The deep venous thrombosis (DVT), a major clinical manifestation of venous thromboembolism (VTE), is a common disease with an incidence of approximately 1 to 2 per 1000 persons per year.1 Patients with DVT have a high risk of developing pulmonary embolism (PE), a complication that is associated with high morbidity and mortality, especially in bedridden or hospitalized patients. Even though major progress has been made towards identifying the association of DVT with several acquired and genetic predisposing factors,2 up to 30 to 50% of the cases remain unexplained.3,4

Therefore, there is a need to identify novel laboratory and clinical correlates associated with DVT that can potentially predict its occurrence and assist in decisions regarding thromboprophylaxis. Consequently, during the last few decades, clinical research has focused on the identification of these correlates and the development of predictive models for VTE.5 Serving this purpose, parameters that are reliable, inexpensive and readily available in routine practice, such as hematological variables obtained with complete blood count (CBC) analysis, seem very promising and should be targeted. The role played by leukocytes and erythrocytes in the process of coagulation has been shown in recent studies6-9 and their presence is clearly observed in the anatomy of a venous clot, which consists of a laminar structure of erythrocytes and fibrin, permeated by large numbers of leukocytes.10 Platelets are also effective cellular elements in thrombus formation and play a major role in the pathogenesis of VTE.11

A number of previous studies have convincingly demonstrated an independent association between the high red cell distribution width (RDW) and DVT.12-17 However, only a limited number of studies have investigated other hematological parameters and their association with DVT. An increased focus on other hematologic variables may therefore be justified and may reveal novel predictors for DVT. Moreover, to the best of our knowledge, studies that have investigated the association of serial values of RDW and other hematologic variables with DVT have not been reported in the literature previously. Furthermore, only one previous study has investigated the association of RDW with the recurrence of DVT.16 Therefore, the aim of the present study was to investigate whether serial values of hematologic variables are associated with DVT among adult patients and also to evaluate the possible correlation between these hematologic variables and classical thrombotic risk factors.

The demographic and clinical characteristics of the study population at the baseline are depicted in Table 1. There was no significant difference between the two groups with regard to age, sex, BMI, frequency of obesity and smoking status (p > 0.05). The median duration (days) of DVT (interval between diagnosis and enrollment into the study) was 7 (IQR: 5 - 12).

Hematological variables of the study population are summarized in Table 2. Both cases and controls were comparable with respect to the RBC count, hemoglobin, hematocrit, MCV, MCH and MCHC at the baseline and all subsequent time points (p > 0.05). However, the RDW was found to be significantly higher in cases, compared to controls, at all time points. No significant differences were evident between the case and control groups, concerning the TLC, ANC, ALC and AMC, at any time point (p > 0.05). Similarly, the platelet count and MPV did not reveal significant differences between the groups at any time point (p > 0.05).

As the RDW was the only hematologic variable which expressed a statistically significant difference between cases and controls at all time points, we determined the percentage change in RDW levels (Δ%RDW) only. The median Δ%RDW (IQR) of cases vs. controls was 3.92 (0.78 - 7.86) vs. 1.37 (-1.60 - 2.52); p < 0.001 from baseline to 3 months, 0.45 (0.07 - 2.45) vs. 0.99 (-0.73 - 1.13); p = 0.677 from 3 months to 12 months, and 4.21 (1.17 - 7.92) vs. 1.45 (-0.98 - 2.63); p < 0.001 from baseline to 12 months.

The association between hematological parameters and the DVT was analyzed with logistic regression method and results of the univariate analysis are shown in Table 3. The only hematologic variable which showed a statistically significant association with DVT at all time points was the RDW.

As the RDW was the only hematologic variable which exhibited a statistically significant association with the DVT in the univariate analysis, we focused on the RDW for further analyses. Results of the univariate and multivariate analyses of the RDW are presented in Table 4. The results obtained in the univariate analysis were not changed to any major degree in the multivariate analysis and the RDW showed an independent association with the DVT at all time points. Moreover, when the RDW was analyzed as a dichotomous variable, high RDW (> 14.0%) compared to lower RDW (≤ 14.0%), it even showed a strong and consistent association with the DVT in both univariate and multivariate analyses at all time points.

When subgroup analysis was performed, the RDW continued to be associated with all subgroups of the DVT, classified according to the etiology and severity of the event, i.e., unprovoked DVT, provoked DVT, DVT alone and DVT plus PE in both univariate and multivariate analyses at all time points, exhibiting even higher odds ratios for high RDW (> 14.0%), compared to lower RDW (≤ 14.0%). Interestingly, the association between high RDW (> 14.0%) and the DVT at the baseline was found to be nearly 2-fold for provoked DVT, compared to unprovoked DVT, and also for DVT plus PE, compared to DVT alone, in both univariate and multivariate analyses (Table 4).

During the 12 months of the follow-up period after the initial event, 10 (6.6%) patients (8 with unprovoked and 2 with provoked incident DVT) experienced a recurrent DVT at different time intervals. Three patients among these were on anticoagulation at the time of recurrence. Interestingly, a significantly higher RDW (IQR) (%) was found at the baseline in patients who developed recurrent DVT during the follow-up than in patients without recurrence of DVT (16.7 [15.2 – 17.1] vs. 15.2 [13.6 – 16.4]; p < 0.001). Patients with high RDW (> 14.0%), compared to lower RDW (≤ 14.0%), at the baseline had a 2.3-fold increased association with the recurrence of DVT during the 12-month follow-up.

The median RDW (IQR) (%) at the baseline was 16.4 (15.4 - 16.9) vs. 15.1 (13.7 - 16.4); p < 0.05, 15.8 (14.9 - 16.3) vs. 15.1 (13.7 - 16.1); p > 0.05, 15.7 (15.4 - 15.9) vs. 15.3 (13.7 - 16.1); p > 0.05, 15.5 (14.8 - 15.9) vs. 15.4 (13.5 - 16.0); p > 0.05, 15.5 (14.0 - 16.2) vs. 15.4 (13.8 - 16.5); p > 0.05, 15.5 (14.3 - 16.0) vs. 15.4 (13.7 - 16.2); p > 0.05 and 15.9 (15.8 - 16.1) vs. 15.3 (13.7 - 16.4); p > 0.05 in patients with vs. without malignancy, surgery, trauma, pregnancy/postpartum state, immobilization, estrogen use and air travel, respectively.

A weak, but statistically significant positive correlation (r = 0.360; p = 0.001), has been shown at the baseline between high RDW and malignancy, while a negligible and statistically insignificant correlation was found between high RDW and other classic thrombotic risk factors (Table 5).

In the present case-control observational study, we demonstrated that the RDW was the only hematologic variable which showed a statistically significant association with the DVT. High RDW (> 14.0%) at the baseline was associated with the presence of DVT, independently of several potential confounders, including clinical and other hematological variables. Our results confirm and extend what has been reported previously in several studies, including small retrospective case-control studies,12,15,17 a large population-based case-control study13 and large population-based cohort studies,14,16 which showed a graded and significant independent association between the RDW and the presence of DVT. In addition, this association remained consistently strong as the strength of the association was not much attenuated when further analyses of the RDW were performed at 3 months and 12 months after the initial DVT event. To the best of our knowledge, this is the first study to report an association between the DVT and serial values of the RDW at different time points after the initial DVT event. In the majority of previous case-control studies, blood samples were collected only once after the thrombotic event, which did not allow for an analysis of potential changes in the RDW over time, and it was unclear whether high RDW (anisocytosis) was an actual risk factor or a consequence of the thrombotic event itself due to the accompanying prolonged inflammatory response, which is known to influence the RDW. In previous population-based cohort studies, the RDW was measured only at the baseline long before the thrombotic event, so it was not known whether results would be different if the RDW was measured during the acute phase. Likewise, RDW values might have fluctuated during the relatively long follow-up before occurrence of the thrombotic event and such fluctuations would tend to change the observed associations with the VTE. At variance with previous studies, our study measured the RDW at the baseline during the acute event and also serially thereafter at approximately every 3-month intervals for 12 months. Furthermore, the Δ%RDW was significantly higher in cases, compared to controls (p < 0.001), from the baseline to 3 months and from the baseline to 12 months, but not from 3 months to 12 months (p > 0.05). This finding indicated that that the association between the RDW and DVT was not much influenced by time after the thrombotic event and persisted even after the acute phase of the DVT had subsided. We were therefore able to rule out the possibility that the association between high RDW and the DVT was only due to the consequence of the thrombotic event itself, suggesting a possible causal relationship between high RDW and the DVT.

Moreover, when we performed the subgroup analysis, the RDW continued to be associated with all the different subgroups of the DVT, similar to other previous studies.13,15-17 Interestingly, we found a stronger association in patients with provoked DVT than unprovoked DVT, as well as in those with DVT plus PE, than in the DVT alone. Our findings are partly consistent with a previous study performed by Lippi et al., which showed that the association of increased RDW was greater in patients with provoked DVT and PE than in those with unprovoked thrombosis, both in unadjusted analysis.17 Ertop et al. found a significantly higher RDW in provoked VTE, compared to unprovoked VTE (p < 0.001), and showed that high RDW levels were related to both provoked and unprovoked VTE, but more significant in provoked VTE.18 Several studies showed that increased RDW levels were seen in cardiovascular disease and heart failure, stroke, chronic renal insufficiency, systemic inflammation, chronic obstructive pulmonary disease, malignancies, peripheral artery disease, pulmonary arterial hypertension and with diabetes-associated complications.18,19 The vast majority of these risk factors have been associated with provoked VTE. The patients who have these diseases, in addition to the effect of VTE itself, have some risk factors that also lead to an elevated RDW. In unprovoked VTE, elevation of the RDW could be caused mainly by the VTE itself, whereas in provoked VTE, the elevated RDW might not only be due to the VTE itself, but also to other concomitant risk factors, and this coexistence could have a synergy and elevate RDW levels more than expected.19

Furthermore, in our study, we found that the high RDW was associated with an increased risk of recurrence of DVT during the 12-month follow-up after the initial event. To the best of our knowledge, this is the second study to explore an association between the RDW and recurrent DVT in patients with the first episode of DVT. A large population-based cohort study performed by Ellingsen et al. was the first to investigate the impact of the RDW on the risk of recurrent VTE after incident VTE in a general population. The VTE patients were followed on average for 4.4 years after the initial event and they found that the RDW was not associated with the risk of VTE recurrence during the first year after the initial event.16 Other previous studies had not consisted of long-term follow-up of DVT patients to examine the effect of the RDW on the recurrence of DVT. Our study was probably underpowered to evaluate this association in view of the small number of events. Further large population-based and prospective studies with long follow-ups are required to confirm this association. It was also intriguing that the recurrent DVT, which occurred mainly in patients with unprovoked DVT (8 out of 10 events), had a higher RDW at the baseline, while at the same time, the provoked DVT had a higher RDW than unprovoked DVT. This could be due to a statistical artefact, or in other words, an unwanted consequence of a very low sample size in this subgroup analysis.

Additionally, in the current study, we found a negligible and statistically insignificant positive correlation between the high RDW and all classic thrombotic risk factors, except malignancy, which also showed a weak, although statistically significant, positive correlation. To our knowledge, this is the first study to investigate the correlation between the high RDW and various classic thrombotic risk factors.

We did not observe a statistically significant association between other CBC parameters and DVT. Various previous studies revealed contradictory results. Compatible with our study, Warny et al.20 did not observe a statistically significant independent association between the hematocrit and VTE. Braekkan et al. failed to exhibit an association between the DVT and MCV.21 Rezende et al. failed to show an association of the TLC, ANC and ALC with the DVT.13 Folsom et al. was not able to show a significant association between the platelet count and VTE.22 In contrast to our findings, other previous studies reported that the DVT was associated with a high RBC count,21 high hemoglobin level,22 high hematocrit level,21,22 high MCV,13 high MCH13 and high AMC.13

In this study, we evaluated the association between serial values of hematologic variables and the DVT. We demonstrated an independent and consistent association of high RDW with the first episode of DVT in adult patients. However, further investigations are needed to clarify a possible causal relationship between high RDW and the DVT, as reverse causation was a possibility. The study was probably underpowered to evaluate the association between high RDW and recurrent DVT in view of the small number of events. Further large population-based and prospective studies with long follow-ups are needed to investigate this association. Other hematologic parameters were not associated with the DVT.

RS, MM and GRP substantially contributed to the research design; GRP performed the study acquisition; GRP and SA analyzed the data and wrote the paper, and; RS, MM, GRP and SA critically revised and approved the final manuscript.