The UK PARAMEDIC-3 trial found that going for intraosseous (IO) access first, rather than intravenous (IV) access first, did not improve 30-day survival in adults with out-of-hospital cardiac arrest. A pooled analysis of three trials, covering 9,332 patients, reached the same answer and suggested IO-first may slightly lower the chance of a sustained pulse. For anyone who gives cardiac arrest drugs, this settles a question practice had been drifting on for years.
Why this matters
Adrenaline in cardiac arrest is time-sensitive. In earlier drug trials, the time from the emergency call to the first dose ranged from 16 to 21 minutes. Much of that delay is spent getting a line in.
IO access, meaning a needle drilled into the bone marrow, usually the upper tibia or upper humerus, is quick and reliable. Observational studies suggested it got drugs in faster than an IV cannula, yet outcomes looked similar or worse. That was hard to interpret, because IO was usually tried only after IV had failed, so those patients had been in arrest for longer.
Guidelines recommended IV first. Despite this, IO use had risen to as much as 60% in some systems. In 2020 the International Liaison Committee on Resuscitation (ILCOR) called for randomised trials to settle the question.
What the researchers did
PARAMEDIC-3 was a pragmatic, open-label randomised trial run by Warwick Clinical Trials Unit and funded by the National Institute for Health and Care Research. It ran from November 2021 to July 2024 across 11 UK emergency medical systems: 10 NHS ambulance services and one air ambulance. Results appeared in the New England Journal of Medicine in October 2024, with Keith Couper as first author.
Adults aged 18 or over in out-of-hospital cardiac arrest who needed vascular access for drugs were eligible. Pregnancy was the only exclusion. When access was needed, the paramedic opened a sealed envelope allocating IO-first or IV-first. After two failed attempts by the allocated route, the paramedic could choose. The anatomical site was also left to the paramedic.
Of 10,723 patients screened, 6,082 were analysed: 3,040 IO and 3,042 IV. The target was 15,000, but recruitment was slow and the trial stopped at the end of its funded period. The primary outcome was survival at 30 days.
The pooled analysis, led by the same group for the ILCOR Advanced Life Support Task Force, was published in Resuscitation in February 2025 (online 30 December 2024). It found three randomised trials with 9,332 patients. One was the Danish IVIO trial of 1,479 adults, published the same day as PARAMEDIC-3.
What they found
In PARAMEDIC-3, 30-day survival was 137 of 3,030 (4.5%) in the IO group and 155 of 3,034 (5.1%) in the IV group. The adjusted odds ratio was 0.945 (95% confidence interval 0.676 to 1.322, p=0.741). Primary outcome data were available for 99.7% of patients.
Survival to discharge with a favourable neurological outcome, defined as a modified Rankin Scale score of 3 or less, was 2.7% with IO and 2.8% with IV. Return of spontaneous circulation (ROSC) at any time was 36.0% with IO and 39.1% with IV (adjusted odds ratio 0.863, 0.765 to 0.974). Sustained ROSC at hospital handover was 21.7% with IO and 24.6% with IV. Only the primary outcome was formally tested, so the ROSC differences are signals rather than proof.
The expected speed advantage did not materialise. First successful access was by the allocated route in 94.4% of the IO group but only 64.6% of the IV group. Even so, median time from the emergency call to drug administration was 24 minutes in both groups. Most IO cannulae went into the proximal tibia (73.5%). One adverse event was reported: mild leg pain in an IO patient, not deemed serious.
The pooled analysis found no improvement in 30-day survival with IO-first (odds ratio 0.99, 0.84 to 1.17, moderate certainty) and no difference in favourable neurological outcome (odds ratio 1.07, 0.88 to 1.30, low certainty). Sustained ROSC was lower with IO-first (odds ratio 0.89, 0.80 to 0.99, moderate certainty). IVIO alone found sustained ROSC in 30% of IO and 29% of IV patients.
Strengths and limitations
This is the largest randomised trial on the question, run by working UK paramedics in real conditions, with almost complete follow-up and balanced groups. Results held across pre-specified subgroups, and the pooled analysis points the same way.
The main weakness is power. The trial reached about 40% of its planned size, so it could not detect a 1% absolute difference in survival. The investigators report that effect estimates were stable when recruitment stopped, but a small benefit either way cannot be excluded. The trial was open-label, did not record CPR quality, and did not standardise the IO site. The ROSC findings were not statistically tested.
Generalisability to remote work is limited. In PARAMEDIC-3, 79% of arrests happened at home, median response time was 8 minutes, and trauma (1.4%) and drowning (0.2%) were rare. An offshore or diver medic working alone, with long evacuation times, is a long way from these conditions.
What it means for your practice
ILCOR’s draft treatment recommendation, dated 16 October 2024, suggests IV rather than IO as the first attempt in adult cardiac arrest (a weak recommendation, low-certainty evidence). If IV access cannot be achieved rapidly within two attempts, it is reasonable to move to IO. UK practice is set by JRCALC, Resuscitation Council UK and your employer. Check them before changing anything.
- Paramedics and ambulance clinicians. IV first where it can be done quickly. Two attempts, then IO, is the pattern the evidence supports. Treat IO as a planned second step, not a failure.
- Offshore and diver medics. The same principle applies, but your patients fall outside the trial population. Where you work alone, an early IO may be the practical choice. Follow your company and topside protocols.
- First responders and FREC 3 clinicians. Vascular access is outside your scope. Your contribution is the part proven to work: early recognition, good compressions, minimal pauses and early defibrillation.
Do not change your IO site preference on this evidence. The trial did not test tibia against humerus. Do not read the ROSC signal as proof that IO drugs are harmful. It was a secondary outcome and IVIO did not reproduce it.
Reflect
- In your last cardiac arrest, how long did vascular access take, and what would have changed if you had switched to IO after two attempts?
- What environmental or patient factors most often slow IV access in your setting, and how could your team reduce them?
- How would you explain to a colleague why a 94% first-pass success rate for IO did not translate into faster drug delivery?
- If you work offshore or in a remote setting, how far do these results apply to your patients, and what does your protocol say?
Read the original
- New England Journal of Medicine: A Randomized Trial of Drug Route in Out-of-Hospital Cardiac Arrest (PARAMEDIC-3), 2024 (open-access author manuscript)
- Resuscitation: Intraosseous and intravenous vascular access during adult cardiac arrest, a systematic review and meta-analysis, 2025
- ILCOR CoSTR: Intravenous and intraosseous drug administration for cardiac arrest in adults, draft for public comment, 2024
- American College of Cardiology: PARAMEDIC-3 trial summary, 2024
- First10EM: IV versus IO for cardiac arrest (PARAMEDIC3 and IVIO), 2024
- Conexiant Cardiology: Cardiac arrest trial questions drug routes, 2025
Published 25 October 2026. Last reviewed October 2026.
These summaries are for education and CPD only. They are not clinical guidelines. Always practise within your scope of practice and follow your employer’s protocols and national guidance such as JRCALC and Resuscitation Council UK.