Effect of non‐recommended doses versus recommended doses of direct oral anticoagulants in atrial fibrillation patients: A meta‐analysis

Abstract Background Several observational studies have shown that the inappropriate dosing use of direct oral anticoagulants (DOACs) in atrial fibrillation (AF) that does not conform to recommendations is becoming a widespread phenomenon. Therefore, we performed a meta‐analysis and systematic review to assess the effect of non‐recommended doses versus recommended doses of DOACs on the effectiveness and safety outcomes among AF patients. Methods The PubMed and Ovid databases were systematically searched to identify the relevant studies until December 2020. The effect estimates were hazard ratios (HRs) and 95% confidence intervals (CIs), which were pooled using a fixed‐effects model (I2 ≤ 50%) or a random‐effects model (I2 > 50%). Results A total of 11 studies were included in this meta‐analysis. Compared with recommended dosing of DOACs, non‐recommended low dosing of DOACs was associated with increased risks of stroke or systemic embolism (SSE, HR = 1.29, 95% CI 1.12–1.49) and all‐cause death (HR = 1.37, 95% CI 1.15–1.62), but not the ischemic stroke, myocardial infarction, gastrointestinal bleeding, intracranial bleeding, and major bleeding. Compared with recommended dosing of DOACs, non‐recommended high dosing of DOACs was associated with increased risks of SSE (HR = 1.44, 95% CI 1.01–2.04), major bleeding (HR = 1.99, 95% CI 1.48–2.68), and all‐cause death(HR = 1.38, 95% CI 1.02–1.87). Conclusion Compared with recommended dosing of DOACs, non‐recommended low dosing of DOACs was associated with increased risks of SSE and all‐cause death. Further study should confirm the findings of non‐recommended high dosing versus recommended dosing of DOACs.


| INTRODUCTION
Atrial fibrillation (AF) is the most common cardiac arrhythmia, resulting in increased risks of stroke and its related mortality and disability. 1 Anticoagulant drugs are effective drugs to prevent stroke risk in patients with AF. 2  (4) low dose of edoxaban (30 mg once a day) if patients have a bodyweight ≤60 kg, creatinine clearance level of 30-50 ml/min, or concomitant P-glycoprotein inhibitors. 3,4 Nevertheless, the use of reduceddose DOACs does not conform to the label-or guideline-recommendations in real-world settings. The off label low or high dose of DOACs is regarded as non-recommended low dose (underdosing) or nonrecommended high dose (overdosing), respectively. The reasons why the off label dose of DOACs is commonly used are still exploratory.
In these years, more and more researchers have explored the effect of non-recommended low dose or non-recommended high dose of DOACs among patients with AF. [5][6][7][8][9][10][11][12][13][14][15] However, the findings of those observational studies are sometimes quite different; and thus, the effectiveness and safety profiles among non-recommended doses of DOACs remain unclear, leaving physicians with difficulties in decision-making regarding the choice of DOAC doses. Therefore, we conducted a metaanalysis and systematic review to examine the effect of non-recommended doses (underdosing or overdosing) versus recommended doses of DOACs on effectiveness and/or safety outcomes among AF patients.

| METHODS
Since it was a meta-analysis based on the published studies, ethical approval was not necessary. The data that support the findings of this study would be available from the corresponding author on the reasonable requests.

| Literature search
The process of this meta-analysis was conducted according to the guidance from the Cochrane Handbook for Systematic Reviews. The results of this meta-analysis were presented according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses). The PubMed and Ovid electronic databases were systematically searched to identify the relevant studies that reporting the inappropriate DOAC dosing in AF patients. The retrieval periods were from January 2009 to December 2020 because the first publication of DOAC (dabigatran) in AF patients was reported in the year of 2009. Table 1, the following index terms and their similar keywords were used in the electronic search: (1) "atrial fibrillation" OR "atrial flutter" AND (2) "direct oral anticoagulants" OR "non-vitamin K antagonist oral anticoagulants" OR "dabigatran" OR "rivaroxaban" OR "apixaban" OR "edoxaban" AND (3) dose OR dosing OR overdosing OR underdosing. To reduce the omission of available and value studies, we scanned the reference lists of relevant meta-analyses and reviews based on the DOAC dosing. 3,[16][17][18][19] We applied no linguistic restrictions in the search.

| Inclusion and exclusion criteria
We included observational studies that reported the effect of nonrecommended doses (underdosing or overdosing) versus recommended doses of at least one DOAC (dabigatran, rivaroxaban, apixaban, or edoxaban) on effectiveness and/or safety outcomes among AF patients. The low or high dose of DOACs was defined as non-recommended low dose (underdosing) or non-recommended high dose (overdosing) respectively if they did not conform to the label-or guideline-recommendations. We applied the dosing criteria of DOACs in each included study. The effect estimates of this study were adjusted hazard ratios (HRs) and 95% confidence intervals (CIs).
If several studies had overlapping data from the same data sources, we included the study with the longest follow-up or highest sample size.

| Effectiveness and safety outcomes
The primary effectiveness outcome was stroke or systemic embolism, whereas the primary safety outcome was major bleeding. Our secondary effectiveness outcomes included ischemic stroke, myocardial infarction, and all-cause death, whereas the secondary safety outcomes were intracranial bleeding and gastrointestinal bleeding. We applied the original definitions of the effectiveness and safety outcomes in the included studies.

| Quality assessment
The Newcastle-Ottawa Scale (NOS) items were used to assess the study quality of observational studies. The NOS tool involved three domains with a total of 9 points: the selection of cohorts (0-4 points), the comparability of cohorts (0-2 points), and the assessment of the outcome (0-3 points

| RESULTS
The flow chart of the literature retrieval of this meta-analysis is shown in Supplementary Figure 1. A total of 11 studies were included in this meta-analysis. [6][7][8]11,14,15,[22][23][24][25][26] The specific baseline characteristics of these included studies are shown in Table 1. All the included studies had a moderate-to-high quality with a NOS score of ≥6 points (Supplemental Table 2).  Figure 3). In addition, re-analyses with an inverse variance heterogeneity or quality effects model suggested similar results as the above-mentioned analysis with a fixed-effects model.

| Secondary outcomes
As presented in Supplementary Figures 4 to 8 Figures 9 to 10).

| Secondary outcomes
As shown in Supplementary Figure 11

| Publication bias
There were no potential publication biases as assessed by inspecting the funnel plots ( Supplementary Figures 12 to 13

DATA AVAILABILITY STATEMENT
Availability of data and materials have been described in the manuscript. They are freely available to any scientist who wishes to use them without breaching participant confidentiality.