Incidence and Risk Factors Associated With Venous Thromboembolism After Orthopaedic Below-Knee Surgery

Reinout R.O. Heijboer, MD; Bart Lubberts, MD; Daniel Guss, MD, MBA; A. Holly Johnson, MD; Christopher W. DiGiovanni, MD


J Am Acad Orthop Surg. 2019;27(10):e482-e490. 

In This Article


Study Design and Participants

Institutional review board approval was attained before pursuing retrospective review of administrative billing data. Two hundred thirty-five Current Procedural Terminology codes and 86 International Classification of Diseases, Ninth Revision (ICD-9-CM), codes were used to identify orthopaedic patients who had undergone below-knee surgery (see Appendix 1, Supplemental Digital Content 1, Orthopaedic surgery was defined as the surgery performed by a US registered orthopaedic surgeon on patients who suffered from an injury or degenerative disease of the musculoskeletal system. Medical record data of patients were retrieved through our Research Patient Data Registry. The Research Patient Data Registry is a centralized clinical data registry that comprises diagnostic (ICD-9-CM) and billing (Current Procedural Terminology) codes, demographic information (eg, sex, date of birth, race), clinical encounters, laboratory values, surgical reports, and radiology records. We included all patients 18 years of age or older who underwent below-knee orthopaedic surgery during a 10-year period between August 2005 and August 2015 at any of three local Level-I trauma centers. Charts were assessed in the same trauma center where the patient received surgery. Exclusion criteria were (1) postoperative follow-up less than 90 days, (2) preoperative thromboembolic event on admission, (3) sustained polytrauma, defined as patients with concomitant trauma to any musculoskeletal area above the ipsilateral knee and/or other systematic injuries, (4) more than one below-knee surgical procedure within the 90-day follow-up period, and (5) nonorthopaedic surgery below the knee. A total of 23,206 patient charts were reviewed. Three thousand one hundred sixty-three of these patients met one of our exclusion criteria, leaving 20,043 patients for inclusion in the subsequent research analysis.

Included patients were divided into four surgical subgroups based on the location of the below-knee surgery: (1) patients who only received forefoot and/or midfoot surgery, (2) patients who only received hindfoot and/or ankle surgery, (3) patients who only received lower leg surgery, and (4) combination surgery—any combination of these procedural groups during a single surgery. Lower leg surgery was specifically defined as the surgery above the ankle but below the knee.

Outcome Measures and Explanatory Variables

The primary outcome variable of interest was the development of symptomatic VTE—specifically deep vein thrombosis (DVT) or pulmonary embolism (PE)—within 90 days of surgery. Because of concerns with the limited positive predictive value of ICD-9-CM codes for PE and DVT,[7] medical records of patients with one of these ICD-9-CM codes were manually reviewed by two research fellows and blinded for explanatory variables to assess whether symptomatic VTE occurrence fulfilled the predefined criteria. A thromboembolic event was defined as symptomatic when suspected DVT was confirmed by a thrombosis in the iliac, femoral, popliteal, or tibial veins with the use of Doppler ultrasonography or magnetic resonance venography, or when suspected PE was confirmed with helical CT or ventilation and perfusion scintigraphy. No patient had routine screening for thromboembolic events. The postoperative day count to any thromboembolic event and the 90-day mortality rate were also recorded.

Risk factors for thromboembolic events such as age at the time of procedure, sex, race, body mass index (BMI), comorbidities as per Charlson Comorbidity Index, tobacco use, known hypercoagulability (ie, factor V Leiden, prothrombin gene mutation, protein C/S deficiency, hereditary thrombophilia, anti-thrombin III deficiency, antiphospholipid syndrome), family history of cardiovascular disease, varicose veins, history of VTE, use of hormone replacement therapy or oral contraceptives within 3 months before surgery, and location of surgery (ie, forefoot/midfoot, hindfoot/ankle, lower leg, or some combination thereof) were considered.

The use of pharmacologic thromboprophylaxis either up to 7 days preoperatively or within 90 days postoperatively was elucidated from surgical reports, discharge reports, and medication lists and then subdivided into three therapy groups: (1) no medical thromboprophylaxis, (2) antiplatelet agent only, or (3) anticoagulant agent with or without an antiplatelet agent. Antiplatelet agents (n = 1,992) were COX inhibitors (96%; n = 1,917), adenosine diphosphate receptor inhibitors (1.9%; n = 37), or a combination of these two (1.9%; n = 38). Anticoagulant agents (n = 8,924) were factor Xa inhibitors (55%; n = 4,950), vitamin K antagonists (5.1%; n = 452), novel oral anticoagulant drugs (0.1%; n = 6), and a combination of factor Xa inhibitors, vitamin K antagonists, and aspirin (39%, n = 3,516). The use of postoperative intermittent lower limb compressive devices was also tracked, and these devices were used in 3,630 (18%) patients.

BMI was calculated using height and weight measurements documented within 1 year prior to surgery. Obesity was defined as BMI > 30. Tobacco use was subdivided into nonsmoker, former smoker, and current smoker. Age was subdivided into less than 40 years, 40 to 65 years, or older than 65 years. The comorbidity status was determined using the modified Charlson Comorbidity Index, a scoring algorithm based on 12 weighted comorbidities. We determined the modified Charlson Comorbidity Index through a previously described algorithm based on ICD-9-CM codes (see Appendix 2, Supplemental Digital Content 2, and Appendix 3,[8–10]

Statistical Analysis

Variables were presented with frequencies and percentages for categorical variables. All continuous variables reported were skewed and were summarized using medians and interquartile ranges (IQRs). To determine the explanatory variables associated with VTE, the Fisher exact test for dichotomous and the chi-squared test for categorical variables were performed. Kruskal-Wallis test was used to determine variation in VTE rates among the three prophylaxis groups (ie, no thromboprophylaxis, antiplatelets, anticoagulants) in high-risk patients. The incidence of symptomatic DVT and PE among prophylactic strategies and the risk ratio of mortality after developing a VTE event were calculated. Differences in patient characteristics between patients who did not receive thromboprophylaxis and patients who received an anticoagulant agent were calculated. Bivariate analyses were performed to create odds ratios (ORs) with 95% CI and P-values for associations between explanatory variables and symptomatic VTE. To adjust for possible confounding factors, multivariable analysis was used to identify the independent effect of variables on the risk of VTE. Variables with a P-value less than 0.10 in the bivariate analysis were included in the multivariable logistic regression analyses. We planned to perform a multivariable analysis for each intervention (ie, no pharmacologic thromboprophylaxis, antiplatelet agent, anticoagulant agent), but an insufficient number of patients receiving an antiplatelet agent limited this analysis to patients receiving no prophylaxis or an anticoagulant. The limited number of VTE events in the antiplatelet group caused concerns about the reliability of the multivariable analysis model, which would require at least 10 events per predictor.[11–13] All analyses were performed with Stata 13.0 (StataCorp, LP), and 2-tailed P-values less than 0.05 were considered significant.