Diagnostic and operative codes are routinely collected for every patient admitted to hospital in the English NHS. Data on post-operative complications following foot and ankle surgery have not previously been available in large numbers. Data on symptomatic venous thromboembolism events and mortality within 90 days were extracted for patients undergoing fixation of an ankle fracture, first metatarsal osteotomy, hindfoot fusions and total ankle replacement over a period of 42 months. For ankle fracture surgery (45 949 patients), the rates of deep-vein thrombosis (DVT), pulmonary embolism and mortality were 0.12%, 0.17% and 0.37%, respectively. For first metatarsal osteotomy (33 626 patients), DVT, pulmonary embolism and mortality rates were 0.01%, 0.02% and 0.04%, and for hindfoot fusions (7033 patients) the rates were 0.03%, 0.11% and 0.11%, respectively. The rate of pulmonary embolism in 1633 total ankle replacement patients was 0.06%, and there were no recorded DVTs and no deaths. Statistical analysis could only identify risk factors for venous thromboembolic events of increasing age and multiple comorbidities following fracture surgery.
Venous thromboembolism following foot and ankle surgery is extremely rare, but this subset of fracture patients is at a higher risk. However, there is no evidence that thromboprophylaxis reduces this risk, and these national data suggest that prophylaxis is not required in most of these patients.
Venous thromboembolic (VTE) events following orthopaedic surgery are rare,1 but fatal pulmonary embolism (PE) can occur. The most effective prophylaxis remains undetermined,2–4 and although low-molecular-weight-heparin (LMWH) appears to reduce rates of symptomatic deep-vein thrombosis (DVT) and PE after arthroplasty of the lower limb, there is no evidence of a reduced incidence of fatal PE and overall mortality following any orthopaedic procedure.5
The National Institute for Health and Clinical Excellence (NICE) produces evidence-based guidance on the appropriate treatment of patients in the NHS in England and Wales.6 In April 2007, they recommended that all orthopaedic in-patients be offered a LMWH for the duration of their stay in hospital.7 These recommendations were recently revised, acknowledging the lack of robust evidence for the widespread use of prophylaxis.8 However, chemical prophylaxis is still recommended if mobility is significantly reduced following surgery or, if casting is used, for the duration of immobilisation. In addition, patients undergoing lower-limb surgery lasting longer than 60 minutes (including anaesthetic time), obese patients, those over 60 years old and patients with comorbidities should be offered chemical prophylaxis (for an unspecified duration).8
Concerns remain among orthopaedic surgeons regarding risk versus the benefit of these recommendations.9–11 Based on these criteria, the vast majority of patients undergoing foot and ankle surgery should be offered prophylaxis. LMWH treatment is costly, delays healing12 and can result in thrombocytopenia, a condition that carries a 15% to 30% mortality risk.13,14 Moreover, there is no proven benefit of the use of LMWH in these patients, and the recommendations are not based on any published evidence.8
The overall incidence of VTE events and mortality after foot and ankle surgery is unknown. This paper aims to establish robust national data on these complications, considering the effect of both age and comorbidities, and with reference to the most recent NICE guidelines.
Materials and Methods
Data for adult patients (18 years and above) who underwent open reduction and internal fixation (ORIF) for a fracture around the ankle, first metatarsal osteotomy, hindfoot fusion or a total ankle replacement were collected from the administrative hospital admissions database (Hospital Episode Statistics), augmented by more recent data from the Secondary Users Service, which reports and analyses the latest data from NHS units to support planning and commissioning health-care. These statistics cover all admissions to English NHS hospitals and includes 14 diagnostic fields, coded using the tenth edition of the International Statistical Classification of Diseases and Related Health Problems (ICD-10 codes),15 and 12 surgical procedure fields coded using the fourth edition of the Office of Population, Censuses and Surveys Classification of Surgical Operations and Procedures (OPCS-4 codes).16 We linked records belonging to the same patient defined using a combination of date of birth, gender and postcode, and noted the number of days between the index operation and any subsequent readmission to any NHS hospital. Patients with missing dates of operation were excluded. By employing the appropriate ICD-10 or OPCS-4 codes, rates of complication could be identified. Rates of PE (fatal and non-fatal), symptomatic DVT and all-cause in-patient mortality within 90 days were extracted, even if the patient had been successfully discharged following the primary procedure. Age and Charlson score17 were recorded for each patient. The Charlson comorbidity index predicts the one-year mortality for a patient who may have a range of comorbid conditions. Each condition is assigned a score of 1, 2, 3 or 6 (Table I⇓), depending on the risk of death associated with it. The combined total score is then used to predict patient-specific overall mortality. Patients with a previous VTE event were excluded (ICD-10 code Z86.7: ‘previous personal history of disease of the circulatory system’). Codes are summarised in Table II⇓. Patients with polytrauma were not excluded from this analysis.
Initially data were extracted for the period between January 2005 and June 2008. In order to determine the effect of the 2007 NICE guidelines, data for the six-month period prior to their publication (October 2006 to March 2007, pre-guideline period) were then analysed and compared with further data collected from a second six-month period (December 2007 to May 2008, post-guideline period). This delayed second period was chosen to allow time for the dissemination of these guidelines and local implementation.
A two-tailed z-score (or critical ratio statistic) with continuity correction, if P2-P1 > 0.5 (1/N1 + 1/N2), where P is the proportion and N is the size of the sample, was calculated with MATLAB software (The MathWorks, Natick, Massachusetts) using the equation 3.5 of Fleiss, Levin and Paik,18 and was used to analyse the differences in VTE events (a composite of PE and DVT data, in order to achieve adequate numbers) and deaths before and after the introduction of the 2007 guidelines. Odds ratios (OR) and 95% confidence intervals (CI)19 were calculated for rates of post-operative complications using the conditions ‘age’ (categorised as 18 to 49, 50 to 59, 60 to 69, 70 to 79 and 80 years and over), ‘Charlson score’ and ‘comorbidities’ for each of the operation groups except total ankle replacement, for which VTE events were too rare. For each condition, one category was used as a reference to which all other categories were compared. The complication tested was VTE within 90 days. The null hypothesis (OR = 1) was tested using the chi-squared test (α = 0.05) for a 2 × 2 contingency table. Statistical significance was taken as p < 0.05.
CHKS (part of the Health Division of Capita Group Plc) specialises in providing clinical benchmarking services to the NHS and independent sectors in the United Kingdom and the Republic of Ireland. For the purposes of this study, CHKS provided the data source, advice on clinical coding and interpretation, and analytical services free of charge. CHKS has approval to re-use Hospital Episode Statistics data with the permission of the Health and Social Care Information Centre.
Hospital episode statistics were available for a total of 88 241 foot and ankle procedures during the 42-month study period. Demographic data are given in Table III⇓. For ankle fracture surgery, DVT and PE rates were 0.12% and 0.17%, respectively. For first metatarsal osteotomy, the rates were 0.01% and 0.02%, and for hindfoot fusions they were 0.03% and 0.11%, respectively. The PE rate in 1633 total ankle replacement patients was 0.06%, with no recorded DVTs. The 90-day DVT, non-fatal and fatal PE, combined VTE and all-cause mortality rates are shown in Table IV⇓. All fatal PEs occurred in patients over 50 years old.
Table V⇓ shows that the risk of a VTE event after ankle fracture surgery was significantly higher in patients over 50 years old and those with a Charlson score of 2 or more (OR = 5.37, 95% CI 2.94 to 9.8). Figure 1⇓ shows that patients with a Charlson score of ≥ 2 have a higher risk of PE in each group. Non-insulin-dependent diabetes had a profound effect on VTE risk (OR = 14.6, 95% CI 9.67 to 21.9). There were no predictors for VTE in the other surgical groups.
There was no significant difference in the VTE rates prior to and following the introduction of the 2007 NICE guidelines in any of the four surgical groups (Table VI⇓). NICE recommend that patients with specific risk factors be offered LMWH as prophylaxis. If those criteria are applied to these national data, the majority of patients undergoing foot and ankle surgery should be offered LMWH, with an estimated drug cost of £10.5 million (based on British National Formulary costs20) for the study period (Table VII⇓). However, based on the risk analysis presented here, only those patients undergoing ankle fracture surgery and aged > 50 or with Charlson score ≥ 2 should be considered for prophylaxis.
Figure 2⇓ shows that 90-day VTE event rates after these four operations are extremely low compared with data on total hip and knee replacement. The background population risk is also shown.
Figure 3⇓ shows low complication rates following ankle fracture surgery, with mortality increasing dramatically in the most elderly patients.
Thromboprophylaxis following foot and ankle surgery is a contentious issue. There have previously been no large population-based studies published on rates of VTE events and there is no good evidence for prophylaxis. Moreover, randomised controlled trials that have compared LMWH to no treatment/placebo in patients immobilised in lower-limb casts failed to identify a reduction in symptomatic events with prophylaxis.8 Yet despite this, NICE continue to recommend thromboprophylaxis for patients undergoing foot and ankle surgery by extrapolating the risk data from joint replacement studies.8
Based on a middle-aged, female non-surgical population, the general background risk of a VTE event was calculated by Sweetland et al21 as 0.058 per 1000 person-months. For a 90-day (three-month) period this is equivalent to a risk of 0.017%. Rates of VTE are eight to 17 times higher than the background risk (0.14% and 0.28%, respectively) after hindfoot fusion and ankle fracture surgery, and only two to three times higher following first metatarsal osteotomy and total ankle replacement. Jameson et al22 have previously shown that the risk of VTE, following lower-limb arthroplasty with the exclusion of fatal PE, is about 2%. This is similar to the findings of Sweetland et al21 in their population-based analysis, representing a risk around 120 times higher than the background risk (Fig. 1⇑). The risk of a VTE event after foot and ankle surgery is therefore extremely low.
Using the NICE criteria for ‘high-risk’ patients (those who should be offered LMWH), all ankle fracture patients, total ankle replacements and hindfoot fusions and over one-third of first metatarsal osteotomies would require thromboprophylaxis (Table VII⇑). However, from this analysis, those at higher risk of VTE events are ankle fracture patients > 50 years old or those who have a Charlson score of ≥ 2. If LMWH were to have an effect, it would be most effective in this group. These findings agree with those of SooHoo et al,23 who proposed thromboprophylaxis for older patients and those with medical comorbidities only. In their study, the rates of PE and death were 0.34% and 1.07%, respectively. Although these rates are higher than those found in our study, their patients were older and had higher Charlson scores. In addition, patients with a known history of peripheral vascular disease, which includes previous VTE, were excluded from our study. A multicentre study performed by Mizel et al24 also showed a low incidence of non-fatal PE of 0.15% in 2733 foot and ankle surgical patients, similar to the data presented here, and concluded that routine thromboprophylaxis was not justified.
Factors associated with an increased risk of venographically or ultrasound-proven VTE in foot and ankle surgery include prolonged tourniquet and operating theatre times, a longer delay to surgery following ankle trauma, hindfoot surgery, post-operative cast immobilisation and non-weight-bearing.24–28 Simon et al29 found that the use of a thigh tourniquet was not associated with an increased risk of VTE following forefoot surgery, and Patil et al30 found that events were rare following cast immobilisation for stable ankle fractures. However, there are few associations with symptomatic VTE events in the literature, presumably because of rarity. A recently published study of a small number of cast-immobilised patients with rupture of tendo Achillis showed a rate of VTE equivalent to the risk with lower-limb arthroplasty.31 None of 208 patients received LMWH, but eight of the 13 patients who developed a symptomatic VTE had factors suggesting an increased risk of an event (family history, blood abnormalities, high body mass index (BMI) and comorbidities). In our study, patients undergoing planned surgery and young ankle fracture patients without comorbidities were at low risk for VTE. Moreover, data on BMI, family history and blood abnormalities were unavailable and, with these factors included in this analysis, the rates in patients who present without any risk is likely to be even lower. Therefore, we believe that cast immobilisation in isolation should not be perceived as a risk factor for VTE disease.
A survey of American and British foot and ankle surgeons performed in 2006 found that 19% of surgeons use thromboprophylaxis in either elective or trauma surgery. Prophylaxis was most commonly used in elective hindfoot surgery and ankle ORIF.32 Following the publication of the 2007 NICE guidelines, 94% of British Orthopaedic Foot and Ankle Surgery (BOFAS) members stated that they prescribed chemical prophylaxis to post-operative elective in-patients immobilised in plaster, and 65% continue thromboprophylaxis for the duration of treatment.33 Despite no evidence of the benefit and no cost-effectiveness analysis,8 the use of thromboprophylaxis appears to be increasing, and in the English NHS surgeons might be under some pressure to adhere to the NICE guidelines, as a result of hospital policy or fears of litigation. Although there are no robust data available on the use of LMWH in these patients, it seems logical that use has increased in the post-guideline era. However, there was no significant beneficial effect on the incidence of VTE events or deaths.
The various specialist societies and government bodies have failed to reach agreement on the type and duration of prophylaxis for different orthopaedic procedures.34–37 The cost-effectiveness of LMWH, which has financial and resource ramifications,38 needs to be considered. Daily injections are painful and inconvenient for patients, costly in terms of staff training, advice and blood monitoring, and resources, and are not without risk of complications. Owing to the lack of evidence, NICE were unable to perform a cost analysis on thromboprophylaxis for foot and ankle surgery. Evidence from several trials in cast-immobilised patients has not shown LMWH to be beneficial over no treatment in reducing symptomatic events.8 NICE does concede that it is “unlikely that prophylaxis will be cost-effective” (p359).8 Despite this, section 1.6.3 of the guidelines states that a surgeon should “consider offering pharmacological VTE prophylaxis to patients with lower-limb plaster casts after evaluating the risks and benefits based on clinical discussion with the patient. Offer LMWH until lower-limb plaster cast removal”.8 This recommendation is vague and at odds with the evidence, and, based on the costs described in Table VII⇑, extremely expensive.
We acknowledge flaws in this observational study. Specifically, we have no data on the use of mechanical or pharmacological prophylaxis, and lack cause of death data for the majority of patients. In addition, the data collection and coding of patients who are diagnosed with DVT as an outpatient may be unreliable, and the figures presented here may underestimate the 90-day DVT rate. Patients who die at home would not be detected, but those who become unwell at home and subsequently die in hospital after readmission would be included. This is consistent across the Hospital Episode Statistics-based literature. However, out-of-hospital deaths account for only 8% of total postoperative mortality.39 The authors acknowledge Hospital Episode Statistics inaccuracies from the past,40 but this system is constantly improving.41 It only provides in-patient data, so we are unable to include patients treated non-operatively.
VTE events are extremely rare following foot and ankle surgery. Moreover, there is no evidence that thromboprophylaxis is of benefit in these patients. We therefore suggest that widespread VTE prophylaxis is not required. However, from this national dataset we have identified a specific group of ankle fracture patients who are at higher risk. Future research into the benefit of thromboprophylaxis in these patients is required.
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.
- Received August 18, 2010.
- Accepted January 15, 2011.
- © 2011 British Editorial Society of Bone and Joint Surgery