Pooled prevalence of deep vein thrombosis among coronavirus disease 2019 patients.

To the editor,

The article by Ren et al. reported that there was an extremely high incidence (85.4%) of lower extremity deep venous thrombosis (DVT) among 48 patients with severe coronavirus disease 2019 (COVID-19) in Wuhan, China [1]. As the global pandemic of COVID-19, there have been several studies on the incidence, risk factors, and preventive strategies of DVT [1-4]. However, the incidence of DVT has been reported diversely among different clinical centers. Thus, we performed a meta-analysis to estimate the pooled prevalence of DVT in confirmed COVID-19 patients.

We searched PubMed, EMBASE, Web of Science, and medRxiv databases until June 22, 2020, for relevant studies, using the keywords (“coronavirus” or “COVID-19” or “SARS-CoV-2” or “2019-nCoV”) and (“thrombosis” or “thrombi” or “thrombus”). In addition, we screened out the relevant potential articles in the references of selected studies. Articles reporting the prevalence of DVT in confirmed COVID-19 patients were included.

The pooled prevalence and its 95% confidence interval (CI) were used to estimate the combined effects. We calculated the prevalence estimates with the variance stabilizing double arcsine transformation [5, 6]. The heterogeneity among studies was assessed with the I2 statistic and Cochran’s Q test. The meta-regression and subgroup analysis were used to investigate the potential heterogeneity sources (such as sample size, prevalence of prophylaxis in COVID-19 patients, location, design of studies, screening methods of DVT, and COVID-19 patients in intensive care unit (ICU)). We chose Egger’s test and Begg’s test to assess publication bias. All analyses were performed using the Stata 11.2 (StataCorp, College Station, TX), and a two-tailed P value < 0.05 was considered to be statistically significant.

A total of 1202 records were initially identified by our searches. We finally included 28 articles in our meta-analysis. The basic characteristics of included studies are shown in Table 1. There were 397 DVT cases in a total of 4138 COVID-19 patients. The pooled estimate of the prevalence for DVT was 16% by using a random-effects model (95% CI 10–23%, P < 0.01, I2 = 96.81, Q = 846.41, P < 0.01) (Fig. 1a). According to patients’ geographic location, the much higher pooled prevalence of DVT was found in COVID-19 patients from China (30%, 95% CI 2–72%, P = 0.02, I2 = 98.73%, Q = 313.90, P < 0.01) compared with those from western countries (13%, 95% CI 8–19%, P < 0.01, I2 = 95.62%, Q = 502.07, P < 0.01) (Fig. 1b). Twenty articles clearly reported the prevalence of DVT in COVID-19 patients treated in ICU or non-ICU. The pooled prevalence of DVT in COVID-19 patients treated in ICU was 23% (95% CI 11–38%, P < 0.01, I2 = 96.44%, Q = 421.29, P < 0.01), which was significantly higher than in COVID-19 patients treated in non-ICU (5%, 95% CI 1–11%, P < 0.01, I2 = 92.17%, Q = 89.42, P < 0.01) (Fig. 1c, d). We found significant publication bias by Egger’s test (P < 0.001) and Begg’s test (P < 0.001). The subgroup analysis showed that none of these factors could explain the significant heterogeneity. However, the meta-regression analysis of multiple covariates indicated that the geographic location of patients could partially explain heterogeneity (P = 0.036).

Table 1

Characteristics of the included studies

Authors	Sample	Age	Male (%)	Location	Design of studies	Screening of DVT	ICU/non-ICU*	Prophylaxis (%)	DVT (%)
Zhang et al. (PMID: 32421381)	143	63 (mean)	74 (51.7)	China	Cross-sectional study	Ultrasound	N/R	53 (37.1)	66 (46.2)
Ren et al. (PMID: 32412320)	48	70 (median)	26 (54.2)	China	Cross-sectional study	Ultrasound	ICU	47 (97.9)	41 (85.4)
Demelo-Rodríguez et al. (PMID: 32405101)	156	68.1 (mean)	102 (65.4)	Spain	Prospective study	Ultrasound	Non-ICU	153 (98.1)	23 (14.7)
Middeldorp et al. (PMID: 32369666)	198	61 (mean)	130 (65.7)	Netherlands	Retrospective study	Ultrasound	ICU/non-ICU	198 (100)	26 (13.1)
Bi et al.†	420	45 (mean)	200 (47.6)	China	Prospective study	N/R	N/R	N/R	6 (1.4)
Klok et al. (PMID: 32291094)	184	64 (mean)	139 (75.5)	Netherlands	Prospective study	Ultrasound	ICU	184 (100)	1 (0.5)
Karmen-Tuohy et al.‡	63	61 (mean)	57 (90.5)	USA	Prospective study	N/R	N/R	N/R	2 (3.2)
Llitjos et al. (PMID: 32320517)	26	68 (median)	20 (76.9)	France	Retrospective study	Ultrasound	ICU	8 (30.8)	14 (53.8)
Lodigiani et al. (PMID: 32353746)	388	66 (median)	264 (68.0)	Italy	Retrospective study	Ultrasound	ICU/non-ICU	307 (79.1)	6 (1.7)§
Helms et al. (PMID: 32367170)	150	63 (median)	122 (81.3)	France	Prospective study	Imaging	ICU	150 (100)	3 (2.0)
Stoneham et al. (PMID: 32423903)	274	N/R	N/R	UK	Prospective study	Imaging	N/R	N/R	5 (1.8)
Galeano-Valle et al. (PMID: 32425261)	785	N/R	N/R	Spain	Prospective study	Ultrasound	Non-ICU	780 (99.4)	13 (1.7)
Xing et al. (PMID: 32345353)	20	N/R	12 (60.0)	China	Retrospective study	Ultrasound	N/R	N/R	7 (35.0)
Beyls et al. (PMID: 32414510)	12	62 (median)	10 (83.3)	France	Retrospective study	Ultrasound	N/R	N/R	6 (50.0)
Poissy et al. (PMID: 32330083)	107	N/R	N/R	France	Retrospective study	Ultrasound	ICU	107 (100)	5 (4.7)
Beun et al. (PMID: 32311843)	75	N/R	N/R	Netherlands	Retrospective study	N/R	ICU	N/R	3 (4.0)
Cattaneo et al. (PMID: 32349132)	64	70 (median)	35 (54.7)	Italy	Retrospective study	Ultrasound	Non-ICU	64 (100)	0 (0.0)
Tavazzi et al. (PMID: 32322918)	54	N/R	N/R	Italy	Retrospective study	Ultrasound	ICU	54 (100)	8 (14.8)
Voicu et al. (PMID: 32479784)	56	N/R	42 (75.0)	France	Prospective study	Ultrasound	ICU	49 (87.5)	26 (46.4)
Hippensteel et al. (PMID: 32484907)	91	56.5 (mean)	53 (58.2)	USA	Retrospective study	Ultrasound	ICU	N/R	11 (12.1)
Fraissé et al. (PMID: 32487122)	92	61 (median)	73 (79.3)	France	Retrospective study	N/R	ICU	92 (100)	6 (6.5)
Desborough et al. (PMID: 32485437)	66	59 (median)	48 (72.7)	UK	Retrospective study	Imaging	ICU	66 (100)	6 (9.1)
Al-Samkari et al. (PMID: 32492712)	400	61.8 (mean)	228 (57.0)	USA	Retrospective study	Imaging	N/R	400 (100)	10 (2.5)
Edler et al. (PMID: 32500199)	80	79.2 (mean)	46 (57.5)	Germany	Prospective study	N/R	N/R	N/R	32 (40.0)
Grandmaison et al. (PMID: 32529170)	58	N/R	N/R	Switzerland	Cross-sectional study	Ultrasound	ICU/non-ICU	N/R	28 (48.3)
Artifoni et al. (PMID: 32451823)	71	64 (median)	43 (60.6)	France	Retrospective study	Ultrasound	Non-ICU	70 (98.6)	15 (21.1)
Nahum et al. (PMID: 32469410)	34	62.2 (mean)	25 (73.5)	France	Prospective study	Ultrasound	ICU	34 (100)	27 (79.4)
Zhang et al. (PMID: 32553905)	23	44.7 (mean)	15 (65.2)	China	Prospective study	N/R	ICU/non-ICU	N/R	1 (4.3)

DVT deep vein thrombosis, ICU intensive care unit, N/R not (clearly) reported

*Articles clearly reported the prevalence of DVT in COVID-19 patients treated in ICU or non-ICU

†doi: 10.1101/2020.04.22.20076190

‡doi: 10.1101/2020.05.07.20094797

§Data missing for patients

Fig. 1

Forest plots of pooled prevalence and its 95% confidence interval (CI) for deep vein thrombosis (DVT) in confirmed coronavirus disease 2019 (COVID-19) patients (a) and subgroup analysis by patients’ geographic location (b) and the severity of disease (c, d)

In conclusion, more attention should be paid to the prevention and clinical management of DVT, especially for COVID-19 patients in ICU, and timely assessment of DVT is essential. However, there was considerable heterogeneity in our meta-analysis. In addition, there was significant publication bias in our meta-analysis, although we searched four databases as many and as carefully as possible. Finally, we included non-survival patients who were seriously ill and may exaggerate the prevalence of DVT in COVID-19 patients.