Background: We are not aware of any previous studies that have compared the rate of venous thromboembolic events in patients who underwent prophylactic intramedullary nailing because of an impending fracture with the rate in patients who underwent intramedullary nailing after a pathological fracture. The objective of the present study was to determine if the rate of venous thromboembolic events varies between patients who are managed with prophylactic fixation and those who are managed with fixation after a pathological fracture.
Methods: We performed a retrospective comparative study in which the Nationwide Inpatient Sample database was used to identify all patients who had undergone femoral stabilization, either for a pathological femoral fracture or for prophylactic fixation of femoral metastatic lesion, over a period of 10 consecutive years (between 2002 and 2011) in the United States. Demographic data, comorbidities, venous thromboembolic event rates, and other common postoperative complications were compared between the 2 groups.
Results: Patients who were managed with prophylactic fixation had significantly higher rates of pulmonary embolism (p < 0.001; adjusted odds ratio, 2.1) and deep-vein thrombosis (p = 0.03; adjusted odds ratio, 1.5). Patients who were managed with fixation after a pathological fracture had a significantly greater need for blood transfusion, higher rates of postoperative urinary tract infection, and a decreased likelihood of being discharged to home (p < 0.001 for all).
Conclusions: Patients with metastatic disease who undergo prophylactic intramedullary nailing have higher observed rates of venous thromboembolic events than those who undergo nailing for the treatment of a pathological fracture and should be actively and vigilantly managed in the postoperative period.
Level of Evidence: Therapeutic Level III. See Instructions for Authors for a complete description of levels of evidence.
Venous thromboembolism (VTE), which includes deep-vein thrombosis (DVT) and pulmonary embolism (PE), is a serious complication in hospitalized patients with cancer, yet there remain substantial knowledge gaps with regard to thromboprophylaxis and VTE treatment in this high-risk group1. Patients with cancer who undergo surgical procedures, especially lower-extremity orthopaedic procedures, are twice as likely to have postoperative DVT and 3 times more likely to have fatal PE than patients without cancer who undergo the same procedures2,3.
Prophylactic intramedullary nailing (IMN) for the treatment of pathological lesions involving long bones prior to fracture has been associated with improved outcomes compared with IMN for the treatment of pathological fractures4-6. Although multiple postoperative variables have been evaluated in studies of patients who have been managed with internal fixation for the treatment of metastatic long-bone disease, we are not aware of any large population-based studies that have compared the rate of VTE between patients undergoing IMN before and after a pathological fracture. PE is such a rare phenomenon that a very large sample size is required to detect a true significant difference. The purpose of the present study was to evaluate differences in the rates of postoperative DVT and PE between these groups with use of a large national database.
Materials and Methods
Design and Setting
A retrospective comparative study was performed with use of the Nationwide Inpatient Sample (NIS) database. This database, which is the largest all-payer inpatient database in the nation, includes patient records from approximately 8 million admissions per year at about 1,000 hospitals in the United States, an approximate 20% stratified sample. The NIS database is part of the Healthcare Cost and Utilization Project (HCUP). The database was queried to collect perioperative data on all patients who underwent treatment either for a pathological femoral fracture or for prophylactic fixation of a femoral metastatic lesion in the United States between January 2002 and December 2011.
The inclusion criteria for the group of patients who underwent prophylactic fixation of the femur (Group A) included a combination of an ICD-9 (International Classification of Diseases, Ninth Revision) procedure code of 78.55 (internal fixation of the femur without fracture reduction) along with an ICD-9 diagnosis code of 198.5 (secondary malignant neoplasm of bone). Patients were excluded if they had a secondary diagnosis code of 733.14-15 (pathological fracture of the femur) in order to provide a stricter criterion for selecting patients with impending fractures.
The inclusion criteria for the group of patients who underwent surgical fixation of a pathological femoral fracture (Group B) included a combination of an ICD-9 procedure code of 79.15 (closed reduction and internal fixation of the femur) along with an ICD-9 diagnosis code of 733.14-15 (pathological fracture of the femur). Patients with an ICD-9 procedure code of 79.35 (open reduction and internal fixation of the femur) were excluded in order to select for intramedullary fixation as opposed to treatment with a plate-and-screw construct.
All patients who were included in the present study underwent IMN fixation for the treatment of a metastatic femoral lesion. Surgical indications, preoperative workup, operative technique, and postoperative protocol were at the discretion of each treating surgeon or medical team and were not reflected in the query.
The main outcome measures that were investigated in the present study were the rates of inpatient postoperative PE, DVT, death, pneumonia, myocardial infarction, cerebrovascular accident, urinary tract infection, and blood transfusion. The average length of hospital stay and discharge disposition were also compared between the 2 groups. Patient-specific covariates included age, sex, race, median household income (classified by zip code), primary payer source, and the prevalence of preexisting medical comorbidities (alcoholism, chronic anemia, congestive heart failure, chronic pulmonary disease, coagulopathy, depression, diabetes, drug abuse, hypertension, hypothyroidism, liver disease, neurological disorders, obesity, peripheral vascular disease, and chronic renal failure) were also compared.
Statistical Methods and Study Size
The differences in covariate data, comorbidity data, and main outcome measures for the 2 groups were analyzed. Two-sample independent t tests were used to compare linear variables, and chi-squared tests were used to compare categorical data. Multivariate analysis taking into consideration differences in age, sex, and preexisting medical comorbidities between the 2 groups was used to adjust for the differences in potential confounding variables. Any variables associated with a p value of < 0.10 in the initial univariate analysis were subsequently incorporated into the multivariate analysis to achieve a higher sensitivity in identifying potential risk factors associated with higher rates of perioperative complications. For the multivariate analysis, the level of significance was set at p < 0.05. A post hoc power analysis was performed with use of the case distribution, total number of patients, and overall VTE rates in the present study. A statistical power of 80% with an alpha of 0.05 was found to detect a ≤1% difference in the overall VTE rate between the prophylactic fixation and pathological fracture fixation groups. All calculations were performed with use of the SPSS software package (version 22; IBM).
Participants and Descriptive Data
Group A included 2,750 patients with a mean age of 65.8 years, and Group B included 2,829 patients with a mean age of 71.9 years. The patients in Group A were, on the average, significantly younger than those in Group B (p < 0.001). Group A was 42.5% male and 57.5% female, whereas Group B was 29.5% male and 70.5% female. There was a higher proportion of female patients in the pathological fracture group and a higher proportion of male patients in the prophylactic fixation group (p < 0.001). There was an even distribution of race and average income between the 2 groups (p = 0.36 and p = 0.21, respectively). Group B had a higher percentage of Medicare patients than Group A (68.9% compared with 55.3%; p < 0.001). The average number of comorbidities was not significantly different between the 2 groups (p = 0.88). A detailed comparison of demographic data is provided in Table I.
PE and DVT Rates
The rates of PE and DVT differed between the prophylactic fixation and pathological fracture groups. Univariate analysis indicated that the prophylactic fixation group had a significantly higher rate of PE than the pathological fracture group (2.1% compared with 1.2%; p = 0.008), with an odds ratio of approximately 2.0. Multivariate analysis also demonstrated a greater risk of PE in the prophylactic fixation group, with an adjusted odds ratio of 2.1 (p = 0.001) (see Appendix). Univariate analysis indicated that the prophylactic fixation group had a significantly higher rate of DVT than those in the pathological fracture group (3.2% compared with 2.2%; p = 0.03). Multivariate analysis also demonstrated a greater risk of DVT in the prophylactic fixation group, with an adjusted odds ratio of 1.5 (p = 0.03) (see Appendix). The results of the univariate and multivariate analyses are presented in Table II and Table III, respectively.
While the average number of comorbidities did not differ between the 2 groups (p = 0.88), the types of comorbidities varied significantly. Patients in the prophylactic fixation group had significantly lower rates of chronic anemia (p < 0.001), congestive heart failure (p < 0.001), coagulopathy (p = 0.001), hypertension (p < 0.001), hypothyroidism (p < 0.001), liver disease (p = 0.005), neurological disorders (p < 0.001), obesity (p = 0.006), and chronic renal failure (p < 0.001). The rates of alcoholism, chronic pulmonary disease, depression, drug abuse, and peripheral vascular disease tended to be lower in the prophylactic fixation group but did not differ significantly between the 2 groups (Table IV).
Univariate analysis indicated that the prophylactic fixation group had significantly fewer postoperative complications overall (p < 0.001), including postoperative urinary tract infections (8.5% compared with 15.7%) and blood transfusions (21.2% compared with 28.6%), and were more likely to be discharged to home (58.4% compared with 30.2%) (Table II). The rates of death, pneumonia, cardiovascular accident, and myocardial infarction were not significantly different between the 2 groups. Multivariate analysis (Table III) reaffirmed significantly lower adjusted odds ratios for blood transfusion (0.74) (see Appendix) and non-home discharge (0.41) (see Appendix) for the patients in the prophylactic fixation group. Multivariate analysis did not demonstrate differences between the 2 groups in terms of the rate of death, cost of hospitalization, or length of stay (see Appendix).
Internal Validation Study
The methods used in the present study were internally validated across a cohort of patients at our own institution (Tables V and VI). The purpose was to determine the accuracy of identifying the surgical group on the basis of ICD-9 diagnosis and procedure coding as compared with the information in the operative report. First, we identified all patients who had undergone internal fixation of the femur without fracture reduction (ICD-9 procedure code, 78.55; n = 101) and closed reduction and internal fixation of the femur (ICD-9 procedure code, 79.15; n = 56) between January 1, 2010, and December 31, 2013. Next, the inclusion diagnosis codes were applied, leaving 41 qualified patients in the prophylactic fixation group and 24 qualified patients in the pathological fixation group. When cross-referenced with medical records, the accuracy was 85% in the prophylactic fixation group and 83% in the pathological fixation group.
The occurrence of a PE is a rare phenomenon even in patients who have cancer, with historical rates in the early postoperative period following mechanical prophylaxis alone ranging from 0.4% to 3%7-11. Historically, DVT rates in patients who have had resection of a musculoskeletal tumor without chemical prophylaxis have tended to be much higher, with a reported prevalence of 40% to 80% for calf vein thrombosis and 10% to 20% for proximal vein thrombosis2,12. The present study demonstrated that patients who underwent prophylactic nailing because of an impending pathological femoral fracture tended to be younger and healthier than those who were managed with nailing after a fracture has occurred. Patients who underwent IMN after a pathological fracture also had more postoperative complications. Surprisingly, despite the aforementioned lower rates of serious medical comorbidities and postoperative complications, the patients in the prophylactic fixation group had significantly higher rates of PE and DVT.
Limitations of Present Study
The present study had a number of limitations, some of which are inherent to retrospective database analyses. Retrospective database analyses provide a very limited picture of only in-hospital events that only permits inference of the association between exposure and disease13. The present study demonstrated a significant difference between the 2 groups in terms of DVT and PE rates but cannot explain why the difference exists. Also, these analyses are subject to coding bias13 because certain fragility fractures may be miscoded by the treating surgeon as pathological fractures, with osteoporosis being the underlying pathological process. To address this potential bias, a validation study was conducted at our institution to cross-reference ICD-9 codes in the NIS with medical records; that study demonstrated that the accuracy of classification and diagnosis of metastatic disease to be acceptable (85% in Group A and 83% in Group B). Previous studies using the NIS database have also demonstrated acceptable accuracy and quality control14,15. If the pathological fracture group were “contaminated” with fragility fractures in osteoporotic patients, one would expect a lower incidence of comorbidities compared with the prophylactically-treated patients because osteoporotic patients tend to be healthier, on the average, than those with metastatic disease.
In addition, several important clinical variables that impact the rates of VTE were unreported in the database, including histological diagnosis, the size of lesion and/or burden of disease, adjuvant chemotherapy or radiotherapy, surgical details (e.g., whether polymethylmethacrylate was utilized or whether reaming was performed prior to nail insertion)16,17, and anticoagulation status. Certain types of malignant lesions, such as lung cancers, are associated with a higher risk of venous thrombosis or embolism8,18. A large tumor burden is associated with higher levels of procoagulant factors12,19-21 and a higher likelihood of thrombotic events22. Chemotherapeutic agents and hormonal treatments used for certain types of metastatic disease contribute to the risk of thrombosis23-26. Similarly, radiotherapy is also associated with DVT when it is used to treat lymphoma and certain gynecological malignant lesions27,28. Therefore, it is possible that 1 of the treatment groups might have had a larger proportion of patients with these confounders.
Another limitation was the lack of specific anticoagulation data during hospitalization and after discharge. Shallop et al.18 demonstrated that patients with metastatic bone lesions had a high rate of VTE after intramedullary nail fixation (7.1%; 24 events in 336 impending or pathological fractures) despite current anticoagulation protocols. Low-molecular-weight heparin was the most commonly used chemoprophylaxis in their overall cohort and in the subset of 19 patients who had a DVT or PE during the perioperative period. Of those 19 patients, 18 had undergone prophylactic fixation because of an impending fracture while 1 had fixation of a pathological fracture, which corroborates our findings. A thorough multivariate analysis was performed in the present study to adjust for the large number of potential confounders. Our large sample size ensures that mean values will not be greatly influenced by coding bias, minimizing type-II error.
Because the database only included information on perioperative complications that occurred during the index admission, we were unable to determine late complication rates or post-discharge mortality and VTE rates. In the study by Shallop et al.18, the majority of VTE cases (74%; 14 of 19) that occurred after IMN fixation for the treatment of metastatic bone lesions occurred during the first 15 days after surgery. Therefore, the results of the present study cannot be generalized to the overall rate of VTE or the long-term risk of VTE after surgery. The NIS database further limits the ability to assess the associated postoperative complications and outcome measures diagnosed after hospital discharge. Therefore, the presented rates of DVT, PE, and death are likely underestimated and may potentially be even higher in both groups with a longer follow-up period. Nevertheless, these results are valuable for examining immediate postoperative rates of VTE and complications after IMN for the treatment of pathological lesions.
The weakness of the present study is the inability to provide definitive explanation for why the prophylactic fixation group had higher rates of DVT and PE. Overall, patients with metastatic femoral disease likely have compromised lung function before surgery as well as multiple medical comorbidities in addition to possible pulmonary parenchymal metastatic disease. Such patients also have hypervascular tumor foci in bone with a larger surface area of vascularity that can intravasate activated inflammatory mediators, fat emboli, and hypercoagulable cancer cells, placing them at higher risk for pulmonary complications. The adult reconstruction literature has demonstrated an association between fat and marrow embolism and VTE, with some authors theorizing that marrow fat intravasation with insertion of the femoral component during total hip arthroplasty leads to endothelial injury and the onset of a coagulation cascade29,30. In the case of metastatic bone lesions, reaming or instrumentation of an intact femur increases the intramedullary pressure and may result in intravasation of fat and thrombogenic material into the systemic circulation, leading to increased endothelial damage and hypercoagulability. This phenomenon has been observed in patients managed with prophylactic IMN for the treatment of impending fractures from metastatic lesions31,32 and in previous cadaveric studies33,34. Accordingly, Wilkens et al.35 suggested making a vent hole to decompress the femoral canal prior to nailing the femur. However, this technique has not been widely investigated clinically.
The higher observed rates of VTE in the prophylactic fixation group in the present study should not dissuade surgeons from prophylactic stabilization of an impending pathological fracture but rather should serve as an important reminder to remain vigilant about VTE prevention. Patients who present with impending fractures for prophylactic fixation tend to be younger and healthier, with a presumed lesser burden of disease, in comparison with patients with pathological fractures, but they remain at a considerable risk for having an acute VTE. The decision to treat a metastatic lesion with prophylactic nailing as opposed to nonoperatively is based on individual clinical judgment, which is outside the scope of the present study. Recent technological advances with computed tomography-based rigidity analysis (CTRA)36,37 may provide an improved method of predicting the risk of pathological fracture resulting from long-bone metastasis as compared with the Mirels scoring system38.
The present study, similar to the literature, demonstrated that prophylactic fixation was associated with a lower rate of acute postoperative comorbidities and a greater likelihood of being discharged to home even after adjusting for sex, age, and comorbidities (p < 0.001). Ward et al.5, in a study of 182 patients, demonstrated that IMN for prophylactic fixation of femoral metastases resulted in decreased morbidity and mortality at 1 and 2 years, with a decreased average blood loss, a shorter hospital stay, a greater likelihood of discharge to home as opposed to an extended-care facility, and a greater likelihood of retaining walking ability without assistive devices as compared with pathological fracture fixation. Multiple other studies also have demonstrated better overall survival at all postoperative time points in patients managed with prophylactic stabilization4,6,39. Janssen et al.40 compared the rates of revisions and complications for 3 types of fixation (IMN, endoprosthetic reconstruction, and open reduction and internal fixation) for metastatic disease or multiple myeloma of the proximal part of the femur. They found a rate of PE (2.3%) similar to that in the present study after IMN procedures, the majority of which were performed for prophylactic fixation. They further revealed no difference in revision and complication rates between the 3 surgical techniques but did not stratify the analysis into prophylactic and post-fracture groups.
Contrary to previous reports that have demonstrated longer hospital stays for patients with pathological femoral fractures4,5, the length of stay did not differ between the 2 groups in our series. Earlier mobilization after fracture fixation with IMN and increased availability of outpatient rehabilitation may explain the earlier discharge. These changes coincide with the economic transformation in healthcare associated with the aim of minimizing inpatient cost. For example, Ristevski et al.4, in a study of patients who were managed with femoral stabilization in Canada during the 1990s, reported a mean length of stay of 19.3 days for the prophylactic treatment group compared with 24.3 for the pathological fracture group. In the present study, which included incentivized hospitals in the U.S., the mean lengths of stay were 7.49 and 7.54 days for the prophylactic fixation and pathological fracture groups, respectively, representing an approximately threefold decrease compared with the values reported by Ristevski et al.4. Last, it is pertinent to note the similarity in racial and socioeconomic background between the 2 groups of patients, indicating that these factors did not likely influence the treatment received by the patient. This finding is in direct contrast to the disparity in care for other orthopaedic procedures in the published literature41,42.
In conclusion, patients who underwent prophylactic treatment of metastatic femoral disease with IMN tended to be younger and healthier than those who were managed with IMN after a pathological fracture. Similar to previous reports in the literature, our study demonstrated that patients with a pathological fracture had increased rate of acute postoperative complications and were less likely to be discharged to home. After adjustment for age, sex, and preexisting comorbidities, the rates of PE and DVT remained higher in the prophylactic nailing group than in the pathological fracture group. The present study is not intended to argue against the use of prophylactic fixation for indicated patients but rather to serve as a caution that patients with metastatic disease who undergo prophylactic nailing should be vigilantly managed to decrease the rate of postoperative symptomatic VTE. Prospective studies are warranted to evaluate and optimize DVT prophylaxis protocols for patients undergoing surgery for the treatment of metastatic bone disease.
Tables showing results of the multivariate analysis are available with the online version of this article as a data supplement at jbjs.org.
Investigation performed at the Department of Orthopaedic Surgery, University of Chicago, Chicago, and the Department of Orthopaedic Surgery, Northwestern University, Chicago, Illinois
Disclosure: The authors indicated that no external funding was received for any aspect of this work. The Disclosure of Potential Conflicts of Interest forms are provided with the online version of this article.
- Copyright © 2017 by The Journal of Bone and Joint Surgery, Incorporated