Publication

• Title: Pulmonary-Artery versus Central Venous Catheter to Guide Treatment of Acute Lung Injury
• Acronym: FACTT (catheter component)
• Year: 2006
• Journal published in: New England Journal of Medicine
• Citation: National Heart, Lung, and Blood Institute Acute Respiratory Distress Syndrome (ARDS) Clinical Trials Network. Pulmonary-artery versus central venous catheter to guide treatment of acute lung injury. N Engl J Med. 2006;354(21):2213-2224.

Context & Rationale

• Background

  • Pulmonary-artery catheterisation (PAC) was widely used in critical illness to estimate filling pressures, cardiac output, and mixed venous oxygenation, with the aim of improving haemodynamic decision-making.
  • Practice was increasingly questioned because procedure-related complications are non-trivial, and prior evidence did not establish patient-centred benefit across heterogeneous ICU populations.
  • In acute lung injury (ALI), fluid and vasopressor decisions are frequent and high-stakes; clinicians often used either PAC-derived measures (e.g., pulmonary-artery occlusion pressure) or central venous pressure (CVP) from a central venous catheter (CVC) to guide therapy.
  • The NHLBI ARDS Network FACTT platform enabled a factorial evaluation of both monitoring modality (PAC vs CVC) and protocolised fluid strategy (liberal vs conservative).
• Research Question/Hypothesis

  • In mechanically ventilated adults with ALI enrolled early after meeting diagnostic criteria, does PAC-guided protocolised management (vs CVC-guided protocolised management) reduce death before hospital discharge home within 60 days?
  • Does catheter type influence key morbidity outcomes (ventilator-free and ICU-free days) and iatrogenic harms (catheter-related complications and infections)?
• Why This Matters

  • Clarifies whether the additional haemodynamic information from PACs translates into improved outcomes in ALI when clinical decisions are protocolised.
  • Quantifies trade-offs between potential benefit and procedure-related harms (notably arrhythmia, pneumothorax, and infection).
  • Informs invasive monitoring policy, training, and resource allocation in critical care practice.

Design & Methods

• Research Question: In adults with acute lung injury, does pulmonary-artery catheter–guided protocolised management improve 60-day death-before-discharge-home (and related morbidity) compared with central venous catheter–guided protocolised management?
• Study Type: Randomised, multicentre, NHLBI ARDS Network, factorial trial (PAC vs CVC; and simultaneously liberal vs conservative fluid strategy), open-label, protocol-driven management in ICU settings.
• Population:
  • Setting: adult ICUs in ARDS Network participating hospitals.
  • Key inclusion features: intubated positive-pressure ventilation; ALI criteria including bilateral infiltrates; PaO₂/FiO₂ ≤ 300; no clinical evidence of left atrial hypertension; enrolled within 48 hours of meeting ALI criteria.
  • Key exclusion features: ALI for >48 hours; pre-existing PAC after onset of ALI; inability to obtain consent; chronic conditions expected to confound survival or weaning (including selected severe chronic cardiopulmonary, renal, neurological, hepatic, or neuromuscular disease) and other protocol-defined exclusions.
• Intervention:
  • Pulmonary-artery catheter insertion (targeted within 4 hours of randomisation).
  • Protocolised haemodynamic management using PAC-derived measures (including pulmonary-artery occlusion pressure and cardiac index) to drive fluid/diuretic/vasopressor/inotrope instructions.
  • Concurrent randomisation to a protocolised fluid strategy (liberal or conservative) applied within the PAC-guided algorithm.
• Comparison:
  • Central venous catheter insertion (targeted within 4 hours of randomisation).
  • Protocolised haemodynamic management using CVP-based targets (and the same clinical protocol framework) to drive fluid/diuretic/vasopressor/inotrope instructions.
  • Concurrent randomisation to the same liberal vs conservative fluid strategy, but implemented using CVC-derived targets.
• Blinding: Unblinded (catheter type is intrinsically difficult to blind); primary and key secondary outcomes were largely objective, reducing (but not eliminating) detection bias.
• Statistics: A total planned sample size of approximately 1000 patients was designed to detect a 10 percentage point reduction in the primary endpoint with 90% power at the 5% two-sided significance level; analyses were primarily intention-to-treat, with assessment for interaction between catheter type and fluid strategy in the factorial design.
• Follow-Up Period: Primary endpoint assessed through 60 days after randomisation; key morbidity outcomes reported through day 28 (e.g., ventilator-free and ICU-free days), with additional in-hospital safety outcomes including catheter complications and infections.

Key Results

This trial was not stopped early. The prespecified enrolment target was achieved (1001 patients randomised; PAC 513, CVC 488).

Outcome	Pulmonary-artery catheter (PAC)	Central venous catheter (CVC)	Effect	p value / 95% CI	Notes
Death before hospital discharge home within 60 days	141/513 (27.4%)	128/488 (26.3%)	Absolute difference (PAC−CVC) 1.1 percentage points	95% CI −4.4 to 6.6; P=0.69	Primary endpoint
Ventilator-free days through day 28	13.2 ± 0.5	13.5 ± 0.5	Difference (PAC−CVC) −0.3 days	95% CI −1.3 to 0.7; P=0.58	Higher is better
ICU-free days through day 28	12.0 ± 0.5	12.5 ± 0.5	Difference (PAC−CVC) −0.5 days	95% CI −1.5 to 0.6; P=0.40	Higher is better
Any catheter-related complication	100/513 (19%)	41/488 (9%)	Not reported	P<0.001	Composite of mechanical, arrhythmic, infectious, and thrombotic complications
Catheter-related arrhythmia requiring treatment	60/513 (12%)	1/488 (0.2%)	Not reported	P<0.001	Dominant driver of excess complications with PAC
Pneumothorax	3/513 (0.6%)	10/488 (2.0%)	Not reported	P=0.047	Mechanical complication; higher absolute count in CVC arm
Packed red-cell transfusion during study period	38%	30%	Not reported	P=0.008	Potential signal of differential procedural/bleeding practice or illness trajectory; mechanism not established in trial report
Positive blood cultures during catheter period	14.2%	9.4%	Not reported	P=0.02	Higher rate in PAC group in reported analysis
Kidney replacement therapy	14%	11%	Not reported	P=0.15	No statistically significant difference reported

• • There was no detectable benefit of PAC over CVC for the primary outcome (27.4% vs 26.3%; absolute difference 1.1 percentage points; 95% CI −4.4 to 6.6; P=0.69) despite early enrolment and protocolised management.
  • Key morbidity outcomes were similar (ventilator-free days 13.2 ± 0.5 vs 13.5 ± 0.5; P=0.58; ICU-free days 12.0 ± 0.5 vs 12.5 ± 0.5; P=0.40).
  • Procedure-related harm was higher with PAC (any catheter complication 19% vs 9%; P<0.001), mainly arrhythmias requiring treatment (12% vs 0.2%; P<0.001).

Internal Validity

• Randomisation and Allocation:
  • Patients were randomised to PAC vs CVC with stratification by hospital and by concurrently assigned fluid strategy within a factorial design.
  • The manuscript describes central, protocolised randomisation but does not detail allocation concealment procedures beyond central assignment.
• Drop out or exclusions:
  • 1001 patients were randomised (PAC 513; CVC 488); loss to follow-up for the primary endpoint was not reported in the manuscript.
• Performance/Detection Bias:
  • Unblinded intervention increases risk of co-intervention and performance bias; however, primary and key secondary endpoints were objective and protocol-defined.
• Protocol Adherence:
  • Catheter protocol use was high (reported as approximately 90% on each study day in both groups).
  • Protocol instructions were followed at similar rates (91 ± 1% in PAC group vs 88 ± 1% in CVC group; P=0.12).
• Baseline Characteristics:
  • Groups were similar on key severity indices (e.g., age 49.9 ± 0.7 vs 49.6 ± 0.7 years; APACHE III 94.7 ± 1.4 vs 93.5 ± 1.4; PaO₂/FiO₂ 158.9 ± 3.8 vs 151.3 ± 3.7).
  • Shock at baseline was common and comparable (37% vs 32%; P=0.11).
• Heterogeneity:
  • No interaction between catheter type and fluid strategy was reported for the primary endpoint (interaction P=0.39).
  • Subgroup analyses by shock at baseline did not demonstrate differential benefit (effect estimates not reported in the manuscript text).
• Timing:
  • Protocol targeted catheter insertion within 4 hours of randomisation; actual insertion timing distribution was not reported.
  • Time from qualification to first protocol instruction differed (3.45 ± 0.17 vs 2.15 ± 0.10 hours; P<0.001), suggesting operational timing differences that did not translate into outcome differences.
• Dose:
  • Duration of catheter use and protocol exposure by group was not reported beyond daily protocol use estimates.
• Separation of the Variable of Interest:
  • Crossover was rare (PAC→CVC: 4 patients; CVC→PAC: 5 patients), supporting strong separation by assigned monitoring modality.
  • Management activity differed modestly (management instructions per day 4.8 ± 0.1 in PAC vs 4.4 ± 0.2 in CVC; P=0.03) and dobutamine was used more often in PAC patients (7% vs 2%; P<0.001).
  • Within PAC patients, haemodynamic abnormalities that could plausibly prompt tailored therapy were present (PAOP >18 mm Hg in 29%; cardiac index <2.5 L/min/m² in 8%; both in 3%), yet outcomes were unchanged.
• Key Delivery Aspects:
  • The trial specifically tested protocolised management tied to haemodynamic targets rather than unrestricted clinician-directed use of catheter-derived information.
  • Fluid/diuretic instruction proportions were similar (fluids 10 ± 1% vs 12 ± 1%; P=0.10; diuretics 27 ± 1% vs 24 ± 1%; P=0.16).
• Crossover:
  • Crossovers were uncommon (total 9/1001), limiting dilution of treatment effect estimates.
• Adjunctive therapy use:
  • Dobutamine use differed (7% vs 2%; P<0.001), without corresponding improvement in outcomes in the PAC group.
• Outcome Assessment:
  • Primary endpoint (death before discharge home by day 60) and ventilator/ICU-free days are standardised, clinically meaningful, and largely objective.
  • Complication ascertainment is partly dependent on reporting fidelity; complication rates per catheter were similar in summary metric (0.08 ± 0.01 vs 0.06 ± 0.01; P=0.35) despite higher absolute complication prevalence in PAC patients (19% vs 9%).
• Statistical Rigor:
  • Two-sided testing was used; factorial interaction was explicitly examined for the primary endpoint.
  • Effect estimates for key continuous outcomes were provided with 95% confidence intervals.

Conclusion on Internal Validity: Overall, internal validity appears strong given central randomisation, very low crossover, high protocol adherence (91 ± 1% vs 88 ± 1%), and objective primary/secondary endpoints; the main limitation is unavoidable lack of blinding with potential co-intervention effects in an open ICU environment.

External Validity

• Population Representativeness:
  • Represents ventilated adults with early ALI in high-resource ICUs, enrolled within 48 hours of meeting diagnostic criteria.
  • Exclusions (including prolonged ALI, pre-existing PAC use, and selected advanced chronic diseases) narrow applicability to patients in whom equipoise around invasive monitoring is plausible.
• Applicability:
  • Findings are most directly applicable to ICUs capable of delivering protocolised fluid/vasopressor management and safe catheter insertion with experienced staff.
  • Generalisation to resource-limited settings is constrained by procedural expertise requirements and differing baseline catheter complication rates.
  • Results may not extend to niche populations where PAC information is uniquely actionable (e.g., complex right ventricular failure physiology) because such phenotypes were not the trial’s explicit target, and subgroup power is limited.

Conclusion on External Validity: External validity is moderate: the trial reflects early ALI/ARDS care in experienced, protocol-capable centres, but broader applicability is limited by selective exclusions and by the fact that the intervention tested was tightly protocolised rather than discretionary PAC-guided management.

Strengths & Limitations

• Strengths:
  • Large, multicentre randomised evaluation (n=1001) with clinically meaningful primary endpoint.
  • Early enrolment (≤48 hours after ALI criteria met) and prespecified insertion window (≤4 hours after randomisation) increased biological plausibility for benefit if present.
  • High protocol fidelity and low crossover (4 vs 5 patients) improve separation of monitoring strategy.
  • Factorial platform enabled explicit assessment for interaction with fluid strategy (interaction P=0.39 for primary endpoint).
• Limitations:
  • Unblinded intervention in an ICU environment introduces potential performance bias and co-intervention variability.
  • Protocolised management may attenuate incremental value of PAC-derived information compared with clinician-directed PAC use; the trial therefore estimates effect under protocol constraints rather than across all real-world PAC use cases.
  • Primary endpoint (death before discharge home by day 60) can be influenced by discharge practices and care pathways, although such effects are likely balanced by randomisation.
  • Trial not designed to fully quantify extremely rare catastrophic PAC complications; however, clinically important adverse events (arrhythmia requiring treatment) were substantially more common with PAC (12% vs 0.2%).

Interpretation & Why It Matters

• Clinical and methodological implications

  • Routine PAC use in ALI did not improve patient-centred outcomes when compared with CVC-based targets within a protocolised management framework.
  • The absence of benefit, coupled with higher complication rates (19% vs 9%; P<0.001), supports a default strategy of using less invasive monitoring unless a patient-specific indication exists.
  • Methodologically, the trial demonstrates that robust, protocol-driven physiological management can be evaluated independently of “more data” assumptions: additional monitoring information did not translate into measurable benefit in this clinical context.

Controversies & Subsequent Evidence

• • Interpretation hinges on what the trial truly “tests”: it evaluates PAC-guided targets embedded in a rigid protocol versus CVP-guided targets embedded in the same protocol, rather than discretionary clinician-directed PAC decision-making.¹
  • The harm signal is clinically meaningful even in expert centres: arrhythmias requiring treatment occurred in 12% with PAC versus 0.2% with CVC (P<0.001), shifting the risk–benefit calculus against routine PAC placement in ALI.
  • Subsequent practice trends in critical care have aligned with these data, with an emphasis on less invasive haemodynamic assessment (including bedside echocardiography) and selective use of advanced invasive monitoring rather than routine PAC insertion.
  • Major follow-up evidence relevant to the broader “PAC value” question is best interpreted alongside other randomised evaluations and systematic syntheses (see Further Reading), which in aggregate have not established routine outcome benefit across ICU populations comparable in severity to ALI/ARDS.

Summary

• PAC-guided protocolised management did not reduce death before hospital discharge home within 60 days versus CVC-guided protocolised management (27.4% vs 26.3%; P=0.69).
• Morbidity outcomes were similar (ventilator-free days 13.2 ± 0.5 vs 13.5 ± 0.5; ICU-free days 12.0 ± 0.5 vs 12.5 ± 0.5).
• Catheter-related complications were more frequent with PAC (19% vs 9%; P<0.001), dominated by arrhythmias requiring treatment (12% vs 0.2%; P<0.001).
• Factorial design showed no evidence that catheter type meaningfully modified the effect of the concurrently assigned fluid strategy on the primary endpoint (interaction P=0.39).
• The results strongly support avoiding routine PAC insertion in ALI/ARDS when adequate CVC-based and clinical monitoring can support protocolised resuscitation and deresuscitation decisions.

Overall Takeaway

In early acute lung injury managed with a protocolised haemodynamic strategy, pulmonary-artery catheters did not improve survival or reduce duration of mechanical ventilation compared with central venous catheters. The incremental physiological information of PACs did not translate into better outcomes, while procedure-related harm—particularly clinically significant arrhythmias—was substantially higher. This trial helped cement the principle that “more monitoring” is not inherently better unless it changes management in ways that improve patient-centred outcomes.

Bibliography

• 1Shure D. Pulmonary-Artery Catheters—Peace at Last? N Engl J Med. 2006;354(21):2273-2274.