications for management, during each inpatient episode. Admissions for which the primary or secondary diagnosis was COVID-19 were selected to prevent morbidity associated with unrelated clinical sequelae in the context of mild cases of SARS-CoV-2 infection being reflected in analyses.

Numerous SARS-CoV-2 positive patients were excluded for this reason and most likely represent mild cases, the implications of which are discussed below.

All participants received a standardized COVID-19 treatment protocol alongside their regular medications, unless suspended for clinical reasons. Due to rapidly evolving guidelines, the protocol was modified during the period we studied. The temporal distribution of patients receiving ACEi/ARBs during the study period tracked with the sample size of the remainder of the cohort, hence these changes are unlikely to have affected our results.

When assessing the effect of inpatient provision of ACEi/ARBs, patients were further excluded from analysis if they were palliated or died within 48 hours of admission. It was deemed unreasonable to assume that the effect of withholding these medications for 48 hours, or that discontinuation of these medications during palliative treatment, was sufficiently contributary to the outcome. 

Participants' demographic characteristics, past medical history and admission data were collated by our search engine for all inpatient admissions meeting the inclusion criteria. Laboratory values, clinical data and details of ACEi/ARB prescription were extracted manually from hospital electronic records. ACEi/ARB prescription was identified from admission documentation for their COVID-19 related admission. This information was cross referenced with, where available, record of recent distribution of ACEi/ARBs from GP records and with inpatient pharmacist "medicine reconciliations" derived from individuals' NHS Summary Care Record (SCR).

Our primary outcome was inpatient mortality. Secondary outcomes were intensive care unit (ICU) admission, length of stay, maximum oxygen requirement (L/min), maximum National Early Warning Score 2 (NEWS-2), maximum C-reactive protein concentration (CRP, mg/L), and maximum acute kidney injury (AKI) stage during admission. AKI was defined according to KDIGO creatinine criteria for Acute Kidney Injury. 16 NEWS-2 score was calculated according to standards set by the Royal College of Physicians. 17

The maximum oxygen requirement was defined as the highest flow rate of oxygen in litres per minute (L/min) required by participants during admission for more than two sets of observations. A threshold of over two sets of observations was selected to account for titration to saturations.

Venturi devices were converted from percentages to approximate L/min as follows: 24% = 3L/min, 28% = 5L/min, 35% = 9L/min, 40% = 11L/min, 60% = 13.5L/min. 18 Patients requiring non-invasive or mechanical ventilation were assigned a maximal score of 15L/min to allow for comparison with the remainder of the cohort.

ACEi/ARB provision as an inpatient was expressed as the proportion of days doses were received compared to the number of days they were required. The number of days required was defined as the number of days as an inpatient, excluding days after palliation and excluding the first day of admission if they were admitted past 9:00am. Counts were adjusted to exclude the days after a palliative decision was made to prevent the discontinuation of ACEi/ARBs after palliation being reflected in further analyses. The first day was excluded if they were admitted after 9:00am as it was assumed that these patients would have taken their ACEi/ARBs at home prior to attendance.

To adjust for hypotension as a cause for withdrawal of ACEi/ARBs, and the morbidity associated with this, the degree of hypotension during admission was estimated for patients taking regular ACEi/ARBs. Significant hypotension was defined as a systolic blood pressure of less than 100mmHg based on clinical data showing an increased mortality below this level and consensus guidelines from the European Society of Intensive Care Medicine. [19] [20] [21] The total number of blood pressure recordings taken during admission, prior to palliation, and the number of readings with a systolic value of less than 100mmHg were recorded. The proportion of hypotensive recordings compared to the total was calculated from this. The days prior to palliation were defined as the days up to and including the assumed date of palliation. The measure was adjusted for this to reflect the degree of hypotension during the period where the primary endpoint (death/discharge) was not yet assured.

Patients were deemed to be palliated on the day that anticipatory medications were prescribed, which are received as standard care for dying patients at our centre. The assumption was made that the decision to palliate was made on the same day.

Categorical variables were expressed as count numbers and percentages. Continuous variables were expressed as means ± standard deviation (±SD). Linear and logistic regressions were used to explore the relationships between our variables. Linear models were employed where the dependent variable was continuous, whereas logistic models were used if the dependent was binary. Results were expressed as odds ratios (OR) with their 95% confidence intervals (CI), and p-values for each component of the regression.

In univariate analyses of mortality, mortality was considered the dependent variable and ORs were calculated as the likelihood of mortality per unit increase in each parameter tested. When comparing patients taking regular ACEi/ARBs with non-users, ORs were calculated as the likelihood of a unit increase in each parameter associated with ACEi/ARB prescription -with ACEi/ARB prescription as the independent variable.

In multivariate analyses, the relationships between outcome measures and long-term ACEi/ARBs were adjusted for age, sex and clinical indications for ACEi/ARB prescription. In multivariate analyses of inpatient ACEi/ARB provision, relevant confounding variables were selected based on the premise that they might be associated with greater or lesser provision of ACEi/ARBs and might also influence morbidity or mortality. The "Enter" method was employed entering variables with a univariate logistic association with our primary outcome measure, mortality, with a p value of less than 0.2 into our models. Ethnicity was excluded in this manner. Having already adjusted for palliative omission of ACEi/ARBs, variables selected for adjustment were age, sex, comorbidities, ICU admission, maximum AKI grade and hypotension.

No sample size calculation was performed because we were unable to find appropriate published data from which to calculate this when data collection was commenced. A two-sided α of less than 0.05 was considered statistically significant. All statistical analyses were performed using SPSS version 27 (SPSS Inc., Chicago, Illinois).

Six hundred and twelve adult inpatient admissions met our inclusion criteria. The average age of our cohort was 69.6 (±17.8) years, of which 354 (57.8%) were male. Eighty-six patients (14.1%) required admission to the ICU. Two hundred and eighty-one patients died (45.9%), an elevated rate compared with our centre's average inpatient mortality due to our inclusion criteria excluding mild or incidental cases. One hundred and fifty-one (24.7%) patients in our cohort were taking ACEi/ARBs prior to admission, of which 98 (16.0%) were prescribed ACEi and 53 (8.7%) were prescribed ARBs [ Table 1 ]. Data was complete for each participant.

Advanced age and male sex were both significantly associated with an excess of mortality. All secondary outcomes were highly predictive of excess mortality in our cohort (p<0.001) apart from length of stay (p=0.629). A significant excess of mortality was also associated with long-term ACEi/ARB use in unadjusted studies. Accordingly, clinical indications for ACEi/ARBs prescriptionnamely hypertension, heart failure, ischaemic heart disease, and diabetes mellitus -displayed similar relationships [ Table 1 ].

Comparison between users of ACEi/ARBs and non-users demonstrated that users were significantly more likely to be elderly and to have a diagnosis of hypertension, heart failure, ischaemic heart disease, or diabetes mellitus. ACEi/ARB use was similar for both sexes (p=0.488). Outcomes in ACEi/ARB users were broadly poorer than non-users. Long term ACEi/ARB use was significantly associated with increased mortality, ICU admission, and unit increases in NEWS-2 score, flowrate of oxygen required, and CRP concentration in unadjusted studies. Non-significant relationships were displayed between ACEi/ARB use and both AKI stage and length of stay [ 

Prior to analysis 21 of the 151 patients taking ACEi/ARBs were excluded as they were palliated or died within 48 hours of admission, the assumption being that provision or non-provision of ACEi/ARBs in this period would not have been a major determinant of clinical outcomes. Fifteen received no doses as an inpatient. For patients that received ACEi/ARBs the average proportion of doses received was 63.3%. It is worth noting that 20 of 29 (69.0%) patients admitted to ICU received no doses of ACEi/ARBs during admission, and of those that did, none received ACEi/ARBs in ICU. It is common practice at our centre to discontinue regular antihypertensive medications in patients admitted to ICU. Hence, all multivariate analyses adjusted for ICU admission to avoid conflating the mortality associated with critical illness requiring ICU admission with the effect of ACEi/ARB provision. The mortality rate of patients admitted to ICU who received at least one dose of their ACEi/ARBs was comparable to those that received no doses, 22.2% and 33.3% respectively, in this small subgroup.

Lower inpatient provision of ACEi/ARBs when adjusted for factors that might prompt discontinuation of these medications was highly predictive of worsened outcomes in COVID- 19 Table 4 ]. The observed effects remain similar when patients that had all of their ACEi/ARB doses omitted are excluded from the analysis, implying that sporadic omission may impact clinical outcomes [ Table 5 ].

Within our cohort of 612 patients hospitalised with COVID-19 mortality was increased in those who were elderly, male, and that had significant comorbidities. Patients who required greater supplemental oxygen, exhibited higher NEWS-2 scores, and that were admitted to ICU were also at higher risk of death. Our subpopulation of 151 patients taking ACEi/ARBs were more likely to be elderly and to have comorbidities associated with ACEi/ARB prescription than the remainder of our cohort. Outcomes were broadly poorer in this subgroup in unadjusted studies.

After adjusting for age, sex, and comorbidities the relationships between ACEi/ARB use and indicators of morbidity diminished but remained significant. Mortality, however, in adjusted studies did not exhibit a significant relationship with ACEi/ARB use. Strikingly, lower inpatient provision of ACEi/ARBs was highly predictive of worsened outcomes in COVID-19. Markedly significant associations were observed between the proportion of ACEi/ARB doses omitted during acute COVID-19 illness and mortality, maximum NEWS-2, and maximum oxygen requirements. These associations remained after adjustment for hypotension, AKI, ICU admission, and palliation.

A large number of studies have examined the impact of long term ACEi/ARB prescription on outcomes in COVID-19. Several meta-analyses have concluded that antecedent ACEi/ARB prescription did not predispose to worsened outcomes in COVID-19. [6] [7] [8] Others concluded that prior ACEi/ARB use reduced the risk of mortality and ICU admission, and decreased length of stay. [9] [10] [11] [12] [13] [14] There is theoretical basis for both deleterious and protective effects of these agents in SARS-CoV-2 infection. 22 Notably however, there is no substantial clinical evidence that ACEi/ARB use worsens outcomes in COVID-19 when compared to appropriately matched controls.

To date, the one peer reviewed publication examining the effects of in hospital ACEi/ARB provision on COVID-19 illness concurs with our findings. 15 Lam et al demonstrate that total omission of ACEi/ARBs in hypertensive patients with SARS-CoV-2 infection is associated with similarly significant (p<0.001) increases in mortality and ICU admission. These effects were adjusted for AKI and hypotension, although the definition of the latter was not defined. We expand on these findings by demonstrating that similarly firm associations are seen for non-hypertensive patients taking ACEi/ARBs, that they are robust to adjustment for ICU admission and palliation, and that partial omission of these agents might also incur morbidity. The first randomised control trials concerning this topic are currently ongoing, the results of which will be the next important step in clarifying the role of ACEi/ARBs in acute COVID-19 illness. [23] [24] [25] [26] [27] Regular ACEi/ARB use in our study was associated with worsened indicators of morbidity. It is possible that this is attributable to the large number of our cohort that had their ACEi/ARBs omitted during admission, an intervention that we find to be strongly associated with poorer outcomes. The average proportion of pre-palliative ACEi/ARB doses received in our cohort was 36.5%. Furthermore, 42.3% of patients taking regular ACEi/ARBs received no doses during their admission. Trials reporting 100% inpatient provision of ACEi/ARBs show significant protective benefits associated with their use. 9 We propose that the mixed picture painted in the wider literature regarding ACEi/ARBs in COVID-19 is in part due to lack of standardisation to provision of these medications during the acute illness. A limitation of studies examining only antecedent use may be in confounding the significant morbidity associated with discontinuing these medications with the overall effect, masking potential benefits.

The findings of this study are derived retrospectively from purely observational data and thus suffer from the same limitations as all retrospective cohorts. It was not possible to perform a sample size calculation at the start of data collection, however, the strength of our findings suggests this was adequate. Our population size was comparable to the wider literature with subgroups reaching acceptable sizes. A selection bias towards moderate to severe cases of COVID-19 was introduced by our inclusion criteria. Unfortunately, no reliable data was available for body mass index (BMI); raised BMI is associated with increased mortality in patients with COVID-19. 28 Medication adherence prior to admission was not feasible to collect and represents a potential source of bias.

Our major finding that omission of ACEi/ARBs in hospital is associated with greater morbidity and mortality must be placed under strict scrutiny. Undeniably the decision to suspend ACEi/ARBs is often associated with clinical sequelae that correlate with morbidity. We believe our attempts to control for these factors are appropriate. Adjustments for both AKI and hypotension are for either event at any point in admission, irrespective of their timing in relation to ACEi/ARB omission, and hence may represent an over-correction. Furthermore, adjustment of the days ACEi/ARBs were required was conducted with conservative parameters to prevent misleading reductions in the proportions of inpatient ACEi/ARB provision. These adjustments were however based on assumptions made about the timing of doses taken at home and palliative decisions.

Our finding that omission of ACEi/ARBs in hospital is associated with greater morbidity and mortality in COVID-19 prompts greater attention to these agents during the acute illness. The cause for omission in this period was infrequently specified. Some doses were missed due to delayed prescription on admission, improper stock of medications and patients refusing doses. In cases where reasoning was documented suspension was often related to AKI. Several omissions were coordinated with a significant drop in blood pressure. Unfortunately, in the majority of cases, the reasoning behind suspension was unclear from electronic documentation.

Five patients in our cohort had their ACEi/ARBs suspended in the absence of AKI and remained persistently hypertensive during admission. Two of these patients received treatment with an alternative antihypertensive agent and three were left untreated. Whilst it is not clear whether there was other reasoning behind suspension, these patients likely represent a subgroup that would have benefited from continuation of their ACEi/ARBs. Clearly in clinical practice there will be instances where omission of ACEi/ARBs will be necessary in patients admitted with COVID-19. It is however well understood that improper management of ACEi/ARB cessation and reintroduction can cause harm. 29, 30 Practically speaking, we believe that our findings warrant active monitoring of decisions to suspend ACEi/ARBs in COVID-19 patients and prompt reintroduction of these agents in the absence of a clear contraindication.

We demonstrate here that lower inpatient provision of ACEi/ARBs is highly predictive of worsened outcomes in COVID-19 in patients established on this therapy. These associations remained after adjustment for common clinical indications for suspension of ACEi/ARBs in hospital -namely hypotension, AKI, ICU admission and palliation. The evolving view of ACEi/ARBs in COVID-19 favours these agents as protective against morbidity and mortality. We propose that the lack of clarity on the subject may relate to an absence of adjustment for inpatient provision in other studies; and confounding the associated morbidity of discontinuing these medications with the overall effect. Our findings prompt active monitoring of decisions to suspend ACEi/ARBs in COVID-19 patients and timely reintroduction of these agents in the absence of a clear contraindication. 

COVID-19) in the UK GOV.uk: Public Health England

Assessment of Deaths From COVID-19 and From Seasonal Influenza

Factors associated with COVID-19-related death using OpenSAFELY

Effect of angiotensin-converting enzyme inhibition and angiotensin II receptor blockers on cardiac angiotensin-converting enzyme 2

Cryo-EM structure of the 2019-nCoV spike in the prefusion conformation

Angiotensin converting enzyme inhibitors and angiotensin II receptor blockers and outcomes in patients with COVID-19: a systematic review and meta-analysis

The Effect of Prior ACEI/ARB Treatment on COVID-19 Susceptibility and Outcome: A Systematic Review and Meta-Analysis

Treatment with ACE inhibitors or ARBs and risk of severe/lethal COVID-19: a meta-analysis

Angiotensin enzyme inhibitors and angiotensin receptor blockers as protective factors in COVID-19 mortality: a retrospective cohort study

Renin-angiotensinaldosterone system inhibitors and the risk of mortality in patients with hypertension hospitalised for COVID-19: systematic review and meta-analysis

ACEI/ARB use and risk of infection or severity or mortality of COVID-19: A systematic review and meta-analysis

Aldosterone System (RAAS) Inhibitors is Associated with a Lower Risk of Mortality in Hypertensive COVID-19 Patients: A Systematic Review and Meta-analysis

The influence of ACE inhibitors and ARBs on hospital length of stay and survival in people with COVID-19

Association of renin-angiotensinaldosterone system inhibitors with mortality and testing positive of COVID-19: Meta-analysis

Continued In-Hospital Angiotensin-Converting Enzyme Inhibitor and Angiotensin II Receptor Blocker Use in Hypertensive COVID-19

Patients Is Associated With Positive Clinical Outcome

Acute Kidney Injury: Definition, Pathophysiology and Clinical Phenotypes

British Thoracic Society Guideline for oxygen use in adults in healthcare and emergency settings

Nontraumatic out-ofhospital hypotension predicts inhospital mortality

The association between systolic blood pressure and in-hospital mortality in older emergency department patients who are hospitalised with a suspected infection

The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3)

ACE-inhibitors, angiotensin receptor blockers and severe acute respiratory syndrome caused by coronavirus

Elimination or Prolongation of ACE Inhibitors and ARB in Coronavirus Disease 2019 (REPLACECOVID) ClinicalTrials.gov: U.S. National Library of Medicine

Continuing versus suspending angiotensin-converting enzyme inhibitors and angiotensin receptor blockers: Impact on adverse outcomes in hospitalized patients with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)--The BRACE CORONA Trial

Effects of discontinuing renin-angiotensin system inhibitors in patients with COVID-19 EU Clinical Trials Register2020

COVID-19 -ACE inhibitors or ARBs discontinuation for Clinical Outcome Risk reduction in patients hospitalized for the Endemic Severe acute respiratory syndrome coronavirus (SARS-CoV-2) infection: the randomized ACORES-2 study EU Clinical Trials Register2020

Obesity in patients with COVID-19: a systematic review and meta-analysis

Stopping RAS Inhibitors to Minimize AKI: More Harm than Good?

Association of Angiotensin-Converting Enzyme Inhibitor or Angiotensin Receptor Blocker Use With Outcomes After Acute Kidney Injury

The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. Data collection was conducted by authors CO, JA, JM, PK, DS, NS, HM, RW and TS. Author CO was responsible for the conception, planning, conduct and reporting of this project. Oversight on all aspects of the project was conducted by authors CO and JD who will act as guarantors. Statistical analysis was performed by CO and both checked and ratified by Dr David Young, lecturer in the Department of Mathematics and Statistics at Strathclyde University. Article writing was conducted by authors CO and JM, and proofread by authors PK, JA and JD. No aid in writing this publication was sought from either a medical writer or medical editor.We disclose no personal communications instrumental in the completion of this manuscript that are not otherwise recognised by authorship.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.