Prophylaxis of Thromboembolism in Critical Care (PROTECT)Trial:
a pilot study Deborah J. Cook MDa,b,*, Graeme Rocker MDc,
Maureen Meade MDa,b, Gordon Guyatt MDa,b, William Geerts MDd, David Anderson MDc, Yoanna Skrobik MDe, Paul Hebert MDf, Martin Albert MDe, Jamie Cooper MDg, Shannon Bates MDa, Christopher Caco MDa, Simon Finfer MDh, Robert Fowler MDd, Andreas Freitag MDa, John Granton MDd, Graham Jones MDa, Stephan Langevin MDi, Sangeeta Mehta MDd, Giuseppe Pagliarello MDf, Germain Poirier MDj, Christian Rabbat MDa, David Schiff MDk, Lauren Griffith MDb, Mark Crowther MDa, for the PROTECT Investigators and the Canadian Critical Care Trials Group A Department of Medicine, McMaster University, Hamilton, Ontario, Canada L8N 3Z5b Department of Clinical Epidemiology & Biostatistics, McMaster University, Hamilton, Ontario, Canada L8N 3Z5c Department of Medicine, Dalhousie University, Halifax, Nova Scotia, Canada B3H 3A7 d Department of Medicine, University of Toronto, Toronto, Ontario, Canada M4N 3M5 eDepartment of Medicine, Universite de Montreal, Montreal, Quebec, Canada H3C 3J7 fDepartment of Critical Care, University of Ottawa, Ottawa, Ontario, Canada K1Y 4E9 gDepartment of Medicine and Surgery, Monash University, Melbourne, Victoria, 3800 Australia hIntensive Therapy Unit, University of Sydney, Sydney, NSW 2006 Australia iDepartment of Medicine, University of Sheerbrooke, Quebec City, Quebec, Canada J1K 2R1 jCharles LeMoyne Hospital, Greenfield Park, Quebec, Canada J4V 2H1 kDepartment of Radiology, McMaster University, Hamilton, Ontario, Canada L8N 3Z5 Received 1 August 2005; revised 31 August 2005; accepted 8 September 2005 Abstract Purpose: There is no randomized trial comparing lowmolecular weight heparin (LMWH) and unfractionated heparin (UFH) for thromboprophylaxis in medical-surgical ICU patients. The primary objective of this randomized pilot study on LMWH vs UFH was to assess the feasibility of conducting a large randomized trial with respect to timely enrollment and blinded study drug administration, practicality of twice-weekly lower limb ultrasounds to screen for deep venous thrombosis, LMWH bioaccumulation and dose adjustment in renal insufficiency, and recruitment rates for a future trial in medical-surgical intensive care unit (ICU) patients. Its additional goals were to evaluate the suitability of 0883-9441/$ see front matter D 2005 Elsevier Inc. All rights reserved.doi:10.1016/j.jcrc.2005.09.010 4 Corresponding author. Departments of Medicine and Epidemiology and Biostatistics, McMaster University Health Sciences Center, Hamilton, Ontario, Canada L8N 3Z5. Tel.: +1 905 525 9140x22900; fax: +1 905 524 3841. E-mail address: debcook@mcmaster.ca (D.J. Cook). Keywords: Low molecular weight heparin; Unfractionated heparin; Thrombopro phylaxis; Deep venous thrombosis Journal of Critical Care (2005) 20, 364 372 the exclusion criteria and to document the range of research activities that precede accrual of patients into a trial to plan multisite management. Materials and Methods: By computerized telephone randomization, we allocated 129 medical-surgical ICU patients to treatment with dalteparin 5000 IU QD SC or that with UFH 5000 IU BID SC. Within each clinical center, only the study pharmacist was not blinded. We performed bilateral lower limb compression ultrasounds within 48 hours of ICU admission, twice weekly, on suspicion of deep venous thrombosis, and 7 days after ICU discharge. Research coordinators and investigators at 7 centers reported the time they engaged in all research activities before the first patient was randomized. Results: Timely complete study drug administration occurred after enrollment. More than 99% of scheduled doses were administered in a blinded fashion. Scheduled ultrasounds were performed without exception. No bioaccumulation of dalteparin was observed when creatinine clearance decreased to lower than 30 mL/min. Average recruitment was 2 patients/center per month before the study exclusion criteria were modified. Study startup activities required, on average, 65.5 hours of combined investigator and research coordinator time at each center. Careful examination of the accrual in the pilot study led to a reexamination of the Prophylaxis of Thromboembolism in Critical Care Trial (PROTECT) study exclusion criteria. Conclusions: This pilot study suggests that a multicenter randomized clinical trial comparing LMWH with UFH in critically ill medical-surgical patients is feasible. Pilot studies can improve the design of larger trials and may enhance successful timely completion. D 2005 Elsevier Inc. All rights reserved. 1. Introduction In this article, we describe the Prophylaxis of Thromboembolism in Critical Care Trial (PROTECT) Pilot Study. The PROTECT was designed as part of a multifaceted research program on venous thromboembolism (VTE) in medical-surgical critically ill patients (Fig. 1). To date, only 2 randomized trials on thromboprophylaxis in critically ill medical-surgical patients have been published [1,2]. One blinded trial allocated 199 medical-surgical ICU patients to subcutaneous unfractionated heparin (UFH) 5000 IU BID or to placebo injections [1]. Using serial fibrinogen lower limb scanning for 5 days, the researchers found the rates of deep venous thrombosis (DVT) to be 13% in the UFH group and 29% in the placebo group (relative risk reduction, . 65; P b .05). Rates of bleeding, pulmonary embolus (PE), and mortality were not reported. The second study was a multicenter trial in which 223 patients with an acute exacerbation of chronic obstructive pulmonary disease requiring mechanical ventilation for at least 2 days were randomized to the low molecular weight heparin (LMWH) nadroparin (3800 anti- Xa U for 45-70 kg or 5700 anti-Xa U for 71-110 kg) once daily or to placebo [2]. Patients were screened with weekly duplex ultrasounds. Venography was attempted in all patients at 21 days or earlier if ultrasound results were Fig. 1 The VTE in Medical-Surgical ICU Patients Research Program. VETEC indicates Venous Thromboembolism in Critical Care; CRIPTEC, Coagulation Risk Profiling for Thromboembolism in Critical Care; BEHAVE, Behavioral Reinforcement of Heparin to Avert Venous Emboli; BITEC, Burden of Illness: Thromboembolism in Critical Care. The Prophylaxis of Thromboembolism in Critical Care Trial 365 positive. The rate of DVT was 16% in the nadroparin group and 28% in the placebo group (relative risk reduction, .45; P b .05). The investigators observed trends toward increased bleeding (25 vs 18 patients; P = .18) and major hemorrhage (6 vs 3 patients; P = .28) in the patients receiving nadroparin. The frequency of PE was not systematically evaluated; 8 patients in each group died (P = .72). An extensive body of literature outside the ICU setting and in critically ill subgroups such as trauma patients [3] suggested that critically ill patients receiving LMWH will have a lower risk of venous thrombosis than similar patients receiving UFH prophylaxis. These 2 types of heparin have not been directly compared in critically ill medical-surgical ICU patients. Extrapolating from randomized trials in other high-risk groups, LMWH may be more effective than UFH.However, because it is more potent, LMWH may bioaccumulate at high doses and cause more bleeding thanUFH. Furthermore, it is more expensive than UFH. As a result, uncertainty [4], equipoise [5], and international practice variation [6] exist regarding which heparin to use for thromboprophylaxis in medical-surgical ICU patients. Only a rigorous and adequately powered randomized trial will resolve this issue.The PROTECT will be a large randomized, stratified, blinded, multicenter trial in medical-surgical ICU patients comparing UFH with LMWH for the prevention of DVT. The scientific objectives of the PROTECT will be to evaluate the effect of LMWH as compared with that of UFH on the primary outcome of lower limb DVT and the secondary outcomes of PE, bleeding, and heparininduced thrombocytopenia (HIT). To optimally preparefor PROTECT, we conducted the PROTECT Pilot Study to test several implementation strategies. In this article, we address 3 overall goals of this pilot study and report the lessons learned from each preparatory component of the study. 2. Objectives 2.1. Measurement of protocol adherence and recruitment The 5 feasibility objectives of the PROTECT Pilot Study were to assess the following: (1) timely enrollment; (2) complete, blinded study drug administration; (3) compliance with twice-weekly lower limb ultrasounds; (4) LMWH bioaccumulation and possible dose adjustment in renal insufficiency; and (5) recruitment rates for a future trial. 2.2. Evaluation of eligibility After completion of the PROTECT Pilot, a secondary objective was to reexamine the need for, and consequences of, conservative exclusion criteria on recruitment. 2.3. Assessment of workload before protocol implementation Beyond the immediate feasibility objectives of the PROTECT Pilot Study, a tertiary objective was to document the range of research activities that precede accrual of patients into a multicenter trial. 3. Methods 3.1. Description of pilot study In this concealed and blinded multicenter randomized pilot trial in 16 closed university-affiliated ICUs (14 in Canada and 2 in Australia), we included patients aged 18 years or older with an expected ICU stay of 72 hours or longer. We excluded patients admitted to the ICU following trauma, orthopedic, cardiac, or neurosurgery; those with severe hypertension, DVT, PE, or hemorrhage within 3 months; those with an international normalized ratio greater than 2 times the upper limit of normal (2_ ULN), partial thromboplastin time greater than 2_ ULN, or a platelet count of less than 100 _ 109/L; those with an estimated creatinine clearance of lower than 30 mL/min [7]; those requiring therapeutic anticoagulation; those who had received more than 2 doses of heparin in the ICU; and those with a contraindication to heparin or blood products. Patients underwent centralized telephone randomization to receive either dalteparin 5000 IU QD and placebo QD SC or UFH 5000 IU BID SC to conceal allocation. We stratified patients by center and by medical vs surgical admission status because surgical ICU patients may have a higher risk of VTE than medical patients. Moreover, our Canadian cross- sectional study showed differences in UFH prophylaxis prescribing between medical and surgical patients in that 1 to 2 doses of heparin were often deferred or omitted immediately postoperatively when patients were still mechanically ventilated [8]. In addition, surgical patients who have an epidural catheter may have been withheld of the study drug out of concern about epidural bleeding associated with LMWH at the time of catheter removal. Patients, families, ICU staff, ultrasound technologists, and research personnel were all blinded to drug allocation. The study pharmacist at each center was the only person who was not blinded. Trough anti-Xa levels were sampled from all patients whose calculated creatinine clearance declined to lower than 30 mL/min. Peak and trough anti- Xa levels were drawn every Tuesday and Thursday regardless of serum creatinine level in 5 centers participatingin an anti-Xa screening substudy, and anti-Xa levelswere also drawn if patients had a bleeding event. Mechanical thromboprophylaxes (antiembolic stockings and pneumatic compression devices) were not used except in the setting of prespecified conditions (severe thrombocytopenia 366 D.J. Cook et al. [platelet count, V50 _ 109/L]), coagulopathy [international normalized ratio of N2_ ULN or partial thromboplastin time of N2_ ULN], bleeding, and suspected or serologically proven [by serotonin release assay] HIT [9]). We disallowed mechanical thromboprophylaxis under other circumstances because no randomized trial has been performed in the ICU, published studies are of poor quality [10], standards for appropriate fitting and methods to monitor compliance do not exist, and use of mechanical thromboprophylaxis is uncommon in Canada [5,6,8]. Patients who developed suspected or serologically proven HIT had the study drug discontinued, all other heparin exposures were stopped, and danaparoid sodium or argatroban was administered. To diagnose proximal DVT, we performed bilateral lower limb venous compression ultrasounds within 48 hours of each patients enrollment, twice weekly, on suspicion of DVT, and 7 days after ICU discharge to diagnose proximal DVT. This required noncompressibility of one or more deep venous segments during ultrasound examination at 6 conventional sites: (1) the trifurcation of the deep calf veins; (2) distal popliteal; (3) proximal popliteal; (4) distal superficial femoral; (5) mid superficial femoral; and (6) common femoral. To ensure a standardized approach, we developed an operations manual and conducted an ultrasound reliability study in one center, documenting excellent agreement between 2 ultrasonographers blinded to each others reading. Suspected PE was evaluated with a protocol including bilateral lower limb ultrasound if this had not been performed already and computed tomographic angiogram in patients without any contraindication to intravenous contrast or transport to the radiology suite. Pulmonary embolus was diagnosed in the presence of an intraluminal filling defect in the main, lobar, or segmental branches of a pulmonary artery. Nondiagnostic computed tomographic angiogram required one or more of the following: a ventilation-perfusion lung scan in patients with normal chest x-ray, serial compression ultrasound, or pulmonary angiography findings, depending on the specific situation. Bleeding was defined as major or minor as per the participating center. In addition, all VTE and bleeding outcomes were adjudicated blinded to study medication using prespecified criteria. By design, the PROTECT Pilot Study was not powered to determine the relative benefit of LMWH vs UFH on the development of DVT. This important scientific question will be answered by the larger multicenter trial. We designed this pilot study on 120 patients to estimate the proportion of patients who would meet our feasibility criteria using confidence intervals. For example, if 110 of the 120 patients received every dose of their study drug, then the estimated rate would be 91.7% (95% confidence interval, 85.2%- 95.9%). We also considered it imperative to conduct a multicenter rather than a single-center pilot trial. Based on our experience with prior studies conducted through the Canadian Critical Care Trials Group [11], we projected that an enrollment of 120 patients and a minimum recruitment of 3 patients in each center would be sufficient to streamline screening and consent procedures, to refine protocol adherence and data collection methods, and to maximize the chance of identifying unforeseen barriers to the conduct of a larger trial across multiple centers. We conducted an intention-to-treat analysis maintaining the blinded group allocation. 3.1.1. Measurement of protocol adherence and recruitment A priori, we specified that the 5 feasibility objectives for the PROTECT Pilot Study would be considered successful if all of the following were met: 1. 115 of the 120 (98.5%) patients received the study drug within 12 hours of randomization; 2. 110 of the 120 (91.7%) patients received every scheduled dose of the study drug in a blinded manner; 3. 110 of the 120 (91.7%) patients had lower limb compression ultrasounds within 48 hours of ICU admission, 110 of the 120 (91.7%) patients had Monday and Thursday ultrasounds within 24 hours, and 90% or more of all surviving patients had a postICU discharge ultrasound within 7 to 10 days; 4.more than 90% of necessary dose adjustments were made appropriately in response to anti-Xa levels; and 5. 2 patients per month, on average, were randomized in each participating center. 3.1.2. Evaluation of eligibility To reevaluate the exclusion criteria for the PROTECT Pilot, we reviewed the first 9 months of screening. Over this period, we randomized 86 patients and excluded 575 patients. The most frequent reasons for exclusion from the PROTECT Pilot Study were the following: (1) creatinine clearance of lower than 30 mL/min (n = 193; 33.6% of reasons); (2) therapeutic anticoagulation (n = 123; 21.4%); (3) hemorrhage on ICU admission (n = 119; 20.7%); (4) platelet count of less than 100_109/L (n = 122; 21.2%); (5) coagulopathy (n = 71; 12.3%); and (6) more than 2 doses of heparin in the ICU (n = 62; 10.8%). We then reevaluated the exclusion criteria and delineated those criteria that could be modified to maximize enrollment and the generalizability of the results. 3.1.3. Assessment of workload before protocol implementation To document all research activities in the participating centers before recruitment, pairs of one research coordinator and the investigator at the 7 participating centers retrospectively reported their activities related to the startup of the pilot from the time of protocol receipt until the commencement of screening. Each pair reported startup activities independent from other pairs. These activities were classified into 8 categories that we established previously [12]: reviewing the protocol; individualizing the protocol to The Prophylaxis of Thromboembolism in Critical Care Trial 367 each research ethics board (REB) specification; adapting consent forms to each institution; having meetings and correspondence with relevant disciplines on logistics; procuring REB signature; responding (written) to scientific and ethical REB comments; having in-services; and holding other activities. 3.2. Research ethics The PROTECT Pilot Study received REB approval from all participating centers. We obtained written informed consent from patients next of kin. The PROTECT Pilot was funded by a peer reviewgranting agency, the Canadian Institutes of Health Research (200202MCT-98778), which had no influence on the design, conduct, analysis, interpretation, and write-up of the study. 4. Results We randomized 129 patients to test trial procedures fully in each center. The mean age of the randomized patients was 59.7 (SD, 14.8) years, with an APACHE II score of 20.9 (SD, 7.3); 42.2% were female; and 67.2% were medical patients. Of the cohort, 6 patients (4.7%) had a previous malignancy, 19 (14.8%) had an active malignancy, and, because 1 patient had both a history of malignancy and an additional active malignancy, 24 (18.75%) patients had malignancy overall. In addition, 108 patients (84.4%) required mechanical ventilation and 56 (43.8%) required inotropes or vasopressors. The baseline characteristics were comparable between groups. In this pilot study, the primary and secondary outcomes were analyzed on the entire cohort of enrolled patients rather than according to the 2 treatment groups. We found that 11 (8.6%) patients had a lower limb DVT. Secondary outcome results were as follows: 2 (1.6%) patients had a PE, 6 (4.7%) had clinically important bleeds, 23 (18%) had minor bleeds, 4 (3.1%) had at least one platelet count of less than 50 _ 109/L, and 2 (1.6%) had a positive HIT test by serotonin release assay, 1 of whom had a bilateral DVT from the trifurcation extending distally. 4.1. Measurement of protocol adherence and recruitment 1. Timely drug administration occurred for all patients within 12 hours of their randomization. 2. Complete study drug administration occurred for 97.8% of scheduled doses. Every dose of study medication was blinded, as reported by each center. However, in one center in which 2 patients were randomized the same day, the study pharmacist reported that study medication was switched in error for 24 hours. 3. The first ultrasound was performed within 48 hours of enrollment for 98.4% of the patients; subsequent ultrasounds occurred as scheduled without exception. 4. We did not observe bioaccumulation of heparin when administered in prophylactic doses in these critically ill patients with a range of renal dysfunction. In the 7 patients overall who developed a creatinine clearance of lower than 30 mL/min, trough anti-Xa levels were very low (0.1 IU/mL [interquartile range, b0.1- 0.1]). These results and data from the substudy patients [13] contributing to anti-Xa levels (4-hour peak, trough, and in the event of bleeding) are presented in Table 1. Only one dose adjustment was necessary according to an anti-Xa level done in a center participating in the substudy of anti-Xa level screening. The peak anti- Xa level of 0.51 IU/mL occurred on study day 2 in a patient weighing 40 kg who had not developed a creatinine clearance of lower than 30 mL/min. The subsequent dose was changed from 5000 to 2500 IU for remaining doses; the patient was transferred to another institution on study day 4 and study medication was discontinued. This led us to plan to include patients weighing 45 kg or heavier in the Table 1 Anti-Xa level results Allocation Trough levels in patients with creatinine clearance V30 mL/min Trough levels in all patients on Tuesdays and Thursdays Peak levels in all patients on Tuesdays and Thursdays Bleeding levels in patients at the time of bleeding events Group A [median (interquartile range); IU/mL] Eight tests in 2 patients Seventeen tests in 9 patients Seventeen tests in 9 patients Seven tests in 4 patients 0.1 (b0.1-0.1) b0.1 (b0.1-0.13) 0.11 (b0.1-0.27) 0.1 (b0.1-0.29) Group B [median (interquartile range); IU/mL] Twenty-three tests in 5 patients Forty-six tests in 12 patients Forty-six tests in 12 patients Ten tests in 8 patients b0.1 (b0.1-0.1) b0.1 (b0.1-0.1) b0.1 (b0.1-0.1) b0.1 (b0.1-0.1) Shown are the anti-Xa levels for patients undergoing anti-Xa testing for 4 reasons in a small substudy of the PROTECT Pilot. Trough anti-Xa levels in this range of 0.1 IU/mL are very low, indicating no evidence of bioaccumulation. Peak Anti-Xa levels in the range of 0.2 to 0.3 IU/mL are in the usual prophylactic range, approximately 4 hours postinjection; these results do not indicate excessive anticoagulant effects in either group. Overall, we found no difference between the UFH and LMWH groups (results remain blinded). 368 D.J. Cook et al. larger trial, as is customary in many thromboprophylaxis trials. 5. Projected recruitment for the future PROTECT is at least 2 patients/center per month and likely 3 to 4 patients per month after modifying 3 exclusion criteria [14,15] as described in the succeeding subsection. 4.2. Evaluation of eligibility To address impediments to recruitment, we modified 3 criteria for the large trial: creatinine clearance of lower than 30 mL/min, platelet count of less than 100 _ 109/L, and more than 2 doses of previous heparin. To clarify whether it was necessary to exclude ICU patients from the PROTECT based on low creatinine clearance, we began a second pilot study to determine whether bioaccumulation of LMWH occurs in patients with renal insufficiency receiving low-dose thromboprophylaxis (Dalteparins Influence on Renal Insufficiency in Critical Care Trial [DIRECT]). The main concern about bioaccumulation of LMWH in patients with renal failure is among those who receive therapeutic, not prophylactic, doses. If bioaccumulation does not occur, we will enroll patients in the PROTECT with a creatinine clearance of lower than 30 mL/min. Second, after informal consultation with the Canadian Critical Care Trials Group community, we reduced the exclusion for low platelet counts to less than 75 _ 109/L. Third, we are planning to recruit patients on weekends using an on-call system for research coordinators because 52% of patients excluded for reasons of prior heparin administration were identified on a Monday morning. We anticipate that these 3 modifications to the pilot study exclusion criteria will increase monthly accrual in the future PROTECT 2-fold to 3-fold [16]. 4.3. Assessment of workload before protocol implementation We identified a range of generic and site-specific activities to help operationalize the PROTECT Pilot protocol before the accrual of patients. These data are reported as the median hours per center (Table 2). Before screening patients for potential recruitment, we found that a mean of 65.6 hours per center was spent in these 7 centers. Research coordinators accounted for 70% of these prerandomization activities; site investigators, for 30%. Translating REB applications and consent forms into French resulted in a small increase in unanticipated workload [17]. 5. Discussion We designed the PROTECT Pilot Study with specific feasibility objectives to replicate most, if not all, aspects of the future trial across multiple sites. We conclude that a randomized clinical trial comparing the efficacy of LMWH with that of UFH for VTE prevention in medical-surgical ICU patients is feasible. The proximal DVT rates remained high in these medical-surgical ICU patients despite universal heparin thromboprophylaxis, underscoring the need for further research to guide improvements in thromboprophylaxis for critically ill patients. The first goal of the PROTECT Pilot Study was to evaluate our 4 specific feasibility objectives. Had delayed or missed drug administration occurred often, we would have had to analyze the process involving the study pharmacist, ICU pharmacist, bedside ICU nurse, and ICU team and then modify our approach accordingly. If centers had been unable to conduct ultrasound evaluations, then fewer events would be detected over the course of the trial, increasing the risk that the larger trial would be underpowered. Had there been inattention to anti-Xa levels and dose adjustments, we would have analyzed and improved the drug delivery system to ensure its safety. Had we not recorded a conservative recruitment rate through careful documenting of screening efforts in relation to recruitment [18] in this multicenter pilot study, we may have fallen into the common trap of overestimating patient accrual, thereby risking trial fatigue, and the threat of inappropriate early stopping [19]. Although safety issues must remain paramount in practice and clinical research, common overstringent exclusion criteria may increase perceived trial safety yet limit the generalizability of trial results and delay answers to Table 2 Prerandomization activities Site A Site B Site C Site D Site E Site F Site G Median (interquartile range) hours Range Protocol review 9 12 51 6 7 8 18 9 (7-18) 6-51 REB preparation 9 9 27 3 1 9 7 9 (3-9) 1-27 Consent adaptation 5 5 1 6 11 4.5 3 5 (3-6) 1-11 Meetings 23 19 17 10 4 16 14 16 (10-19) 4-23 Signatures 7 1 1 1 2 2 1.5 1.5 (1-2) 1-7 REB responses 10.5 3 2 4 6 5 1 4 (2-6) 1-10.5 In-services 19.5 5 2 4 6 11 24 6 (4-19.5) 2-24 Others 13 6 2 2 6 0 1.5 2 (1.5-6) 0-13 Total time 96 60 103 36 43 55.5 70 36-103 Total = 463.5 h (mean 66.2 h/site) Shown are the types of research activities engaged in by research coordinators and investigators before patients were randomized in the PROTECT Pilot. The Prophylaxis of Thromboembolism in Critical Care Trial 369 important clinical questions. Reevaluation of the PROTECT Pilot exclusion criteria will facilitate timely completion and enhance the applicability of the larger PROTECT study in several ways. The PROTECT Pilot indicated the need for another pilot study (DIRECT) to determine the safety of dalteparin 5000 IU SC OD among patients with severe renal insufficiency (creatinine clearance, b30 mL/min). We assessed this by measuring the proportion of patients with trough anti-Xa levels greater than 0.40 IU/mL after 3 F 1, 10 F 1, and 17 F 1 days of dalteparin prophylaxis treatment and by estimating the risk of major bleeding during the treatment period. We considered DIRECT to be a valuable contribution to the literature because experts recommend anti-Xa level measurement when patients with renal insufficiency receive treatment doses of LMWH [20], and our observations do not suggest that this is necessary for prophylactic doses of LMWH. Furthermore, our systematic review of observational studies on LMWH thromboprophylaxis in nonICU patients with dialysis-independent renal insufficiency showed that evidence supporting the need to monitor anti-Xa levels at a creatinine clearance of 30 mL/min or lower is neither strong nor consistent [21]. Reexamination of the PROTECT Pilot Study exclusion criteria underscores the importance of unhurried, careful planning before embarking on a large multicenter trial so that the ultimate study enrolls the most broadly suitable type of patients. Attention to detail of the research administration during the PROTECT Pilot provided several useful findings: foremost was the diverse range of time and costconsuming activities necessary before starting to recruit patients into this trial. Documenting prerandomization research activity can help plan project timelines, justify costs incurred for prerecruitment activities, and identify where such funding should be allocated. Strengths of the approach to examining prerandomization activities were the documentation of a full range of research activities and independent blinded data collection methods in geographically diverse centers. We subsequently presented our data on prerandomization activities for the PROTECT Pilot in the grant application for the future PROTECT. However, these activities were retrospectively self-reported and unverified, thus subject to random error and recall bias. Our data reflect the activities of only one pilot trial; however, we hope that the methods may serve as a template for analyzing other pilot studies with different designs in other settings [22]. The term pilot study is an informal term describing an investigation that serves to prepare for future clinical [23] or economic [24] studies. Pilot studies can be distinguished from pilot work in that the latter type is less likely to be standardized (ie, protocol) and less likely to lead to generalizable, published knowledge. Examples of pilot, or preparatory, work include reliability testing of survey instruments, focus groups to generate potential end points in long-term outcome studies, and refinement of complex interventions in preparation for randomized trials. Examples of pilot studies are investigations testing hypotheses about mechanisms of action [25], surveys of stated practice [26], chart reviews to estimate the burden of illness of a target condition [27], prospective observational studies to estimate expected event rates [28], a series of multicenter observational studies [23] to inform the design of a large randomized trial, systematic reviews of prior randomized trials to estimate relative risk reductions [29], and more useful multicenter randomized trials with specific feasibility objectives to replicate most, if not all, aspects of the future trial across several sites. The objectives of pilot studies vary widely, in keeping with their diverse designs. Examples of pilot study objectives are to estimate the safety of LMWH for stroke patients [30], to document the diagnostic accuracy of myocardial infarction for prehospital thrombolysis [31], and to learn about end-of-life informational needs and decision-making preferences through interviewing elderly acutely ill hospitalized patients [32]. To test the feasibility of rapid enrollment, recruitment efficiency, and 6-month retention, the CRASH investigators randomized 1000 patients in 52 emergency departments in 14 hospitals to thoroughly test their international emergency medicine megatrial protocol [33]. Pilot studies may promote clinical research ethics by maximizing the chance that future studies are designed rigorously, conducted safely and efficiently, and completed as planned. We conclude that pilot studies with explicit feasibility objectives and a priori success criteria are important foundation steps in preparing for large trials [23], for megatrials [33,34], and for research programs [35]. Ongoing formal review of the multifaceted issues inherent in the design and conduct of pilot studies can provide invaluable feasibility and scientific data for clinical trialists. We endorse completion of comprehensive pilot studies in preparation for multicenter randomized trials, particularly in light of the large number of international peer review ICU collaborations planned in the future [36,37]. Acknowledgments This trial was funded by the Canadian Institutes for Health Research (Ottawa, Ontario) and the Father Sean OSullivan Research Center (Hamilton, Ontario). Study medication was provided by Pharmacia, which played no role in the design, conduct, analysis, or interpretation of the results. We are grateful to the expert PROTECT Pilot Study research associates and Canadian Critical Care Trials Group for their key role in this work. We thank Nicole Zytaruk for her expert coordination of this study. We also thank John Marshall, Kang-Hoe Lee, and Fun-Gee Chen for discussions 370 D.J. Cook et al. about this pilot trial. D. Cook is the research chair of the Canadian Institutes for Health Research; M. Meade, an investigator of the Canadian Institutes for Health Research; M. Crowther, a career investigator of the Heart and Stroke Foundation of Ontario. Appendix A PROTECT Writing Committee: Deborah Cook, Graeme Rocker, Maureen Meade, Jamie Cooper, Yoanna Skrobik, Paul Hebert, and Martin Albert. PROTECT Pilot Research Coordinators and Investigators: Carlos Martinez and Sangeeta Mehta (Mount Sinai Hospital, Toronto, Ontario); Andrea Matte-Martyn, Marilyn Steinberg, Patrice Bret, Stephanie Wilson, and John Granton (University Health Network, General Site, Toronto, Ontario); Michelle Kho, Christine Wynne, Mark Duffett, Christopher Caco, and Andreas Freitag (Hamilton Health Sciences- McMaster University Medical Center, Hamilton, Ontario); Ellen McDonald, Terry Minuk, and Graham Jones (Hamilton Health Sciences-Henderson General Hospital, Hamilton, Ontario); Ellen McDonald, France Clarke, David Schiff, and Deborah Cook (St. Josephs Healthcare, Hamilton, Ontario); Julia Gibson, Ian Doris, and Maureen Meade (Hamilton Health Sciences-Hamilton General Hospital, Hamilton, Ontario); Andrea McNeil, Susan Pleasance, Derrick McPhee, and Graeme Rocker (Queen Elizabeth Health Sciences Center, Halifax, Nova Scotia); Craig Dale, Mel Keogh, Alan Moody, and Robert Fowler (Sunnybrook and Womens Hospital); Johanne Harvey, Michel Dube, Gaetan Barrette, Martine LeBlanc, and Yoanna Skrobik (Maissoneuve Rosemont, Montreal, Quebec); Louise Provost, Sylvie Groleau, and Germain Poirier (Charles LeMoyne, Montreal, Quebec); Anik Rioux, Genevieve Larouche, Jean-Marc Turcotte, Francois LeBlanc, and Stephan Langevin (LEnfant Jesus, Quebec City, Quebec); Carson Davidson, Irene Watpool, Tracy McArdle, and Paul Hebert (Ottawa Hospital General Site, Ottawa, Ontario); Mary-Jo Lewis, Julia Foxall, Bernard Lewandowski, and Giuseppe Pagliarello (Ottawa Hospital Civic Site, Ottawa, Ontario); Carole Sirois, Carole Nadon, Maria Danet, and Martin Albert (Hopital Sacre-Coeur, Montreal, Quebec); Catherine Harry, Kathryn Moulden, Samantha Ellis, and Jamie Cooper (Alfred Hospital, Melbourne Australia); and Julie Potter, Anne OConnor, Naresh Ramakrishnan, Charles Fisher, Christopher Ward, and Simon Finfer (Royal North Shore Hospital, Sydney, Australia). PROTECT Methods Center: Nicole Zytaruk (Project Coordinator), Suzanne Duchesne, Lauren Griffith, Christian Rabbat, Shannon Bates, Diane Heels-Ansdell, and Stephen Walter (McMaster University, Hamilton, Ontario). PROTECT Pilot Steering Committee: Deborah Cook (Principal Investigator), Mark Crowther, Maureen Meade, Gordon Guyatt, and Lauren Griffith (McMaster University, Hamilton, Ontario); David Anderson (Dalhousie University, Halifax, Nova Scotia); and William Geerts (University of Toronto, Toronto, Ontario). References [1] Cade JF. High risk of the critically ill for venous thromboembolism. Crit Care Med 1982;10:448- 50. [2] Fraisse F, Holzapfel L, Couland JM, et al. Nadroparin in the prevention of deep vein thrombosis in acute decompensated COPD. Am Rev Respir Crit Care Med 2000;161:1109- 14. [3] Geerts WH, Jay RM, Code KI, et al. A comparison of low-dose heparin with lowmolecular weight heparin as prophylaxis against venous thromboembolism after major trauma. N Engl J Med 1996;335:701- 7. [4] Geerts W, Pineo GF, Heit JA, Bergqvist D, Lassen MR, Colwell CW, et al. Prevention of venous thromboembolism. American College of Chest Physicians (ACCP) Antithrombotic Consensus Conference. Chest 2004;126(3):338S - 400S. [5] Cook DJ, et al, for the Canadian ICU Directors Group. Prevention and diagnosis of venous thromboembolism in critically ill patients: a Canadian survey. Crit Care 2001;5(6):336 - 42. [6] Lacherade JC, et al, for the French and Canadian ICU Directors Groups. Prevention of venous thromboembolism (VTE) in critically ill medical patients: a Franco-Canadian cross-sectional study. J Crit Care 2003;18(4):228- 37. [7] Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976;16(1):31 - 41. [8] Cook DJ, et al, for the Canadian ICU Directors Group. Prevention of venous thromboembolism in critically ill surgery patients. J Crit Care 2001:161- 6. [9] Warkentin TE, Levine ME, Hirsh J, et al. Heparin-induced thrombocytopenia in patients treated with low molecular weight heparin or unfractionated heparin. N Engl J Med 1995;332:1330 - 5. [10] Handoll HH, Farrar MJ, McBirnie J, et al. Heparin, low molecular weight heparin and physical methods for preventing deep vein thrombosis and pulmonary embolism following surgery for hip fractures. Cochrane Database Syst Rev 2000:CD000305. [11] Cook DJ, Todd TRJ. The Canadian Critical Care Trials group: a collaborative educational organization for the advancement of clinical ICU research. Intensive Care World 1997;14(2):68 - 70. [12] Clarke F, McDonald E, Rocker G, Cook DJ. Research coordinator activities in the ICU: an observational study. Abstract #108. Crit Care Med 2004;31(12(Suppl)):A26. [13] McDonald E, et al, for the PROTECT Investigators and Canadian Critical Care Trials Group. Anti-Xa levels in critically ill patients receiving dalteparin for thromboprophylaxis. Blood 2004;104(ii):407. [14] Zytaruk N, et al, for the Canadian Critical Care Trials Group. Prophylaxis for Thromboembolism in Critical Care Trial: a pilot study. Am J Respir Crit Care Med 2004;169(7):A666. [15] McDonald E, et al, for the PROTECT Investigators and Canadian Critical Care Trials Group. PROphylaxis for ThromboEmbolism in Critical care Trial. Blood 2004;104(ii):A1784. [16] McDonald E, et al, for the Canadian Critical Care Trials Group. Multicenter RCT pilot studies: exclusion criteria revisited. Am J Respir Crit Care Med 2004;169(7):A257. [17] Clarke FJ, et al, for the Canadian Critical Care Trials Group. Prerandomization multicenter research activity: a pilot trial. Am J Respir Crit Care Med 2004;169(7):A625. [18] Foster D, et al, for the Canadian Critical Care Trials Group. Use of a screen log to audit patient recruitment into multiple randomized trials in the ICU. Crit Care Med 2000;28(3):867 - 71. [19] Meade MO. Is it acceptable to stop large multicentre randomized controlled trials at interim analysis for futility? The hazards of stopping for futility. Crit Care 2005;9(1):35 - 6. [20] Weitz JI. Low molecular weight heparins. N Engl J Med 1997;337: 688- 98. The Prophylaxis of Thromboembolism in Critical Care Trial 371 [21] Nagge J, Crowther M, Hirsh J. Is impaired renal function a contraindication to the use of low molecular weight heparin? Arch Intern Med 2002;162:2605 - 9. [22] Cook DJ, Brower R, Cooper J, Brochard L, Vincent J-L. Multicenter clinical research in critical care. Crit Care Med 2002; 30:1636- 43. [23] Cook DJ, Heyland D, Marshall J, for the Canadian Critical Care Trials Group. On the need for observational studies to design and interpret randomized trials in ICU patients: a case study in stress ulcer prophylaxis. Intensive Care Med 2001;27: 347- 54. [24] Drummond M, Coyle D. The role of pilot studies in the economic evaluation of health technologies. Int J Technol Assess Health Care 1998;14(3):405- 18. [25] Quyyumi AA, Diodati JG, Lakatos E, Bonow RO, Epstein SE. Angiogenic effects of low molecular weight heparin in patients with stable coronary artery disease: a pilot study. J Am Coll Cardiol 1993; 22(3):635- 41. [26] Hebert PC, et al, for the Transfusion Requirements in Critical Care Investigators and The Canadian Critical Care Trials Group. A Canadian survey of transfusion practices in critically ill patients. Crit Care Med 1998;26(3):482- 7. [27] McIntyre LA, et al, for The Canadian Critical Care Trials Group. Are delays in the recognition and initial management of patients with severe sepsis associated with hospital mortality? Crit Care Med 2004;31(12(Suppl)):A75. [28] Cook DJ, Crowther M, Meade M, Rabbat C, Griffith L, Schiff D, et al. Deep venous thrombosis in medical-surgical critically ill patients: prevalence, incidence and risk factors. Crit Care Med 2005;33(7): 1565-1571. [29] Cook DJ, Reeve BK, Guyatt GH, Griffith LE, Heyland DK, Buckingham L, et al. Stress ulcer prophylaxis in critically ill patients: resolving discordant meta-analyses. JAMA 1996;275(4):308 - 14. [30] Kay R, Wong KS, Woo J. Pilot study of low molecular weight heparin in the treatment of acute stroke. Stroke 1994;25:684- 5. [31] Roth A, Barbash GI, Hanoch HOD, et al. Should thrombolytic therapy be administered in the mobile intensive care unit in patients with evolving acute myocardial infarction? A pilot study. J Am Coll Cardiol 1990;15(5):932 - 6. [32] Heyland DK, Tranmer J, Feldman-Stewart D. End of life in seriously ill hospitalized patients: an organizing framework and results of a preliminary study. J Palliat Care 2000;16:S31-9. [33] The CRASH Trial Pilot Study Collaborative Group. The MRC CRASH Trial: study design, baseline data and outcome in 1000 randomised patients in the pilot phase. Emerg Med J 2002;19:510- 4. [34] The SAFE Investigators. A comparison of albumin and saline for fluid resuscitation in the intensive care unit. N Engl J Med 2004; 350:2247- 56. [35] Cook DJ, et al, for the Level of Care Study Investigators and the Canadian Critical Care Trials Group. Levels of care in the intensive care unit: a research program [submitted for publication]. [36] Cook DJ. Is albumin SAFE? [Editorial]N Engl J Med 2004;350(22): 2294- 6. [37] Cook DJ, Finfer S. Our ICU in 2015: organizing clinical critical care research and implementing the results. In:Vincent JL, editor.Yearbook of intensive care and emergency medicine. Springer Verlag; 2005 [in press]. 372 D.J. Cook et al.