The effect of aspirin and low-molecular-weight heparin on venous thromboembolism after knee replacement

A non-randomised comparison using National Joint Registry Data

S. S. Jameson, P. N. Baker, S. C. Charman, D. J. Deehan, M. R. Reed, P. J. Gregg, J. H. Van der Meulen


We compared thromboembolic events, major haemorrhage and death after knee replacement in patients receiving either aspirin or low-molecular-weight heparin (LMWH). Data from the National Joint Registry for England and Wales were linked to an administrative database of hospital admissions in the English National Health Service. A total of 156 798 patients between April 2003 and September 2008 were included and followed for 90 days. Multivariable risk modelling was used to estimate odds ratios adjusted for baseline risk factors (AOR). An AOR < 1 indicates that risk rates are lower with LMWH than with aspirin. In all, 36 159 patients (23.1%) were prescribed aspirin and 120 639 patients (76.9%) were prescribed LMWH. We found no statistically significant differences between the aspirin and LMWH groups in the rate of pulmonary embolism (0.49% vs 0.45%, AOR 0.88 (95% confidence interval (CI) 0.74 to 1.05); p = 0.16), 90-day mortality (0.39% vs 0.45%, AOR 1.13 (95% CI 0.94 to 1.37); p = 0.19) or major haemorrhage (0.37% vs 0.39%, AOR 1.01 (95% CI 0.83 to 1.22); p = 0.94). There was a significantly greater likelihood of needing to return to theatre in the aspirin group (0.26% vs 0.19%, AOR 0.73 (95% CI 0.58 to 0.94); p = 0.01). Between patients receiving LMWH or aspirin there was only a small difference in the risk of pulmonary embolism, 90-day mortality and major haemorrhage.

These results should be considered when the existing guidelines for thromboprophylaxis after knee replacement are reviewed.

Venous thrombosis is considered to be common following knee replacement, with the incidence of asymptomatic deep-vein thrombosis (DVT) being estimated to be as high as 60%.1 However, the reported incidences of symptomatic venous thromboembolism (VTE, 1.8%2) and fatal pulmonary embolism (PE, 0.15%3) within 90 days of surgery are much lower. In order to reduce the occurrence of VTE, the National Institute for Health and Clinical Excellence (NICE) recommend combined mechanical and pharmacological prophylaxis post-operatively for ten to 14 days for all patients undergoing knee replacement, unless there are contraindications.1 Recommended agents include low-molecular-weight heparins (LMWHs), direct factor Xa inhibitors, fondaparinux sodium and unfractionated heparin (UFH). NICE state that the ‘protective effect of aspirin against VTE is insufficient’, and therefore do not recommend aspirin and other antiplatelet medications for pharmacological prophylaxis. Although this has also been the stance of the American College of Chest Physicians (ACCP)4 and the International Consensus Group,5 the American Academy of Orthopaedic Surgeons (AAOS) in their guidelines were unable to find conclusive evidence to recommend one chemical prophylaxis agent over any other.6

Recommendations against the use of aspirin are based on extrapolations from mostly small, historical studies.7-9 Peri-operative care has subsequently evolved, and this evidence may now be outdated. More recent publications with larger numbers have suggested that the risk of VTE following either aspirin or injectable prophylaxis against VTE (LMWH/fondaparinux) is equivalent,10 and the routine use of potent anticoagulants (LMWH/direct factor Xa inhibitors) does not reduce the overall mortality or the proportion of deaths due to PE.11 Aspirin is an attractive method of prophylaxis as it is inexpensive, orally administered, and does not require laboratory monitoring. Prior to the NICE guidance almost a quarter of all patients undergoing knee replacement were given aspirin for VTE prophylaxis.12

We have previously compared the effects of aspirin and LMWH on VTE after hip replacement,13 showing a small survival benefit for those receiving LMWH, but no differences were seen for the risk of VTE and bleeding events. We now examine the role of these two chemical agents following total knee replacement (TKR) in a similar national series of patients.

Patients and Methods

We used records from the National Joint Registry for England and Wales (NJR) linked at patient level to records from the Hospital Episode Statistics (HES) database. The NJR aims to collect data prospectively on all patients who undergo knee replacement in England and Wales.14 The HES database includes all patients admitted to NHS hospitals in England, or to private hospitals funded by the NHS (no Registry data from Wales was included in this analysis).15 HES records contain patient details and diagnostic information coded using the International Classification of Diseases, tenth revision (ICD-10),16 and operative procedure codes using the United Kingdom Office for Population Censuses and Surveys classification, fourth revision (OPCS-4).17 The HES database also includes date of death from the Office for National Statistics.

This analysis was based on data described in the Seventh NJR Annual Report.12 We identified 227 668 patients for whom a primary knee replacement (total, patellofemoral and unicompartmental) was recorded in the NJR between 1 April 2003 and 30 September 2009, and who could be linked to HES. The linkage was carried out based on a hierarchy of deterministic criteria, including NHS number, year of birth and gender, as detailed in the Seventh NJR Annual Report.12 Of these, 156 798 were given either aspirin or LMWH as the sole chemical thromboprophylaxis, irrespective of whether mechanical prophylaxis had also been used. Patients receiving more than one type of pharmacological prophylaxis were excluded. All patients were followed for at least 90 days.

Information regarding the method of thromboprophylaxis was extracted from NJR records with relevant details of the operation and patient characteristics that were considered risk factors for VTE or haemorrhage. We used the Royal College of Surgeons Charlson Score18 to search the HES database for comorbid conditions.

HES data were also used to identify thromboembolic events (PE, DVT) and death within 90 days of surgery, major haemorrhage (cerebrovascular accident or gastrointestinal haemorrhage) and return to theatre for wound complications within 30 days of surgery. Minor haemorrhage was not considered, as it is unlikely to be accurately coded in the HES database.

Statistical analysis

Logistic regression was used to assess the effect of treatment on outcome, with adjustment for patient characteristics that are risk factors for VTE or haemorrhage. Results are presented as odd ratios (OR) with 95% confidence intervals (CI); an OR < 1 indicates that rates are lower with LMWH than with aspirin. The likelihood ratio test was used as a basis for p-values, and p < 0.05 was considered to indicate statistical significance. All risk models included the type of pharmacological prophylaxis, age (grouped into three equal-sized categories with cut-off values rounded to the nearest multiple of five), gender, American Society of Anesthesiologists (ASA) grade,19 number of comorbid conditions according to the Charlson score, indication for surgery (osteoarthritis or other), whether regional anaesthesia (epidural or spinal procedures) was used, type of prosthesis type (cemented TKR, cementless TKR, hybrid TKR, and unicompartmental or patellofemoral knee replacements), use of mechanical thromboprophylaxis, and provider type (NHS hospital, NHS treatment centre, independent hospital, or independent treatment centre).

Use of regional anaesthesia was the only variable with missing values. In order to address this we used multiple imputation by chained equations.20 Five datasets were created containing imputed values for missing values. Rubin’s rules21 were used to combine the estimation results based on each of these datasets.


Of the 156 798 patients who were eligible for inclusion, aspirin was used as thromboprophylaxis in 36 159 (23.1%) and LMWH in 120 639 (76.9%). Patients in the aspirin group were more likely than those in the LMWH group to receive mechanical prophylaxis and undergo treatment in an NHS hospital, but were less likely to have regional anaesthesia (Table I). The two groups were otherwise very similar.

View this table:
Table I

Characteristics of the patients and their treatment according to pharmacological prophylaxis with aspirin or low-molecular-weight heparin (LMWH)

Without adjustment for potential risk factors, there was a significantly greater risk of return to theatre within 30 days in the aspirin group (OR 0.71 (95% CI 0.56 to 0.91)), although the absolute rate was low (aspirin 0.26% vs LMWH 0.19%). We found no statistically significant differences in the rate of thromboembolic events, 90-day mortality or major haemorrhage between the two groups (Table II). The rate of PE was 0.49% with aspirin and 0.45% with LMWH (unadjusted OR 0.91). The 90-day mortality was 0.39% with aspirin and 0.45% with LMWH (unadjusted OR 1.16). The differences in the other outcomes were small, and all unadjusted ORs varied around 1.

View this table:
Table II

Effect of aspirin and low-molecular-weight heparin (LMWH) on outcomes, with adjustment based on the multivariable risk model. An odds ratio (OR) < 1 indicates that rates are lower with LMWH than with aspirin (CI, confidence interval)

After risk adjustment, return to theatre within 30 days remained significantly higher in the aspirin group (adjusted OR 0.73; p = 0.01). The impact of the risk adjustment on the differences in the other outcomes was small, and the groups had similar outcomes in terms of PE, DVT, 90-day mortality and major haemorrhage (Table II).


In this large national cohort of patients undergoing knee replacement no significant differences were observed in the rates of VTE, 90-day mortality and major (cerebrovascular and gastrointestinal) haemorrhage when aspirin was compared with LMWH for prophylaxis. Within the constraints of the study design, we found that a significantly greater number of patients in the aspirin group returned to theatre for the management of wound complications within 30 days of the operation.

The multivariable model included all available risk factors that were thought to be associated with the outcome, including adjusting risk for the presence or absence of mechanical prophylaxis. However, a number of unmeasured variables, known to influence VTE risk, could confound this analysis: obesity, cancer, previous VTE, family history, admission to intensive care, smoking, dehydration, thrombophilia, contraceptive use, and post-operative immobility.1 If these higher-risk patients were more likely to have LMWH our results might underestimate the effect of LMWH. Unmeasured risk factors for bleeding may also influence the choice of prophylaxis. In addition, the NJR collects thromboprophylaxis data immediately post-operatively on an ‘intention to treat’ basis. The agent given is not recorded, nor the dose, the duration of treatment or compliance. Data were unavailable on patients’ use of gastro-protective drugs.

Patients who had a NJR record of knee replacement but lacked a matching record in the HES were excluded, thereby reducing the population studied. Furthermore, there are limitations in the ICD-10 and OPCS-4 coding systems,22 and coding errors may underestimate event rates (as missing an event is more likely than incorrectly identifying one). Coding inaccuracies are especially relevant for DVT and wound infection, both of which are difficult to diagnose within the current limitations of HES.23 We therefore used ‘return to theatre’ codes as a surrogate for wound complications, including infection. However, rates of minor wound problems that did not require operative intervention, such as prolonged oozing, could not be quantified using the data available. Results from administrative data analyses must therefore be interpreted carefully. Regardless of these coding limitations, mortality as an outcome is well defined and therefore robust.

Although asymptomatic DVT may occur in most patients after knee replacement, it is estimated that only around 1 in 21 of these will result in symptomatic VTE.2 Reported rates of symptomatic VTE within 90 days of knee replacement are < 2%.2,3 We found lower risks of DVT and PE, which may be explained by improvements in peri-operative care during the last decade. However, under-reporting may have played a role, as many DVTs are diagnosed in the outpatient clinic or in the community.

There is currently limited evidence on the effects of thromboprophylaxis on the rates of PE following knee replacement,10 and there is no significant evidence relating to fatal PE and death.1 Moreover, no chemical thromboprophylaxis agent has been shown to be superior to others in terms of either efficacy or safety.11 Guidelines are therefore based on evidence extrapolated from complex network analyses that provide indirect comparisons of symptomatic DVT rates that are difficult to interpret, and rely on the assumption that a reduction in DVT reduces the risk of fatal PE. In an independent systematic review of the same trials used by the ACCP to formulate their guidelines, Brown,24 in contrast to the ACCP, concluded that aspirin was superior to LMWH. He found that the use of potent anticoagulants significantly increased the risk of post-operative bleeding without reducing clinically relevant symptomatic VTE and fatal PE rates compared to aspirin. Many of the studies analysed are historical and pre-date modern peri-operative care pathways. Admission on the day of surgery, early mobilisation and reduction in length of stay are all likely to contribute to a reduction in the risk of VTE.10 The outcomes in these older VTE studies may not be representative of the current situation. Owing to the decreased baseline risks, the balance between harm and benefit of prophylaxis regimens will have changed. In addition, PE is not the leading cause of death after joint replacement, as was previously thought. In Scottish patients who underwent knee replacement between 1992 and 2001, acute myocardial infarction accounted for 36% of deaths compared to 18% for PE,3 and when cause of death following almost 100 000 joint replacements was analysed, cardiopulmonary disease (excluding PE) accounted for 48%, compared with 25% for PE, irrespective of thromboprophylaxis agent used.11

We have previously reported findings from a similar comparison study of aspirin and LMWH in hip replacement patients.13 Although a very small survival benefit was seen in the LMWH group, no differences in VTE events and major haemorrhage were found. There are only a few direct comparisons of the effects of aspirin and LMWH. Westrich et al9 reported a non-significant difference in DVT rates in their randomised comparison of aspirin (DVT rate 17.8%) and LMWH (14.1%) when used in conjunction with mechanical prophylaxis. More recently Bozic et al10 retrospectively compared rates of VTE, bleeding, infection and mortality in 93 840 patients undergoing knee replacement using one of three pharmacological agents (aspirin n = 4719; LMWH/fondaparinux n = 37 198; warfarin n = 51 923). Once adjusted for case mix there were no significant differences for any outcomes between the aspirin and LMWH/fondaparinux groups. In NICE’s network analysis of the risk reduction (RR) for DVT, the indirect comparison showed that aspirin and LMWH were equivalent when used in conjunction with mechanical devices (aspirin RR = 0.15, LMWH RR = 0.14). Although aspirin had the lowest bleeding risk of all prophylactic agents (aspirin: RR = 0.44 (95% CI 0.17 to 1.04); LMWH: RR = 1.23 (95% CI 0.91 to 1.68); fondaparinux: RR = 2.21 (95% CI 1.27 to 3.94)), the difference between aspirin and LMWH was non-significant owing to the limited evidence on which the calculations were based.1

In this large cohort study of over 150 000 knee replacements, patients receiving aspirin or LMWH had similar risks of pulmonary embolus, 90-day mortality and major haemorrhage. Despite a shift away from the use of aspirin as thromboembolic prophylaxis over the last ten years, it remains a viable alternative to LMWH for the prevention of VTE and mortality after knee replacement. Despite the limitations, this study provides further evidence for those engaged in the review of thromboprophylaxis guidelines for patients undergoing knee replacement.


  • The authors would like to thank the patients and staff of all the hospitals in England and Wales who have contributed data to the National Joint Registry. We are grateful to the Healthcare Quality Improvement Partnership (HQIP), the NJR steering committee and the staff at the NJR centre for facilitating this work.

    The National Joint Registry for England and Wales is funded through a levy raised on the sale of hip and knee replacement implants. The cost of the levy is set by the NJR Steering Committee. The NJR Steering Committee is responsible for data collection. This work was funded by a fellowship from the National Joint Registry. The authors have conformed to the NJR’s standard protocol for data access and publication. The views expressed represent those of the authors and do not necessarily reflect those of the National Joint Register Steering committee or the Health Quality Improvement Partnership (HQIP) who do not vouch for how the information is presented.

    No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.

  • Supplementary material. Two tables detailing i) the International Statistical Classification of Diseases and Related Health Problems (10th revision) (ICD-10) and Office of Population, Censuses and Surveys Classification of Surgical Operations and Procedures (4th revision) (OPCS-4) codes and ii) the Charlson score are available with the electronic version of this article on our website

  • Received January 13, 2012.
  • Accepted February 23, 2012.


View Abstract