Andrew F. Mah

Extended Out-of-Hospital Low-Molecular-Weight Heparin Prophylaxis against Deep Venous Thrombosis in Patients after Elective Hip Arthroplasty: A Systematic Review

The use of accurate, objective venographic testing to detect deep venous thrombosis in patients who undergo hip arthroplasty has led to randomized trials of various prophylactic regimens against venous thromboembolism (123). The need for in-hospital prophylaxis has been firmly established (24, 25) and accepted in clinical practice. Evidence-based medicine guidelines (26) based on venographic end points recommend low-molecular-weight heparin (LMWH) prophylaxis or warfarin prophylaxis for 7 to 10 days in patients who undergo elective hip surgery (25). These guidelines are considered a grade 1A recommendation, which indicates a strong recommendation for a therapy that has a clear benefit; the recommendation is based on randomized clinical trials that do not have important limitations and that can apply to most patients in most circumstances without reservation (26). Recent surveys indicate that more than 90% of patients who have undergone elective hip surgery have received thromboprophylaxis (27, 28). The results of randomized trials in Europe indicate the need for extended out-of-hospital prophylaxis in patients undergoing hip arthroplasty (2934). In contrast, on the basis of relatively low rates of symptomatic venous thromboembolism observed in descriptive studies in North America with long-term follow-up, investigators have inferred that extended prophylaxis is not required (3539). For these reasons, the reports of the Fifth (24) and Sixth (25) American College of Chest Physicians Consensus Conferences stated that extended out-of-hospital prophylaxis by using LMWH may offer additional protection. This is a 2A recommendation because of uncertainty regarding the riskbenefit ratio (24) and cost-effectiveness (25). A grade 2A recommendation indicates unclear benefit based on randomized clinical trials without important limitations and is an intermediate-strength recommendation (26). A possible interpretation of the North American data (3539) is that extended prophylaxis is unnecessary for patients in the United States and Canada because of differences in clinical practice, a shorter length of hospital stay, and earlier patient ambulation compared with Europe. A recent epidemiologic study (27) used a linked hospital discharge database provided by the State of California to report the outcomes in 19 586 patients undergoing total-hip arthroplasty and 24 059 patients undergoing knee arthroplasty. Of the patients having elective hip surgery who had symptomatic venous thromboembolism, the median time of the event was postoperative day 17, whereas the median time in patients having knee surgery was postoperative day 7; most patients (>90%) received in-hospital prophylaxis. These findings strongly suggest a need for extended out-of-hospital prophylaxis in patients undergoing total-hip replacement but not for patients undergoing total-knee replacement. Given the uncertainty about the need for extended prophylaxis, we performed a systematic review to provide clinicians with a practical pathway for translating clinical research into practice. Methods To ensure high methodologic quality, we adhered to the 15 criteria for systematic review outlined by McAlister and colleagues (40, 41). The first 10 criteria assess methodologic rigor, and the last 5 criteria assess the scientific basis of treatment recommendations (40). We also adhered to the QUOROM (Quality of Reporting of Meta-analyses) guidelines (42) for the reporting of meta-analyses of randomized trials. We systematically identified published and unpublished articles for inclusion in this analysis, described variations in study design and execution, evaluated study quality (43), and quantified the relative benefits of extended prophylaxis with LMWH (44). We excluded studies that did not use venography to assess the presence or absence of deep venous thrombosis because previous studies have shown that noninvasive tests, including duplex ultrasonography, are relatively insensitive for detecting thrombosis in patients who have undergone hip replacement (25). Study Identification We attempted to identify all published and unpublished randomized trials that compared extended prophylaxis with LMWH versus out-of-hospital placebo in patients undergoing hip arthroplasty. A strategy was developed for locating all studies in the PubMed and MEDLINE databases that were published between January 1976 and May 2001; the search was not restricted to English-language journals. We augmented our MEDLINE search by manually reviewing the reference lists of original articles and review articles. We also reviewed abstracts in conference proceedings and through the Cochrane Library Database and contacted investigators and pharmaceutical companies. Study Eligibility Two investigators independently evaluated studies for inclusion in the systematic review, and any disagreements were resolved by discussion between these two investigators. Investigators were not blinded to journal titles, author names, or institutional affiliations. Studies were included if they 1) enrolled patients undergoing elective hip arthroplasty, 2) randomly assigned participants to treatment groups, 3) investigated the extended posthospital discharge efficacy of once-daily subcutaneous LMWH compared with out-of-hospital placebo for prevention of deep venous thrombosis, 4) objectively documented the presence or absence of all episodes of deep venous thrombosis and proximal venous thrombosis by using bilateral ascending contrast venography, and 5) used objective methods for assessing bleeding complications (2932, 45, 46). Abstracts that reported full methods and results were eligible for inclusion. Deep venous thrombosis was defined as constant intraluminal filling defects in the deep veins; proximal venous thrombosis was defined as constant intraluminal filling defects in the popliteal deep veins or in the more proximal deep veins. Data Extraction One study investigator collected data on the following study-level factors: 1) the type of LMWH prophylaxis used, 2) whether a high-risk dose, approved by a regulatory affairs authority, was used, 3) the frequency of administration of LMWH, 4) the length of in-hospital stay, 5) the time interval after surgery when venography was performed [in days], and 6) venographic findings. For the last factor, we noted new out-of-hospital findings on venography or combined in-hospital and out-of-hospital findings on venography; where both findings were reported, we analyzed new out-of-hospital findings, which were more conservative and more recent. Two investigators independently extracted data on the major outcomes, which were the frequency of 1) all episodes of deep venous thrombosis, 2) proximal venous thrombosis, 3) symptomatic deep venous thrombosis and pulmonary embolism, and 4) major-bleeding complications. They also recorded data on other variables, including death, minor bleeding, wound hematomas, and thrombocytopenia. After the two investigators made their respective independent selection of studies for inclusion in the analysis, we compared their selections and calculated the percentage of agreement between them and the coefficient (47). Investigator disagreements were resolved by discussion. Assessment of Study Quality We assessed the quality and strength of each study by examining four key issues: 1) true randomization by using a random-numbers table or a computer program; 2) the masking of the allocation sequences from the investigators, staff, and patients involved in the study; 3) double-blinding [45]; and 4) the proportion of patients with successful (adequate) venography. One investigator extracted these data from the primary studies. When details were not reported, we requested additional information from the authors. Data Synthesis and Statistical Analysis For each of the major outcomes, we calculated summary treatment effects as the relative risk and the number needed to treat for benefit (NNTB) to prevent one thromboembolic event. The relative risk was used as the primary measure of treatment effect. We considered a P value less than 0.05 to be statistically significant for all statistical tests. Analyses were performed by using the metan procedure (48) of Stata software, release 6.0 (Stata Corp., College Station, Texas). To assess the validity of combining results from individual studies, we used the MantelHaenszel test for statistical heterogeneity (49). The outcome values were combined in both fixed-effects and random-effects models to estimate treatment effects on outcomes for all the studies. The relative risk ratios were consistent among studies for the treatment effect of preventing venographically documented deep venous thrombosis (all episodes and cases of proximal deep venous thrombosis); however, the 95% CIs for the relative risk ratios within studies were relatively wide. Therefore, we combined the data to provide more precise estimates of relative risk and NNTB. Results for NNTB were based on random-effects analysis of risk. Sensitivity Analysis We performed a sensitivity analysis for each of the three major outcomes. To uncover possible publication bias, we created inverted funnel plots for the major outcomes by plotting odds ratios against the sample size for each study (50). Moreover, to identify any studies that exerted a disproportionate influence on the summary treatment effect, we performed repeated calculations in which the data from each individual study were deleted, one at a time. Trials that met all inclusion criteria except for onethe use of bilateral ascending venography to assess end points at the end of the out-of-hospital study intervalwere included in a secondary meta-analysis of symptomatic venous thromboembolism and major bleeding end points. Results Study Identification and Selection Our search strategies identified 206 potentially relevant studies. After an initial scanning of titles and abstracts, we excluded 184 studies: 160 studies did