Introduction

Venous thromboembolism (VTE) (including deep vein thrombosis [DVT] and pulmonary embolism [PE]) remains one of the major causes of morbidity, mortality, and costs in total hip and knee arthroplasty (THA/TKA)1. Compliance with clinical practice guidelines (CPG) has been inconsistent due to concerns with the quality and recency of evidence, the relevance of symptomatic versus asymptomatic VTE, and the need to balance the risks and benefits of VTE prophylaxis against the risk of bleeding and surgical site infection2,3,4.

The Australian Orthopaedic Association (AOA) have endorsed the Arthroplasty Association of Australia guidelines for VTE prophylaxis for THA/TKA5. Many surgeons use the AOA VTE guidelines6, so it is important to investigate the level of compliance with these guidelines, and whether non-compliance is associated with VTE. Our study group has previously shown that non-compliance with the (then current) National Health and Medical Research Council (NHMRC) VTE prevention clinical guidelines7 was associated with higher rates of VTE after THA/TKA. The NHMRC VTE clinical guideline is a high quality guideline developed using rigorous methods8,9. The quality of the AOA guideline cannot be determined as there is no published information on the methods used or quality evidence underpinning the recommendations. The key difference between the two guidelines is that the NHMRC guidelines recommended only potent anticoagulants as all THA and TKA recipients are considered at high risk of VTE, whereas the AOA guidelines provides criteria to stratify patient risk, and recommends either potent anticoagulants or aspirin (an antiplatelet agent) with a sequential compression device for people assessed as being at routine VTE risk5,7. Using the same prospective cohort, this study aims to determine the magnitude of non-compliance with AOA VTE prophylaxis CPG and determine if there is an association between non-compliance and VTE after THA/TKA5.

Methods

Registration and data collection

A prospective observational cohort study of people undergoing elective primary THA/TKA for osteoarthritis in one of 19 high-volume institutions in Australia was performed to examine the relationship between non-compliance with several VTE prophylaxis and antibiotic CPG, and patient outcomes. In this paper we explored the association with the AOA VTE CPG and did not analyse any antibiotic data. Eligible sites included private and public Australian hospitals with high annual surgical volume (> 275 cases) of THA/TKA. Participant inclusion criteria were adults (≥ 18 years) with a primary diagnosis of osteoarthritis, undergoing primary THA/TKA; sufficient English to comprehend the protocol; and available to participate in 12 month follow-up.

Data were collected prospectively prior to surgery, during the acute admission and via telephone follow at 35, 90, and 365 days post-surgery. The accuracy of patient-reported and acute complications was verified by medical record audits at all sites and by contacting surgeons, primary care, and hospitals. The study protocol was registered prior to commencement10 and ethical approval was obtained from nine relevant human research ethics committees. The study has been described in accordance with strengthening the reporting of observational studies in epidemiology guidelines11.

Criteria for primary outcome and compliance with clinical guidelines

The primary outcome was any symptomatic VTE event (PE/DVT) up to 90 days post-surgery. The exposure was non-compliance with the risk-stratified recommendations of the AOA VTE clinical guidelines for THA/TKA5 (Table 1). Using criteria identified in the AOA VTE clinical guidelines the level of risk was categorised for each participant as routine or high for VTE, or high for bleeding. There were three dichotomous elements of compliance (Table 1): patient-level risk-stratified compliant prophylaxis, compliant duration of prophylaxis and compliance with other general recommendations (Table 1)5. For people who did not have a VTE or had a VTE after the recommended period of prophylaxis, the duration of prophylaxis was considered compliant if they received at least one prophylactic drug for at least 21 days from the day of surgery. For people who had a VTE during the recommended period of prophylaxis, the duration was calculated up until the day of VTE diagnosis, as further use of an anticoagulant was considered treatment rather than prophylaxis. If they started VTE treatment the same day or day after prophylaxis ceased, they were considered compliant for duration. The AOA recommended duration was three to six weeks, however we allowed a minimum of three weeks as compliant, without penalising for longer term use. Overall guideline compliance required the person to be compliant with all three elements. The primary predictor was the dichotomous overall AOA VTE clinical guidelines compliance variable.

Table 1 Criteria for determining risk and compliance with Arthroplasty Association of Australia (AOA) Prevention of VTE Guideline recommendations (2018).
Full size table

Anaesthesia was dichotomised into neuraxial anaesthesia received or not from nine combinations of general anaesthetic (GA), spinal, epidural, regional nerve block and sedation. Compliance with anaesthetic recommendations included people who received neuraxial anaesthesia or were taking warfarin preoperatively, where use of neuraxial anaesthesia may have been considered contraindicated. Some elements of the recommendations were excluded if they were not relevant for this study or if the data was unavailable, such as current metastatic cancer and individual normalised ratios.

Computer-based algorithms were developed to automatically generate compliance results to ensure the consistent application of the criteria for risk assessment and guideline compliance12 (https://doi.org/10.26190/mddw-by48). The a priori sample size calculations for the primary study were used for the purpose of this secondary data analysis10.

Data analyses

All data were entered into a designated REDCap database13. All analyses were conducted using the R Environment for Statistical Computing (version 3.6.1)12. Descriptive statistics were calculated to profile site-level and participant-level characteristics. Results were presented as median and inter-quartile values. Some variables (bilateral joint, smoking status, American Society of Anesthesiology score [ASA], education) were collapsed to allow for adequate group size or clinically meaningful groups. The dataset was reduced to include only those for whom risk-stratified compliance could be determined.

Bivariable (unadjusted) analyses were performed including chi-squared tests or t-tests, using Fisher’s exact test. We conducted multivariable logistic regression to explore associations between overall AOA VTE CPG non-compliance and the risk of VTE outcomes. The primary predictor and other factors identified on bivariable (unadjusted) analysis with a p-value < 0.25 were entered into a backwards, stepwise multivariable logistic regression model (using the Akaike information criterion–AIC) forcing only the main predictor into the model (Supplementary Table 4).

The rate of missing data was low (1%) and appeared to be random. Missing data were imputed using multiple imputation by chained equations and model selection was performed using one of the imputed datasets. Effect estimates were pooled estimates from the five imputed datasets. Using the final model, interaction terms for the primary non-compliance predictor against each other variable were tested for each analysis. Sensitivity analyses were performed using THA and TKA separately, excluding participants with premorbid history of VTE, and excluding participants who received a routine doppler ultrasound (DUS). Further sensitivity analyses were performed using complete case analysis and using Bayesian information criterion (BIC) for model selection. A final sensitivity analysis included using the three elements of compliance (prophylaxis, duration, general recommendations) as separate predictors, instead of the overall AOA VTE compliance variable. A deidentified version of the data set and the full R code for all analyses are available (https://doi.org/10.26190/mddw-by48).

Ethics approvals and consent to participate

Ethical approval was obtained from nine ethics committees. Austin Health Human Research Ethics Committee (EC00204)—Approval LNR/14/Austin/208; Barwon Health Human Research Ethics Committee (EC00208)—Approval LR13/82; Calvary Health Care Adelaide Human Research Ethics Committee (EC00302)—Approval 13_CHREC-E007; Calvary Health Care Tasmania Clinical and Research Ethics Committee—Approval 2:05:13 and 5:13:12; Epworth HealthCare Human Research Ethics Committee (EC00217) Approval LR138- 13; Hunter New England Human Research Ethics Committee (EC00403) Approval LNR/HNE/390; St Vincent's Health and Aged Care HREC (EC00324) Approval HREC #13/10; Sydney Adventist Hospital Group Human Research Ethics Committee (EC00141)—Approval 2013-016; The Prince Charles Hospital HREC (EC00168) Approval HREC 13-141; Prior to data collection informed written consent was obtained from eligible participants and the signed consent form was witnessed by the site coordinator.

Results

Sample ascertainment

In total, 3285 patients were screened and 2529 (77%) were eligible. Of the 2143 who provided consent, acute care data were received for 1905 participants. Participants were excluded if they did not proceed to surgery or no acute-care data were received (238 participants), they did not have any post-acute follow up (30 people, 1.6%) so 1875 participants were included in the sample. A further 38 participants (2%) were excluded as compliance could not be calculated. Data were analysed for 1837 (97.9%) consenting participants (Fig. 1). Missing data was less than 2.2% for all variables.

Figure 1
figure 1

Participant recruitment, eligibility, and participation.

Full size image

Sites, surgeon and participant characteristics

Site, surgeon, and participant characteristics are provided in Table 2.

Table 2 Participant characteristics.
Full size table

The use of multiple medications for post-operative VTE prophylaxis was common (N = 738, 40.2%) (Table 3, Supplementary Table 1). The duration of prophylaxis was shorter than recommended for 18.8% (N = 612) THA and 14.7% (N = 558) TKA participants7.

Table 3 Site, surgeon, surgical and acute care characteristics.
Full size table

VTE non-compliance

Most participants were assessed as routine VTE risk (N = 1554, 82.9%), with 12.5% (N = 230) at high VTE risk and 2.9% (N = 53) at high bleeding risk, according to the AOA VTE CPG criteria (Table 4). The rate of compliance with patient-level risk-stratified compliant prophylaxis AOA VTE clinical guideline recommendations was 85.9% (N = 1570), including 1411 (76.8%) of the 1554 people considered at routine VTE risk who received routine VTE risk compliant prophylaxis and 168 (9.1%) of the 230 people considered at high VTE risk who received high VTE risk compliant VTE prophylaxis. No patient considered at high bleeding risk received care compliant for high bleeding risk prophylaxis. As overall compliance required people to receive care considered compliant with all three elements, only 14.0% (N = 258) of the whole cohort received care considered compliant with AOA recommended VTE prophylaxis. When considered by the assessed level of VTE and bleeding risk, nearly 30% of people at high VTE risk, and Forty-four people considered high VTE risk received 100–300 mg aspirin alone, which is only recommended for people at routine VTE risk. Nearly two thirds (N = 1137, 61.9%) of the cohort received prophylaxis for the minimum recommended duration of 21 days, and 42 (2.3%) people ceased prophylaxis to commence VTE treatment. Over two thirds 67.8% (N = 1245) were non-compliant with other general recommendations.

Table 4 Patient VTE risk and compliance with AOA VTE Prevention Guidelines (2018) (N = 1838).
Full size table

Participant outcomes: surgical complications up to 90 days

The incidence of symptomatic VTE was 2.6% (N = 48), with nearly two-thirds of the cases experiencing DVT alone (Table 5). Bleeding complications requiring readmission or reoperation were rare (17, 0.92%) (Table 5, Supplementary Table 2). Four participants died from surgical complications, and seven died from medical causes.

Table 5 Prevalence of VTE events up to 90-days.
Full size table

Association between VTE prophylaxis non-compliance and VTE outcomes

In unadjusted analyses, non-compliance with AOA VTE prevention clinical guidelines was not significantly associated with lower symptomatic 90-day VTE (Supplementary Table 3).

In adjusted analyses, overall non-compliance with AOA VTE prevention clinical guidelines recommendations was not significantly associated with the risk of symptomatic VTE at 90 days (AOR = 0.93, 95%CI = 0.4 to 1.3, p = 0.86)5 (Table 6, Supplementary Table 4). There was no collinearity between variables and no significant interaction terms. The effect estimates for VTE prevention clinical guidelines non-compliance remained nonsignificant when people with high bleeding risk were excluded (AOR = 0.94, 95%CI = 0.44 to 2.34, p = 0.89), when people with a history of VTE were excluded (AOR = 1.17, 95%CI = 0.50 to 3.46, p = 0.74), when people who received a routine DUS were excluded (AOR = 0.85, 95%CI = 0.37 to 2.31, p = 0.73), and when people with THA or TKA were analysed separately (THA: AOR = 0.44, 95%CI = 0.12 to 2.06, p = 0.24; TKA: AOR = 1.39, 95%CI = 0.54 to 4.70, p = 0.54).

Table 6 Final regression model for association between AOA VTE prevention clinical guideline non-compliance and symptomatic 90-day VTE outcomes.
Full size table

Discussion

Non-compliance with the AOA VTE prevention clinical guidelines was not significantly associated with the risk of VTE. We reported higher levels of non-compliant prophylaxis for people assessed as being at high VTE risk (27.0%) and high bleeding risk (100%), compared to those at routine risk (9.2%) (using AOA criteria). All people considered at high bleeding risk received chemical prophylaxis, despite this not being recommended for this group.

Adherence to AOA VTE prevention clinical guideline recommendations does not appear to reduce the risk of VTE after THA and TKA. These findings are consistent with the US Surgical Care Improvement Project (SCIP), which reported a higher rate of PE in those who received care compliant with the VTE process measures (doctor ordered VTE prophylaxis, and the patient received VTE prophylaxis at the right time)14,15. This program uses process of care elements that are also very broad and may mask any differential outcome associated with different types of chemoprophylaxis16. Compliance with AOA recommended VTE chemical prophylaxis is flexible and surgeons are given discretion to use aspirin (100–300 mg), low molecular weight heparins or DOACs for people at routine VTE risk, although in this study aspirin was also used for people assessed as high VTE risk5. While it was not within the scope of this study, further research should explore why some surgeons used this flexibility with discretion to accommodate individual patient needs, while others prescribed noncompliant prophylaxis. These results may inform future revision of the AOA VTRE Prevention Guidelines.

While nonsignificant, we reported a trend towards a reduction in risk associated with non-compliant VTE prophylaxis. This contrasts with our previous analysis demonstrating noncompliance with the NHMRC VTE guideline was significantly associated with increased risk of VTE17. The main difference in the recommended prophylaxis is that aspirin is considered compliant for AOA VTE prevention clinical guidelines, but was not considered compliant for NHMRC VTE prevention clinical guideline5,7. The use of aspirin remains controversial although it is now recommended in several VTE clinical guidelines when used as part of a comprehensive VTE prophylaxis program. VTE, SSI and bleeding outcomes for aspirin prophylaxis have been shown to be no different to potent anticoagulants18,19,20. However other studies have reported aspirin is associated with an increased incidence of DVT and wound complications, but a lower risk of bleeding21,22,23. There has been a call for high quality RCT to confirm the efficacy and safety profile of aspirin, and several are now underway1,18,24,25.

This study has several limitations. The AOA guidelines target PE rather than DVT26; we included symptomatic DVT as both are associated with higher costs, lower patient satisfaction and are a target for Australian hospital accreditation27,28. The low event rate for PE prevented sensitivity analyses for PE alone. Calculation of patient risk was constrained by the available data; alternative criteria for compliance may yield different results. We may not have accounted for all the variation in prophylactic regimens. We did not assess the quality of the AOA VTE prevention clinical guidelines as the lack of peer-reviewed evidence regarding the methods or impact of the AOA VTE clinical guidelines precluded formal assessment. Research has demonstrated variable quality and issues with implementing other VTE clinical guidelines9,29, even though the patient and system benefits of increasing compliance14,30 can only be realised if the guidelines are of the highest quality8,9. Finally, our sample included a higher rate of THA and surgery performed in public hospitals than national registry data and the findings may not be generalisable to other populations5.

A key strength of this study lies in the comprehensive prospective clinical data to assess risk stratified VTE prophylaxis and compliance with the AOA VTE prevention clinical guidelines. Patient appropriate variations were considered in assessing compliance31,32. It is unclear whether the use of aspirin was associated with the increased VTE events we saw with people who experienced VTE while taking prophylaxis. Other potential factors are the AOA VTE prevention clinical guidelines recommendations are too broad to impact risk or a consequence of the varying recommendations between different guidelines or other factors28,33. Further research is recommended to explore factors influencing individualised risk assessment and prophylaxis, and the association with economic and patient reported outcomes27.

Conclusion

The aim of clinical guidelines is to help clinicians apply evidence from clinical trials into real-world practice and improve patient outcomes31. In contrast to this aim, we found non-compliance with AOA VTE prevention clinical guidelines was not associated with the risk of symptomatic 90-day VTE after THA/TKA, suggesting that these guidelines are not effective at reducing VTE events. The lack of data regarding the methodological quality and evidence underpinning the AOA VTE prevention clinical guidelines is concerning8,9. VTE is associated with high cost and burden for patients and the health system34. Definitive studies are needed to validate individualised VTE risk assessments and confirm effective personalised VTE prophylaxis strategies, particularly for those considered high risk, to improve patient outcomes and the value of arthroplasty surgery.