Adults who are unconscious or severely ill and need a breathing tube connected to a ventilator are at high risk of developing a lung infection (pneumonia) within the first few days in the intensive care unit. This early pneumonia affects up to 30 to 50 % of certain high-risk patients, prolongs the time on the ventilator and in hospital, and increases the use of antibiotics.

Two strategies are commonly used today to try to prevent this infection: a short, three-day course of an intravenous antibiotic, and removal of secretions from the airway with a sterile suction catheter. Both have limitations - antibiotics can favour the growth of resistant bacteria, and catheter suctioning is uncomfortable and may injure the airway.

PIRÁMIDES is a small (60-patient) pilot study that compares the current practice with two non-invasive, mechanical alternatives for keeping the airway clear: a continuous low-pressure suction system built into a special breathing tube, and a device that produces a gentle, programmed "artificial cough" through the ventilator. Adult patients who are intubated for severe trauma, severe brain injury, stroke, resuscitated cardiac arrest or other causes of decreased consciousness are randomly assigned, in equal numbers, to one of the three approaches and followed for 14 days, with a final visit at day 90.

The main goal is to find out which of the three strategies best prevents early pneumonia, and which provides the best overall result for patients when survival, severity of infection, need for additional antibiotics and side effects are considered together. To make these comparisons as fair as possible in an open-label study, an independent committee of doctors not involved in patient care reviews each suspected pneumonia case without knowing which strategy the patient received. The results will help design a larger trial to confirm which approach is safest and most effective for preventing early pneumonia in critically ill patients on a ventilator.

Background. Patients with structural coma are at high risk of so-called early onset pneumonia (EOP). Incidence rates of up to 50% have been reported in patients with head trauma or stroke. The usual causative microorganisms belong to the normal upper airway flora like Streptococcus pneumoniae, Staphylococcus aureus, Haemophilus influenza and Moraxella catarrhalis. EOP typically is not present at admission and develops after 2 to 7 days after endotracheal intubation.

A short course of systemic antibiotic therapy and aspiration of subglottic secretions (ASS) are associated with significant reductions in EOP, although ASS does not prevent late-onset pneumonia. Non-invasive mechanical methods may avoid the use of prophylactic antibiotics and pain and injury to the tracheal mucosa caused by the conventional suctioning catheter.

PIRÁMIDES is a single-centre, pilot, open-label, randomised controlled trial (1:1:1) comparing three approved strategies for the prevention of early-onset ventilator-associated pneumonia (early VAP) in adult intubated patients at high risk of this complication (severe trauma, severe traumatic brain injury, ischaemic or haemorrhagic stroke, post-cardiac-arrest syndrome and other acute causes of decreased consciousness). Sixty patients (20 per arm) are recruited at the Intensive Care Unit of Hospital Clínico San Carlos (Madrid, Spain). The trial compares the standard 3-day short antibiotic course (ceftriaxone 2 g IV every 24 h) plus sterile-catheter suctioning, continuous subglottic secretion suctioning for 7 days through a dedicated endotracheal-tube channel, and 7 days of programmed sessions of a CE-marked bionic cough simulator (BCS3, Yaguo). The unit-standard topical selective digestive decontamination regimen is applied in all three arms. Intervention duration is 7 days, the main observation period extends to day 14 and final outcomes are collected at day 90. Because the interventions are visually distinct at the bedside, the trial is open-label for the treating team and patients.

The current protocol version (v2.0, April 2026) incorporates two substantial amendments. First, the objectives have been restructured: a single primary endpoint, the cumulative incidence of bacterial early VAP through day 14, replaces the original co-primary structure, and a hierarchical "Desirability of Outcome Ranking" (DOOR) endpoint at day 14 is introduced as the key secondary endpoint. The DOOR endpoint is applied identically to the three arms and integrates, in a clinically ordered five-category hierarchy, mortality, the occurrence and severity of VAP, exposure to rescue antibiotics and serious intervention-related adverse events. Categories range from (1) alive at day 14 without VAP, rescue antibiotic or related serious adverse event, to (5) death before day 14, with intermediate categories for rescue-antibiotic use without confirmed VAP, non-severe VAP, and severe VAP or major intervention-related complication.

Second, the amendment establishes an independent blinded Adjudication Committee (two intensivists specialized in ICU-acquired infections external to the trial team, one chest radiologist and one microbiologist) responsible for the final classification of every potential pneumonia episode and for the assignment of each patient to the corresponding DOOR category. The Committee reviews case-report forms, chest radiographs and lung ultrasound images, microbiology reports and other relevant clinical documentation, with all fields identifying the assigned arm removed or masked; decisions are reached by consensus.

The trial is sized as a pilot and is not powered for confirmatory testing; effect estimates and 95 % confidence intervals are exploratory and hypothesis-generating. The primary endpoint is analyzed in the intention-to-treat population by Fisher's exact test. The DOOR endpoint is analyzed in two pre-specified steps: a global ordinal comparison across the three arms using a proportional-odds model, followed by pairwise win-ratio comparisons on the same pre-specified hierarchy, with win odds as a sensitivity analysis and a pre-specified continuous tiebreaker (ventilator-free days through day 14, then ICU length of stay) where ties are abundant. Multiplicity between the two main pairwise comparisons (each mechanical strategy versus control) is addressed through a hierarchical testing procedure pre-specified in the Statistical Analysis Plan. A pre-specified electrical impedance tomography sub-study in five patients per arm explores the acute effects of each strategy on regional ventilation and compliance.

Inclusion criteria

1. Endotracheal intubation with an anticipated duration > 48 hours.

2. High risk of early respiratory infection associated with a diagnosis of:

   1. Severe trauma.
   2. Severe traumatic brain injury.
   3. Ischaemic or haemorrhagic stroke.
   4. Other causes of impaired consciousness: post-resuscitated cardiac arrest status, intoxications, acute infections or diseases of the central nervous system, seizures.

3. Informed consent signed by the patient or, when impossible due to clinical status, by their legal representative, with re-consent by the patients themselves upon regaining capacity (section 15).

Exclusion criteria

1. Intubation with an anticipated duration < 48 hours.
2. Foreseeable ominous prognosis within < 7 days.
3. Already established indication for systemic antibiotic therapy, either for suspected aspiration pneumonia with radiological pulmonary infiltrate or for suspected non-respiratory source infection.
4. Active haemoptysis or pulmonary haemorrhage.
5. Unstable chest.
6. Undrained pneumothorax (inclusion may be considered once drained).
7. Known allergy or intolerance to beta-lactam antibiotics.

Upon presentation of an analysis plan and local approvement by ERB.

From publication of study onwards. No time limit.

Contact with principal investigator.