Publication

• Title: Dalteparin versus Unfractionated Heparin in Critically Ill Patients
• Acronym: PROTECT
• Year: 2011
• Journal published in: The New England Journal of Medicine
• Citation: PROTECT Investigators for the Canadian Critical Care Trials Group and the Australian and New Zealand Intensive Care Society Clinical Trials Group. Dalteparin versus unfractionated heparin in critically ill patients. N Engl J Med. 2011;364(14):1305-14.

Context & Rationale

• Background

  • Critically ill adults have high baseline risk of venous thromboembolism (VTE) due to immobility, inflammation, invasive devices, organ failure, and frequent contraindications/delays to pharmacological prophylaxis.
  • Unfractionated heparin (UFH) and low-molecular-weight heparins (LMWHs) were both used for ICU thromboprophylaxis, but comparative evidence in ICU populations was limited to small trials and indirect evidence.
  • Clinicians needed clarity on whether LMWH provides superior protection against clinically important events (particularly pulmonary embolism) without increasing bleeding, and whether it alters heparin-induced thrombocytopenia (HIT) risk.
  • Routine screening ultrasonography in ICU trials permitted capture of largely asymptomatic deep-vein thrombosis (DVT), enabling power for vascular endpoints where clinically apparent PE is relatively infrequent.
• Research Question/Hypothesis

  • Does once-daily dalteparin, compared with twice-daily UFH, reduce incident proximal lower-limb DVT in critically ill patients receiving protocolised screening compression ultrasonography?
  • Key secondary hypothesis: LMWH might reduce pulmonary embolism and/or HIT without excess bleeding.
• Why This Matters

  • ICU thromboprophylaxis is a near-universal intervention; even modest effect differences can translate to meaningful population-level benefit or harm.
  • Choosing between LMWH and UFH has practical implications (dosing frequency, monitoring in renal dysfunction, line patency heparin exposure, and HIT risk).
  • A definitive, blinded, international ICU RCT could resolve persistent equipoise and shape practice guidelines and procurement decisions.

Design & Methods

• Research Question: In critically ill adults, does dalteparin (5000 IU subcutaneously once daily) reduce incident proximal leg DVT compared with UFH (5000 IU subcutaneously twice daily) during ICU stay?
• Study Type: Randomised, multicentre, investigator-initiated, international, double-blind, double-dummy, parallel-group trial conducted in ICUs (67 ICUs in Canada, United States, United Kingdom, Australia, Brazil, Saudi Arabia).
• Population:
  • Adults admitted to ICU expected to remain in ICU for at least 3 days.
  • Key exclusions: major trauma; neurosurgery; orthopaedic surgery; need for therapeutic anticoagulation; receipt of heparin in ICU for ≥3 days; contraindication to heparin or blood products; pregnancy; limitations of life support; enrolment in a related trial.
  • Baseline leg ultrasonography planned within 2 days after ICU admission, enabling identification of prevalent DVT at enrolment.
• Intervention:
  • Dalteparin 5000 IU subcutaneously once daily.
  • Plus once-daily placebo injection (double-dummy) to maintain indistinguishable twice-daily administration.
  • Study drug continued for duration of ICU stay, with protocol allowances for withholding during bleeding/thrombocytopenia and use of mechanical prophylaxis where indicated.
• Comparison:
  • Unfractionated heparin 5000 IU subcutaneously twice daily (matching injection schedule).
  • Co-interventions (e.g., aspirin, clopidogrel, warfarin, low-dose heparin for line patency, pneumatic compression) permitted and recorded.
• Blinding: Double-blind with a research pharmacist at each centre preparing identically appearing syringes; investigators, bedside clinicians, patients, outcome assessors, and adjudicators were blinded to allocation.
• Statistics: A total of 3618 patients were required to detect a 30% relative reduction in incident proximal DVT (from 8.0% to 5.6%) with 80% power at the 5% significance level; primary analyses were intention-to-treat (excluding patients who withdrew consent), using unadjusted Cox regression to estimate hazard ratios and 95% confidence intervals, with prespecified subgroup analyses.
• Follow-Up Period: Screening compression ultrasonography within 2 days of ICU admission and twice weekly while in ICU; clinical events assessed until death or discharge, with data censored at death or discharge (or at 100 days if still hospitalised).

Key Results

This trial was not stopped early. Enrolment completed at 3764 randomised patients (intention-to-treat cohort 3746 after withdrawals of consent).

Outcome	Dalteparin	UFH	Effect	p value / 95% CI	Notes
Incident proximal DVT (primary endpoint)	96/1873 (5.1%)	109/1873 (5.8%)	HR 0.92	95% CI 0.68 to 1.23; P=0.57	Time-to-event analysis; proximal DVT detected by scheduled compression ultrasonography.
Any DVT	146/1873 (7.8%)	148/1873 (7.9%)	HR 0.98	95% CI 0.78 to 1.25; P=0.82	Includes distal + proximal DVT.
Pulmonary embolism (definite or probable)	23/1873 (1.2%)	39/1873 (2.1%)	HR 0.53	95% CI 0.30 to 0.92; P=0.02	Clinically suspected and adjudicated; includes “definite” and “probable” categories.
Any VTE (DVT or PE)	154/1873 (8.2%)	186/1873 (9.9%)	HR 0.83	95% CI 0.68 to 1.03; P=0.10	Directionally favoured dalteparin; not statistically significant.
Major bleeding	103/1873 (5.5%)	105/1873 (5.6%)	HR 1.00	95% CI 0.75 to 1.34; P=0.96	Major bleeding definition included critical-site bleeding, transfusion ≥2 units within 24 hours, or surgical intervention.
Any bleeding	440/1873 (23.5%)	459/1873 (24.5%)	HR 0.95	95% CI 0.84 to 1.09; P=0.36	Broad bleeding outcome; no signal for excess bleeding with dalteparin.
Heparin-induced thrombocytopenia	5/1873 (0.3%)	12/1873 (0.6%)	HR 0.47	95% CI 0.16 to 1.37; P=0.16	Rare outcome; diagnosis supported by heparin–PF4 testing with central serotonin-release assay availability.
Death in ICU	253/1873 (13.5%)	284/1873 (15.2%)	HR 0.87	95% CI 0.74 to 1.02; P=0.09	No statistically significant mortality difference.
Death in hospital	406/1873 (21.7%)	445/1873 (23.8%)	HR 0.90	95% CI 0.78 to 1.03; P=0.12	Followed until discharge (censoring at 100 days if still hospitalised).

• Despite very large sample size and protocolised ultrasound surveillance, dalteparin did not reduce the primary endpoint of incident proximal DVT (5.1% vs 5.8%).
• Dalteparin was associated with fewer adjudicated definite/probable pulmonary emboli (1.2% vs 2.1%), without an increase in major bleeding (5.5% vs 5.6%).
• Directionally favourable signals for “any VTE” and mortality did not meet conventional statistical significance thresholds in the prespecified analyses.

Internal Validity

• Randomisation and Allocation:
  • Central randomisation with allocation concealment; stratified by centre and type of ICU admission (medical vs surgical).
  • Double-dummy preparation by research pharmacy supported concealment at bedside.
• Drop out or exclusions (post-randomisation):
  • Randomised: 3764; intention-to-treat cohort: 3746 (1873 per group) after withdrawals of consent.
  • Supplementary as-treated analysis excluded randomisation errors and patients receiving no study drug; per-protocol analysis excluded prevalent VTE treated, ≤1 ultrasound, or only 1 day of study drug (substantial exclusions for ultrasound-based endpoints). ²
• Performance/Detection Bias (blinding feasibility):
  • Blinding was strong: identical syringes, twice-daily injections in both groups, and blinded adjudication of clinical events.
  • Primary endpoint relied on scheduled imaging (less vulnerable to clinician suspicion bias) but vulnerable to missing imaging and to clinical importance debates.
• Protocol Adherence:
  • Doses of study drug were missed on 3.3% of study days; nonstudy anticoagulant used on 1.0% of study days.
  • Median duration of exposure to study drug was 7 days (IQR 4 to 12) in both groups.
  • Key concomitant drug/device exposures were similar (e.g., antiplatelet therapy 9.3% vs 9.7%; renal replacement therapy 7.5% vs 7.1%).
• Baseline Characteristics:
  • Groups were well balanced: age 61.1±16.5 vs 61.7±16.1; APACHE II 21.4±7.8 vs 21.7±7.8; mechanical ventilation 89.3% vs 91.1%; vasopressors 43.2% vs 46.8%.
  • Medical admissions predominated (75.2% vs 75.9%), supporting relevance to general mixed ICUs.
• Heterogeneity:
  • Prespecified subgroup analyses for the primary endpoint did not show effect modification (all interaction P values >0.05). ²
  • Examples: medical admissions HR 0.89 (95% CI 0.64 to 1.25) vs surgical HR 1.02 (95% CI 0.60 to 1.71); P for interaction 0.41. ²
• Timing:
  • Baseline ultrasonography was protocolised within 2 days of ICU admission, supporting early identification of prevalent DVT and early onset surveillance.
  • Scheduled twice-weekly surveillance helped standardise DVT detection across centres, but per-protocol exclusions for limited ultrasonography imply nontrivial missingness in practice. ²
• Dose:
  • Dalteparin: 5000 IU once daily; UFH: 5000 IU twice daily (both standard prophylactic regimens in many ICUs).
  • The trial did not evaluate higher-frequency UFH regimens (e.g., three times daily), which may limit inference where such dosing is standard.
• Separation of the Variable of Interest:
  • Distinct anticoagulant class and dosing frequency by design; background co-interventions broadly similar (e.g., aspirin 6.2% vs 6.7%; unfractionated heparin for line patency 29.6% vs 28.1%).
  • Mechanical prophylaxis use was also similar: antiembolic stockings 17.6% vs 16.6%; pneumatic compression 10.8% vs 11.4%.
• Outcome Assessment:
  • Objective vascular endpoints: compression ultrasonography for DVT; adjudicated PE with graded certainty (definite/probable/possible).
  • Major bleeding was prespecified and clinically anchored (critical-site bleeding, transfusion ≥2 units RBC within 24 hours, or surgical intervention).
• Statistical Rigor:
  • Sample size target met (randomised 3764 vs planned 3618), with prespecified Cox-model time-to-event analyses and clearly reported effect estimates and confidence intervals.
  • Primary endpoint neutral; statistically significant PE reduction emerged for a secondary endpoint, requiring cautious multiplicity-aware interpretation.

Conclusion on Internal Validity: Overall, internal validity appears strong: robust allocation concealment and blinding, balanced baseline characteristics, high protocol separation, and prespecified analyses. The main constraints are reliance on an imaging-detected primary endpoint with incomplete surveillance in a minority (as reflected by per-protocol exclusions) and interpretive caution around statistically significant secondary outcomes.

External Validity

• Population Representativeness:
  • Broad, mixed ICU population dominated by medical admissions, with high illness severity (APACHE II ~21) and very high mechanical ventilation rates (~90%).
  • Key exclusions (major trauma, neurosurgery, orthopaedic surgery, therapeutic anticoagulation needs, pregnancy) limit applicability to several high-risk ICU subgroups.
• Applicability:
  • Clinical outcomes (PE, bleeding, mortality) are broadly transportable across systems; scheduled ultrasonography surveillance is less representative of routine care.
  • Centres had capacity for systematic screening and trial pharmacy blinding; implementation in lower-resource settings may differ, particularly where LMWH cost or monitoring constraints exist.
  • Renal dysfunction considerations remain practice-defining; subgroup results did not show interaction by dialysis status, but dalteparin pharmacokinetics and institutional LMWH policies vary.

Conclusion on External Validity: Generalisability is moderate: the trial is highly informative for adult mixed medical–surgical ICUs using UFH twice daily as standard prophylaxis, but extrapolation to trauma/neurosurgical populations and to settings without routine imaging surveillance should be cautious.

Strengths & Limitations

• Strengths:
  • Large sample size (n=3764) with adequate power for the prespecified primary endpoint.
  • International, multicentre ICU trial with rigorous double-blind, double-dummy design.
  • Systematic, protocolised ultrasonography enabling robust ascertainment of proximal DVT and separation from “clinically triggered” imaging.
  • Clinically relevant secondary outcomes (PE, bleeding, mortality) with adjudication processes.
• Limitations:
  • Primary endpoint is an imaging-detected surrogate (proximal DVT), which may not align perfectly with outcomes that drive ICU harm (PE, death).
  • PE outcome incorporated an adjudication schema; statistically significant reduction was demonstrated for “definite or probable” PE, but PE remains a relatively infrequent event and is a secondary endpoint.
  • Comparator was UFH 5000 IU twice daily; results may not generalise to practices using more frequent UFH dosing.
  • Exclusion of several high-risk ICU subgroups (major trauma, neurosurgical, orthopaedic) limits applicability to those populations.

Interpretation & Why It Matters

• Clinical practice signal

  • For mixed ICU populations receiving UFH twice daily, dalteparin did not improve proximal DVT prevention but reduced adjudicated definite/probable PE (1.2% vs 2.1%) without bleeding penalty.
  • This creates a credible “clinical outcome” argument for LMWH use when bleeding risk is acceptable and renal function policies allow.
• Mechanistic and methodological insight

  • The dissociation (no proximal DVT reduction yet fewer PEs) highlights complexities of ICU thromboembolism biology and the limitations of leg-ultrasound–centred endpoints.
  • PROTECT exemplifies how a neutral surrogate primary endpoint can coexist with a potentially important secondary clinical benefit, demanding careful interpretation rather than simplistic “negative trial” labelling.
• Operational considerations

  • Once-daily LMWH may reduce administration burden compared with twice-daily UFH, but institutional decisions must also incorporate renal impairment pathways, cost, and HIT surveillance.

Controversies & Subsequent Evidence

• Primary endpoint choice (screen-detected proximal DVT) vs clinical outcomes:
  • Correspondence highlighted that asymptomatic, screen-detected DVT can be a contested surrogate for patient-important outcomes, and that interpretation should focus on PE, bleeding, and mortality. ¹
  • The authors’ reply in the same correspondence emphasised that most PEs were clinically suspected (rather than incidental) and associated with worse outcomes, supporting PE as a meaningful endpoint. ¹
• Multiplicity and interpretive hierarchy:
  • The trial’s primary endpoint was neutral; the PE reduction, although clinically compelling and statistically significant (HR 0.53; 95% CI 0.30 to 0.92; P=0.02), is a secondary outcome and should be interpreted within the prespecified analytic framework and multiplicity context.
• HIT signal and analytical sensitivity:
  • In the main trial report, HIT was numerically lower with dalteparin (0.3% vs 0.6%) but not statistically significant (P=0.16).
  • Per-protocol analysis in the supplementary appendix suggested a larger relative difference (3/1566 [0.2%] vs 12/1561 [0.8%]; HR 0.27; 95% CI 0.08 to 0.98; P=0.046), but this rests on a reduced analysis population and a small number of events. ²
• Subgroup credibility and consistency:
  • Prespecified subgroup analyses for the primary endpoint did not show effect modification (interaction P values all >0.05), limiting arguments for targeted LMWH use based solely on the trial’s subgroup signals. ²

Summary

• PROTECT randomised 3764 ICU patients to dalteparin 5000 IU once daily vs UFH 5000 IU twice daily in a double-blind, double-dummy international ICU trial.
• The primary endpoint (incident proximal DVT detected by scheduled compression ultrasonography) was not reduced (5.1% vs 5.8%; HR 0.92; 95% CI 0.68 to 1.23; P=0.57).
• Dalteparin reduced adjudicated definite/probable PE (1.2% vs 2.1%; HR 0.53; 95% CI 0.30 to 0.92; P=0.02) without increasing major bleeding (5.5% vs 5.6%; P=0.96).
• Mortality differences did not meet statistical significance (ICU death 13.5% vs 15.2%; P=0.09; hospital death 21.7% vs 23.8%; P=0.12).
• Methodological debate focused on the surrogate nature of the ultrasound-detected primary endpoint and the interpretation of clinically important secondary outcomes (notably PE).

Overall Takeaway

PROTECT is a landmark ICU thromboprophylaxis trial because it combined robust blinding and international scale with systematic outcome ascertainment, providing high-certainty comparative evidence for LMWH versus UFH in critically ill adults. Although the primary surrogate endpoint (proximal DVT) was neutral, the observed reduction in adjudicated pulmonary embolism without increased bleeding materially shaped how clinicians weigh LMWH versus UFH in routine ICU care.

Bibliography

• Dalteparin in critically ill patients. N Engl J Med. 2011;365(2):179-180.
• Supplementary Appendix to: Dalteparin versus unfractionated heparin in critically ill patients. N Engl J Med. 2011;364(14):1305-1314.