Deep vein thrombosis (DVT) and pulmonary embolism (PE), collectively known as venous thromboembolism (VTE), are significant potential post-operative complications.[[1]] Prevention of VTE is important because 20–50% of patients with symptomatic DVTs develop post-thrombotic syndrome, and 5.4–14.6% of in-hospital deaths are caused or contributed to by PEs.[[2–6]] As antithrombotic agents significantly increase the likelihood of bleeding, risk stratification must be employed to target those with a favourable risk-to-benefit ratio.[[7–9]] While numerous risk factors are involved in the pathogenesis of VTE, a significant one is major orthopaedic surgery. In the absence of pharmacological thromboprophylaxis, the incidence of symptomatic VTE after total hip joint replacement (THJR), total knee joint replacement (TKJR) and hip fracture surgery is 4.3%.[[10]]

Effective strategies to reduce the incidence of post-operative VTE include pharmacological and mechanical prophylaxis in the form of intermittent pneumatic compression and early mobilisation. A systematic review, published in JAMA, of 47 studies involving 44,844 patients taking pharmacological prophylaxis found the incidence of symptomatic VTE following hip or knee joint arthroplasty prior to discharge was 0.53% and 1.09% respectively.[[11]] Their incidence of DVTs was 0.26% and PEs 0.14% following hip arthroplasty, and 0.63% and 0.27% after knee arthroplasty. Meta-analysis of 27 studies from Kwok et al. found intermittent pneumatic compression reduced the incidence of DVT by 59% and PE by 58% for elective orthopaedic patients.[[12]]

With these advances in thromboprophylaxis, the incidence of VTE following orthopaedic surgeries is relatively low. Shahi et al. analysed 173,591 revision arthroplasties and found the incidence of VTE, DVT and PE following revision of THJR to be 1.34%, 1.06% and 0.37%, and for revision of TKJR 1.16%, 0.88% and 0.34%.[[13]] McNamara et al. analysed 5,300 patients admitted for neck of femur (NOF) fractures and found their incidence for VTE within 1 year of admission was 2.2%, DVTs was 1.5% and PEs 0.7%.[[14]] The incidence of VTE was highest following intramedullary nailing at 3.3%, followed by sliding hip screw at 2.9%, hemiarthroplasty at 1.7%, multiple screws at 1.6%, then non-surgical treatment at 1.1%. More locally in New Zealand, Dixon et al. reviewed 5,046 procedures performed in at Waitematā District Health Board (DHB) and found the incidence of VTE in the 90 days following total hip joint replacement to be 1.5%, knee joint replacement 5.3% and NOF fracture surgery 4.2%.[[15]] Lapidus et al. investigated the incidence of VTE in the 6 weeks following 45,968 orthopaedic procedures. They found the incidence of VTE was 12% following internal fixation of pelvic fractures, 3.6% after ankle fractures internal fixation, 0.3% after spinal surgeries and 0.4% after upper limb surgeries.[[16]]

The incidence of VTE following orthopaedic surgery in Tauranga, New Zealand is currently unknown. This study aims to investigate the incidence of VTE in Tauranga Public Hospital following various orthopaedic procedures. Secondary aims are to investigate the relative risk of VTE risk factors in our cohort and the time from operation to VTE diagnosis.

Method

All patients undergoing orthopaedic surgery at Tauranga Public Hospital admitted in a 5-year period (31 July 2016 to 1 August 2011 ) were identified from hospital discharge data. Patients discharged from orthopaedics who were treated non-operatively were excluded. Patients who were managed operatively routinely used Thrombo-Embolus Deterrent (TED) stockings and mechanical prophylaxis in the form of calf compression or foot pumps. Arthroplasty patients typically mobilised the day after surgery.

Patients diagnosed with symptomatic VTE at Tauranga Public Hospital from 1 August 2011–29 October 2016 were identified by ICD coding (I26.0, I26.9, I74.3, I74.9, I80.2, I80.3, I82.8, I82.9, I97.8, T81.7, T85.86). Private radiology clinics performing compression ultrasounds in Tauranga were contacted for a list of National Health Indexes (NHIs) of patients diagnosed with DVTs within the same timeframe. These NHIs were crossmatched with the list of patients who had undergone orthopaedic surgeries. These notes were then formally reviewed to ensure they had a DVT diagnosed by ultrasound, or PE diagnosed by CT pulmonary angiogram (CTPA) within 90 days after their orthopaedic surgery. If the VTE was diagnosed within 90 days of multiple procedures, the most recent lower limb joint arthroplasty or revision arthroplasty was considered causative.

Incidence and prevalence data were analysed using Wilson’s score method assuming a binomial distribution. Risk ratios were calculated using the Wald test with a small sample adjustment and bootstrap method. Risk ratios were calculated comparing the prevalence of risk factors in all surgical patients to VTE patients. Additionally, these were calculated comparing all hip and knee arthroplasty and revision arthroplasty patients to those undergoing the same procedures who developed VTE. These were calculated using the statistical software “R” using the prop.test function without Yates' continuity correction and the risk.ratio command, respectively.[[17]]

Results

Table 1 summarises the basic characteristics of subjects included in this study. Over the 5-year period, 11,394 orthopaedic procedures were performed on 9,328 patients, 78 of whom developed VTE. 48.4% of patients were female, while 54.8% of VTE patients were female. 42.9% of surgical patients and 67.9% of VTE patients were 65 years old or older. Most patients were of European descent.

The incidence of symptomatic VTE within 90 days of their respective procedures is summarised in Table 2 and Figure 1. The incidence of VTE following TKJRs and revision of TKJRs was higher than their THJRs counterparts, although this difference was only statically significant for unilateral primary arthroplasties. For NOF fracture surgeries, intramedullary nails had the highest incidence of VTE, followed by hip hemiarthroplasty, cannulated screws and dynamic hip screws, although these differences did not reach statistical significance. The incidence of VTE following ankle fracture surgeries are comparable to that of lower limb total joint arthroplasties. The operations associated with the lowest incidence of VTE were spinal, upper limb and paediatric surgeries.

View Tables 1–4 & Figures 1 & 2.

Table 2 summarises the relative risk of various risk factors. Risk factors for VTE include advanced age, comorbid cardiovascular disease and being admitted to the intensive care unit post operatively. Only the latter two were statistically significant risk factors for VTE in lower limb arthroplasty and revision arthroplasty patients.

The time from the most recent orthopaedic procedure to the diagnosis of VTE is illustrated in Figure 2. Thirty-eight point five percent of VTE were diagnosed within 1 week of the patients’ operations, 66.7% within 2 weeks and 75.6% within 3 weeks.

The medication charts were available for 70 (89.7%) of VTE patients in this study. Table 4 shows the proportion of these patients on each pharmacological thromboprophylactic agent at the time of VTE diagnosis. At least 67.9% of patients were on pharmacological thromboprophylaxis, 43.6% of whom were taking aspirin monotherapy, and the remaining 24.4% were on more potent antithrombotics. We did not have access to the thromboprophylactic regimens of patients who did not develop VTE.

Discussion

These results show that VTE are relatively rare complications of orthopaedic surgery at Tauranga Hospital. The highest rates were seen in those undergoing total lower limb arthroplasty, revision of lower limb arthroplasty, NOF and ankle fracture surgery. Surprisingly, more patients were diagnosed with PEs than DVTs. We identified comorbid cardiovascular disease and being transferred to ICU post operatively as risk factors for VTE. The majority of VTEs were diagnosed within 1 month of surgery. At the time of VTE diagnosis, almost half of these patients were taking aspirin and over a quarter were on more potent antithrombotic regimens. This highlights that using anticoagulants cannot eliminate the risk of post-operative VTE.

The rates of VTE following lower limb joint arthroplasty and revision arthroplasty in this study are higher than the rates reported in other studies.[[11,13]] However, it is important to note that these studies analysed VTE developing from surgery to discharge, while our rates are within 90 days of surgery. Employing their methodology excludes 52% of our VTE cases. Another major difference between our study and this meta-analysis is that they only included patients receiving pharmacological thromboprophylaxis. Not all patients in this study received anticoagulants due to contraindications and clinical judgement. The decision to prescribe pharmacological thromboprophylaxis is not always straightforward, as demonstrated by two of our patients who developed haemarthrosis following TKJR, prompting cessation of anticoagulants—then they subsequently developed PEs. Another factor contributing to our higher incidence of VTE was that five of our patients also underwent non-orthopaedic surgery within 90 days of their VTE diagnosis, which likely contributed to its development. A study from New Zealand with similar methods had results comparable to ours, with a VTE incidence of 1.5% following hip joint arthroplasty and 5.3% following knee joint replacement.[[15]]

One weakness of this study is not having data on the thromboprophylactic regimens of patients who did not develop VTE. Consequently, we were unable to determine what antithrombotic agent was most effective. Another weakness is our sparse number of bilateral lower limb arthroplasty cases. Consequently, we cannot determine whether these patients are at higher risk than those undergoing unilateral arthroplasty. Additionally, this study is at risk of type II error due to small event numbers. Finally, we relied on hospital coding to identify patients diagnosed with venous thromboembolism. This may have failed to identify some cases. A major strength of this study is its high capture incidence. In Tauranga, CTPA is only performed at Tauranga Public Hospital, so all patients diagnosed with PE in Tauranga will be included in our data. While ultrasounds can also be performed at eight private radiology clinics in Tauranga, six were able to provide us with data, identifying six more DVT patients. Another plausible explanation as to why more PEs were diagnosed than DVTs is that the former carries a higher incidence of morbidity and mortality, therefore in patients with signs of both diseases, clinicians may opt to investigate for PEs first. Because the management of patients with PE does not change if they have a DVT, PE patients were less likely to be investigated for concurrent DVT. There is also evidence that PEs can occur in the absence of DVTs.[[18]]

This study highlights those with the greatest risk of post-operative VTE. Patients undergoing major lower limb surgery with comorbid cardiovascular disease and those requiring ICU admission post operatively have the greatest to gain from thromboprophylaxis. This suggests that the use of more potent anticoagulation for these patients may be appropriate, especially in the first 2 to 4 weeks after their procedure. In future research we plan to investigate the sequelae of VTEs in orthopaedic patients.