or

608

Coagulation and Fibrinolysis

Probability of developing proximal deep-vein thrombosis and/or pulmonary embolism after distal deep-vein thrombosis Andrei Brateanu1; Krishna Patel1; Kevin Chagin2; Pichapong Tunsupon1; Pojchawan Yampikulsakul1; Gautam V. Shah1; Sintawat Wangsiricharoen1; Linda Amah1; Joshua Allen1; Aryeh Shapiro1; Neha Gupta1; Lillie Morgan1; Rahul Kumar1; Craig Nielsen1; Michael B. Rothberg1 1Medicine

Institute, Cleveland Clinic, Cleveland, Ohio, USA; 2Quantitative Health Sciences, Cleveland Clinic, Cleveland Clinic, Cleveland, Ohio, USA

Summary Isolated distal deep-vein thrombosis (DDVT) of the lower extremities can be associated with subsequent proximal deep-vein thrombosis (PDVT) and/or acute pulmonary embolism (PE). We aimed to develop a model predicting the probability of developing PDVT and/or PE within three months after an isolated episode of DDVT. We conducted a retrospective cohort study of patients with symptomatic DDVT confirmed by lower extremity vein ultrasounds between 2001–2012 in the Cleveland Clinic Health System. We reviewed all the ultrasounds, chest ventilation/perfusion and computed tomography scans ordered within three months after the initial DDVT to determine the incidence of PDVT and/or PE. A multiple logistic regression model was built to predict the rate of developing these complications. The final model included 450 patients with isolated DDVT. Within three months, 30 (7 %) patients developed an episode of PDVT and/or PE. Only two factors Correspondence to: Andrei Brateanu MD Cleveland Clinic, Mail Code NA10 9500 Euclid Avenue Cleveland, OH 44195, USA Tel.: +1 216 407 4374 E-mail: [email protected]

predicted subsequent thromboembolic complications: inpatient status (OR, 6.38; 95 % CI, 2.17 to 18.78) and age (OR, 1.02 per year; 95 % CI, 0.99 to 1.05). The final model had a bootstrap bias-corrected c-statistic of 0.72 with a 95 % CI (0.64 to 0.79). Outpatients were at low risk (< 4 %) of developing PDVT/PE. Inpatients aged ≥ 60 years were at high risk (> 10 %). Inpatients aged < 60 were at intermediate risk. We created a simple model that can be used to risk stratify patients with isolated DDVT based on inpatient status and age. The model might be used to choose between anticoagulation and monitoring with serial ultrasounds.

Keywords Deep-vein thrombosis, distal deep-vein thrombosis, proximal deepvein thrombosis, pulmonary embolism

Received: June 20, 2015 Accepted after major revision: November 5, 2015 Epub ahead of print: December 10, 2015 http://dx.doi.org/10.1160/TH15-06-0503 Thromb Haemost 2016; 115: 608–614

Introduction

Methods

Distal deep-vein thrombosis (DDVT) of the lower extremities is confined to the infra-popliteal veins (peroneal, posterior tibial, anterior tibial veins). Observational studies of patients with isolated DDVT describe rates ranging from 0 to 31 % for subsequent proximal deep-vein thrombosis (PDVT) extension and 0 to 20 % for pulmonary embolism (PE) (1). However, the true incidence of these complications is not well known. Consequently, there is controversy over whether patients with DDVT should receive anticoagulation or only be monitored with serial ultrasounds of the lower extremities. The American College of Chest Physicians (ACCP) suggests that both approaches are acceptable (2). However, no guidance is offered as to which patients are at high risk or may benefit from anticoagulation. The primary objective of this study was to identify predictors of subsequent PDVT and/or PE within three months after an episode of isolated DDVT of the leg, and develop a model to risk stratify patients.

We conducted a retrospective cohort study of patients with isolated deep-vein thrombosis (DVT) diagnosed in the inpatient and outpatient settings between 2001 and 2012 in the Cleveland Clinic Health System (CCHS). CCHS is a not-for-profit multispecialty academic medical centre located in Cleveland, Ohio that includes eight community hospitals and 18 community-based health centres. CCHS uses a common electronic health record system (Epic Systems Corporation) allowing for the medical data to be easily extracted. The Cleveland Clinic serves an estimated population of > 1.5 million people, with 75 % of patients coming from the seven counties adjacent to Cleveland, and 25 % from other counties in Ohio, states or countries. We extracted from the electronic medical records (EMR) the results of the lower extremity vein ultrasounds ordered in patients with suspected DVT. The patient status was defined as outpatient when the initial diagnosis was made in the outpatient setting or emergency department and

© Schattauer 2016

Thrombosis and Haemostasis 115.3/2016 Downloaded from www.thrombosis-online.com on 2017-01-30 | ID: 1000466832 | IP: 128.205.114.91 For personal or educational use only. No other uses without permission. All rights reserved.

609

Brateanu et al. Thromboembolism after distal deep-vein thrombosis

the patient had follow-up visits with physicians employed by our health system, so the information could be extracted from our EMR. Similarly, the status was defined as inpatient when the diagnosis was made during a hospitalisation and follow-up visits were scheduled with our employed physicians. We reviewed all the ultrasound reports extracted from EMR and from this sample, all patients, 18 years or older, with the diagnosis of isolated DDVT of the lower extremities were included in the study. Patients with simultaneous PDVT, upper extremity DVT, or acute PE at the time of DDVT diagnosis, history of venous thromboembolism in the previous three months, and pregnant women were excluded from the study. For each patient we collected the demographic data (age, sex and race), alcohol consumption and cigarette smoking status, body mass index (BMI), number of chronic medical conditions considered to be risk factors for the thromboembolic disease (diabetes, varicose veins, nephrotic syndrome, inflammatory bowel disease, chronic obstructive pulmonary disease, congestive heart failure, history of ischaemic stroke or transient ischaemic attack), idiopathic versus secondary DDVT aetiology (active cancer, immobilisation, recent surgery in the past three months, or hormonal use), patient status at the time of diagnosis (inpatient vs outpatient) and following the diagnosis (home vs long-term care facility), number of distal deep veins affected, therapeutic anticoagulation use and duration of treatment (3, 4). We reviewed all the lower extremity ultrasounds, and chest computed tomography or ventilation-perfusion scans performed within three months of the initial DDVT episode to detect the incidence of PDVT, and/or PE, respectively. Blinding was not performed since this was a retrospective study. The Cleveland Clinic Institutional Review Board approved the study protocol (# 12–961).

Statistical analysis A multiple logistic regression model was built to predict the probability of developing PDVT and/or PE within three months after an isolated episode of DDVT. For the univariate analysis, the statistical difference was calculated using the Chi-square test for the categorical, and the Wilcox Sign Rank Test for the continuous variables, respectively. Determining the missing values was done randomly and multiple imputation using chained equations (MICE) was used to compute the missing data. The model started with 13 covariates and was assessed for collinearity and overly inflated estimates, including standard errors. Variables were removed from the analysis if they presented too much association with other independent variables or estimates were high (standard error above 0.4). Once a stable model was identified, variable reduction using a stepdown approximation as described by Frank E. Harrell was conducted (5). This process ranks the variables in order of importance starting from the full model and using a bootstrap bias-corrected concordance index as the stopping criteria. Overfitting was assessed on the final model using the shrinkage factor and the slope/intercept of the bootstrap predicted outcome. Since the purpose of the study was to create the most accurate predictive model, variables with individual p-values > 0.05 were left in the final

Figure 1: Study design.

model after the variable reduction process as they offered information to improve the overall model accuracy. The model’s performance was analysed using the concordance statistic (c-statistic) for measuring discrimination and calibration curves. All reported values were internally validated using 1000 bootstrap resamples to correct for model bias.

Results We reviewed 9,412 reports of lower extremity vein ultrasounds ordered in patients with suspected DVT, and identified 520 with isolated DDVT. We excluded 38 already receiving therapeutic anticoagulation for conditions other than thromboembolism, 22 inpatients that died before being discharged from the hospital, and 10 with no follow up information, leaving 450 patients in our final sample (▶ Figure 1). Baseline characteristics are presented in ▶Table 1. Within three months, 30 (7 %) patients developed a thromboembolic event (22 PDVT, 7 PE and 1 both PDVT and PE). Eleven events occurred within one week, and more than half (18) within one month after the initial DDVT (▶ Figure 2). In the bivariate analysis, only older age, inpatient status and discharge to a skilled nursing facility were associated with a thromboembolic event. In the multivariable analysis, two factors available at the time of initial DDVT diagnosis were found to contribute to the prediction model (▶ Table 2). Inpatient status and age increased the risk (odds ratio [OR], 6.38; 95 % confidence interval [CI], 2.17 to 18.78; p< 0.001, and OR, 1.02 per year; 95 % CI, 0.99 to 1.05; p= 0.19, respectively). The treatment of the isolated DDVT with therapeutic anticoagulation and the duration of treatment were associated with a


© Schattauer 2016

610


Variables

Total cohort

PDVT and/or PE within 3 months of an isolated DDVT

P-value

450

Yes = 30 (7 %)

No = 420 (93 %)

N (%) or Median (IQR)



61 (52, 71)

64 (58, 76)

60 (51, 71)

0.04

Caucasian

283 (63 %)

24 (79 %)

259 (77 %)

0.78

African American

83 (18 %)

6 (21 %)

77 (23 %)

235 (52 %)

17 (57 %)

218 (52 %)

0.62

60 (13 %)

2 (7 %)

58 (14 %)

0.40

162 (36 %)

10 (33 %)

152 (37 %)

0.67

28.8 (25, 33.3)

27.4 (22.9, 33)

28.9 (25, 33.3)

0.19

1 (0, 3)

1 (0, 2)

0.46 < 0.001

Table 1: Characteristics of the study population.

Age (years) Racea

Gender Male Active Smokerb Yes Active

Drinkingc

Yes BMId in kg/m2 Number of chronic medical

conditionse 1 (0, 2)

Patient status at the time of diagnosisf Outpatient

218 (48 %)

4 (13 %)

214 (52 %)

Inpatient

226 (50 %)

26 (87 %)

200 (48 %)

Idiopathic

118 (26 %)

8 (27 %)

110 (26 %)

Secondary

331 (74 %)

22 (73 %)

309 (73 %)

1 (1, 2)

1 (1, 2)

0.21

141 (31 %)

6 (20 %)

135 (32 %)

0.17 0.46

DDVT

aetiologyg

Number of distal deep veins initially

0.97

involvedh

1 (1, 2) Therapeutic anticoagulation after the isolated DDVT Yes Duration of

anticoagulationi

< 3 months

72 (16 %)

4 (14 %)

68 (17 %)

≥ 3 months

43 (10 %)

1 (3 %)

42 (11 %)

No anticoagulation

309 (68 %)

24 (83 %)

285 (72 %)

Patient status following the

diagnosisj

Home

324 (72 %)

16 (53 %)

308 (76 %)

Long-term care facility

109 (24 %)

14 (47 %)

95 (24 %)

< 0.01

Missing data for: a, 84 patients; b, 9 patients; c, 13 patients; d, 34 patients; e, 4 patients; f, 6 patients; g, 1 patient; h, 5 patients; i, 26 patients; j, 17 patients.

© Schattauer 2016


611


Figure 2: The cumulative incidence of the proximal deep-vein thrombosis (PDVT) and/ or pulmonary embolism (PE) events within three months after the isolated episode of distal deep-vein thrombosis (DDVT).

lower risk of a later PDVT and/or PE, but the association did not reach statistical significance. Patients who received treatment were more likely to be younger, heavier, and diagnosed with DDVT in the outpatient setting (▶ Table 3). The final model had a bootstrap bias-corrected c-statistic of 0.72 with a 95 % CI (0.64 to 0.79). The calibration of the model measuring the relationship between the model’s predicted probability against the actual probability is demonstrated in ▶ Figure 3. Outpatients had uniformly low rates of PDVT and/or PE (< 4 %). Inpatients aged ≥ 60 years were at high risk (> 10 %), whereas those aged < 60 years were at intermediate risk. Sensitivity analyses were performed and model was fitted excluding patients treated with anticoagulation. The direction of the relationship was maintained but the reliability was low due to the small number of proximal PDVT and/or PE events.

Discussion Isolated DDVT of the lower extremities can be associated with subsequent episodes of PDVT and/or acute PE. We found that a simple model consisting of inpatient status and age could identify patients at high risk of subsequent thromboembolism. To our knowledge, this is the first reported tool to predict the probability of developing PDVT and/or PE within three months of an isolated episode of DDVT. This model has the advantage of using only two clinical predictors easily obtained at the time of the initial event, and providing the associated risk of a subsequent thromboembolic complication.

In our study, the prevalence of isolated DDVT in patients with suspected DVT was approximately 5 %, in line with the lower end of the spectrum of previously reported rates of 4–15 % (6, 7). The incidence of subsequent PDVT/PE in these patients was 7 %. This rate is comparable to that reported in a recent meta-analysis (5-year cumulative risk of 7.6 %) (8). The risk factors for developing symptomatic isolated DDVT and proximal DVT are different. While the DDVT is more often associated with transient risk factors such as surgery, immobilisation, travel, proximal DVT is associated with chronic medical conditions and age > 75 years, respectively (9). However, little is known about the predictors of venous thromboembolism recurrence after an isolated DDVT. One study looking only at patients with a first isolated distal DVT who did not have a cancer at the time of diagnosis found that age > 50 years, unprovoked DDVT, and number of distal veins involved influences the risk of recurrence (10). The increased risk of recurrence in patients with unprovoked versus provoked DVT was most likely the result of inTable 2: Multivariable logistic regression analysis.

Variables

Odds Ratio (95 % CI)

P-value

Intercept

< 0.01 (0.0008–0.0437)

< 0.0001

Patient status at the time of diagnosis Outpatient Inpatient Age

1.00 6.38 (2.17–18.78)

< 0.001

1.02 (0.99–1.05)

0.19


© Schattauer 2016

612


Table 3: Patients’ characteristics by anticoagulation treatment.

Variables

Patients treated with anticoagulation after the isolated DDVT

P-value

What is known about this topic?

•

Yes = 141 (31 %) No = 309 (69 %) N (%) or Median (IQR)


57 (49, 67)

63 (53, 73)

< 0.001

100 (76 %)

183 (78 %)

0.74

•

Age (years) Racea Caucasian

African American 31 (24 %) Gender Male Active Active

166 (84 %)

0.35

17 (12 %)

43 (14 %)

0.51

67 (48 %)

95 (32 %)

< 0.01

29.7 (26.0, 35.5)

28.4 (24.3, 32.9)

< 0.01

1 (0,2)

0.04 < 0.01

in

kg/m2

Number of chronic medical

conditionse

1 (0, 2) Patient status at the time of

diagnosisf

Outpatient

83 (61 %)

135 (44 %)

Inpatient

54 (39 %)

172 (56 %)

Idiopathic

23 (16 %)

95 (31 %)

Secondary

118 (84 %)

213 (69 %)

DDVT

•

A simple model consisting of inpatient status and age can identify patients at high risk of subsequent thromboembolism. Patients with isolated DDVT can be risk stratified. Outpatients do not appear to require treatment; inpatients < 60 years of age could be monitored with serial ultrasounds. For inpatients aged ≥ 60 years, anticoagulation may be prudent.

Drinkingc

Yes BMId

69 (49 %)

Smokerb

Yes

What does this paper add?

•

52 (22 %)

Isolated distal deep-vein thrombosis (DDVT) of the lower extremities can be associated with subsequent episodes of proximal deep-vein thrombosis (PDVT) and/or acute pulmonary embolism (PE). However, the true incidence of these complications is not well known. There is controversy over whether patients with DDVT should receive anticoagulation or only be monitored with serial ultrasounds of the lower extremities. The American College of Chest Physicians suggests that both approaches are acceptable.

etiologyg < 0.01

Number of distal deep veins initially involvedh 1 (1, 2)

1 (1, 2)

0.67

110 (81 %)

214 (72 %)

0.05

26 (19 %)

83 (28 %)

Patient status following the Home Long-term care facility

diagnosisj

Missing data for: a, 84 patients; b, 9 patients; c, 13 patients; d, 34 patients; e, 4 patients; f, 6 patients; g, 1 patient; h, 5 patients; j, 17 patients.

cluding only transient risk factors (such as fracture of lower extremity, immobilisation for more than three days and surgery in the previous three months) in the analysis. In our study, age was also found to be associated with subsequent PDVT/PE, but the number of distal veins involved and DDVT aetiology were not. The discrepancy could be explained by the inclusion in our study of patients with an active malignancy, such as breast, prostate, pancreatic, lung, bladder, rectal, colon cancer, as well as leukaemia and lymphoma. Generally, the risk of recurrent VTE is increased twoto four- fold in patients with an active cancer (11), comparable © Schattauer 2016

with the risk of recurrence after an unprovoked DVT (hazard ratio [HR], 3.1; 95 % CI, 1.4–6.9) reported by Galanaud in the OPTIMEV study (10). We found that patients diagnosed in the inpatient setting with DDVT have a higher chance of a subsequent thromboembolic event. The fact that majority of these patients, while in the hospital, receive preventive antithrombotic measures that include heparin could explain why inpatients were less likely to be treated with full dose anticoagulation when compared with patients seen in the outpatient setting. Yet, therapeutic anticoagulation after the isolated DDVT was not associated in the final model with the risk of recurrence. The mechanism for the association between inpatient status at the time of diagnosis and the increased risk of recurrence could be explained by the presence of a hypercoagulable state that predisposes the inpatients to develop the thrombotic index event, even in the presence of prophylactic doses of anticoagulation. Alternatively, the association might be explained by the fact that in the inpatient setting patients have a persistence of the risk factors, such as immobilisation. In line with our results, a retrospective analysis of only hospitalised patients with isolated DDVT found the rate of progression to PDVT to be higher (12) than that reported from studies including patients from the outpatient setting (13). In the univariate analysis, patients discharged to a long-term care facility following the DDVT diagnosis had a higher risk of VTE recurrence than patients sent home. However, this association was not maintained on the multivariate analysis. The optimal approach for the follow up and treatment of isolated DDVT remains controversial. ACCP recommends either a two-week follow-up with serial ultrasounds or treatment with anticoagulation (2). In our study, anticoagulation treatment, regardless of duration, was not associated with a lower risk of subsequent PDVT/PE. However, our sample was too small to conclude that anticoagulation is not helpful. For patients at high risk of subThrombosis and Haemostasis 115.3/2016

Downloaded from www.thrombosis-online.com on 2017-01-30 | ID: 1000466832 | IP: 128.205.114.91 For personal or educational use only. No other uses without permission. All rights reserved.

613


Figure 3: The calibration of the model, which measures the relationship between the model’s predicted probability against the actual probability. C-Statistic: 0.72 with 95 %CI (0.64, 0.79).

sequent thromboembolism (i. e. inpatients age ≥ 60 years), anticoagulation may be prudent. In contrast, for patients at low risk (i. e. outpatients), no treatment would appear necessary. For the remaining patients, serial ultrasounds may be reasonable, at least in the first month, when the risk is highest. There are few limitations of our study. The retrospective nature and the relatively small sample size can lead to bias and errors. To reduce the potential errors, all the radiology reports were fully reviewed to identify the DDVT diagnosis and the charts of all patients with the index diagnosis were reviewed. In addition, the small sample size might represent an atypical patient population with high rate of subsequent events. However, the majority of patients were diagnosed with DDVT or PDVT/RE when symptomatic, therefore the true incidence of the events could have been underestimated. Similarly, our model cannot be applied to asymptomatic patients. The model itself has also some limitations, tending to underestimate the low and intermediate risk patients and overestimate the high-risk ones. In conclusion, we created a simple tool to risk stratify patients with isolated DDVT. Based on our findings, outpatients appear to have a lower risk of developing proximal PDVT and/or PE. Depending on the individual’s risks and benefits of anticoagulation, treatment may not be necessary; inpatients < 60 years of age are at intermediate risk, and could be monitored with serial ultrasounds.

For inpatients aged ≥ 60 years, who are at high risk of developing PDVT and/or PE, anticoagulation may be prudent. Yet, before it can be routinely used in clinical practice, the model needs prospective validation, ideally followed by a management trial. Author contributions

Andrei Brateanu MD, takes responsibility for the content of the manuscript, including the data and analysis. Andrei Brateanu MD, Krishna Patel MD, Kevin Chagin MS, Gautam Shah MD and Michael Rothberg MD, MPH had full access to all of the data in the study and take responsibility for the integrity of the data and accuracy of the data analysis, including and especially any adverse effects. Andrei Brateanu MD, Krishna Patel MD, Kevin Chagin MS, Pichapong Tunsupon MD, Pojchawan Yampikulsakul MD, Gautam V Shah MD, Sintawat Wangsiricharoen MD, Linda Amah MD, Joshua Allen MD, Aryeh Shapiro MD, Neha Gupta MD, Lillie Morgan MD, Rahul Kumar MD, Craig Nielsen MD and Michael B. Rothberg MD, MPH contributed substantially to the study design, data analysis and interpretation, and the writing of the manuscript. Conflicts of interest

None declared.


© Schattauer 2016

614


References 1. De Martino RR, Wallaert JB, Rossi AP, et al. A meta-analysis of anticoagulation for calf deep venous thrombosis. J Vasc Surg 2012; 56: 228–237 e1. 2. Kearon C, Akl EA, Comerota AJ, et al. Antithrombotic therapy for VTE disease: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest 2012; 141 (2 Suppl): e419S-94S. 3. Ciuti G, Grifoni E, Pavellini A, et al. Incidence and characteristics of asymptomatic distal deep vein thrombosis unexpectedly found at admission in an Internal Medicine setting. Thromb Res 2012; 130: 591–595. 4. Anderson FA, Jr., Spencer FA. Risk factors for venous thromboembolism. Circulation 2003; 107 (23 Suppl 1): I9–16. 5. Harrell FE. Regression modeling strategies: With applications to linear models, logistic regression, and survival analysis. 45. New York: Springer; 2001. pp. 1–67. 6. Palareti G, Schellong S. Isolated distal deep vein thrombosis: what we know and what we are doing. J Thromb Haemost 2012; 10: 11–19.

© Schattauer 2016

7. Galanaud JP, Quenet S, Rivron-Guillot K, et al. Comparison of the clinical history of symptomatic isolated distal deep-vein thrombosis vs. proximal deep vein thrombosis in 11 086 patients. J Thromb Haemost 2009; 7: 2028–2034. 8. Baglin T, Douketis J, Tosetto A, et al. Does the clinical presentation and extent of venous thrombosis predict likelihood and type of recurrence? A patient-level meta-analysis. J Thromb Haemost 2010; 8: 2436–2442. 9. Galanaud JP, Sevestre-Pietri MA, Bosson JL, et al. Comparative study on risk factors and early outcome of symptomatic distal versus proximal deep vein thrombosis: results from the OPTIMEV study. Thromb Haemost 2009; 102: 493–500. 10. Galanaud JP, Sevestre MA, Genty C, et al. Incidence and predictors of venous thromboembolism recurrence after a first isolated distal deep vein thrombosis. J Thromb Haemost 2014; 12: 436–443. 11. Fahrni J, Husmann M, Gretener SB, et al. Assessing the risk of recurrent venous thromboembolism – a practical approach. Vasc Health Risk Manag 2015; 11: 451–459. 12. Sales CM, Haq F, Bustami R, et al. Management of isolated soleal and gastrocnemius vein thrombosis. J Vasc Surg 2010; 52: 1251–1254. 13. Schwarz T, Buschmann L, Beyer J, et al. Therapy of isolated calf muscle vein thrombosis: a randomized, controlled study. J Vasc Surg 2010; 52: 1246–1250.
