ORIGINAL INVESTIGATION Pharmacological Venous Thromboembolism Prophylaxis in Hospitalized Medical Patients A Meta-analysis of Randomized Controlled Trials Lironne Wein; Sara Wein; Steven Joseph Haas, BPharm, BPharmSci(Hons), MSHPA; James Shaw, MBBS, PhD, FRACP; Henry Krum, MBBS, PhD, FRACP Background: There is uncertainty regarding which phar- confidence interval [CI], 0.26-0.42) and pulmonary em- macological agents most effectively prevent venous throm- bolism (RR, 0.64; 95% CI, 0.50-0.82), as was LMWH (RR, boembolism in hospitalized medical patients. We there- 0.56; 95% CI, 0.45-0.70; and RR, 0.37; 95% CI, 0.21- fore performed a meta-analysis to determine this. 0.64, respectively). A UFH dosage of 5000 U 3 times daily was more effective in preventing DVT than a UFH dos- Methods: MEDLINE, EMBASE, and the Cochrane Cen- age of 5000 U twice daily when compared with the con- tral Register of Controlled Trials were searched from 1950, trol (RR, 0.27; 95% CI, 0.20-0.36; vs RR, 0.52; 95% CI, 1966, and 1800, respectively, through June 30, 2006, for 0.28-0.96). Neither UFH nor LMWH reduced mortal- randomized controlled trials that involved medical pa- ity. When directly compared with UFH, LMWH was as- tients comparing unfractionated heparin (UFH) or low- sociated with a lower risk of DVT (RR, 0.68; 95% CI, 0.52- molecular-weight heparin or heparinoid (LMWH) with 0.88) and injection site hematoma (RR, 0.47; 95% CI, a control, LMWH with UFH, or selective factor Xa in- 0.36-0.62), but no difference was seen between the 2 hibitors with a comparator. Study selection, validity as- agents in the risk of bleeding or thrombocytopenia. sessment, and data abstraction were performed by 2 in- dependent reviewers (L.W. and S.W.). Data synthesis was Conclusions: Both UFH and LMWH reduce venous undertaken by 1 blinded investigator (S.J.H.). thromboembolic risk in hospitalized medical patients, but neither agent alters mortality. When directly compared, Results: Thirty-six studies were included. Compared with LMWH is more effective in preventing DVT. the control, UFH was associated with a reduced risk of deep venous thrombosis (DVT) (risk ratio [RR], 0.33; 95% Arch Intern Med. 2007;167(14):1476-1486 ENOUS THROMBOEMBOLISM ting,13-15 even though medical patients rep- (VTE), which consists of resent most hospitalized patients and at deep venous thrombosis least 75% of fatal PEs occur in this group.16 (DVT) and pulmonary em- Unfractionated heparin (UFH), low- bolism (PE), is a major and molecular-weight heparin or heparinoid often unrecognized cause of morbidity and (LMWH), and selective factor Xa inhibi- V 1-9 mortality in hospitalized patients. Ap- tors are all used for the prevention of VTE. proximately 10% of hospital deaths can be Current International Consensus State- ment11 and American College of Chest Phy- Author Affiliations: National 17 Health and Medical Research See also pages sicians guidelines recommend the use of Council Centre of Clinical 1451 and 1471 UFH or LMWH in medical patients at risk Research Excellence in for VTE. We performed a meta-analysis of Therapeutics, Department of 10 randomized controlled trials to compare Epidemiology and Preventive attributed to pulmonary emboli. It is the efficacy and safety of the various agents Medicine, Faculty of Medicine, widely accepted that reliance on the diag- available for thromboprophylaxis. Nursing, and Health Sciences, nosis and treatment of an established event Monash University, Alfred is an inappropriate way to approach VTE METHODS Hospital (Ms L. Wein, Mr Haas, because diagnosis is often difficult and mas- and Drs Shaw and Krum), sive PE may be the first clinical manifesta- Baker Heart Research Institute tion of the disease.11,12 Prevention is there- STUDY SEARCH (Dr Shaw), and Faculty of Medicine, Dentistry, and Health fore of paramount importance. MEDLINE (via PubMed), EMBASE, and the Sciences, The University of Thromboprophylaxis is routinely used Cochrane Central Register of Controlled Trials Melbourne (Ms S. Wein), in surgical patients. However, it is not as were searched from 1950, 1966, and 1800, re- Melbourne, Victoria, Australia. widely practiced in the medical set- spectively, through June 30, 2006. MEDLINE (REPRINTED) ARCH INTERN MED/ VOL 167 (NO. 14), JULY 23, 2007 WWW.ARCHINTERNMED.COM 1476 ©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/30/2021 and the Cochrane Central Register of Controlled Trials were DVT. In the case of DVT, the total number of patients was deemed searched using the medical subject heading terms heparin, ve- to be the number who completed the trial as set out in the in- nous thrombosis, and pulmonary embolism. EMBASE was searched dividual study protocols for that outcome. The reason for this using the EMTREE keywords heparin and venous thromboem- discrepancy is that in some of the trials, several randomized pa- bolism. These databases were also searched using the term fac- tients did not undergo assessment for DVT. Therefore, the pro- tor Xa inhibitor. All searches were restricted to trials that in- portion of patients who underwent investigation who were found volved humans and were published in English. In addition, the to have DVT would more closely reflect the true rate of DVT than reference lists of all relevant trials were hand searched. the proportion of all patients who started treatment. STUDY SELECTION, VALIDITY ASSESSMENT, QUANTITATIVE DATA SYNTHESIS AND DATA ABSTRACTION Formal quantitative data synthesis was undertaken in a blinded Only prospective randomized controlled trials were consid- manner by 1 investigator (S.J.H.) in consultation with 2 other ered for inclusion in this meta-analysis. Studies were eligible investigators (L.W. and S.W.). Data analysis was conducted with if they compared (1) UFH with the control, (2) LMWH with STATA statistical software, version 8.2 (StataCorp, College Sta- the control, (3) LMWH with UFH, or (4) a selective factor Xa tion, Texas) using the Mantel-Haenszel fixed-effect method and inhibitor with a placebo. Studies were considered appropriate the DerSimonian and Laird random-effects method19,20 of meta- for inclusion only if they involved hospitalized medical pa- analysis for binary outcomes. The fixed-effects model calcu- tients and reported the occurrence of DVT, PE, and/or mortal- lates an average of the outcome statistic from each study, whereas ity after the administration of therapy. the random-effects model additionally considers the variabil- Studies with fewer than 30 patients were excluded. Studies ity among the studies analyzed. Results were presented as es- were also excluded if they examined thromboembolism in sur- timates of relative risk (RR) for each outcome measure, with gical, trauma, or critical care patients only or if the study au- 95% confidence intervals (CIs), in which relative weights were thors did not provide a subgroup analysis for medical patients assigned to each study on the basis of treatment group size and regarding the occurrence of thromboembolism. In addition, trials number of observed events.19,20 Heterogeneity ␹2 tests were per- that studied thromboembolism associated with central ve- formed in each analysis, and statistical significance was as- nous catheters in patients with cancer and trials that involved sumed at the .05 level via the use of statistical 2-way z tests. patients admitted to intensive care units were excluded. Sensitivity analyses were performed to assess the impact of each Study selection, validity assessment, and data abstraction were individual trial on the final pooled estimate for each outcome performed by 2 independent reviewers (L.W. and S.W.) in an measure. Funnel plots were constructed via RevMan Analy- unblinded standardized manner. Two other investigators (H.K. ses, version 1.0.5 (the Nordic Cochrane Centre, Copenhagen, and J.S.) were consulted whenever the need arose for further dis- Denmark) to investigate the potential for publication bias. cussion about the eligibility of a trial for inclusion in this meta- analysis or about results reported by individual studies. This pro- RESULTS cess was undertaken in accordance with the guidelines of the Quality of Reporting of Meta-analyses statement.18 SEARCH RESULTS OUTCOME MEASURES Our search for studies conducted in accordance with the AND APPLIED DEFINITIONS Quality of Reporting of Meta-analyses statement18 is sum- We were able to conduct a meta-analysis for a particular out- marized in Figure 1. We identified 936 potentially suit- come if it was investigated in 3 or more trials. DVT, PE, mortal- able articles. Of these, 54 examined the relevant end points 21-29 ity, and total bleeding were analyzed in studies that compared in appropriate patient groups. Nine of these trials were LMWH with the control, UFH with the control, and LMWH with excluded because they did not compare (1) UFH with the UFH. Results of all studies that compared UFH, LMWH, or fac- control, (2) LMWH with the control, (3) LMWH with UFH, tor Xa inhibitor with the control were combined to produce an or (4) a selective factor Xa inhibitor with a comparator. analysis of prophylaxis vs no prophylaxis for these outcomes. Four studies30-33 were excluded because they were not ran- Studies that compared LMWH with the control and LMWH with domized controlled trials, and 2 studies34,35 were ex- UFH had also provided further details regarding other outcome cluded because of potential biases from the study random- measures, which have been analyzed in this investigation: ma- ization process. One article36 was excluded because it was jor bleeding, minor bleeding, thrombocytopenia, and injection an abstract that outlined research later published in an- site hematoma. In addition, UFH dosages of 5000 U twice daily 37 38 and 5000 U 3 times daily were analyzed separately by compari- other journal. One additional study was excluded be- son with the control for the outcome of DVT.