Publication

• Title: Supine body position as a risk factor for nosocomial pneumonia in mechanically ventilated patients: a randomised trial
• Acronym: Not applicable
• Year: 1999
• Journal published in: The Lancet
• Citation: Drakulovic MB, Torres A, Bauer TT, Nicolas JM, Nogué S, Ferrer M. Supine body position as a risk factor for nosocomial pneumonia in mechanically ventilated patients: a randomised trial. Lancet. 1999;354(9193):1851-1858.

Context & Rationale

• Background

  Nosocomial pneumonia (including ventilator-associated pneumonia) is a frequent, morbid complication of invasive mechanical ventilation, with plausible mechanistic links to microaspiration of colonised oropharyngeal and/or gastric contents (particularly during sedation, impaired airway reflexes, and enteral feeding).
• Research Question/Hypothesis

  Does nursing invasively ventilated ICU patients at 45° (semirecumbent) reduce clinically suspected and microbiologically confirmed nosocomial pneumonia compared with 0° (supine) positioning?
• Why This Matters

  Head-of-bed elevation is a low-cost, broadly scalable “system intervention” that (if effective) could prevent a common ICU complication; however, it competes with haemodynamic instability, procedures, and pressure-injury risk, so credible trial evidence was needed to justify widespread implementation and define the target angle.

Design & Methods

• Research Question: In intubated, mechanically ventilated ICU patients, does a semirecumbent nursing position (45°) reduce the incidence of nosocomial pneumonia compared with a supine position (0°)?
• Study Type: Prospective, parallel-group, randomised controlled trial in two adult ICUs of a single tertiary hospital; investigator-initiated; unblinded; one planned interim analysis.
• Population:
  • Setting: 6-bed respiratory ICU and 8-bed medical ICU at Hospital Clínic, University of Barcelona (Spain).
  • Inclusion: Consecutive intubated and mechanically ventilated ICU patients (randomised after intubation or ICU admission) with informed consent from next-of-kin.
  • Key exclusions: Abdominal surgery within previous 7 days; neurosurgical intervention within previous 7 days; shock refractory to vasoactive drugs or volume therapy; previous endotracheal intubation within previous 30 days.
  • Protocol termination: First weaning trial, extubation, death, or permanent change in body position >45 minutes; surveillance continued for 72 hours after termination.
• Intervention:
  • Semirecumbent position targeted at 45° (backrest elevated); staff instructed to maintain position unless medically required to change.
  • Co-interventions (both groups): Sterile tracheal suctioning; ventilator circuits not routinely changed; stress-ulcer prophylaxis (sucralfate for those tolerating enteral feeding; ranitidine/omeprazole if parenteral nutrition); enteral nutrition per unit practice; pressure-injury prophylaxis with a water cushion.
• Comparison:
  • Supine position targeted at 0°; otherwise identical ICU care and infection surveillance strategy.
• Blinding: Unblinded (nursing position not practically blindable); implications include risk of performance bias and differential threshold for “clinical suspicion” triggering microbiological sampling.
• Statistics: Planned sample size 182 to detect a 50% relative reduction in clinically suspected nosocomial pneumonia (assumed 40% in the supine group) with 80% power at the 5% significance level (95% confidence); one interim analysis planned at 50% enrolment with stochastic curtailment correction; primary analyses were not pure intention-to-treat (4/90 excluded post-randomisation), effectively a modified intention-to-treat/per-protocol analysis of 86 patients.
• Follow-Up Period: From randomisation until protocol termination (weaning trial, extubation, death, or >45-minute change of position), plus 72 hours thereafter.

Key Results

This trial was stopped early. It was stopped after the planned interim analysis at ~50% of the intended sample (86 analysed of 182 planned) because clinically suspected nosocomial pneumonia was significantly lower in the semirecumbent group (P=0.003).

• Effect size was large for both clinically suspected (8% vs 34%) and microbiologically confirmed pneumonia (5% vs 23%), supporting internal consistency across a more subjective and a more objective endpoint.
• Clinical outcomes beyond pneumonia were not convincingly changed: ICU mortality was 18% vs 28% (P=0.289), and mean ICU stay was similar at baseline (9.3 ± 7.2 days vs 9.7 ± 7.8 days).
• Enteral nutrition appeared to concentrate risk: pneumonia occurred in 50% (14/28) of enterally fed supine patients versus 9% (2/22) of enterally fed semirecumbent patients.

Internal Validity

• Randomisation and allocation: Randomisation list was computer-generated with allocation implemented by an independent individual using a randomisation table; allocation concealment beyond this is not described in detail.
• Dropout/exclusions (post-randomisation): 90 randomised (43 semirecumbent; 47 supine); 4/90 (4.4%) excluded from analysis: 1 death ~2 hours after protocol initiation (semirecumbent) and 3 protocol violations due to reintubation (all semirecumbent), yielding an analysed cohort of 86 (39 semirecumbent; 47 supine).
• Performance and detection bias: Unblinded positioning creates plausible performance bias (e.g., feeding interruptions, suctioning vigilance) and detection bias because the diagnostic pathway to “microbiologically confirmed” pneumonia was triggered by clinical suspicion (new infiltrate + clinical criteria).
• Protocol adherence: Position was checked daily, but continuous measurement of achieved backrest angle was not reported; protocol termination occurred if body position was permanently changed for >45 minutes (7/86; 8%), suggesting potential dilution of exposure contrast late in follow-up.
• Baseline characteristics and residual confounding: Groups were broadly comparable but with potentially relevant imbalances (supine vs semirecumbent): age 67 ± 14 vs 63 ± 16 years; APACHE II 23.8 ± 6.1 vs 21.3 ± 6.0; fatal/ultimately fatal underlying disease 94% vs 80%; large-bore nasogastric tube 87% vs 72%; ranitidine use 62% vs 41%; parenteral nutrition 21% vs 8%.
• Heterogeneity: Mixed case-mix (COPD, other pulmonary disease, post-operative, drug overdose/neurological emergencies) in a small single-centre cohort increases the risk that chance imbalances influence the estimate of effect.
• Timing: Intervention began after intubation/ICU admission and continued until weaning/extubation/death/position change; pneumonia was predominantly late-onset (15/19; 79%), aligning with duration of ventilation as an independent risk factor (≥7 days OR 10.9; 95% CI 3.0 to 40.4).
• Dose and feasibility of the intervention: The target was 45° versus 0°; actual achieved angles (and therefore delivered “dose”) were not quantified beyond daily checks.
• Separation of the variable of interest: Assigned positioning was 45° vs 0° (large theoretical separation), but achieved separation in degrees over time was not reported.
• Adjunctive therapy use: Enteral nutrition exposure was common and similar (60% vs 56%), but stress ulcer prophylaxis differed (sucralfate 75% vs 85%; ranitidine 62% vs 41%), which could influence gastric colonisation and aspiration-related risk.
• Outcome assessment: Clinical suspicion of pneumonia is partly subjective; microbiological confirmation improves specificity, but microbiological sampling depended on clinician suspicion in an unblinded trial.
• Statistical rigour: The trial did not reach its planned sample size (182) due to early stopping at interim analysis; although correction for interim looks was used, early-stopped trials are at risk of overestimating effect size, particularly for subjective outcomes.

Conclusion on Internal Validity: Overall, internal validity is moderate: randomisation and concordant microbiologically confirmed outcomes support a real signal, but open-label design, early stopping, post-randomisation exclusions (all in the intervention arm), and unmeasured achieved angle introduce meaningful risk of bias and effect-size inflation.

External Validity

• Population representativeness: Medical/respiratory ICU ventilated patients were included, but key exclusions (recent abdominal or neurosurgery; refractory shock; recent intubation) limit applicability to trauma, neurocritical care, and early postoperative populations where higher head-of-bed angles may be constrained.
• Applicability across systems: The intervention is low-cost and implementable in most ICUs, but the comparator (0° supine) is less reflective of contemporary “usual care” where some elevation is common; therefore, effect size may not translate directly to comparisons such as 30° vs 45°.
• Feasibility: Sustained 45° is often difficult to maintain in routine practice due to procedures, haemodynamic instability, and patient comfort; the trial did not quantify achieved angle, limiting transferability of the “dose” of head elevation.

Conclusion on External Validity: External validity is moderate: the physiological rationale and direction of effect are broadly generalisable, but the magnitude of benefit is sensitive to what “control” positioning looks like and whether 45° elevation is achievable in everyday practice.

Strengths & Limitations

• Strengths:
  • Clinically simple, low-cost intervention with clear mechanistic plausibility (aspiration prevention).
  • Randomised design with prespecified interim analysis and a parallel microbiologically confirmed endpoint supporting consistency of effect direction.
  • Reported interaction analysis highlighting enteral nutrition as a major effect modifier (important for bedside implementation).
• Limitations:
  • Early stopping with a small analysed sample (86/182 planned) increases risk of exaggerated effect estimates.
  • Unblinded intervention and partially subjective primary endpoint with sampling triggered by clinical suspicion (detection bias risk).
  • Post-randomisation exclusions (4/90), including 3 protocol violations all in the intervention arm, compromising strict intention-to-treat inference.
  • Achieved backrest angle over time not quantified; degree of exposure separation is uncertain.
  • Single-centre design and control arm at 0° may overstate benefit relative to modern practice where some elevation is routine.

Interpretation & Why It Matters

• Clinical practice implication

  When clinically feasible, avoid sustained 0° supine positioning in invasively ventilated patients; the largest absolute differences in pneumonia occurred in enterally fed patients (50% at 0° vs 9% at 45°), supporting prioritisation of head-of-bed elevation during enteral feeding.
• Bundle logic

  This trial provided an early RCT foundation for incorporating head-of-bed elevation into VAP prevention bundles; subsequent work shifted the emphasis from “45° always” to “achieve ≥30° reliably and monitor adherence” in real-world settings.
• Methodological lesson

  Large early-stopped effects in unblinded pragmatic ICU trials warrant caution: replication, angle-achievement data, and implementation science are needed to translate efficacy signals into dependable effectiveness.

Controversies & Subsequent Evidence

• Baseline imbalances and early stopping: The accompanying Lancet commentary highlighted that several baseline differences (e.g., severity/underlying disease and stress-ulcer prophylaxis patterns) could have biased results against the supine arm, and noted the inherent uncertainty in an early-stopped single-centre trial despite the striking effect size.¹
• Feasibility of maintaining 45°: A later randomised study incorporating backrest-angle monitoring found that target semirecumbency was difficult to achieve and did not reproduce a clear reduction in VAP, reframing the question from “does 45° work?” to “what angle is achievable and effective in practice?”.²
• Implementation and adherence as a core determinant: Continuous monitoring work demonstrated that maintaining head-of-bed elevation ≥30° is not reliably achieved without systems support (visual cues/alerts), reinforcing that “position” is a process-of-care variable rather than a one-off intervention.³
• Competing risks (pressure injury, clinical constraints): Observational work linked factors associated with semi-recumbent compliance and pressure ulcers during invasive mechanical ventilation, underscoring the need to integrate aspiration prevention with skin integrity and haemodynamic considerations rather than treat positioning as universally benign.⁴
• Evidence grading remained cautious despite biological plausibility: An evidence-based recommendation synthesising clinical data and feasibility considerations supported head elevation (often framed as ≥30°) but judged the strength of recommendation as cautious due to limited high-quality trial evidence and delivery challenges.⁵
• Synthesis of the broader trial network: A network meta-analysis of positioning strategies suggested reduced VAP with semirecumbent positioning compared with supine, but the evidence base is small and heterogeneous (definitions, achieved angles, and co-interventions), so precision of effect estimates is limited.⁶
• Guideline translation: Modern prevention guidance continues to recommend head-of-bed elevation (typically 30–45°) as part of a broader prevention strategy, shifting emphasis to reliable implementation and monitoring rather than assuming continuous 45° is achievable for all patients.⁷

Summary

• Randomised single-centre ICU trial comparing semirecumbent 45° versus supine 0° positioning in invasively ventilated patients.
• Stopped early after interim analysis (86 analysed of 182 planned) due to lower clinically suspected pneumonia at 45° (8% vs 34%; P=0.003).
• Microbiologically confirmed pneumonia was also lower at 45° (5% vs 23%; P=0.018), supporting internal consistency across endpoints.
• No statistically significant mortality difference was shown (ICU mortality 18% vs 28%; P=0.289), and the trial was not powered for mortality.
• Enteral feeding strongly modified risk: pneumonia was highest in enterally fed supine patients (50%; 14/28) versus enterally fed semirecumbent patients (9%; 2/22).

Overall Takeaway

This early-stopped randomised trial reported a large reduction in nosocomial pneumonia when ventilated patients were nursed semirecumbent at 45° rather than supine at 0°, with the strongest signal in enterally fed patients. While later evidence emphasised feasibility limits and uncertainty about the precise “effective angle”, Drakulovic et al. remains a landmark by crystallising head-of-bed elevation as a core, testable ICU prevention strategy and anchoring its adoption into VAP prevention bundles.

Bibliography

• 1Webster NR. Importance of position in which patients are nursed in intensive-care units. Lancet. 1999;354:1835-1836.
• 2van Nieuwenhoven CA, Vandenbroucke-Grauls C, van Tiel FH, et al. Feasibility and effects of the semirecumbent position to prevent ventilator-associated pneumonia: a randomized study. Crit Care Med. 2006;34(2):396-402.
• 3Wolken RF, Woodruff RJ, Smith J, Albert RK, Douglas IS. Observational study of head of bed elevation adherence using a continuous monitoring system in a medical intensive care unit. Respir Care. 2012;57(4):537-543.
• 4Llaurado-Serra M, Ulldemolins M, Fernandez-Ballart J, et al. Related factors to semi-recumbent position compliance and pressure ulcers in patients with invasive mechanical ventilation: an observational study (CAPCRI study). Int J Nurs Stud. 2016;61:198-208.
• 5Niël-Weise BS, Gastmeier P, Kola A, et al. An evidence-based recommendation on bed head elevation for mechanically ventilated patients. Crit Care. 2011;15(2):R111.
• 6Pozuelo-Carrascosa DP, Herráiz-Adillo Á, Alvarez-Bueno C, et al. Body position for preventing ventilator-associated pneumonia: a systematic review and network meta-analysis. J Intensive Care. 2022;10(1):9.
• 7Klompas M, Branson R, Cawcutt K, et al. Strategies to prevent ventilator-associated pneumonia, ventilator-associated events, and nonventilator hospital-acquired pneumonia in acute care hospitals: 2022 update. Infect Control Hosp Epidemiol. 2022;43(6):687-713.