Introduction

Patients affected by SARS-CoV-2 infection have a wide range of clinical presentations from being asymptomatic to suffer from an acute respiratory distress syndrome which is associated with a high mortality rate [1]. In these patients, the presence of hypoxia, inflammation, platelet activation, endothelial dysfunction, and stasis may predispose to venous thromboembolic events (deep vein thrombosis [DVT] and/or pulmonary embolism [PE]) [2]. Indeed, since early, a high incidence of venous thromboembolism (VTE) was observed in hospitalized patients with coronavirus disease 19 (COVID-19) [3]. These complications are considered as predictors of poor prognosis and may contribute to morbidity and mortality [4].

Clinical guidance on the prevention of VTE in COVID-19 patients were promptly published but, as based on relatively low evidences, the recommended strategies were not univocal [2, 5,6,7,8,9]. The use of heparin (unfractionated [UFH] or low molecular weight [LMWH]) at prophylactic doses was recommended by some guidelines and the use of intermediate (any dosage between the standard and the therapeutic one) or therapeutic (full anticoagulation) doses in others.

Noteworthy, patients with COVID-19 might be at risk of excess bleeding due to the imbalances in platelet production and destruction, coagulation factor consumption in the setting of severe inflammation, and use of antiplatelet or anticoagulant agents [2].

Recently published studies comparing prophylactic, intermediate or therapeutic heparin regimens in patients hospitalized for COVID-19 in different clinical settings reported inconsistent reduction of adverse outcomes; thus, the use of higher than prophylactic heparin regimens remains controversial [10,11,12,13]. Indeed, these studies were mainly focused on reduction in mortality or on composite outcomes (eg. venous or arterial thrombosis, treatment with extracorporeal membrane oxygenation, or mortality).

We performed a meta-analysis of published studies on the efficacy (prevention of VTE) and safety of standard or higher heparin regimens for thromboembolic prophylaxis in hospitalized patients with COVID-19.

Methods

A protocol for this study was developed detailing the specific objectives, criteria for study selection, approach to assess study quality, outcomes, and statistical methods (PROSPERO registration number CRD42021252550).

Data sources and searches

We performed an unrestricted search in PubMed, ClinicalTrial.gov, BioRxiv and MedRxiv, from inception through June 13th, 2022. No language restriction was applied. Reference lists of retrieved articles and review articles were manually searched for other relevant studies. The search strategy is reported in the Supplementary material.

Study selection

Two reviewers (M.C.V. and M.G.) performed study selection independently, with disagreements solved through discussion and the opinion of a third reviewer (C.B.). Studies on patients with COVID-19 were considered potentially eligible for the meta-analysis if they met the following predetermined criteria: (a) were randomized clinical trials (RCTs) or observational cohorts (prospective or retrospective); (b) included and reported data on outcome events in patients hospitalized due to SARS-CoV2 infection; (c) included both groups of patients receiving heparin at standard prophylactic dose and at high-dose (intermediate or therapeutic doses); d) reported VTE or bleeding events by groups.

Studies were included in the meta-analysis if the following data were available: (i) number of patients hospitalized due to confirmed COVID-19 who received standard-dose vs. high-dose (intermediate or therapeutic) heparin prophylaxis; (ii) outcome events separately reported by treatment group.

Studies were not eligible for the analysis if: (a) reported on the use of thromboprophylaxis agents other than heparin; (b) included fewer than 10 patients.

The primary study outcome was in-hospital VTE. Secondary study outcomes were in-hospital major bleeding, all-cause-death, fatal pulmonary embolism (PE) and fatal bleeding.

Study outcomes events were reported according to the definition used in the individual studies (e-Table 1).

For duplicate publications, the most complete was considered.

Data extraction and quality assessment

Data were extracted and presented according to the Providing Innovative Service Models and Assessment (PRISMA) criteria [14].

For each study, the following data were extracted: general data (study design, year of publication), population characteristics (mean age, gender), setting (intensive care unit [ICU], non-ICU) and thromboprophylaxis regimen (standard dose, intermediate or therapeutic dose). Information on the following outcomes was collected: VTE, major bleeding and death (all-cause, fatal PE, fatal bleeding).

Study quality was assessed by two reviewers (M.C.V. and M.G.) using Cochrane Collaboration’s tool to assess risk of bias in randomized trials, which cover the following bias domains: selection bias, performance bias, detection bias, attrition bias and reporting bias [15]. High quality was defined when at least 6 out of 7 criteria were satisfied (e-Table 2). Similarly, two investigators evaluated the risk of bias in observational studies using the Newcastle–Ottawa quality assessment scale for cohort studies [16]. This scale assesses the representativeness of the sample, ascertainment of the exposure, control of confounding variables, assessment of outcome and adequacy of follow-up, which provides a score ranging from 0 (lowest grade) to 9 (highest grade). High-quality studies were considered when at least 8 out of 9 criteria were satisfied. We resolved disagreements in study data extraction and quality assessment by consensus or by discussion with a third reviewer (C.B.).

Statistical analysis

Study outcomes in patients who received the high-dose (intermediate or therapeutic) were compared with patients receiving standard-dose heparin prophylaxis using a random effect model. To evaluate statistical heterogeneity, we calculated the I2-index: a value of 25% was defined as low heterogeneity, 50% as moderate heterogeneity, and 75% as high heterogeneity [17]. In case of heterogeneity, subgroup analyses (according to study design, hospital setting, dose of prophylaxis, study quality) and metaregression analyses were performed (according to study design, study quality, hospital setting, dose of prophylaxis).

We expressed comparison between treatment groups by risk ratios (RR) with 95% confidence intervals (CIs) and forest plots. Cells including zero were replaced with 0.5. Main results were also reported as number needed to treat (NNT), number needed to harm (NNH) and likelihood of being helped or harmed (LHH calculated as NNH/NNT) [18].

Subgroup analyses according to study design (RCTs, cohort studies) and hospital setting (ICU or non-ICU) were performed. Separate analyses comparing studies reporting on intermediate-dose vs. standard-dose and comparing therapeutic-dose vs. standard-dose heparin prophylaxis were also performed. In addition, analyses limited to high-quality studies by quality assessment and by study design (retrospective studies excluded) were performed.

Publication bias was assessed by using Egger’s regression test and considered significant if p-value was < 0.10.

To assess agreement between reviewers for study selection, we used the kappa statistic, which measures agreement beyond chance.

Analysis was performed with StatsDirect 3.2.10 (StatsDirect Ltd, Wirral, UK).

Results

Overall, 3598 studies were found and 124 were selected as potentially relevant. At the end of the selection process, 33 studies (11,387 patients) satisfied criteria for inclusion in the meta-analysis (Table 1) [10,11,12,13, 19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47]. Flow diagram for study selection is reported in e-Fig. 1. Inter-observer agreement for study selection was good (k = 0.82). Ten studies were RCTs, 4 were prospective and 19 were retrospective cohort studies. Sites of MBs are reported in e-Table 1.

Table 1 Baseline characteristics of the studies included in the meta-analysis
Full size table

Venous thromboembolism

Twenty-three studies (8428 patients) reported on in-hospital VTE: a significant reduction of VTE events was observed in patients receiving heparin high-dose compared to those receiving standard-dose prophylaxis (RR 0.71, 95% CI 0.55–0.90, I2 48.8%) (Fig. 1A). Main results are reported in Table 2. Moderate heterogeneity was observed, metaregression analysis showed not significant influence of study design (p = 0.092), study quality (p = 0.212), hospital setting (p = 0.131), while it seemed associated to dose of heparin prophylaxis (p = 0.025). The separate analyses of the different doses of heparin prophylaxis showed persistent heterogeneity (I2 51.5%) for studies reporting on high-dose heparin prophylaxis (not sortable), while for studies reporting on the intermediate and for the therapeutic doses it was low (I2 1.4% and 0.0%, respectively).

Fig. 1
figure 1

Risk of venous thromboembolism (A) and major bleeding (B) in patients receiving high-dose or standard-dose heparin prophylaxis

Full size image
Table 2 Study outcomes according to study design and settings
Full size table

Heterogeneity was reduced by exclusion of retrospective studies (RR 0.50, 95% CI 0.36 to 0.69, I2 14.8%) and remained unchanged after exclusion of low-quality studies (RR 0.65, 95% CI 0.50–0.85, I2 38.9%).

In the subgroup analyses of studies reporting separately data in patients admitted to the ICU and to the non-ICU settings, results were similar to those observed in the main analysis (RR 0.70, 95% CI 0.52–0.93, I2 52.1% and RR 0.48, 95% CI 0.26–0.87, I2 0.0%) (Fig. 2A, B). Moderate heterogeneity was found in the sub-analysis on ICU setting that persisted after removing retrospective or low-quality studies. A separate analysis of randomized controlled studies was reported in e-Table 3 and e-Fig. 2.

Fig. 2
figure 2

Risk of venous thromboembolism (A = ICU; B = non-ICU) and major bleeding (C = ICU; D = non-ICU) in patients receiving high-dose or standard-dose heparin prophylaxis according to hospital setting

Full size image

No significant differences were observed when intermediate heparin doses were compared to standard-dose prophylaxis, but a significant reduction in VTE events was observed when the therapeutic doses were used (e-Table 4).

Egger’s test did not reveal publication bias. The NNT was 33, 36 and 78 in the overall population, in the ICU and in the non-ICU patients, respectively. In the separate analysis of intermediate or therapeutic doses NNT was 111 and 33, respectively.

Major bleedings

Overall, 28 studies (10,283 patients) reported on major bleeding. Major bleedings were nearly doubled in patients receiving high-dose heparin regimens compared to those receiving standard-dose (RR 1.94, 95% CI 1.47–2.56, I2 18.1%) (Fig. 1B). Low heterogeneity was observed. Metaregression analyses showed no influence of study deign (p = 0.272), hospital setting (p = 0.304), study quality (p = 0.994), dose of heparin prophylaxis (p = 0.880) on heterogeneity. A significant increase of major bleedings was confirmed in the subgroup analyses according to study design, in the ICU setting and when the therapeutic dose was used (Table 2, Fig. 2C, 2D, e-Table 3, e-Table 4 and e-Fig. 3) and when retrospective or low-quality studies were excluded.

Egger’s test did not reveal publication bias. The NNH was 50, 33 and 125 in the overall population, in the ICU and in the non-ICU patients, respectively. In the separate analysis of intermediate or therapeutic doses NNH was 40 and 59, respectively. The LHH was 1.5, 0.9, 1.6 in the overall population, in the ICU and in the non-ICU patients, respectively. In the separate analysis of intermediate or therapeutic doses LHH was 0.4 and 1.8, respectively.

Mortality

Overall, 21 among selected studies (7849 patients) reported data on mortality. All-cause death occurred in 17.7% of patients who received high-dose regimens and in 20.3% of patients receiving standard-dose (RR 0.84, 95% CI 0.68–1.03, I2 72.5%) (e-Fig. 4). Metaregression analyses showed no influence of study deign (p = 0.968), hospital setting (p = 0.340), study quality (p = 0.836), dose of heparin prophylaxis (p = 0.693) on heterogeneity.

Heterogeneity was reduced by excluding retrospective studies (RR 0.94, 95% CI 0.80–1.11, I2 32.1%) and remained unchanged after excluding low quality studies (RR 0.82, 95% CI 0.63–1.06, I2 79.2%). No differences were observed when subgroups by study design and clinical setting were analysed (Table 2, e-Table 3, e-Table 4, e-Fig. 5 and e-Fig. 6). In the non-ICU sub-analyses, heterogeneity persisted after exclusion of retrospective or low-quality studies, while in the ICU sub-analyses heterogeneity disappeared after the exclusion of retrospective studies (RR 1.04, 95% CI 0.92–1.16, I2 0.0%).

Egger’s test revealed publication bias. Data on fatal PEs were reported in 3 studies and no events were observed in the two treatment groups. Fatal bleedings were higher in the high-dose regimen (0.28%) compared to the standard-dose heparin prophylaxis (0.04%) but the difference was not statistically significant (Table 2).

Discussion

This meta-analysis showed that in patients hospitalized with COVID-19, high-dose prophylactic heparin (intermediate or therapeutic) was associated with significantly lower rates of VTE (risk reduced by 29%) compared to standard-dose, with the trade off a significant increase of MB (risk increased by 51%). In particular, rates of MB were significantly doubled when heparin was used at therapeutic compared to standard prophylaxis doses. The efficacy to safety profile of high-dose heparin regimens expressed as LHH was lower in ICU patients than in non-ICU patients (0.9 and 1.6, respectively) in comparison to standard heparin prophylaxis. For the high-dose heparin regimens, a significant 52% risk reduction in VTE and a non-significant 55% risk increase in MB was observed for non-ICU patients, while a significant 30% risk reduction of VTE and a significant 53% risk increase of MB was observed in ICU patients. No significant differences were observed for all-cause-mortality.

A high incidence of thromboembolic complications was reported in patients hospitalized for COVID-19 that seemed to persist despite the use of standard heparin prophylaxis [1, 48]. Indeed, SARS-CoV-2 infection can induce excessive and aberrant hyper-inflammatory host immune response that is associated with a so-called "cytokine storm" and a prothrombotic derangement of the hemostatic system [2]. This condition is mainly described in patients with critical COVID-19 disease. A close interconnection between thrombosis and inflammation is well known [2]. The two processes mutually reinforce each other and the net effect of the excess of thrombin generation and fibrinolytic shutdown may induce a profound hypercoagulable state. These processes have been shown to result in diffuse microthrombosis and endotheliitis of pulmonary vessels in patient with severe COVID-19 [2, 49]. These events can be the basis for the disproportionately high incidence of overt thromboembolic events (DVT, PE overt and incidental subsegmental, arterial events) associated with COVID-19 despite the use of standard thromboprophylaxis [3]. Based on these data, many clinicians started to use increased doses of heparin to treat patients with COVID-19 and several studies have been conducted aimed at assessing the benefit of this approach. Indeed, international societies refrained physician from using high-dose heparin for prevention of thromboembolism in the lack of evidence from clinical trials [5].

Our meta-analysis shows that increasing heparin dose is associated with a reduction in VTE. However, this result is obtained at the cost of increased incidence of bleeding complications. Additionally, we observed a non-significant increase in fatal bleedings with high-dose heparin in comparison to standard heparin prophylaxis.

Overall, these results are not surprising. The concept that increasing anticoagulant effect results in bleeding is already known in many clinical scenarios as acute coronary syndromes, ischemic stroke and prophylaxis of VTE in medical patients [3, 50, 51]. Moreover, in patients admitted for COVID-19, those in the ICU setting have higher bleeding risks than those in the non-ICU setting [10,11,12,13].

A pre-COVID-19 Cochrane Review on the role of heparin prophylaxis in more than 7,000 acute medical patients showed a 0.6% incidence of major bleeding with the use of heparin prophylaxis [51]. This rate is similar to that observed in our meta-analysis in non-ICU patients receiving standard heparin dose (0.9%). In our meta-analysis non-ICU patients seemed to receive the best risk reduction (52%) in VTE from high-dose heparin regimen, despite a quite low absolute event rate (2.2%) and at no significant increase in MB.

We observed no effect of high-dose vs. standard-dose of heparin in mortality in hospitalized patients for COVID-19. This result is clinically relevant as the majority of randomized studies assessing the efficacy and safety of the two heparin regimens had death or duration of need for organ support and not VTE as primary outcome. The rationale for this outcome is based on the concept that prevention of microthrombosis could prevent refractory ARDS and death. However, meta-analyses focused on mortality outcome showed controversial results. Parisi et al. reported an advantage of the therapeutic anticoagulation compared with the prophylactic anticoagulation especially in patients admitted to ICU (RR 0.30, 95% CI 0.15–0.60), while Ortega-Paz et al. showed no differences in all-cause death between the two heparin regimens (RR 0.96, 95% CI 0.78–1.18) [52, 53]. Some points should be underlined: in the first meta-analysis no RCTs were included, while the second consisted of only RCTs counting also a recent study comparing therapeutic rivaroxaban or heparin to standard-dose heparin prophylactic anticoagulation.

Differently from previous meta-analyses we focused on VTE as primary outcome and performed a strict selection of studies excluding those reporting on anticoagulants different from heparin. This may have led to a selection of a sicker population; however, the main findings were confirmed in the sub-analyses of the RCTs.

In our study, despite the beneficial effect of high-dose prophylactic heparin in preventing VTEs, no advantage on mortality was observed and a safety concern raised. This finding is in keeping with recent findings from three large RCTs. The first study by Sadeghipour et al. showed a not significant difference in the primary outcome (venous or arterial thrombosis, treatment with extracorporeal membrane oxygenation, or mortality) and no difference in VTE events when assessed separately in patients admitted to ICU who received heparin at intermediate dose compared to standard prophylaxis [10]. Similarly, Goligher et al. found that patients in the ICU setting who received therapeutic anticoagulation with heparin did not have a greater probability of organ support free days or survival compared to those receiving usual care pharmacologic thromboprophylaxis [11]. As expected, major thrombotic and PE events were reduced (6.4% vs 10.4%, and 2.5% vs 7.5%, respectively) and major bleedings increased (3.8% vs 2.3%) in patients who received therapeutic compared to those receiving standard heparin thromboprophylaxis. Interestingly, in patients in the non-ICU setting, therapeutic anticoagulation with heparin increased the probability of survival without organ support as compared with usual care thromboprophylaxis [12]. Major thrombotic events and PEs were reduced (1.1% vs 2.1%, and 0.9% vs 1.8%, respectively) with a not significant increase in the risk of MB (1.9% vs 0.9%) in patients receiving therapeutic heparin doses as compared with usual care thromboprophylaxis. Differently, in the study by Spyropoulos et al. therapeutic-dose LMWH reduced major thromboembolism and death (RR 0.68, 95% CI, 0.49–0.96) compared with standard heparin thromboprophylaxis in hospitalized patients with COVID-19 (but not in ICU patients) [13]. However, patients could have been included in case of very elevated D-dimer levels and only 257 patients were included out of the 11,694 screened (548 declined participation in the study).

Our results are in line with those of Valeriani et al. [54]. Differently from that meta-analysis we included both observational and RCTs. We believe our findings can increase knowledge on this topic and extend the results of previous studies including only RCTs. Indeed, previous meta-analyses on RCTs only, have reported conflicting results. For example, the use of full dose anticoagulation was suggested regardless of clinical setting in the study by Loffredo et al. [55], while it was discouraged by Sholzberg et al. [56] in critically ill patients.

Our study has some limitations in addition to those intrinsic to the meta-analytic approach, which combines heterogeneous datasets. First, the definition of standard thromboprophylaxis is heterogeneous among studies. In some studies [10,11,12,13, 20, 26, 30, 34, 40, 42] for a small number of patients, standard-heparin doses were adjusted in case of high body weight or BMI. In particular, in one study by the multiplatform [11] many patients in the standard-heparin group actually received intermediate-dose thromboprophylaxis upon admission to the ICU due to a change in the national (United Kingdom) practice guidelines during the trial. However, when this study was removed from our main analysis, results remained unchanged (e-Table 2). Similarly, no differences were observed with the main findings in the sub-analyses were data from the HEP-COVID trial were excluded [13]. In that study, 39% of the patients assigned to the standard heparin group received the intermediate-dose heparin prophylaxis. Second, as this is an aggregated data meta-analysis no adjustment for age, comorbidities and severity of disease, concomitant treatments were performed. Third, high heterogeneity was observed in the majority of the analyses on death, but this result should be regarded by taking into account the multiple causes of death, about 30% of these patients died due to respiratory failure and 30% due to sepsis. Moreover, in the VTE study outcome, metaregression analysis showed influence of heparin prophylaxis on heterogeneity. Indeed, in the separate analyses on intermediate and on therapeutic doses no heterogeneity was observed. Fourth, the open-label design may have introduced bias in the ascertainment of thrombotic events in the RCTs. Fifth, the search strategy may not be complete as some databases have been lost (eg EMBASE). However, references of the included studies were selected and only one study was found and added to the meta-analysis. At last, publication bias could not be excluded for mortality outcome, indeed study selection was focused on studies reporting VTE and bleeding events.

Some strengths of this meta-analysis that should be underlined include the following : (i) the selection of studies using heparin as thromboprophylaxis treatment; (ii) the focus on VTE as primary outcome; (iii) the high number of included studies and patients; (iv) the inclusion of the recently published RCTs; (v) the subgroups analyses of the different settings (ICU and non-ICU); (vi) additional data on fatal PEs and on fatal bleedings.

Conclusion

The use of heparin prophylaxis at high-dose reduces the risk of VTE but increased the risk of MB compared to the standard-dose. No clinical benefit for heparin high-dose was observed for ICU setting, but its role in the non-ICU should be further evaluated.