Prevalence of bias attributable to composite outcome in clinical trials: a systematic review
============================================================================================

* José Mário Nunes da Silva
* Juliana Ferreira Souza Conceição
* Paula C. Ramírez
* Christian Leonardo Diaz-León
* Fredi Alexander Diaz-Quijano

## Abstract

**Objective** To investigate the prevalence of bias attributable to composite outcome (BACO) in clinical trials.

**Study design and setting** We searched PubMed for randomized clinical trials where the primary outcome was a binary composite that included all-cause mortality among its components from January 1, 2019, to December 31, 2020. For each trial, the BACO index was calculated to assess the correspondence between effects on the composite outcome and that on mortality. This systematic review was registered in PROSPERO (CRD42021229554).

**Results** After screening 1,076 citations and 171 full-text articles, 91 studies were included from 13 different medical areas. The prevalence of significant or suggestive BACO among the 91 included articles was 25.2% (n=23), including 12 with p<0.005 and 11 with p between 0.005 and <0.05. We observed that in 17 (73.9%) of these 23 studies, the BACO index value was between zero and <1, indicating an underestimation of the effect. The other six studies showed negative values (26.1%), indicating an inversion of the association with mortality. None of the studies showed significant overestimation of the association attributable to the composite outcome.

**Conclusion** These findings highlight the need to predefine guidelines for interpreting effects on composite endpoints based on objective criteria such as the BACO index.

**Key Findings**

*   The study found that 25.2% of the included clinical trials exhibited significant or suggestive bias attributable to composite outcomes (BACO).

*   In 73.9% of these cases, the BACO index was less than 1, indicating an underestimation of the effect. 26.1% of the studies showed an inversion of the association with mortality.

*   No significant overestimation of the association due to composite outcomes was observed.

**What This Adds to What Was Known?**

*   This study contributes to the existing knowledge by quantifying the prevalence of bias attributable to composite outcomes in clinical trials.

*   It highlights that a significant proportion of trials may underestimate the effect or even show an inversion of the association with mortality when composite outcomes are used.

*   This finding emphasizes the need for careful consideration and objective criteria, like the BACO index, in the design and interpretation of clinical trials involving composite outcomes.

**What Is the Implication and What Should Change Now?**

*   Researchers and clinicians should be cautious about relying solely on composite outcomes without assessing the potential biases they introduce.

*   The study suggests a need for predefined guidelines and objective criteria, such as the BACO index, for interpreting the effects of composite outcomes.

Keywords
*   Composite outcome
*   Mortality
*   Bias
*   Clinical trials
*   Research methodology
*   BACO

## 1. Introduction

Composite outcomes are often used in randomized trials to assess the efficacy of a new intervention compared to standard treatment [1,2]. Their use involves analyzing a greater number of outcomes over shorter follow-up periods. Typically, this is expected to increase the power of the study, reduce costs, and provide a quicker response to a research question [3]. However, composite outcomes can lead to misleading conclusions when the individual components, which may vary in importance and frequency, are affected differently or oppositely by the interventions being evaluated [4,5].

In this context, the Bias Attributable to Composite Outcome (BACO) Index is a recently developed strategy that aids in interpreting the effects on a composite outcome [6]. This index corresponds to the ratio of the logarithms of the measures of association between the composite outcome and mortality. BACO index values different from one indicate that using a composite outcome affects the prognosis as follows: overestimated (BACO index > 1), underestimated (BACO index between 0 and < 1), or inverted (BACO index < 0), using the effect on mortality as the reference point [6].

Despite the frequent use of composite outcomes, especially in cardiovascular clinical trials, the frequency and direction of BACO have not been widely quantified. Therefore, we aim to investigate the prevalence of BACO in clinical trials published in PubMed between 2019 and 2020.

## 2. Methods

### 2.1 Protocol and registration

We conducted the review according to the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [7]. We registered the protocol in PROSPERO (registration number: CRD42021229554).

### 2.2 Eligibility Criteria

#### 2.2.1 Inclusion criteria

We included randomized clinical trials where the primary outcome was a binary composite outcome that included all-cause mortality among its components [6].

#### 2.2.2 Exclusion criteria

We excluded cluster randomized trials, secondary analyses, subgroup analyses, and studies with fewer than five fatal events. Additionally, we did not include four articles that lacked data on the frequency of composite outcomes or mortality.

### 2.3 Search strategy

We searched PubMed for articles published electronically in English, Portuguese, and Spanish between January 2019 and December 2020 (updated on April 5, 2021). We used the following terms: Composite AND primary AND (endpoint OR outcome OR (“end-point”)) AND (mortality OR death) AND (randomized OR randomised) AND (trial).

### 2.4 Selection of studies

Pairs of two independent reviewers (J.F.S.C and P.C.R; J.M.N.S and C.L.D.L) screened titles and abstracts of all citations retrieved during the literature search. Subsequently, reviewers read potentially eligible articles in full to determine if they met the eligibility criteria. Any discrepancies between reviewers that arose at each stage of the study selection process were resolved through consensus and, if needed, arbitration by a third reviewer (F.A.D.Q). Prior to both steps of study selection, we conducted a pilot-tested using random samples of 10 full articles.

### 2.5 Data extraction and management

We used a pre-defined standardized protocol where two independent reviewers (J.F.S.C and P.C.R) extracted data from included studies, compared information, and resolved disagreements through discussion. We extracted the following data: article title, year of publication, first author’s name, journal published, randomization and blinding process, follow-up period, experimental and control group interventions, sample size in each group, sample loss, components of the composite outcome, number of composite endpoints and deaths, measure of association used, and information on intention-to-treat analysis. We also recorded whether the study’s conclusion was based on the composite outcome and if authors addressed discrepancies between the composite outcome result and mortality, as well as protocol registration on a platform.

### 2.6 Data analysis

We calculated the BACO index (Bias Attributable to Composite Outcome) (6) defined as [6]: where <pC e <pd are the relative risks for the composite outcome (RRc) and mortality (RRd), respectively.

A BACO index equal to one indicates no bias attributable to using a composite outcome, with mortality as the reference measure of association [6].

We set a predefined significance level of 0.005. However, we used the term suggestive for p-values between 0.005 and 0.05. The 95% confidence intervals for the BACO index and hypothesis testing were conducted following the methods described in the original study [6]. We performed the analyses using Microsoft Office Excel 2019 (Microsoft Corporation, Redmond, Washington).

### 2.7 Risk of bias in individual studies

For studies suggestive of BACO or with significant BACO, we conducted a bias assessment by two independent reviewers (J.F.S.C and P.C.R) using the updated version of the Cochrane tool (RoB2) [8]. This tool assesses bias risk across the following domains: bias arising from the randomization process, bias due to deviations from intended interventions, bias from missing outcome data, outcome measurement bias, and bias in the reporting of results. Discrepancies were reassessed by a third reviewer (J.M.N.S), and consensus was reached through discussion.

## 3. Results

### 3.1 Search results

The initial search provided us a total of 1,076 studies. After reviewing titles and abstracts, we excluded 905 studies. Of the 171 eligible full-text articles, 123 met the study’s inclusion criteria. However, we excluded 32 studies due to fewer than 5 total deaths. Ultimately, we selected 91 studies for this review (Fig. 1).

![Figure 1.](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/07/18/2024.07.18.24310633/F1.medium.gif)

[Figure 1.](http://medrxiv.org/content/early/2024/07/18/2024.07.18.24310633/F1)

Figure 1. 
Flow chart of the study identification and selection.

### 3.2 Study characteristics

All articles were written in English and were published in 44 different journals covering 13 different thematic areas, according to Scopus classification. The most prevalent categories were Cardiology and Cardiovascular Medicine (37/91; 40.7%) and Medicine (various, 33/91; 36.2%). The manuscripts were most frequently published in The New England Journal of Medicine (17/91; 18.7%), Circulation (12/91; 13.2%), and JAMA (10/91; 11%) (Table 1).

View this table:
[Table 1.](http://medrxiv.org/content/early/2024/07/18/2024.07.18.24310633/T1)

Table 1. Characteristics of the articles included (n=91).

The majority (84/91; 92.3%) of included articles had registered protocols, and the same number conducted intention-to-treat analysis. Masked intervention was evaluated in most studies (76/91; 83.5%), and allocation was based on specific software randomization sequences in 64/91 studies (70.3%). Six articles reported discrepancies between the composite outcome association and mortality (6/91; 6.6%), and conclusions were not based on the composite outcome in two manuscripts (2/91; 2.2%).

### 3.3 Prevalence of BACO

Out of the 91 included articles, 23 (25.2%) had a significant or suggestive BACO, including 12 with p<0.005 and 11 with p between 0.005 and <0.05 (Tables 2 and 3). These 23 studies included a total of 21,285 participants in the experimental group, with sample sizes ranging from 46 to 5,523 patients, a median of 417 (IQR: 182-1,050); presenting 24 to 975 composite outcomes, with a median of 89 (IQR: 69-203); and 2 to 119 deaths, with a median of 15 (IQR: 7-36). In the control groups, there were a total of 20,048 participants, with sample sizes ranging from 54 to 5,493, a median of 409 (IQR: 190-1,038). In these groups, the number of composite events ranged from 19 to 924, with a median of 88 (IQR: 60-176); and fatal events ranged from 1 to 100, with a median of 11 (IQR: 6-38) (Table 2).

View this table:
[Table 2.](http://medrxiv.org/content/early/2024/07/18/2024.07.18.24310633/T2)

Table 2. General description of the study population and outcomes of clinical trials selected (n=23).

View this table:
[Table 3.](http://medrxiv.org/content/early/2024/07/18/2024.07.18.24310633/T3)

Table 3. Relative risks of composite and death and BACO index in clinical trials (n=23).

We observed that in 17 (73.9%) of these 23 studies, the BACO index value was between zero and <1; the remaining six had negative values, indicating a reversal of the association concerning a fatal event (mortality). None of the studies showed significant overestimation of the association attributable to the composite outcome (Table 3). More information on all 91 included articles can be found in the supplementary material.

### 3.4 Methodological quality of included studies

Of the 23 studies with significant or suggestive BACO, almost all had a low risk of bias (n=11) or some concerns (n=11). Only one article presented a high risk of bias, mainly due to the randomization process and outcome measurement (supplementary material).

## 4. Discussion

In our study, a quarter of the articles selected for review had a significant or suggestive BACO index. In all these cases, the BACO was significantly less than 1, indicating that the use of composite outcomes underestimated the association between the intervention and the prognosis, and in some cases even inverted the association.

The underestimation of the effect can be interpreted as a dilution of the association due to the inclusion of events affected differently by the intervention [6]. This phenomenon leads to a contradictory situation because composite outcomes are often used to obtain a higher number of events and, thus, greater statistical power. However, if the use of composite outcomes leads to a dilution of the effect, it paradoxically results in a reduction of the study’s power.

This explains situations such as the one observed in the study by Onland et al. [22], where hydrocortisone therapy was associated with a significant reduction in mortality among premature infants, while the effect on the composite outcome was not statistically significant (Table 3). Another example, recently recognized but beyond the scope of this review, is the study by Sesso et al. [32], where the use of a composite outcome underestimated the effect of cocoa extract supplementation on prognosis. This intervention did not reduce the incidence of the first cardiovascular event but was associated with a 27% reduction in mortality from that cause.

In another direction, underestimation can also lead to the failure to identify a harmful effect of the intervention. This occurred in the study by Stone et al. [12], where percutaneous coronary intervention (PCI) significantly increased mortality compared to coronary artery bypass grafting, while the association with the composite outcome was not statistically significant.

In other cases, the underestimation of the effect does not qualitatively change the conclusions but can lead to quantitatively significant differences, as seen in the study by Dangas et al. [9]. In this study, the effect of rivaroxaban was evaluated, finding that the excess risk for mortality was 67%. In contrast, the excess risk for the composite outcome was only 33%.

In six of the studies with a significant or suggestive BACO, the index value was negative, indicating an inversion of the association between the composite outcome and mortality. Beyond our review, we found an example of this trend in the study by Laurens et al. [33], which evaluated the effect of stopping prophylactic use of cotrimoxazole in adults with human immunodeficiency virus (HIV) infection. The authors compared standard prophylaxis using cotrimoxazole with its discontinuation, either alone or combined with chloroquine. The associations of the primary and secondary composite outcomes were in the opposite direction to mortality, with BACO indices of −0.48 (95% CI: −1.79 to 0.83; p = 0.03) and −0.69 (95% CI: −2.31 to 0.93; p = 0.04) for the primary and secondary composite outcomes, respectively [34]. This finding suggests that caution should be exercised when interpreting results, as they include effects in opposite directions. In such cases, it would be more appropriate to evaluate and interpret the components of the composite outcome individually.

There are some recommendations for constructing a composite outcome, such as considering the relevance of the outcomes for the patient, the frequency of the components of the composite outcome, and the similarity of the treatment effect for each of the events analyzed [3]. However, we believe it is challenging to construct a composite outcome that effectively represents the prognosis and allows for efficient evaluation of intervention effects. In this regard, the BACO index provides the possibility to test and guide interpretation, which can be predefined without affecting the study’s objectivity.

As limitations, our study is restricted to a relatively short time period and included primarily studies initiated before the pandemic. Therefore, we cannot infer that the prevalence and direction of BACO indices will be similar in other scenarios. However, a recent study focusing on clinical trials in COVID-19 patients estimated the BACO index in 28 effect estimates on composite outcomes [35]. In most studies, the composite outcome estimate was closer to the null value than that of mortality, and the BACO index was significantly less than 1 in five studies. Similar to the present study, there was no statistically significant overestimation of the effect associated with composite outcomes. Thus, we consider that in various scenarios, underestimation of the effect is likely the most prevalent bias resulting from the use of composite outcomes.

## 5. Conclusions

In a quarter of the selected studies, we observed suggestive results or statistically significant BACO. In most cases, the bias consisted of underestimating the effects, and in others, there was an inversion of the direction of the RR. These findings highlight the need to predefine guidelines for interpreting effects on composite endpoints based on objective criteria, such as the BACO index.

## CRediT authorship contribution statement

José Mário Nunes da Silva: Methodology, Supervision, Data curation, Formal analysis, Writing - original draft, Final approval of the version to be submitted. Juliana Ferreira Souza Conceição: Investigation, Formal analysis, Final approval of the version to be submitted. Paula C. Ramírez: Methodology, Investigation, Formal analysis, Writing - review & editing, Final approval of the version to be submitted. Christian Leonardo Diaz-León: Investigation, Formal analysis, Final approval of the version to be submitted. Fredi A. Diaz-Quijano: Conceptualization, Methodology, Project administration, Formal analysis, Writing - review & editing, Final approval of the version to be submitted.

## Declaration of competing interest

The authors declare that there are no conflicts of interest.

## Supplementary data

Supplementary data for this article can be found in the appendix.

## Ethical statement

Local ethics review board approval was not required.

## Data Availability

All data produced in the present study are available upon reasonable request to the authors

## Supplementary Material

![Figure S1.](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/07/18/2024.07.18.24310633/F2.medium.gif)

[Figure S1.](http://medrxiv.org/content/early/2024/07/18/2024.07.18.24310633/F2)

Figure S1. 
Assessment of bias risk in the 23 studies with significant or suggestive BACO.

*   Received July 18, 2024.
*   Revision received July 18, 2024.
*   Accepted July 18, 2024.


*   © 2024, Posted by Cold Spring Harbor Laboratory

This pre-print is available under a Creative Commons License (Attribution-NoDerivs 4.0 International), CC BY-ND 4.0, as described at [http://creativecommons.org/licenses/by-nd/4.0/](http://creativecommons.org/licenses/by-nd/4.0/)

## References

1.  [1].McCoy C. Understanding the Use of Composite Endpoints in Clinical Trials. West J Emerg Med 2018;19:631–4. doi:10.5811/westjem.2018.4.38383.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.5811/westjem.2018.4.38383&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30013696&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

2.  [2].Palileo-Villanueva LM, Dans AL. Composite endpoints. J Clin Epidemiol 2020;128:157–8. doi:10.1016/j.jclinepi.2020.07.017.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.jclinepi.2020.07.017&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

3.  [3].Montori VM, Permanyer-Miralda G, Ferreira-González I, Busse JW, Pacheco-Huergo V, Bryant D, et al. Validity of composite end points in clinical trials. Br Med J 2005;330:594–6. doi:10.1136/bmj.330.7491.594.
    
    [FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiRlVMTCI7czoxMToiam91cm5hbENvZGUiO3M6MzoiYm1qIjtzOjU6InJlc2lkIjtzOjEyOiIzMzAvNzQ5MS81OTQiO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyNC8wNy8xOC8yMDI0LjA3LjE4LjI0MzEwNjMzLmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 

4.  [4].McKenna SP, Heaney A. Composite outcome measurement in clinical research: the triumph of illusion over reality? J Med Econ 2020;23:1196–204. doi:10.1080/13696998.2020.1797755.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1080/13696998.2020.1797755&link_type=DOI) 

5.  [5].Freemantle N, Calvert M, Wood J, Eastaugh J, Griffin C. Composite Outcomes in Randomized Trials: Greater Precision but with Greater Uncertainty? J Am Med Assoc 2003;289:2554–9. doi:10.1001/jama.289.19.2554.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.289.19.2554&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=12759327&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000182976500024&link_type=ISI) 

6.  [6].Diaz-Quijano FA. Estimating and testing an index of bias attributable to composite outcomes in comparative studies. J Clin Epidemiol 2021;132:1–9. doi:10.1016/j.jclinepi.2020.12.003.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.jclinepi.2020.12.003&link_type=DOI) 

7.  [7].Moher D, Liberati A, Tetzlaff J, Altman DG, for the PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ 2009;339:b2535–b2535. doi:10.1136/bmj.b2535.
    
    [FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiRlVMTCI7czoxMToiam91cm5hbENvZGUiO3M6MzoiYm1qIjtzOjU6InJlc2lkIjtzOjE3OiIzMzkvanVsMjFfMS9iMjUzNSI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDI0LzA3LzE4LzIwMjQuMDcuMTguMjQzMTA2MzMuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 

8.  [8].Sterne JAC, Savović J, Page MJ, Elbers RG, Blencowe NS, Boutron I, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ 2019:l4898. doi:10.1136/bmj.l4898.
    
    [FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiRlVMTCI7czoxMToiam91cm5hbENvZGUiO3M6MzoiYm1qIjtzOjU6InJlc2lkIjtzOjE3OiIzNjYvYXVnMjhfMi9sNDg5OCI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDI0LzA3LzE4LzIwMjQuMDcuMTguMjQzMTA2MzMuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 

9.  [9].Dangas GD, Tijssen JGP, Wöhrle J, Søndergaard L, Gilard M, Möllmann H, et al. A Controlled Trial of Rivaroxaban after Transcatheter Aortic-Valve Replacement. N Engl J Med 2020;382:120–9. doi:10.1056/NEJMoa1911425.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1056/NEJMoa1911425&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

10. [10].Yasuda S, Kaikita K, Akao M, Ako J, Matoba T, Nakamura M, et al. Antithrombotic Therapy for Atrial Fibrillation with Stable Coronary Disease. N Engl J Med 2019;381:1103–13. doi:10.1056/NEJMoa1904143.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1056/NEJMoa1904143&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

11. [11].Schuetz P, Fehr R, Baechli V, Geiser M, Deiss M, Gomes F, et al. Individualised nutritional support in medical inpatients at nutritional risk: a randomised clinical trial. The Lancet 2019;393:2312–21. doi:10.1016/S0140-6736(18)32776-4.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S0140-6736(18)32776-4&link_type=DOI) 

12. [12].Stone GW, Kappetein AP, Sabik JF, Pocock SJ, Morice M-C, Puskas J, et al. Five-Year Outcomes after PCI or CABG for Left Main Coronary Disease. N Engl J Med 2019;381:1820–30. doi:10.1056/NEJMoa1909406.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1056/NEJMoa1909406&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=31562798&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

13. [13].Vardeny O, Kim K, Udell JA, Joseph J, Desai AS, Farkouh ME, et al. Effect of High-Dose Trivalent vs Standard-Dose Quadrivalent Influenza Vaccine on Mortality or Cardiopulmonary Hospitalization in Patients With High-risk Cardiovascular Disease: A Randomized Clinical Trial. JAMA 2021;325:39. doi:10.1001/jama.2020.23649.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2020.23649&link_type=DOI) 

14. [14].Lanz J, Kim W-K, Walther T, Burgdorf C, Möllmann H, Linke A, et al. Safety and efficacy of a self-expanding versus a balloon-expandable bioprosthesis for transcatheter aortic valve replacement in patients with symptomatic severe aortic stenosis: a randomised non-inferiority trial. The Lancet 2019;394:1619–28. doi:10.1016/S0140-6736(19)32220-2.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S0140-6736(19)32220-2&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

15. [15].Lomivorotov V, Kornilov I, Boboshko V, Shmyrev V, Bondarenko I, Soynov I, et al. Effect of Intraoperative Dexamethasone on Major Complications and Mortality Among Infants Undergoing Cardiac Surgery: The DECISION Randomized Clinical Trial. JAMA 2020;323:2485. doi:10.1001/jama.2020.8133.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2020.8133&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=32573670&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

16. [16].Vermeersch K, Gabrovska M, Aumann J, Demedts IK, Corhay J-L, Marchand E, et al. Azithromycin during Acute Chronic Obstructive Pulmonary Disease Exacerbations Requiring Hospitalization (BACE). A Multicenter, Randomized, Double-Blind, Placebo-controlled Trial. Am J Respir Crit Care Med 2019;200:857–68. doi:10.1164/rccm.201901-0094OC.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1164/rccm.201901-0094OC&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

17. [17].Frith P, Sladek R, Woodman R, Effing T, Bradley S, Van Asten S, et al. Pragmatic randomised controlled trial of a personalised intervention for carers of people requiring home oxygen therapy. Chron Respir Dis 2020;17:147997311989727. doi:10.1177/1479973119897277.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/1479973119897277&link_type=DOI) 

18. [18].Wilson AM, Clark AB, Cahn T, Chilvers ER, Fraser W, Hammond M, et al. Effect of Co-trimoxazole (Trimethoprim-Sulfamethoxazole) vs Placebo on Death, Lung Transplant, or Hospital Admission in Patients With Moderate and Severe Idiopathic Pulmonary Fibrosis: The EME-TIPAC Randomized Clinical Trial. JAMA 2020;324:2282. doi:10.1001/jama.2020.22960.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2020.22960&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

19. [19].Koch CG, Sessler DI, Duncan AE, Mascha EJ, Li L, Yang D, et al. Effect of red blood cell storage duration on major postoperative complications in cardiac surgery: A randomized trial. J Thorac Cardiovasc Surg 2020;160:1505–1514.e3. doi:10.1016/j.jtcvs.2019.09.165.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.jtcvs.2019.09.165&link_type=DOI) 

20. [20].Willems S, Tilz RR, Steven D, Kääb S, Wegscheider K, Gellér L, et al. Preventive or Deferred Ablation of Ventricular Tachycardia in Patients With Ischemic Cardiomyopathy and Implantable Defibrillator (BERLIN VT): A Multicenter Randomized Trial. Circulation 2020;141:1057–67. doi:10.1161/CIRCULATIONAHA.119.043400.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1161/CIRCULATIONAHA.119.043400&link_type=DOI) 

21. [21].Futier E, Garot M, Godet T, Biais M, Verzilli D, Ouattara A, et al. Effect of Hydroxyethyl Starch vs Saline for Volume Replacement Therapy on Death or Postoperative Complications Among High-Risk Patients Undergoing Major Abdominal Surgery: The FLASH Randomized Clinical Trial. JAMA 2020;323:225. doi:10.1001/jama.2019.20833.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2019.20833&link_type=DOI) 

22. [22].Onland W, Cools F, Kroon A, Rademaker K, Merkus MP, Dijk PH, et al. Effect of Hydrocortisone Therapy Initiated 7 to 14 Days After Birth on Mortality or Bronchopulmonary Dysplasia Among Very Preterm Infants Receiving Mechanical Ventilation: A Randomized Clinical Trial. JAMA 2019;321:354. doi:10.1001/jama.2018.21443.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2018.21443&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30694322&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

23. [23].Thiele H, Kurz T, Feistritzer H-J, Stachel G, Hartung P, Eitel I, et al. Comparison of newer generation self-expandable vs. balloon-expandable valves in transcatheter aortic valve implantation: the randomized SOLVE-TAVI trial. Eur Heart J 2020;41:1890–9. doi:10.1093/eurheartj/ehaa036.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/eurheartj/ehaa036&link_type=DOI) 

24. [24].Karaye KM, Sa’idu H, Balarabe SA, Ishaq NA, Sanni B, Abubakar H, et al. Selenium supplementation in patients with peripartum cardiomyopathy: a proof-of-concept trial. BMC Cardiovasc Disord 2020;20:457. doi:10.1186/s12872-020-01739-z.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/s12872-020-01739-z&link_type=DOI) 

25. [25].Lee MMY, Petrie MC, Rocchiccioli P, Simpson J, Jackson CE, Corcoran DS, et al. Invasive Versus Medical Management in Patients With Prior Coronary Artery Bypass Surgery With a Non-ST Segment Elevation Acute Coronary Syndrome: A Pilot Randomized Controlled Trial. Circ Cardiovasc Interv 2019;12:e007830. doi:10.1161/CIRCINTERVENTIONS.119.007830.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1161/CIRCINTERVENTIONS.119.007830&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

26. [26].De Araújo CAL, Ray JG, Figueiroa JN, Alves JG. BRAzil magnesium (BRAMAG) trial: a double-masked randomized clinical trial of oral magnesium supplementation in pregnancy. BMC Pregnancy Childbirth 2020;20:234. doi:10.1186/s12884-020-02935-7.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/s12884-020-02935-7&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=32316938&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

27. [27].Zhang H, Zhang X, Yin Y, Zhou C, Deng W, Zhang J, et al. An abluminal biodegradable polymer sirolimus-eluting stent versus a durable polymer everolimus-eluting stent in patients undergoing coronary revascularization: 3-year clinical outcomes of a randomized non-inferiority trial. Sci Rep 2019;9:18549. doi:10.1038/s41598-019-54964-8.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41598-019-54964-8&link_type=DOI) 

28. [28].De Luca G, Damen SA, Camaro C, Benit E, Verdoia M, Rasoul S, et al. Final results of the randomised evaluation of short-term dual antiplatelet therapy in patients with acute coronary syndrome treated with a new-generation stent (REDUCE trial). EuroIntervention 2019;15:e990–8. doi:10.4244/EIJ-D-19-00539.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.4244/EIJ-D-19-00539&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

29. [29].Tong DC, Quinn S, Nasis A, Hiew C, Roberts-Thomson P, Adams H, et al. Colchicine in Patients With Acute Coronary Syndrome: The Australian COPS Randomized Clinical Trial. Circulation 2020;142:1890–900. doi:10.1161/CIRCULATIONAHA.120.050771.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1161/CIRCULATIONAHA.120.050771&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

30. [30].Sekiziyivu BA, Bancroft E, Rodriguez EM, Sendagala S, Nasirumbi MP, Najjengo MS, et al. Task Shifting for Initiation and Monitoring of Antiretroviral Therapy for HIV-Infected Adults in Uganda: The SHARE Trial. JAIDS J Acquir Immune Defic Syndr 2021;86:e71–9. doi:10.1097/QAI.0000000000002567.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1097/QAI.0000000000002567&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=33230029&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

31. [31].Johnston SC, Amarenco P, Denison H, Evans SR, Himmelmann A, James S, et al. Ticagrelor and Aspirin or Aspirin Alone in Acute Ischemic Stroke or TIA. N Engl J Med 2020;383:207–17. doi:10.1056/NEJMoa1916870.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1056/NEJMoa1916870&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F07%2F18%2F2024.07.18.24310633.atom) 

32. [32].Ramírez PC, Diaz-Quijano FA. Bias attributable to the use of a composite outcome in evaluating a cocoa extract supplement. Am J Clin Nutr 2022;116:1452. doi:10.1093/ajcn/nqac185.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/ajcn/nqac185&link_type=DOI) 

33. [33].Laurens MB, Mungwira RG, Nampota N, Nyirenda OM, Divala TH, Kanjala M, et al. Revisiting Co-Trimoxazole Prophylaxis for African Adults in The Era of Antiretroviral Therapy: A Randomized Controlled Clinical Trial. Clin Infect Dis 2021. doi:10.1093/cid/ciab252.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/cid/ciab252&link_type=DOI) 

34. [34].Ramírez PC, Diaz-Quijano FA. Using a Composite Outcome in Estimating the Effect of Co-trimoxazole Prophylaxis on Prognosis in African Adults Receiving Antiretroviral Therapy. Clin Infect Dis 2021;73:1550–1. doi:10.1093/cid/ciab362.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/cid/ciab362&link_type=DOI) 

35. [35].Martins PN, Lourenço MHT, Mota GPS, Cavalcanti AB, Antonio ACP, Diaz-Quijano FA. Composite endpoints in covid-19 randomized controlled trials: a systematic review. Clin Trials J Soc Clin Trials n.d.:in press.