key: cord-0852288-n2vh391p authors: Kurdi, A.; Weir, N.; Mueller, T. title: An umbrella review and meta-analysis of the use of renin-angiotensin system drugs and COVID-19 outcomes: what do we know so far? date: 2022-03-21 journal: nan DOI: 10.1101/2022.03.20.22272664 sha: b89a9b41b8e5aadb30fa8f178f780ee48abfe820 doc_id: 852288 cord_uid: n2vh391p Backgrounds Evidence from several meta-analyses are still controversial about the effects of angiotensin-converting enzyme inhibitors (ACEIs)/angiotensin-receptor blockers (ARBs) on COVID-19 outcomes. Purpose Umbrella review of systematic reviews/meta-analysis to provide comprehensive assessment of the effect of ACEIs/ARBs on COVID-19 related outcomes by summarising the currently available evidence. Data Source Medline (OVID), Embase, Scopus, Cochrane library and medRxiv from inception to 1st February 2021. Study Selection Systematic reviews with meta-analysis that evaluated the effect of ACEIs/ARBs on COVID-19 related clinical outcomes Data Extraction Two reviewers independently extracted the data and assessed studies risk of bias using AMSTAR 2 Critical Appraisal Tool. Data Synthesis Pooled estimates were combined using the random-effects meta-analyses model including several sub-group analyses. Overall, 47 reviews were eligible for inclusion. Out of the nine COVID-19 outcomes evaluated, there was significant associations between ACEIs/ARBs use and each of death (OR=0.80, 95%CI=0.75-0.86; I2=51.9%), death/ICU admission as composite outcome (OR=0.86, 95%CI=0.80-0.92; I2=43.9%), severe COVID-19 (OR=0.86, 95%CI=0.78-0.95; I2=68%), and hospitalisation (OR=1.23, 95%CI=1.04-1.46; I2= 76.4%). The significant reduction in death/ICU admission, however, was higher among studies which presented adjusted measure of effects (OR=0.63, 95%CI=0.47-0.84) and were of moderate quality (OR=0.74, 95%CI=0.63-0.85). Limitations The effect of unmeasured confounding could not be ruled out. Only 21.3% (n=10) of the studies were of moderate quality. Conclusion: Collective evidence from observational studies indicate a good quality evidence on the significant association between ACEIs/ARBs use and reduction in death and death/ICU admission, but poor-quality evidence on both reducing severe COVID-19 and increasing hospitalisation. Our findings further support the current recommendations of not discontinuing ACEIs/ARBs therapy in patients with COVID-19. Registration The study protocol was registered in PROSPERO (CRD42021233398). Funding Source None Introduction treatment with RAAS inhibitors (i.e. ACEIs and/or ARBs) compared to those not exposed to RAAS 117 inhibitors. Reviews conducting a comparison between patients exposed to ACEIs 118 and patients exposed to ARBs were also eligible for inclusion. Outcomes of interest were COVID-119 19 infection risk and COVID-19 related clinical outcome, including but not limited to: death; severity of 120 COVID-19 infection; admission to intensive care unit (ICU); hospitalisation; hospital discharge; 121 ventilator use; length of hospital stay; hospital re-admission; dialysis; acute respiratory distress 122 syndrome; septic shock; acute kidney injury; cardiac injury; pneumonia severity; as well as other 123 relevant outcomes identified iteratively throughout study selection and data extraction. 124 125 Search strategy 126 The databases Medline, EMBASE, Scopus, Cochrane, and medRxiv were searched in February 127 2021. Publications were searched from 2019 onwards to reflect the date with which COVID-related 128 reviews could have been published. The search was limited to the English language and for 129 systematic review articles. Search terms for "renin-angiotensin system", "angiotensin-converting 130 enzyme inhibitors", "angiotensin II receptor antagonists", "COVID-19" were used with various 131 synonyms, truncation codes and Boolean operators (Supplementary file 1). When full texts were not 132 obtainable the author(s) were contacted up to two times to request full texts. The reference lists of 133 included reviews were also screened to identify eligible reviews. 134 135 Article selection 136 Article selection was conducted using Covidence software (9). To ensure consistency in the study 137 selection process 10% of the articles' titles/abstracts and full texts were randomly selected and 138 screened independently by two researchers (NW and TM). The percentage of agreement was 139 calculated for all independent validation, with >80% considered adequate (10) . Where dubiety arose 140 over an article's eligibility a third reviewer was consulted (AK). 141 142 Data extraction 143 Data were extracted from the reviews using Microsoft Excel. A data extraction template was piloted 144 with 10% of reviews by NW and agreed for use by all authors. 10% of reviews were randomly 145 selected and underwent independent data extraction by NW and TM; the percentage of agreement 146 was calculated. Again, agreement >80% was considered adequate (10) . Where dubiety arose over 147 data extraction a second reviewer was consulted (AK). Data extracted from the reviews included: title; 148 authors; year review published; study design; sample size; setting; population; exposure 149 (e.g. ACEIs/ARBs, ACEIs, or ARBs); and outcomes (e.g. death, COVID-19 infection, 150 hospitalisation). Data was extracted from the published reviews only; the primary studies were not 151 referred to and authors were not contacted for further data. 152 153 Quality Assessment 154 Quality assessment was conducted independently by NW and TM using the AMSTAR 2 tool 155 (11).Studies were categorised as having high, moderate, low and critically low confidence in the 156 results based on the number of 'critical domains'. Critical domains related to each review containing: 5 The random-effects meta-analysis model was used to statistically combine the measure of effects for 163 those outcomes that were reported by more than one study to obtain one pooled estimate for each 164 outcome, stratified by the three level of exposure (ACEIs/ARBs, ACEIs, ARBs). We used random-165 effects model because it allows the results to be generalisable to other populations as well as 166 addresses the likely heterogeneity between the included studies; hence it is the most commonly used 167 meta-analysis model (12) . In order to explore the potential source of heterogeneity as well as the 168 effect of potential confounders on the sensitivity and robustness of the combined pooled estimates, 169 we conducted several sub-group analyses based on numerous variables including: whether the 170 reported measure of effects was crude or adjusted, the study was peer-reviewed or not, and the 171 study's methodological quality as per the quality assessment. Furthermore, to assess the impact of 172 ACEIs/ARBs among patients with hypertension (the most common indication for ACEIs/ARBs), we 173 also conducted sub-group analysis based on whether the studies had included either patients with 174 hypertension only or at least had hypertension as one of the comorbidities versus those studies which 175 did not recorded the hypertension status of their study population. The combined pooled estimates 176 were presented as odds ratios and 95%CI and graphically as forest plots. I 2 statistic (13) was used to 177 assess heterogeneity between the studies, to check whether the variability is more likely to be due to 178 chance or heterogeneity in the studies; I 2 values ranged between 0%-100% with 0% indicating lack of 179 heterogeneity, whereas 25%, 50%, and 75% indicating low, moderate and high heterogeneity, 180 respectively (13). Publication bias was assessed using funnel plots and Egger's asymmetry test (14) 181 for those outcomes where >10 studies were included in the analysis as recommended by Cochrane 182 guidelines (15). Furthermore, we evaluated the influence of individual reviews on the summary pooled 183 estimate for each outcome by conducting influential analyses (16) whereby the pooled meta-analysis 184 estimates for each outcome were computed by omitting one study at a time. Data were analysed 185 using STATA 12. 186 187 Results 188 Out of an initial 157 publications, 66 systematic reviews underwent full text screening; after further 189 exclusions based on pre-specified criteria, 47 studies were identified to be relevant for this project 190 (Figure 1 ) (4) (5) (6) Effect of ACEIs/AEBs (as a one group) on the study outcomes 248 Overall, the effect of ACEIs/ARBs on nine COVID-19 related clinical outcomes were evaluated (Table 249 1). The combined pooled meta-analysis estimates indicated that ACEIs/ARBs used was associated 250 with a significant reduction in three clinical outcomes including death (OR=0.80, 95%CI=0.75-0.86; I 2 251 = 51.9%) (Figure 2 ) death/ICU admission as composite outcome (OR=0.86, I2= 252 43.9%) (Figure 3 ) and severe COVID-19 infection (OR=0.86, 95% CI=0.78-0.95; I2 = 68%) (Figure 4) ; 253 on the other hand, ACEIs/ARBs was associated with a significant increase in hospitalisation 254 (OR=1.23, 95%CI=1.04-1.46; I2= 76.4%) ( Figure 5) . However, there was insignificant association with 255 each of ICU admission (Figure 6) , risk of acquiring COVID-19 infection (Figure 7) , use of mechanical 256 ventilator (Figure 8) , risk of SARS (Figure 9) , and risk of severe pneumonia (Figure 10) . 257 258 However, the sub-group analyses indicated different results for some of the outcomes ( Table 2) . 259 Firstly, despite the consistent significant reduction in death in association with ACEIs/ARBs use 260 regardless of studies' crude/adjusted measure of effects, peer-review status and hypertension use 261 status, there was a trend toward lower protective effective of ACEIs/ARBs on death as the quality of 262 the studies enhanced from critically low (OR=0. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) 162) (Note) *Indicates that the pooled estimate is the same as the overall analyses because all the studies were in one group; NA: not applicable indicating that no studies were available to perform meta-analyses for these outcomes; Discussion 322 This umbrella review for the first time combined all the available evidence so far from observational 323 studies on the impact of ACEIs/ARBs on COVID-19 clinical outcomes (47 systematic review studies 324 which reported 213 meta-analyses) into one pooled estimate using an umbrella review and meta-325 analysis approach. The collective, combined pooled estimates indicated evidence of statistically 326 significant reduction in mortality, death/ICU admission (as a composite endpoint) and severe 19 infection in association with ACEIs/ARBs use, but significant increase in the risk of hospitalisation 328 (Table 1) . Interestingly, when analysing ACEIs and ARBs as a two separate groups, there was no 329 evidence of any significant association between ACEIs, or ARBs and any of the nine COVID-19 330 related clinical outcomes analysed in our study. 331 332 Although the magnitude of observed impact of ACEIs/ARBs use on reducing mortality was decreasing 333 as the quality of studies improved (ranged from 25% reduction death-OR=0. effects. In contrast, the quality of the evidence for the impact of ACEIs/ARBs use on severe COVID-347 19 was low since a significant reduction was only observed among critically-low quality studies (31% 348 reduction-OR=0.69; 95%CI: 0.53, 0.92) and in fact, the significant association disappeared as the 349 quality of the studied enhanced from critically low quality to either low or moderate quality. 350 351 In terms of the impact of ACEIs/ARBs on hospitalisation, the quality of the evidence was low because 352 the significant association was not apparent when the data were analysed by the quality of the 353 studies, even though the magnitude of the effect was almost consistent across the various quality of 354 the studies; besides, the significant increase in hospitalisation was observed only among: studies that 355 reported adjusted measure of effects (33% increase-OR=0. between ARBs and ACEIs in their impact on COVID-19 clinical outcomes has been suggested to be 366 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 21, 2022. ; https://doi.org/10.1101/2022.03.20.22272664 doi: medRxiv preprint due to the increased level of angiotensin-II, which occurs following ARBs treatment but not ACEIs, 367 which in turn imposes an increased substrate load on ACE2 enzyme (the key cell entry point for 368 requiring its upregulation (62); hence facilitates COVID-19 virus cell entry and its 369 subsequent infectivity/pathogenicity (63). Furthermore, the increase in ACE2 activity demonstrated in 370 patients with hypertension, either due to the pathophysiology of hypertension itself (64) or 371 administration ACEIs/ARBs as antihypertensive medications (65), could at least partially explain some 372 of our study findings as why ACEIs/ARBs had significant impact on certain COVID-19 clinical 373 outcomes only among studies that included patient with hypertension. 374 375 Several hypotheses have been suggested to explain the potential negative and positive effects of 376 ACEIs/ARBs use on COVID-19 clinical outcomes. The negative effects are hypothesised to be due to 377 ACEIs/ARBs induced upregulation of ACE2 expression; hence enhancing viral binding and cell entry 378 (65); whereas the positive protective effects could be through ACEIs/ARBs effects on angiotensin II 379 expression leading to subsequent increase in the protective angiotensin 1-7 and 1-9 which have anti-380 inflammatory and vasodilatory effects; hence potentially attenuating the cardiac and pulmonary 381 damages (2). Genetic ACE2 polymorphism among some individuals has been also suggested as 382 potential factor explaining, at least partially, the harmful effects on ACEIs/ARBs on COVID-19 383 outcomes (66). 384 385 Our study findings are in contrast to the findings from a recent randomised clinical trial (RCT) (67) 386 which found insignificant differences in the mean number of days alive and out of the hospital 387 between those assigned to discontinue vs continue ACEIs or ARBs. However, there are certain points 388 that should be considered when interpreting the findings from this clinical trial in comparison to our 389 study findings. First, this RCT was designed to evaluate the impact of continuing ACEIs or ARBs vs. 390 their discontinuation after contracting COVID-19 rather than evaluating ACEIs/ARBs use vs. non-use 391 of these medication which was the focus of most of the observational studies involved in our current 392 study. Secondly, the RCT included only patients with mild or moderate COVID-19 with more than half 393 of the participants (57%; n=376) having mild COVID-19, and evaluated only two COVID-19 related 394 clinical outcomes, namely days alive (mortality) and out of hospital days; hence leaving a big gap in 395 the evidence around ACEIs/ARBs' impact on other important COVID-19 clinical outcomes such is ICU 396 admission, hospitalisation, acquiring COVID-19 infection and severe COVID-19 as well as limiting the 397 findings' external validity (generalisability) to patients with severe COVID-19. Furthermore, although 398 the RCT's participants were all hypertensive patients, about one-third (~31%) and ~1% had diabetes 399 and heart failure, respectively, which further limits the generalisability of the RCT's findings to these 400 conditions for which ACEIs/ARBs are commonly indicated. Moreover, the RCT's participants were all 401 from Brazil and hence extending the findings to other races or ethnicities will be limited; this is 402 particularly importantly because there are evidence demonstrating that there are potential genetic 403 variants of renin, angiotensinogen, ACE, angiotensin II and ACE2 among various populations that 404 influence the function of the renin-angiotensin aldosterone system; hence affecting someone' 405 response to the COVID-19 infection (68). Finally, it is not entirely clear how long it takes for the ACE2 406 upregulation (induced by ACEIs/ARBs treatment) to return to its normal level after discontinuing 407 ACEIs/ARBs therapy, suggesting that measuring any clinical outcome within 30 days might not be 408 long enough for the ACE2 level to return back to its pre-ACEIs/ARBs treatment level (i.e., ACE2 level 409 would be comparable between those continued or discontinued ACEIs/ARBs treatment) which could 410 potentially explain the insignificant difference in the study outcomes between the two groups in the 411 RCT; however, this requires further investigation. 412 413 It is rather surprising and unusual to have such high number of published systematic reviews and 414 meta-analysis (47 studies) on the same topic. Circumstances associated with the pandemic may have 415 influenced researchers' decisions and overall study quality. For example, researchers may have 416 decided not to submit a published protocol to quicken the review process for rapid dissemination of 417 results to clinicians and COVID-19 policy makers (41). 418 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 21, 2022. ; https://doi.org/10.1101/2022.03.20.22272664 doi: medRxiv preprint 419 Strengths and limitations 420 This review presents the most comprehensive and systematic overview on the impact using RAAS 421 inhibitors on COVID-19 related clinical outcomes, with a wide range of sensitivity (sub-group) 422 analyses to assess the strength, validity and robustness of the evidence while accounting for potential 423 confounding variables. Furthermore, none of the pooled meta-analysis estimates for the nine studied 424 outcomes was affected/dominated by a single individual study. Although most of the included studies 425 were classified as 'low' or 'critically low' quality when assessed using AMSTAR 2 tool, it is widely 426 acknowledged that the AMSTAR 2 tool has a high standard with most reviews rated as 'critically low' 427 (69, 70) . The AMSTAR 2 tool is also prone to subjective biases (71) , and assessment results are at 428 the discretion of the reviewers regarding what is a "comprehensive" literature search or "satisfactory" 429 explanation of heterogeneity or risk of bias assessment (71); therefore, quality assessment was 430 conducted fully independent in this review and further criteria were set by the assessors to ensure 431 inter-rater consistency. Alternatives tools to AMSTAR 2 exist such as the ROBIS tool, however the 432 measurement categories are found to be broadly similar with the AMSTAR 2 tool considered more 433 reliable (71). Additionally, we accounted for this issue by conducting a sub-group analysis based on 434 the level of studies' quality. 435 436 Conclusion 437 Collective evidence so far from observational studies indicate a good quality evidence on the 438 significant association between ACEIs/ARBs use and reduction in death and death/ICU admission (as 439 a composite outcome). Additionally, ACEIs/ARBs use was found to be associated with a significant 440 reduction in severe COVID-19 but a significant increase in hospitalisation; however, the evidence for 441 these two outcomes was of poor quality; hence, cautious interpretation of these findings is required. 442 Interestingly, findings for some of the clinical outcomes were dependent on whether the included 443 patients had hypertension or not. Overall, our study findings further support the current 444 recommendations of not discontinuing ACEIs/ARBs therapy in patients with COVID-19 due to the lack 445 of good quality evidence on their harm but rather it could be beneficial to patients. 446 447 448 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 21, 2022. ; https://doi.org/10.1101/2022.03.20.22272664 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 21, 2022. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 21, 2022. ; https://doi.org/10.1101/2022.03.20.22272664 doi: medRxiv preprint C o V -2 I n f e c t i o n o r H o s p i t a l i z a t i o n W i t h C O V I D -1 9 D i s e a s e : A S y s t e m a t i c R e v i e w a n d M e t a -A n a l y s i s . 599 A m J T h e r . 2 0 2 2 ; 2 9 ( 1 ) : e 7 4 -e 8 4 . 600 5 4 . U s m a n M S , S i d d i q i T J , K h a n M S , A h m e d A , A l i S S , M i c h o s E D , e t a l . A m e t a -a n a l y s i s o f t h e 601 r e l a t i o n s h i p b e t w e e n r e n i n -a n g i o t e n s i n -a l d o s t e r o n e s y s t e m i n h i b i t o r s a n d C O V I D -1 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 21, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 ACEIs/ARBs use 683 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 21, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 Supplementary file 11. Results of the influential analyses 756 757 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 21, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 NOTE: Weights are from random effects analysis Odds ratio . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 21, 2022. Odds ratio Severe COVID-19 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 21, 2022. ; https://doi.org/10. 1101 /2022 Supplementary file 6A. Forest plot depicting sub-group analyses pooled estimates for the association 705 between mortality and ACEIs use sub-grouped by A) type of analyses (crude vs B) peer-706 review status; C) methodological quality; and D) hypertension stats Supplementary file 6B. Forest plot depicting sub-group analyses pooled estimates for the association 709 between mortality and ARBs use sub-grouped by A) type of analyses (crude vs B) peer-710 review status; C) methodological quality; and D) hypertension stats Forest plot depicting sub-group analyses pooled estimates for the association 714 between death/ICU admission (as a composite outcome) and ACEIs/ARBs use sub-grouped by A) 715 type of analyses (crude vs B) peer-review status; C) methodological quality; and D) 716 hypertension stats 717 718 Supplementary file 7A. Forest plot depicting sub-group analyses pooled estimates for the association 719 between death/ICU admission (as a composite outcome) and ACEIs use sub-grouped by A) type of 720 analyses B) peer-review status; C) methodological quality; and D) hypertension Forest plot depicting sub-group analyses pooled estimates for the association 724 between death/ICU admission (as a composite outcome) and ARBs use sub-grouped by A) type of 725 analyses (crude vs B) peer-review status; C) methodological quality; and D) hypertension Forest plot depicting sub-group analyses pooled estimates for the association 730 between severe COVID-19 and ACEIs/ARBs use sub-grouped by A) type of analyses B) peer-review status; C) methodological quality; and D) hypertension stats 732 733 Supplementary file 8A. Forest plot depicting sub-group analyses pooled estimates for the association 734 between severe COVID-19 and ACEIs use sub-grouped by A) type of analyses B) peer-review status; C) methodological quality; and D) hypertension stats Supplementary file 8B. Forest plot depicting sub-group analyses pooled estimates for the association 738 between severe COVID-19 and ARBs use sub-grouped by A) type of analyses (crude vs B) peer-review status; C) methodological quality; and D) hypertension stats Forest plot depicting sub-group analyses pooled estimates for the association 742 between hospitalisation and ACEIs/ARBs use sub-grouped by A) type of analyses B) peer-review status; C) methodological quality; and D) hypertension stats 744 745 Supplementary file 9A. Forest plot depicting sub-group analyses pooled estimates for the association 746 between hospitalisation and ACEIs use sub-grouped by A) type of analyses B) 747 peer-review status; C) methodological quality; and D) hypertension stats 748 749 Supplementary file 9B. Forest plot depicting sub-group analyses pooled estimates for the association 750 between hospitalisation and ARBs use sub-grouped by A) type of analyses B) 751 peer-review status; C) methodological quality; and D) hypertension stats Supplementary file 10. Publication bias funnel plot for the outcomes with >=10 studies