Searching PubMed and Web of Science databases for conducting literature search till December 1, 2023 (see the flow chart in Figure 1 for the complete searching strategy), with no restrictions on publication status. For a comprehensive literature search, we also manually search for the references included in studies, review articles, and editorials/letters.

Inclusion criteria: (1). Retrospective or prospective cohort study; (2). Patients with AF and concurrent thrombocytopenia; (3). The study reported the platelet level of the patient, which was in line with the diagnostic criteria for thrombocytopenia provided by the international consensus report (17): Thrombocytopenia is usually defined as platelet count less than 100 × 10⁹/L in Asian; In Europe and America, thrombocytopenia is usually defined as platelet count less than 150 × 10⁹/L. (4). Studies have reported the correlation between thrombocytopenia as well as one or more of the following outcomes: ischemic stroke/systemic embolism, bleeding events or mortality in AF.

Exclusion criteria: (1). Full text cannot be obtained; (2). Non-English literature; (3). The definition of thrombocytopenia is not international standardized; (4). No clinical outcomes were reported in AF.

Data and literature were collected and extracted independently by two reviewers (QD and WX), and if there was any dispute, it would be discussed and resolved by a third researcher to reach a consensus. From each study, we extracted the study type, treatment, study population, average age, male proportion, country, follow-up period and outcome indicator (ischemic stroke/systemic embolism, bleeding events, mortality etc.).

The Newcastle-Ottawa scale was used to measure three aspects of a study: selection of subjects, comparability between groups, and outcome measurement. Except for comparability, which can score up to 2 points, the rest are 1 point, the highest score is 9 points, 7 to 9 means high quality, 4 to 6 means medium quality, and the score below 4 means low quality.

The analysis was performed by using REVMAN 5.4 and Stata 16.0 software, and CI was used as the combined effect size of the binary classification. Odds ratios (OR) were calculated for all outcomes using the pooled effect of each therapy according to the Mantel–Haenszel method for random effects. [I²] was used to evaluate the heterogeneity among the trials (26), I² ≤ 50% indicated that there was no statistical heterogeneity among the studies, I² > 50% suggested that there was heterogeneity among studies. The publication bias of the literature included for outcome measures was assessed by Egger's test (27). P < 0.05 was considered statistically significant.

The searching strategy had identified 5,223 studies (Figure 1). Of these, 37 potentially relevant studies were identified for full text assessment. After excluding 25 studies, all the rest 12 studies met the inclusion criteria. In the report OR research, 10 studies examined the association between thrombocytopenia and ischemic stroke/systemic embolism, 9 studies examined the association between thrombocytopenia and major bleeding, and 7 studies examined the association between thrombocytopenia and all-cause mortality.

The 12 studies included in this systematic review has been analyzed, including 8 studies on Asian populations, 4 studies on European and American populations. A total of 73,824 patients with atrial fibrillation and thrombocytopenia were analyzed, with average age of 72.67, 57.3% male, and the average follow-up ranging from 87 days to 55 months. The baseline characteristics of 12 studies included was shown in Table 1, The quality evaluation of 12 studies included was used by the Newcastle-Ottawa scale as Supplementary Table S1.

A total of 10 studies with 63,005 participants involved examined the association between thrombocytopenia and ischemic stroke/systemic embolism (Figure 2). Thrombocytopenia was significantly associated with a lower risk of ischemic stroke/systemic embolism in patients with AF [OR: 0.79, 95% CI: (0.69, 0.91), P = 0.0008], with low heterogeneity ([I²] = 10%). The Egger's test revealed no evidence of publication bias (Supplementary Table S2, P = 0.0891). Given that the impact of different anticoagulant regimens on outcomes, we performed a subgroup analysis of ischemic stroke/systemic embolism according to the anticoagulant regimen [OR: 0.79, 95% CI: (0.68 −0.91), P = 0.002] (Supplementary Figure S1), which are concurred with the overall analysis.

A total of 9 studies with 57,055 participants involved examined the association between thrombocytopenia and major bleeding (Figure 3). Thrombocytopenia was significantly associated with a higher risk of major bleeding [OR: 1.51, 95% CI: (1.20, 1.79), P = 0.0002], with not high heterogeneity ([I²] = 41%). Given the mild heterogeneity, we performed a sensitivity analysis to explore a cause of heterogeneity by excluding the study [Varun Iyengar et al. (10)], and found that thrombocytopenia compared with control was associated with a significant increased risk of major bleeding [OR: 1.34; 95% CI: (1.18, 1.52), P < 0.00001] (Supplementary Figure S4), with not heterogeneity [(I²) = 0%]. Meanwhile, we performed a subgroup analysis of major bleeding according to the anticoagulant regimen [OR: 1.47, 95% CI: (1.20 −1.79), P = 0.0002] (Supplementary Figure S2), which is concurred with the overall analysis. The Egger's test revealed no evidence of publication bias (Supplementary Table S2, P = 0.8912).

A total of 6 studies with 28,799 participants involved examined the association between thrombocytopenia and combined bleeding events (Figure 4). Thrombocytopenia was significantly associated with a higher risk of combined bleeding events in patients with AF [OR: 1.51, 95% CI: (1.27, 1.81), P < 0.00001], with not high heterogeneity [(I²) = 43%]. The Egger's test showed publication bias (Supplementary Table S2, P = 0.0049), therefore, the result may be not very reliable in combined bleeding events.

A total of 4 studies with 5,135 participants involved examined the association between thrombocytopenia and minor bleeding (Figure 5). Overall, the risk of minor bleeding in AF combined with thrombocytopenia was similar to that in AF [OR: 1.59, 95% CI: (0.79, 3.21), P = 0.19], with significant heterogeneity [(I²) = 68%]. Due to high heterogeneity, we performed the sensitivity analysis of minor bleeding, and found that thrombocytopenia was associated with a significant increased risk of minor bleeding [OR: 1.14; 95% CI: (0.79, 1.66), P = 0.49] compared with control group, with significant heterogeneity [(I²) = 0%] by excluding the study (28) (Supplementary Figure S5). The Egger's test revealed publication bias (Supplementary Table S2, P = 0.0024), therefore, the result is not very reliable in minor bleeding.

A total of 3 studies with 1,637 participants involved examined the association between thrombocytopenia and gastrointestinal hemorrhage (Figure 6). Overall, the risk of gastrointestinal hemorrhage in AF combined with thrombocytopenia was similar to that in AF [OR: 1.26, 95% CI: (0.46, 2.94), P = 0.74], with low heterogeneity [(I²) = 34%].The Egger's test showed no evidence of publication bias (Supplementary Table S2, P = 0.5394).

A total of 4 studies with 32,449 participants involved examined the association between thrombocytopenia and intracranial hemorrhage (Figure 7). Overall, the risk of intracranial hemorrhage in AF combined with thrombocytopenia was similar to that in AF [OR: 1.21, 95% CI: (0.85, 1.71), P = 0.29], with no heterogeneity [(I²) = 0%]. The Egger's test showed no evidence of publication bias (Supplementary Table S2, P = 0.9933).

A total of 3 studies with 1637 participants involved examined the association between thrombocytopenia and clinically relevant non-major bleeding (Figure 8). Overall, the clinically relevant risk of non-major bleeding in AF combined with thrombocytopenia was similar to that in AF [OR: 1.59, 95% CI: (0.54, 4.73), P = 0.40]. Heterogeneity was relatively high [(I²) = 53%]. Due to the high heterogeneity, we also conducted a sensitivity analysis by excluding the study [Eitaro Kodani et al. (29)] and found that thrombocytopenia was associated with a nonsignificant increased risk of clinically relevant non-major bleeding [OR: 0.99; 95% CI: (0.65, 1.52)] compared with control group (Supplementary Figure S6). The Egger's test displayed no evidence of publication bias (Supplementary Table S2, P = 0.0997).

A total of 7 studies with 26,430 participants involved examined the association between thrombocytopenia and all-cause mortality (Figure 9). Thrombocytopenia was significantly associated with a higher risk of all-cause mortality [OR: 1.40, 95% CI: (1.23, 1.61), P < 0.00001], with no significant heterogeneity [(I²) = 12%]. The Egger's test showed no publication bias (Supplementary Table S2, P = 0.2872). To better examine the relationship between thrombocytopenia and all-cause mortality, we also performed a subgroup analysis according to the anticoagulant regimen. We found that thrombocytopenia was significantly associated with a higher risk of all-cause mortality [OR: 1.40, 95% CI: (1.23 −1.61), P < 0.00001] (Supplementary Figure S3), which is concurred with the overall analysis.

Platelets play a crucial role in the complex clotting process. They release clotting factors and activate the clotting pathway by aggregating in injured blood vessels, and eventually form thrombus (32). However, excessive or rapid formation of thrombus may cause vascular obstruction, affect blood circulation, even lead to ischemic stroke/systemic embolism, myocardial infarction and other serious thrombotic diseases (33). On the contrary, low platelet count or decreased coagulation function will increase the risk of bleeding (34). A study on the correlation between PLT counts and stroke occurrence and prognosis found that PLT counts were positively correlated with the risk of ischemic stroke, elevated platelet count (>300 × 10⁹/L) was a risk factor for ischemic stroke, and low PLT count (<100 × 10⁹/L) was associated with a reduced risk of ischemic stroke (35). Another study has found out the similar results, low platelet count (<50 × 10⁹/L) could reduce the risk of venous thromboembolism (36). This meta-analysis also found that AF patients with thrombocytopenia had a significantly reduced risk of ischemic stroke/systemic embolism compared to general AF, which is consistent with the conclusions of the above studies.

However, other studies observed that thrombocytopenia could not reduce thromboembolic events, especially in the case of AF complications (37, 38). Thrombocytopenia might be a causal factor in an ischemic stroke, a risk factor for hemorrhagic stroke (39). Due to limited data on AF patients with thrombocytopenia, whether low PLT can protect AF patients against ischemic stroke/systemic embolic events remains to be further studied.

The central role of platelets in the clotting process is self-evident. The decrease in platelet number or function will increase the clotting time, and the main clinical consequence is bleeding caused by impaired primary hemostatic function and platelet embolism formation (40). An observational study has confirmed that a lower platelet count was related with an increased risk of spontaneous bleeding (41). The key determinants of the anticoagulant bleeding rate in AF are age, previous thrombosis and bleeding events (42). Then, could platelet level be another determinant of bleeding events in AF? A previous research have verified that platelet level was an independent predictor of bleeding events (43). Moreover, the degree of thrombocytopenia was positively correlated with the risk of bleeding, more severe thrombocytopenia (<75 × 10⁹/L) was associated with an increased risk of major bleeding (10). Our meta-analysis similarly displayed that AF patients with thrombocytopenia had a significantly increased risk of bleeding events compared to AF patients without thrombocytopenia, which is consistent with the conclusions of previous cohort studies (44, 45). We also found that AF patients with thrombocytopenia had a greater tendency to minor bleeding, gastrointestinal hemorrhage, intracranial hemorrhage, and clinically relevant non-major bleeding, although the correlation was not statistically significant, which might be due to the small number of studies we included, limited by the insufficient population. Future studies need to enroll more literature and conduct subgroup analysis according to the severity of thrombocytopenia, so as to obtain more accurate data for stratified prediction of bleeding risk in AF patients combined with thrombocytopenia.

Thrombocytopenia can be divided into primary and acquired thrombocytopenia. Primary thrombocytopenia is mostly caused by chronic blood diseases. The causes of acquired thrombocytopenia may include trauma, surgery, infection, drugs, radiotherapy, immune dysfunction, and nutritional disorders (46). Multiple studies have shown a significant association between thrombocytopenia and increased risk of all-cause mortality (47–51). Particularly, another study showed that chronic kidney disease, active cancer, and liver cirrhosis, bone marrow disorders which are frequent features of patients with AF were significantly associated with thrombocytopenia (52).

The risk of mortality in patients with AF is 1.5 to 1.9 times the mortality of the normal population (53), and the types of mortality can be divided into cardiovascular death and non-cardiovascular death. The risk factors mainly include infection, cancer, blood uric acid, diabetes, hypertension, myocardial infarction and heart failure (54, 55). Although no studies investigated the specific reasons of mortality in AF with thrombocytopenia, the correlation between thrombocytopenia and mortality in AF may be mediated by thrombocytopenia affecting these major risk factors. The present meta-analysis found that AF patients with thrombocytopenia had a significant higher risk of mortality than AF patients without thrombocytopenia, which was consistent with previous studies. In addition, according to the research (56), the bleeding of AF patients was closely related to subsequent major unscrupulous cerebrovascular events (MACCE) and death, and preventing major bleeding could significantly avoid cardiovascular events and death. Our analysis has also shown similar results that AF complicated with thrombocytopenia had a significantly increased risk of bleeding events, major bleeding, and potentially increased risk of fatal bleeding such as intracranial hemorrhage, which might be one of the reasons of significantly increased risk of mortality in AF complicated with thrombocytopenia.

Therefore, thrombocytopenia increases the all-cause mortality. If AF is accompanied by thrombocytopenia, there is significant increase the additional mortality. Traditionally, the focus of improving the prognosis of AF is on preventing stroke, and anticoagulation therapy has been shown to significantly reduce thromboembolism and stroke-related mortality in AF (57). Our analysis suggests that screening and attention to comorbidities, such as thrombocytopenia, may be one of the effective ways to reduce the risk of mortality in patients combined with AF. What's more, the complex association between atrial fibrillation with thrombocytopenia and mortality needs further investigation.