Journal Logo

ORIGINAL PAPERS: Views on COVID-19 infection

Association between renin–angiotensin–aldosterone system blockers and outcome in coronavirus disease 2019: analysing in-hospital exposure generates a biased seemingly protective effect of treatment

Lahens, Alexandrea; Mullaert, Jimmyb,c; Gressens, Simond; Gault, Nathaliec; Flamant, Martina,e; Deconinck, Laurèned; Joly, Véroniqued; Yazdanpanah, Yazdanb,d; Lescure, François-Xavierb,d; Vidal-Petiot, Emmanuellea,e

Author Information
doi: 10.1097/HJH.0000000000002658
  • Free

Abstract

INTRODUCTION

The membrane-bound angiotensin-converting enzyme type 2 (ACE2) is the receptor for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Very early into the coronavirus disease 2019 (COVID-19) pandemic, studies have shown a high prevalence of cardiovascular comorbidities in patients with COVID-19, especially in those with severe forms of the disease [1–3]. In addition, some animal data [4,5] suggest that ACE2 expression might be increased in patients treated with renin–angiotensin–aldosterone system (RAAS) blockers. This led several authors to hypothesize that angiotensin-converting enzyme inhibitors (ACEI) and angiotensin receptor blockers (ARBs) may facilitate viral entry into host cells and therefore increase the risk for SARS-CoV-2 infection, and to raise the question of temporary treatment cessation during the pandemic or in patients with confirmed infection [6–9]. These widely shared viewpoints, including in general and social media, led most scientific societies to take position – all in favour of continued use of both ACEI and ARBs in the absence of clear evidence of harm [10].

Conversely, ACE2 is a key counter-regulator of ACE, by converting angiotensin II to angiotensin (1–7), a peptide with vasodilating, antifibrotic and anti-inflammatory properties, opposite to those induced by angiotensin 2 after binding to its type 1 angiotensin II receptor. As previously shown for SARS-CoV, SARS-CoV-2 viral infection may downregulate ACE2 and thereby lead to a deregulation of the ACE-angiotensin II/ACE2-angiotensin (1–7) homeostasis in favour of angiotensin II, aggravating lung injury [11,12]. Therefore, other authors have advocated the use of ARBs to avoid angiotensin II-induced deleterious effects during COVID-19 [10,13].

Meanwhile, given the public health emergency to provide evidence-based data, several observational studies have attempted to establish whether RAAS blockers increased the risk to get infected with COVID-19 [14–18] and/or modified the course and prognosis of the disease in infected patients [15–41]. For the former, the exposure variable is always ‘chronic exposure to RAAS blockers’, derived from electronic medical records. For the latter, the exposure definition is more variable. Some studies also collected chronic treatment, hence treatment before diagnosis or hospital admission [15–28]. Other studies analysed in-hospital exposure [33,35,36], or, very similarly, prehospital exposure continued after admission [29,32,34]. In some cases, exposure was unclear, but was likely collected after admission [31,37,39,40]. A few studies analysed both prior and in-hospital treatment, with little data on differences generated by the different procedures of exposure measurement [30,38].

Although study design and adjustment for confounders differed, studies based on chronic exposure to RAAS blockers yielded similar results. Indeed, although some of them showed a higher risk of severe COVID-19 in patients chronically treated with RAAS-blockers in unadjusted analyses, nearly all of them concluded to the absence of association between RAAS blockers and outcome of the disease after adjustment. Conversely, a common and systematic finding of all analyses based on in-hospital exposure to treatment was a marked protective effect of RAAS blockers, with significantly lower risks of severity of the disease and/or mortality [29–31,33–36]. We suspected that a bias related to the reason of continuation or discontinuation of treatment after hospital admission might explain these discrepant results.

The aim of the study was to collect RAAS blocker prescription before and during the first 7 days after admission in a population of patients hospitalized for COVID-19, and to establish the associations between exposure to RAAS blockers prior to or during hospitalization and outcome of the disease. Reasons for treatment discontinuation or initiation were very carefully recorded to assess how much analyses based on in-hospital exposure may be biased.

METHODS

Study design and oversight

Data from all patients admitted for COVID-19 in the department of infectious diseases of Bichat Hospital (Paris, France) from 23 January to 29 April 2020 were retrospectively analysed. Baseline characteristics, comorbidities, medications, test results, and clinical outcomes of all patients were extracted manually from electronic medical charts by four trained doctors of the unit. Prior exposure to ACEI or ARBs (hence chronic treatment, before hospital admission), prescription of these drugs at days 1, 3, and 7 posthospital admission, and reason for treatment modification and date of modification, if relevant, were recorded. All data regarding exposure to RAAS blockers and outcomes were independently verified by a second member of the team.

The current study was approved by the institutional review board (IRB 00006477) of Assistance-Publique Hôpitaux de Paris, AP-HP.Nord.

Study population and diagnosis confirmation

The retrospective cohort study includes all adult patients (≥18 years old) admitted with a confirmed or probable diagnosis of COVID-19. Confirmed diagnosis was based on a positive real-time reverse-transcriptase PCR test from a nasopharyngeal swab. Probable diagnosis relied on the combination of epidemiological context, recent medical history compatible with known natural history of the disease and evocative lesions of computed tomography scan (analysed through a predefined algorithm by trained radiologists), despite negative PCR test. We excluded from the probable diagnosis group patients with secondarily revised diagnosis after initial treatment, such as isolated left-sided heart failure.

Study outcomes

The primary outcome was in-hospital mortality within 30 days of admission. The secondary outcome was a severe disease defined as the need for at least 9 l/min of oxygen (the threshold for high-concentration mask in our unit), ICU admission, or death, within 30 days of admission. In case of patients transferred to another hospital, telephone contact was made with the medical team to ensure completeness of data. There were no missing outcome data.

Exposure definitions

Chronic exposure to RAAS blockers was based on the chronic treatment of the patients, prior to hospitalization, independently of whether treatment was continued or not in-hospital.

We also defined four different patterns depending on exposure to RAAS blockers across hospital admission (using a 7-day window after admission).

  • (1) Patients never exposed.
  • (2) Patients who were not treated prior to admission and initiated treatment during the first 7 days of hospital admission.
  • (3) Patients who were previously exposed and discontinued treatment within the first 7 days of hospital admission. If treatment was later resumed after temporary cessation they were still analysed in this discontinued treatment group.
  • (4) Patients who were previously exposed and continued treatment.

A patient who died or was discharged alive while still treated within the first 7 days after hospital admission was included in the ‘continued treatment’ arm.

In-hospital exposure corresponded to patients from groups 2 and 4 and absence of in-hospital exposure corresponded to patients from groups 1 and 3.

In a sensitivity analysis, these four patterns were analysed using a 3-day window after hospital admission.

Statistical analysis

Data are reported as number and percentage for categorical data and median and interquartile range (IQR) for continuous variables. Baseline characteristics of patients with or without exposure to RAAS blockers prior to hospitalization were compared using Fisher exact test or Wilcoxon test, as appropriate. Mortality rates among patients with severe disease and among those admitted in ICU were compared using chi-squared tests.

The associations between exposure and outcomes were analysed with event rates, unadjusted and adjusted odds ratios (ORs), and their 95% confidence intervals (CIs). Event rates for the four patterns of prescription were calculated. Exact CI for event rates were calculated using binomial distribution. Covariates in the adjusted logistic regression model were selected a priori as potential confounders and included age, sex, smoking, history of diabetes, hypertension, coronary artery disease, heart failure, dyslipidaemia, and chronic kidney disease (defined as estimated glomerular filtration rate of less than 60 ml/min per 1.73 m2). There were no missing data except for smoking (in 23 patients, which were counted as never smokers because this was the most likely explanation for the missing information).

All analyses were performed with the software R version 3.6.1 (R Foundation, Vienna, Austria).

RESULTS

A total of 347 patients were included in the study. Median age was 61 years (IQR, 51–72), 209 (60%) were male, 169 (49%) had a history of treated hypertension, 98 (28%) had a history of diabetes, and 18 (6%) were active smokers (Table 1). The diagnosis of COVID-19 was confirmed by a positive PCR in 320 patients and probable in 27 patients. Median delay between onset of symptoms and hospital admission was 7 [3–10] days. During the 30 days following admission, 125 (36%) patients developed a severe form of the disease and 47 (14%) died. The median delay between hospital admission and death was 10 days (IQR, 6–17).

TABLE 1 - Baseline characteristics of the patients, in the total population (N = 347), or in patients previously treated or not previously treated with renin–angiotensin–aldosterone system blockers
Parameter All patients RAAS blockers before admission, N = 117 No RAAS blockers before admission, N = 230 P value
Age (years) 61 [51–72] 65 [58–75] 58 [48–71] <0.001
Male sex 209 (60) 69 (59) 140 (61) 0.73
Smoking
 Current 18 (6) 8 (7) 10 (5) 0.49
 Former 68 (21) 25 (23) 43 (20)
 Never 238 (73) 78 (70) 160 (75)
 Treated hypertension 169 (49) 110 (94) 59 (26) <0.001
 Diabetes 98 (28) 57 (49) 41 (18) <0.001
 Obesity 98 (39) 38 (45) 60 (37) 0.32
 Coronary artery disease 35 (10) 20 (17) 15 (7) 0.0039
 History of Heart failure 26 (7) 15 (13) 11 (5) 0.0096
 COPD 17 (5) 7 (6) 10 (4) 0.60
 Asthma 25 (7) 6 (5) 19 (8) 0.38
 Chronic kidney disease 57 (16) 31 (26) 26 (11) <0.001
 Solid organ transplant 10 (3) 2 (2) 8 (3) 0.51
 Dyslipidaemia 76 (22) 43 (37) 33 (14) <0.001
Treatment before admission
 Beta-blockers 69 (20) 39 (33) 30 (13) <0.001
 Hydrochlorothiazide 44 (13) 37 (32) 7 (3) <0.001
 Calcium channel blocker 78 (22) 44 (38) 34 (15) <0.001
 Insulin 34 (10) 20 (17) 14 (6) 0.0019
 Statin 75 (22) 52 (44) 23 (10) <0.001
 Oral antidiabetic agent 63 (18) 40 (34) 23 (10) <0.001
 NSAIDs 6 (2) 1 (1) 5 (2) 0.67
 Immunosuppressant therapy 13 (4) 5 (4) 8 (3) 0.77
 Antiplatelet agent 56 (16) 32 (27) 24 (10) <0.001
Data are in n (%) for categorical variables and median [interquartile range] for continuous variables. All parameters were available for all 347 patients except smoking habit which was missing in 23 patients and BMI (used to define obesity), which was missing in 101 patients (29%, similar proportion in both groups). COPD, chronic obstructive pulmonary disease; RAAS, renin–angiotensin–aldosterone system.

Prescription of renin–angiotensin–aldosterone system blockers prior to hospitalization and outcome

Chronic treatment included an ACEI in 55 patients and an ARB in 62 patients. One patient received dual RAAS blockade. Nonmutually exclusive indications for ACEI were hypertension in 93% (95% for ARB), heart failure in 16% (10% for ARB), coronary artery disease in 20% (16% for ARB), and proteinuria in 11% (5% for ARB) of the patients. Compared with patients who did not receive a RAAS blocker prior to hospitalization, patients who received a RAAS blocker prior to hospitalization were older, had a higher frequency of hypertension, diabetes, dyslipidaemia, coronary artery disease, previous heart failure, or chronic kidney disease (Table 1).

Event rates and ORs for mortality and severe disease are reported in Table 2. Among patients with prior exposure to RAAS blockers, 17% died within 30 days of hospital admission versus 12% in patients without prior exposure to RAAS blockers [OR = 1.60 (95% CI, 0.82–2.89), P = 0.17 and adjusted OR 0.62 (95% CI, 0.25–1.48), P = 0.28] and 33% developed a severe disease versus 37% in patients without prior exposure to RAAS blockers [OR 0.84 (95% CI, 0.52–1.33), P = 0.46 and adjusted OR 0.39 (95% CI, 0.20–0.74), P = 0.005].

TABLE 2 - Association of exposure to renin–angiotensin–aldosterone system blockers, prior to hospitalization or during hospital stay, with 30-day mortality (primary outcome) and disease severity (secondary outcome)
30-Day mortality, N = 47 Severe disease within 30 days, N = 125
Event rate % (95% CI) OR (95% CI) P value Adj OR (95% CI) P value Event rate % (95% CI) OR (95% CI) P value Adj OR (95% CI) P value
All, N = 347 14 (10–18) 36 (31–41)
No prior prescription, N = 230 12 (8–17) 1 (–) 1 (ref) 37 (31–44) 1 (–) 1 (–)
Prior prescription, N = 117 17 (11–25) 1.60 (0.82–2.89) 0.17 0.62 (0.25–1.48) 0.28 33 (25–43) 0.84 (0.52–1.33) 0.46 0.39 (0.20–0.74) 0.005
No in-hospital exposure, N = 263 15 (11–20) 1 (–) 1 (ref) 40 (34–46) 1 (–) 1 (–)
In-hospital exposure, N = 84 8 (3–16) 0.51 (0.20–1.11) 0.11 0.25 (0.09–0.65) 0.007 23 (14–33) 0.43 (0.24–0.75) 0.0038 0.23 (0.11–0.45) <0.001
Adj OR, adjusted odds ratio; CI, confidence interval; OR, odds ratio.

Mortality rates among patients admitted to ICU were significantly higher in patients with prior exposure to RAAS blockers (11/24, 46%) than in patients without prior exposure to RAAS blockers (10/59, 17%, P = 0.014). Likewise, mortality rates among those who developed severe illness tended to be higher in patients with prior exposure to RAAS blockers (20/39, 51%) than in patients without prior exposure to RAAS blockers (27/86, 31%, P = 0.053).

Patterns of prescription after hospital admission and corresponding event rates

The flow chart detailing patterns of RAAS blocker prescription before and after hospital admission, and corresponding outcomes, is shown in Fig. 1. Among the 117 patients who received a RAAS blocker prior to hospitalization, treatment was continued for at least 7 days after hospital admission in 78 (67%) patients. Within the same timeframe, treatment was discontinued in 39 patients (33%), in the vast majority (33 of them, 85%) within the first 48 h of admission. Among the 230 patients who were not treated with RAAS blockers prior to hospitalization, treatment was initiated de novo within a week of admission in six patients (3%). The remaining patients (n = 224) were never exposed to RAAS blockers.

FIGURE 1
FIGURE 1:
Flow chart of the study, event rates associated with each prescription subgroup and potential biases. Or discharged alive within 7 days, with treatment continued (no patient died within the first 7 days while still treated). ∗∗ N = 19 stopped at admission, N = 14 stopped before day 3, N = 6 stopped before day 7. The different potential biases are indicated in italic, below the step where they occur. £Source of imprecision, attenuates the potential real effect of treatment in analysis based on prior exposure. AKI, acute kidney injury; ICU, intensive care unit; RAAS, renin-angiotensin-aldosteron system.

The reason underlying treatment discontinuation was ICU admission or decision of palliative care in 16 patients, acute kidney injury or hypotension in 20 patients, and was a seemingly arbitrary medical decision in three patients (Fig. 1). In five patients, discontinuation was temporary, and treatment was resumed by day 7 (all of these patients survived).

In the four groups defined by patterns of exposure to treatment before and after hospital admission, outcomes varied greatly (Fig. 1). Mortality rate was 9% when treatment was maintained for 7 days after hospital admission and 0% in patients with de novo prescription (8% for these two groups analysed together). It was 33% when treatment was discontinued, and 12% in patients who were never exposed to RAAS blockers, either before or during hospital-stay. Rates for severe disease followed a similar pattern: 22% of the patients who continued treatment developed a severe disease versus 56% for those who discontinued treatment. In those who were never exposed or initiated treatment, the rates of severe disease were 38 and 33% (Fig. 1).

If the timeframe of exposure had been a 3-day window (versus 7), event rates in the group of patients who continued treatment would have been 26 (versus 22%) and 12% (versus 9%) for severe disease and mortality, respectively.

Association between in-hospital exposure and outcomes

In-hospital exposure tended to be associated with a lower risk of mortality [rates of 8 versus 15%, OR 0.51 (95% CI 0.20–1.11), P = 0.11], and this protective effect became significant after adjustment [adjusted OR 0.25 (95% CI 0.09–0.65), P = 0.007]. In addition, in-hospital exposure was associated with less severe disease [23 versus 40%, OR 0.43 (95% CI 0.24–0.75), P = 0.038, adjusted OR 0.23 (95% CI 0.09–0.65), P < 0.001], as reported in Table 2.

DISCUSSION

In this retrospective study of 347 patients hospitalized for COVID-19 in the department of infectious diseases of Bichat Hospital (Paris, France), 117 (34%) received a RAAS blocker as part of their chronic treatment. A multivariable adjusted logistic regression model found no significant association between RAAS blocker prescription prior to hospitalization and in-hospital mortality, whereas adjusted risk to develop severe illness was lower in patients with prior exposure to RAAS blockers. After hospital admission, a third of the patients chronically treated with RAAS blockers discontinued their treatment, for reasons directly or indirectly related to disease severity in the vast majority of the cases. Thereby, in-hospital exposure to RAAS blockers was influenced by reverse causality and was spuriously associated with markedly lower adjusted risks of mortality and severe disease.

A key finding of our study is the demonstration that analyses based on in-hospital exposure, although they better capture treatments actually received during this important phase of the disease, are just as much impacted by baseline confounding as are those based on chronic exposure, but even more importantly although less recognized, they are prone to a specific bias arising from a different source of confounding, and generating a spurious protective association of RAAS blockers with outcome of COVID-19, away from the null hypothesis. In our dataset, analyses based on in-hospital exposure yielded adjusted ORs of 0.25 (95% CI 0.09–0.65) and 0.23 (95% CI 0.11–0.45) for mortality and severe illness, respectively. Effects of the same range order were found in previous studies which captured in-hospital exposure to treatment [29–31,33–36].

Although these studies more or less cautiously discussed these results as showing a potential protective effect of treatment, our detailed examination of the reason for treatment cessation provides a different explanation for these seemingly beneficial results. Our analysis indeed demonstrated that outcomes associated with in-hospital treatment continuation, discontinuation or initiation are highly biased, both by an immortal-time bias (continued treatment arm) and by reverse causality (discontinued treatment arm). Defining the exposed arm based on a certain treatment duration indeed generates an immortal time bias [42,43]. Patients who continued treatment during this timeframe could not have had an adverse event requiring treatment cessation during that time. If the window of exposure had been 3 days (instead of 7), event rates would have been slightly less favourable for patients who continued treatment: the longer the timeframe, the greater the immortal time bias. The mirror bias is that occurring in patients who discontinued treatment within the first 7 days in hospital. In the current study, a third of previously exposed patients interrupted treatment during the first 7 days of hospital admission, the vast majority of them within the first 48 h. In nearly half of these patients, treatment cessation was directly related to disease severity (severe illness caused treatment cessation, and not the reverse, a typical case of reverse causality), and in the remaining patients, it was discontinued for other reasons (acute kidney injury, hypotension without alteration of renal function or seemingly arbitrary medical decision), most of which were early markers of a severe disease [44]; hence, a discontinuation bias also influenced by reverse causality [45]. The bias would have been even worse if patients who resumed treatment after initial discontinuation (who were left in the discontinuation group in our study) had been analysed in the continued treatment arm, which some previous studies may have done.

Overall, in observational pharmacoepidemiological studies trying to analyse effect of treatment, any treatment cessation related directly or indirectly to the later development of the outcome, and analysed in the nonexposed arm, artificially worsens the prognosis of nonexposed patients, and conversely treatment initiation or continuation occurs in healthier patients and will improve the outcome of the exposed patients, the so-called healthy user-sick stopper bias (Fig. 1), which may erroneously lead to conclude that treatment is beneficial [45,46].

The healthy user-sick stopper bias is a very likely explanation as to why all studies which captured ‘in-hospital’ treatment [29–31,33–36], unlike those which captured chronic (prehospital or prediagnosis) treatment [15–25,27,28], showed a favourable outcome associated with RAAS blockers.

For instance, in a multicentric study in nine hospitals in Hubei (China), Zhang et al. collected in-hospital exposure to RAAS blockers and reported a propensity score-matched and adjusted hazard ratio of 0.37 (95% CI 0.15–0.89) for 28-day mortality in patients exposed versus those nonexposed to treatment [35]. In another multicentre study conducted in 17 hospitals from the same region, Zhou et al. analysed data from 906 patients treated with RAAS blockers matched with 1812 nontreated patients, all hospitalized for COVID-19, and reported that in-hospital use of ACEI/ARB was associated with a significantly lower risk of 28-day all-cause mortality of COVID-19 (adjusted hazard ratio of 0.39; 95% CI, 0.26–0.58; P < 0.001) [36]. In a smaller single-centre Italian study of 397 patients hospitalized for COVID-19 in Milan, patients who continued chronic RAAS blocker treatment during hospitalization had an adjusted OR for mortality of 0.14 (95% CI 0.03–0.66) compared with those who did not receive in-hospital treatment (patients never exposed, or treatment discontinued) [29]. It is unfortunate that these studies and a few smaller scaled other ones with similar biases [31,33,34] have been analysed together with studies based on chronic exposure in meta-analyses [22,47,48], leading these to conclude to a beneficial effect of treatment.

This bias is far from being anecdotal. In our study, a third of the previously exposed patients had their treatment interrupted within a week of hospital admission. To our knowledge, three other studies reported the rates of treatment discontinuation after admission [27,29,30]. In the above-mentioned single-centre study conducted in 397 patients hospitalized for COVID-19 in Milan, RAAS blockers were interrupted in 68% of the previously treated patients upon hospital admission [29]. In a case series of patients admitted to 12 hospitals in the New York area, among 2411 patients with data available for home and in-hospital treatment, more than half of the patients discontinued treatment during their hospital stay [27]. Finally, Chaudhri et al. reported a 39% rate of treatment discontinuation in a cohort study of patients hospitalized with COVID-19 in a University Hospital of New York state (USA) [30]. These very high rates of treatment discontinuation after hospital admission reported in different countries and settings confirm that this event can majorly impact the results of studies based on in-hospital exposure. Our results highlight that in pharmacoepidemiological studies, the way exposure variable is documented and analysed should be very carefully explained and unambiguous.

Importantly, the ‘healthy user-sick stopper’ bias, inherent to in-hospital exposure measurement, should not be confused with confounding due to different baseline characteristics between the exposed and nonexposed groups, or indication bias, which impacts both analyses based on chronic and on in-hospital treatment. As expected, and as shown in previous studies [14,16,19,20,24,25], adjusting for age and comorbidities markedly reduced the value of the ORs for the associations between RAAS blocker exposure and adverse outcomes, and this was true for mortality as well as severe disease, and for chronic as well as in-hospital prescription (Table 2). Confounding due to different baseline characteristics between treatment groups (older age and comorbidities are associated to both exposure and outcome, Fig. 2) is inherent to all observational studies, and avoiding confounding is the main purpose of pseudorandomization methods such as propensity score analyses in observational research or random treatment allocation in trials. It is highly relevant in the specific context of RAAS blockers and COVID-19, as comorbidities treated with RAAS blockers, as well as older age, have been shown to be associated with more severe forms of COVID-19 [1,2,27].

FIGURE 2
FIGURE 2:
Causal diagram illustrating bias due to confounding and bias due to conditioning on a collider (selection bias). In this specific setting, exposure variable is associated with the collider (hospitalization) through age and comorbidities which also act as a confounder. Hence, adjusting for baseline confounding may also attenuate the collider bias.

Our study illustrates another limitation of observational studies, namely background noise due to cross-over between treatment arms. Some patients with treatment prior to hospitalization were no longer exposed during hospital stay and therefore could not benefit or be harmed by treatment, and conversely for cases of treatment initiation. Another source of background noise is nonadherence. Even in the setting of randomized trials with regular in-person visits, approximately 20% of the patients are poorly or nonadherent [49]. In real-life conditions, about half of the patients are not fully adherent to the prescribed antihypertensive treatment [50]. Therefore, the information regarding previous exposure to RAAS blockers derived from medical records is expectedly highly inaccurate, by overestimating patients actually exposed to the treatment (and conversely, by incomplete self-declared chronic treatment). Cross-over patterns and poor adherence attenuate the ability to show a real treatment-related effect (bias towards the null hypothesis) in analyses based on treatment prior to hospital admission, which may in part have contributed to the lack of a significant association between RAAS blockers and outcome in the vast majority of studies based on chronic exposure.

Interestingly, in our study, although adjusted analyses revealed no significant association between chronic exposure to RAAS blockers and mortality, the risk to develop severe illness (high-flow oxygen, ICU admission, or death) was lower [adjusted OR 0.39 (95% CI 0.20–0.74)] in patients with prior exposure to RAAS blockers. This result was unexpected because a potential protective effect associated with chronic treatment has rarely been reported so far. Bean et al.[38] analysed data from 1200 patients admitted at two hospitals in London, of whom 388 (33%) were taking RAAS blockers prior to hospitalization. After adjustment for age, sex, and comorbidities, the OR for the primary outcome (death or transfer to a critical care unit within 21 days of symptom onset) was 0.63 (95% CI 0.47–0.84) [38]. However, these authors report that they obtained similar results in a sensitivity analysis using only in-hospital medication, which is unexpected and raises questions regarding actual exposure measurement. Very interestingly, the only other study which showed a protective effect of treatment [adjusted OR for mortality 0.46 (95% CI 0.26–0.84) in patients exposed to ACEI, no significant effect for ARBs] was performed in a very particular setting, as it included 324 elderly patients admitted to ICU [41]. One may interpret these results as being in favour of a protective effect of RAAS blockers. We suggest an alternative explanation, namely, a bias due to conditioning on a collider – hospital admission (Fig. 2) (or ICU admission in the case of the study by Jung et al.[41]). This selection bias, inherent to studies restricting the population of interest to patients hospitalized for COVID-19, may induce a spurious protective association between exposure and outcome, as more fragile patients such as those treated with RAAS blockers may be admitted for more benign forms of the disease (Fig. 2) [51]. One could indeed anticipate that restricting the analysis to patients admitted in ICU may amplify this phenomenon.

This hypothesis of a selection bias is supported by the findings of Mehta et al.[17] (conducted in two centres of the Cleveland Clinic Health System in Ohio and Florida), who found a higher rate of hospital admission in the 214 of 1735 patients diagnosed with COVID-19 who had chronic exposure to RAAS blockade, although rates of mechanical ventilation did not differ. The authors hypothesized that there may have been a lower threshold of severity to admit the patients exposed to RAAS blockers due to underlying comorbid conditions. Our hypothesis is also supported by comparison of our results with the retrospective cohort from Fosbøl et al.[16] conducted in 4480 patients with COVID-19, only half of whom were hospitalized. Although in the ACEI/ARB users, 18% died, a rate very similar to that observed in our study (17%), in nonusers, only 7% died (compared with 12% in our study). Likewise, the outcome of mortality or severe disease occurred in 32% of users and 14% of nonusers in the study by Fosbøl et al.[16] compared with 33 and 37% in our study. This imbalance of event rates between our study and theirs shows that nonusers had to have relatively more severe forms of the disease, compared with RAAS blocker users, to be admitted in hospital, generating the selection bias.

Importantly, in most previous studies, severity was defined by the combined endpoint of mortality or transfer to ICU, leaving aside patients with severe pneumonia, requiring high-flow oxygen, but refused ICU admission. As this situation may occur more frequently in older patients with comorbidities, we included the need for high-flow oxygen in our definition of severe illness to avoid generating a bias by modifying the number of severe outcomes in an unbalanced manner in both groups.

Overall, the adjusted OR observed for severe disease even for chronic exposure needs to be interpreted with great caution until randomized trials are able to analyse the potential protective effect associated with the blockade of this pharmacological pathway in the context of SARS-CoV-2 infection.

In any case, our results do not support previous concerns on the role of RAAS blockers during the COVID-19 pandemics and are thereby in line with previous studies [19–24,28,38], including several conducted in both inpatients and outpatients [14–16,18,37], which overall did not show an increased risk of mortality or severe illness in patients chronically exposed to RAAS blockers. In the study by Reynolds et al.[15] based on New York University Langone Health electronic health record, after careful adjustment, there was no significant association between chronic exposure to RAAS blockers and severe illness (ICU admission, mechanical ventilation, or death) in 5894 patients diagnosed with COVID-19. Mancia et al.[14] in another large study conducted in Lombardy, showed no increase in the adjusted risk of developing a critical or fatal form of the disease in those whose prescription included ACEI/ARBs in 2019. Likewise, Fosbøl et al.[16] found no independent association between prior use of ACEI/ARBs and mortality or severe disease in 4480 patients in Denmark. de Abajo et al.[21] analysed data from 1139 patients requiring hospital admission in Madrid, matched with 11 390 controls, and concluded that RAAS blockers did not increase the risk of COVID-19 requiring admission to hospital, including fatal cases or admission to ICU. Iaccarino et al.[24] collected data from 1591 patients admitted in 26 hospitals in Italy and found that RAAS blockers were not independently associated with mortality. Smaller scaled studies yielded similar conclusions [18,20,22,23,28].

However, observational studies, even large-sized, carefully adjusted and designed to limit all types of biases, are merely hypotheses-generating. Final proof on the role of RAAS blockers in the course of COVID-19 will be provided by randomized trials. Two different types of clinical trials are ongoing. Some are randomizing COVID-19 patients previously treated with RAAS blockers and admitted to hospital for treatment continuation or discontinuation (ACORES-2 in France, NCT04329195, CORONACION in Ireland, NCT04330300, REPLACECOVID in the United States, NCT04338009), whereas others (more than 10 trials in the United States and in Europe) are randomizing patients to receive either an ARB or a placebo.

Our study has several limitations. First, it is a monocentric and fairly small-scaled study, but data regarding treatment exposure were collected in much more detail than in previous studies, with critical analysis of reasons underlying treatment modifications, allowing to demonstrate for the first time that treatment discontinuation in patients hospitalized for COVID-19 is a frequent event, occurring in patients with the worst outcomes, and reflects reverse causality. Second, we were not able to quantify adherence to chronic treatment due to the retrospective, observational design of the study. Furthermore, since we did not have access to a larger dataset of unselected patients tested for COVID-19, including outpatients and inpatients, we cannot quantify the magnitude of the selection bias due to the setting of hospitalized patients. Finally, our results need to be confirmed in other settings and countries. The discontinuation bias would likely be much higher in studies conducted in ICUs, since RAAS blockers are typically interrupted upon ICU admission. It would be of particular interest for previous studies based on in-hospital exposure to collect treatment prior to hospitalization and analyse whether this potentially modifies their findings, and for studies with no or less selection bias, to analyse how restricting analyses to hospitalized patients would have influenced their results (a way to quantify the selection bias).

In conclusion, our study did not find an association between chronic use of RAAS blockers and mortality in patients with COVID-19, while the inverse association with disease severity might reflect a selection bias. Upon admission, discontinuation of treatment is frequent and occurs in those with the worst outcomes, and conversely for treatment continuation, generating a spurious protective association between in-hospital exposure to RAAS blockers and outcome of COVID-19. In light of our results, conclusions of previous observational studies showing a protective effect of ‘in-hospital’ exposure to RAAS blockers need to be reconsidered.

ACKNOWLEDGEMENTS

Conflicts of interest

E.V.-P. received fees and travel funding from Servier, outside the submitted work. F.-X.L. perceived consulting fees from Gilead and MSD, and travel funding from Astellas, Eumedica, and MSD, all outside the submitted work. Other authors have nothing to disclose.

REFERENCES

1. Wu Z, McGoogan JM. Characteristics of and important lessons from the coronavirus disease 2019 (COVID-19) outbreak in China: summary of a report of 72 314 cases from the Chinese Center for Disease Control and Prevention. JAMA 2020; 323:1239–1242.
2. Grasselli G, Zangrillo A, Zanella A, Antonelli M, Cabrini L, Castelli A, et al. Baseline characteristics and outcomes of 1591 patients infected with SARS-CoV-2 admitted to ICUs of the Lombardy Region, Italy. JAMA 2020; 323:1574–1581.
3. Clerkin KJ, Fried JA, Raikhelkar J, Sayer G, Griffin JM, Masoumi A, et al. COVID-19 and cardiovascular disease. Circulation 2020; 141:1648–1655.
4. Ferrario CM, Jessup J, Chappell MC, Averill DB, Brosnihan KB, Tallant EA, et al. Effect of angiotensin-converting enzyme inhibition and angiotensin II receptor blockers on cardiac angiotensin-converting enzyme 2. Circulation 2005; 111:2605–2610.
5. Kreutz R, Algharably EAE, Azizi M, Dobrowolski P, Guzik T, Januszewicz A, et al. Hypertension, the renin–angiotensin system, and the risk of lower respiratory tract infections and lung injury: implications for COVID-19. Cardiovasc Res 2020; 116:1688–1699.
6. Fang L, Karakiulakis G, Roth M. Are patients with hypertension and diabetes mellitus at increased risk for COVID-19 infection? Lancet Respir Med 2020; 8:e21.
7. Esler M, Esler D. Can angiotensin receptor-blocking drugs perhaps be harmful in the COVID-19 pandemic? J Hypertens 2020; 38:781–782.
8. Diaz JH. Hypothesis: angiotensin-converting enzyme inhibitors and angiotensin receptor blockers may increase the risk of severe COVID-19. J Travel Med 2020; 27:taaa041.
9. Aronson JK, Ferner RE. Drugs and the renin–angiotensin system in covid-19. BMJ 2020; 369:m1313.
10. Bavishi C, Maddox TM, Messerli FH. Coronavirus disease 2019 (COVID-19) infection and renin angiotensin system blockers. JAMA Cardiol 2020; 5:745–747.
11. Imai Y, Kuba K, Rao S, Huan Y, Guo F, Guan B, et al. Angiotensin-converting enzyme 2 protects from severe acute lung failure. Nature 2005; 436:112–116.
12. Kuba K, Imai Y, Rao S, Gao H, Guo F, Guan B, et al. A crucial role of angiotensin converting enzyme 2 (ACE2) in SARS coronavirus-induced lung injury. Nat Med 2005; 11:875–879.
13. Vaduganathan M, Vardeny O, Michel T, McMurray JJV, Pfeffer MA, Solomon SD. Renin–angiotensin–aldosterone system inhibitors in patients with Covid-19. N Engl J Med 2020; 382:1653–1659.
14. Mancia G, Rea F, Ludergnani M, Apolone G, Corrao G. Renin–angiotensin–aldosterone system blockers and the risk of Covid-19. N Engl J Med 2020; 382:2431–2440.
15. Reynolds HR, Adhikari S, Pulgarin C, Troxel AB, Iturrate E, Johnson SB, et al. Renin–angiotensin–aldosterone system inhibitors and risk of Covid-19. N Engl J Med 2020; 382:2441–2448.
16. Fosbøl EL, Butt JH, Østergaard L, Andersson C, Selmer C, Kragholm K, et al. Association of angiotensin-converting enzyme inhibitor or angiotensin receptor blocker use with COVID-19 diagnosis and mortality. JAMA 2020; 324:168–177.
17. Mehta N, Kalra A, Nowacki AS, Anjewierden S, Han Z, Bhat P, et al. Association of use of angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers with testing positive for coronavirus disease 2019 (COVID-19). JAMA Cardiol 2020; doi:10.1001/jamacardio.2020.1855.
18. Son M, Seo J, Yang S. Association between renin–angiotensin–aldosterone system inhibitors and COVID-19 infection in South Korea. Hypertension 2020; 76:742–749.
19. Jung SY, Choi JC, You SH, Kim WY. Association of renin–angiotensin–aldosterone system inhibitors with COVID-19-related outcomes in Korea: a nationwide population-based cohort study. Clin Infect Dis 2020; doi:10.1093/cid/ciaa624.
20. Andrea C, Francesco M, Antonio N, Evgeny F, Marzia S, Fabio C, et al. Renin–angiotensin–aldosterone system inhibitors and outcome in patients with SARS-CoV-2 pneumonia: a case series study. Hypertension 2020; 76:e10–e12.
21. de Abajo FJ, Rodríguez-Martín S, Lerma V, Mejía-Abril G, Aguilar M, García-Luque A, et al. Use of renin–angiotensin–aldosterone system inhibitors and risk of COVID-19 requiring admission to hospital: a case-population study. Lancet 2020; 395:1705–1714.
22. Gao C, Cai Y, Zhang K, Zhou L, Zhang Y, Zhang X, et al. Association of hypertension and antihypertensive treatment with COVID-19 mortality: a retrospective observational study. Eur Heart J 2020; 41:2058–2066.
23. Holt A, Mizrak I, Lamberts M, Lav Madsen P. Influence of inhibitors of the renin–angiotensin system on risk of acute respiratory distress syndrome in Danish hospitalized COVID-19 patients. J Hypertens 2020; 38:1612–1613.
24. Iaccarino G, Grassi G, Borghi C, Ferri C, Salvetti M, Volpe M, et al. Age and multimorbidity predict death among COVID-19 patients: results of the SARS-RAS study of the Italian Society of Hypertension. Hypertension 2020; 76:366–372.
25. Liabeuf S, Moragny J, Bennis Y, Batteux B, Brochot E, Schmit JL, et al. Association between renin–angiotensin system inhibitors and COVID-19 complications. Eur Heart J Cardiovasc Pharmacother 2020; doi:10.1093/ehjcvp/pvaa062.
26. Pan W, Zhang J, Wang M, Ye J, Xu Y, Shen B, et al. Clinical features of COVID-19 in patients with essential hypertension and the impacts of renin–angiotensin–aldosterone system inhibitors on the prognosis of COVID-19 patients. Hypertension 2020; 76:732–741.
27. Richardson S, Hirsch JS, Narasimhan M, Crawford JM, McGinn T, Davidson KW, et al. Presenting characteristics, comorbidities, and outcomes among 5700 patients hospitalized with COVID-19 in the New York City Area. JAMA 2020; 323:2052–2059.
28. Tedeschi S, Giannella M, Bartoletti M, Trapani F, Tadolini M, Borghi C, et al. Clinical impact of renin–angiotensin system inhibitors on in-hospital mortality of patients with hypertension hospitalized for coronavirus disease 2019. Clin Infect Dis 2020; 71:899–901.
29. Cannata F, Chiarito M, Reimers B, Azzolini E, Ferrante G, My I, et al. Continuation versus discontinuation of ACE inhibitors or angiotensin II receptor blockers in COVID-19: effects on blood pressure control and mortality. Eur Heart J Cardiovasc Pharmacother 2020; doi:10.1093/ehjcvp/pvaa056.
30. Chaudhri I, Koraishy FM, Bolotova O, Yoo J, Marcos LA, Taub E, et al. Outcomes associated with the use of RAAS Blockade in hospitalized patients with SARS-CoV-2 infection. Kidney360 2020; 1:801–809.
31. Feng Y, Ling Y, Bai T, Xie Y, Huang J, Li J, et al. COVID-19 with different severities: a multicenter study of clinical features. Am J Respir Crit Care Med 2020; 201:1380–1388.
32. Li J, Wang X, Chen J, Zhang H, Deng A. Association of renin–angiotensin system inhibitors with severity or risk of death in patients with hypertension hospitalized for coronavirus disease 2019 (COVID-19) infection in Wuhan, China. JAMA Cardiol 2020; 5:825–830.
33. Meng J, Xiao G, Zhang J, He X, Ou M, Bi J, et al. Renin–angiotensin system inhibitors improve the clinical outcomes of COVID-19 patients with hypertension. Emerg Microbes Infect 2020; 9:757–760.
34. Yang G, Tan Z, Zhou L, Yang M, Peng L, Liu J, et al. Effects of angiotensin II receptor blockers and ACE (angiotensin-converting enzyme) inhibitors on virus infection, inflammatory status, and clinical outcomes in patients with COVID-19 and hypertension: a single-center retrospective study. Hypertension 2020; 76:51–58.
35. Zhang P, Zhu L, Cai J, Lei F, Qin JJ, Xie J, et al. Association of inpatient use of angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers with mortality among patients with hypertension hospitalized with COVID-19. Circ Res 2020; 126:1671–1681.
36. Zhou F, Liu YM, Xie J, Li H, Lei F, Yang H, et al. Comparative impacts of angiotensin converting enzyme inhibitors versus angiotensin II receptor blockers on the risk of COVID-19 mortality. Hypertension 2020; 79:e15–e17.
37. Bravi F, Flacco ME, Carradori T, Volta CA, Cosenza G, De Togni A, et al. Predictors of severe or lethal COVID-19, including angiotensin converting enzyme inhibitors and angiotensin II receptor blockers, in a sample of infected Italian citizens. PLoS One 2020; 15:e0235248.
38. Bean DM, Kraljevic Z, Searle T, Bendayan R, Kevin OG, Pickles A, et al. ACE-inhibitors and angiotensin-2 receptor blockers are not associated with severe SARS-COVID19 infection in a multisite UK acute Hospital Trust. Eur J Heart Fail 2020; 22:967–974.
39. Guo T, Fan Y, Chen M, Wu X, Zhang L, He T, et al. Cardiovascular implications of fatal outcomes of patients with coronavirus disease 2019 (COVID-19). JAMA Cardiol 2020; 5:811–818.
40. Huang Z, Cao J, Yao Y, Jin X, Luo Z, Xue Y, et al. The effect of RAS blockers on the clinical characteristics of COVID-19 patients with hypertension. Ann Transl Med 2020; 8:430.
41. Jung C, Bruno RR, Wernly B, Joannidis M, Oeyen S, Zafeiridis T, et al. Inhibitors of the renin–angiotensin–aldosterone system and Covid-19 in critically ill elderly patients. Eur Heart J Cardiovasc Pharmacother 2020; doi:10.1093/ehjcvp/pvaa083.
42. Suissa S. Immortal time bias in pharmaco-epidemiology. Am J Epidemiol 2008; 167:492–499.
43. Wolkewitz M, Allignol A, Harbarth S, de Angelis G, Schumacher M, Beyersmann J. Time-dependent study entries and exposures in cohort studies can easily be sources of different and avoidable types of bias. J Clin Epidemiol 2012; 65:1171–1180.
44. Cheng Y, Luo R, Wang K, Zhang M, Wang Z, Dong L, et al. Kidney disease is associated with in-hospital death of patients with COVID-19. Kidney Int 2020; 97:829–838.
45. Krumme AA, Glynn RJ, Schneeweiss S, Choudhry NK, Tong AY, Gagne JJ. Defining exposure in observational studies comparing outcomes of treatment discontinuation. Circ Cardiovasc Qual Outcomes 2018; 11:e004684.
46. Cohen JB, Hanff TC, South AM, Sparks MA, Hiremath S, Bress AP, et al. Response by Cohen et al. to letter regarding article, ‘Association of inpatient use of angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers with mortality among patients with hypertension hospitalized with COVID-19’. Circ Res 2020; 126:e140–e141.
47. Zhang X, Yu J, Pan L, Jiang H. ACEI/ARB use and risk of infection or severity or mortality of COVID-19: a systematic review and meta-analysis. Pharmacol Res 2020; 158:104927.
48. Guo X, Zhu Y, Hong Y. Decreased mortality of COVID-19 with renin–angiotensin–aldosterone system inhibitors therapy in patients with hypertension: a meta-analysis. Hypertension 2020; 76:e13–e14.
49. Blaschke TF, Osterberg L, Vrijens B, Urquhart J. Adherence to medications: insights arising from studies on the unreliable link between prescribed and actual drug dosing histories. Annu Rev Pharmacol Toxicol 2012; 52:275–301.
50. Burnier M, Egan BM. Adherence in hypertension. Circ Res 2019; 124:1124–1140.
51. Cole SR, Platt RW, Schisterman EF, Chu H, Westreich D, Richardson D, et al. Illustrating bias due to conditioning on a collider. Int J Epidemiol 2010; 39:417–420.
Keywords:

angiotensin-converting enzyme inhibitors; angiotensin receptor blockers; coronavirus disease 2019; renin–angiotensin–aldosterone system blockers; severe acute respiratory syndrome coronavirus 2

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.