Adult cardiac arrest

It helps to keep four quantities apart. Pressure is the force per unit area that blood exerts at one place, such as inside the aorta. A gradient is the difference between the pressures at two places. Resistance is how much the vessels between those places oppose flow. Flow is the volume of blood that actually moves through tissue, and it is set by the gradient divided by the resistance, not by any single pressure.

Resistance is why the same gradient can deliver different flow: coronary resistance is high while the compressed myocardium squeezes its own vessels and lower during release, and a vasoconstrictor such as epinephrine raises aortic pressure largely by raising peripheral resistance, which helps the heart and brain only if their own beds are not constricted to the same degree.

The heart is a muscle that never rests in health, and it extracts much of the oxygen from the blood that reaches it even at baseline, so it has little reserve to draw on when supply falls. The blood inside its chambers nourishes little beyond the innermost layer, because the ventricular walls are too thick for oxygen to diffuse far from the cavity. The myocardium therefore depends on its own circulation. That circulation begins at the aortic root, where the right and left coronary arteries arise from small openings just above the cusps of the aortic valve. From there the arteries run over the surface of the heart, send branches down into the muscle, divide into arterioles, and end in a dense bed of capillaries where oxygen leaves the blood. The blood is then collected by cardiac veins, most of which converge on the coronary sinus, a short venous channel in the groove on the back of the heart that empties into the right atrium. Some blood takes other routes, through anterior cardiac veins that enter the right atrium directly and small Thebesian veins that open into the chambers themselves, but for this model the main path is the one that ends in the right atrium.

That path explains why two pressures in such different places belong in the same calculation. Aortic pressure is the pressure at the inlet of the coronary circulation. Right atrial pressure is the pressure in the compartment that receives most of the coronary venous outflow. The right atrium supplies nothing to the coronary arteries; it enters the model only as the receiving side, the pressure that blood leaving the heart muscle has to drain against. If you trace a single red cell, it leaves the aortic root, crosses the myocardium, and arrives in the right atrium, and the pressure difference between where it started and where it ended is what moved it. A rough analogy is water running between two tanks through a pipe: the flow depends on the difference in level between the tanks and on how narrow the pipe is, not on how high the upper tank sits. The analogy fails in ways that matter here, because the coronary vessels are collapsible, their resistance changes within every compression cycle, and the flow is pulsatile rather than steady.

Aortic pressure is the pressure of the blood inside the aorta, expressed relative to atmospheric pressure. A transducer reads zero when it is open to air at the level of the right atrium, the phlebostatic level, so a reading of 35 mm Hg means the blood in the aorta is pushing 35 mm Hg harder than the atmosphere at that height. In a beating circulation this pressure comes from the left ventricle ejecting blood into an aorta that is elastic, or compliant. Each beat stretches the aortic wall, and the stretched wall stores some of the ejected energy. When the aortic valve closes, the wall recoils and keeps pushing blood onward into the arteries, including the coronary arteries, while pressure gradually falls as blood runs off into the periphery. That is why arterial pressure does not drop to zero between beats.

During CPR the ventricle is not doing that work, and the chest is. Compression raises pressure partly by squeezing the heart between the sternum and the spine and partly by raising the pressure of everything inside the chest, which pushes blood out of the heart, lungs, and great vessels toward the periphery. How much each contributes probably varies with the patient, the chest, and the technique, and neither mechanism alone accounts for the observed pressures in all settings. What matters for this argument is that compression fills and stretches the aorta, as a heartbeat would. On release, the aortic valve typically closes against the falling pressure in the ventricle, the stretched aorta recoils, and some pressure persists in the aortic root while blood continues to run off. That residual pressure is what the coronary arteries see during release. In this text it is called relaxation-phase aortic pressure; monitors and studies of CPR usually report the same quantity as arterial or aortic diastolic pressure.

Right atrial pressure is the pressure of the blood inside the right atrium, read against the same zero. It reflects how much blood is arriving from the veins, how well the right ventricle is emptying the atrium, how stiff or compliant the chamber wall is, and what is pressing on the heart from outside. That last factor is easy to forget. A catheter in the atrium measures intracavitary pressure, the pressure inside the chamber relative to atmosphere. The pressure that actually distends the chamber is the transmural pressure, the inside pressure minus the pressure surrounding the heart. When intrathoracic pressure rises, the measured intracavitary pressure rises with it even if the chamber holds no more blood.

Positive-pressure breaths are one example. Each breath raises intrathoracic pressure, and some of that rise is transmitted to the right atrium; in anesthetized patients outside of CPR, adding positive end-expiratory pressure raised measured right atrial pressure, and the animal data on ventilation rate during CPR are gathered in 5.1. The human measurement was not made during CPR and the CPR data are from animals, so the example shows the direction of the effect rather than its size in a given patient. A higher right atrial reading therefore does not mean more volume, better preload (the stretch that the blood filling a ventricle puts on its muscle just before it contracts), or more forward flow; it may simply mean that the chamber is being pressed on. Right atrial pressure in this model is not a gauge of volume. It is the pressure on the downstream side of the coronary bed.

Coronary artery pressure and coronary perfusion pressure sound alike but are different ideas. Coronary artery pressure is a local value, the pressure at one point inside a coronary artery. Coronary perfusion pressure is a difference between two locations, the inlet and the outlet of the coronary circulation. There is no conversion from aortic pressure to coronary inlet pressure, because the coronary arteries open directly from the aortic root and the two spaces are continuous; at the inlet they are essentially the same pressure. Pressure then falls along the vessels as blood meets resistance, most steeply in the small arteries and arterioles, so a reading at the inlet says nothing about the pressure beyond a narrowing. A patient with a tight stenosis can have a normal aortic root pressure and very little pressure in the territory past the obstruction.

With that picture in place, the equation is a short summary of it. Coronary perfusion pressure equals relaxation-phase aortic pressure minus the right atrial pressure measured at the same moment. Both are in millimeters of mercury, both are read against the same zero, and both must be taken during the release phase of the same cycle. Mixing phases misleads: the peak aortic pressure during compression set against the right atrial pressure during release gives a number that describes no moment the coronary bed ever experienced. Three illustrative pairs, chosen to show the arithmetic and not as targets, make the point. In pair A, aortic pressure is 35 and right atrial pressure is 10, so the gradient is 25 mm Hg. In pair B, aortic pressure is again 35 but right atrial pressure is 25, so the gradient is 10 mm Hg. Pair B is worse than A even though the inlet pressure is identical, because the outlet pressure has risen and the difference across the bed has narrowed. Pair C is hypothetical, with relaxation-phase pressures above those usually measured in human CPR: aortic pressure is 50 and right atrial pressure is 40, so the gradient is again 10 mm Hg. C looks better on the arterial line, but the higher inlet pressure is matched by a higher outlet pressure and the gradient is no wider than in B. The arithmetic shows the gradients; it does not by itself tell you the flow in any of these patients or what happens to them, because flow also depends on resistance, which these numbers do not capture.

Phase matters for a second reason, which has to do with the vessels themselves. The small coronary vessels run through the muscle, and when the chest is compressed the surrounding myocardium is squeezed along with them. Pressure at one point inside those vessels may be high, but the vessels are collapsible, and what moves blood forward is a gradient through vessels that are open. During compression the intramyocardial vessels are compressed and forward coronary flow is small. On release the muscle relaxes, the vessels reopen, and the gradient from the aortic root to the right atrium can drive blood through them. The release phase of each compression cycle is therefore when the coronary bed fills. It is a different thing from a pause in CPR. A pause stops compressions altogether, so the aorta is no longer refilled and relaxation-phase aortic pressure decays, while the release phase is a part of every cycle in which the stretched aorta is still pushing.

For the heart, the gradient that matters is coronary perfusion pressure, and it exists during the relaxation phase. In a classic human study, return of spontaneous circulation was rare when the maximal gradient during CPR stayed below about 15 mm Hg. That is an observed association, not a treatment target, and a gradient above it did not guarantee return of circulation. Two things therefore have to happen on every cycle. The push has to be deep enough to raise aortic pressure, and the release has to be complete enough to let the chest and the heart re-expand. A deep compression followed by incomplete recoil undoes the second half, and overventilation acts on the same phase, though not in an identical way; the pig data for each are set out in 3.2, and how much either matters in a given patient depends on everything else that is happening at the same time.

The model has limits, and they are the places where the number and the flow part company. So far the argument has leaned on the gradient and kept resistance in the background. In practice it is only half of the relation; coronary resistance, set by vascular tone, by any fixed obstruction, by the pressure within the myocardium, and by pressure inside the ventricles, determines how much flow a given gradient buys. Right atrial pressure is a reasonable stand-in for the downstream pressure because most coronary venous blood drains there, but it may not always be the effective back pressure. As an evidence note: in animal studies of ischemic myocardium outside of CPR, small coronary vessels behaved as if they closed at a pressure above the venous outlet pressure, a so-called waterfall, and when that happens flow depends on the inlet pressure minus that closing pressure rather than minus right atrial pressure. Whether and how much this operates during CPR has not been measured directly, so it is a physiological inference, not an established feature of resuscitation.

In the beating heart, perfusion of the inner layer of the left ventricle is often described with a different difference, aortic diastolic pressure minus left ventricular end-diastolic pressure, because the subendocardial vessels are compressed by the pressure in the left ventricle around them. That expression comes from a different physiology, and left ventricular end-diastolic pressure is not interchangeable with right atrial pressure.

This is why the 2025 AHA adult algorithm specifies the whole cycle, not just the push: at least 2 inches (5 cm) deep, 100 to 120 per minute, full recoil, and no leaning. Compressing on a firm surface is preferred when it is feasible and does not delay compressions. More is not automatically better. Observational data have linked depths greater than 6 cm (about 2.4 inches) with lower survival, and faster rates tend to cost depth and recoil. Compression depth also falls after about 90 to 120 seconds of continuous CPR as the compressor tires, which is why the compressor changes about every 2 minutes.

What you can observe at the bedside is limited. You can see depth, rate, and recoil, and end-tidal CO2, the partial pressure of CO2 at the end of each exhaled breath (Part VI), gives an indirect signal of flow. You usually cannot see the gradient itself, so neither compressions that look vigorous nor a single arterial number proves that blood is moving through the heart muscle. A high arterial reading can sit on top of a high right atrial pressure, and a good-looking push can be followed by a release that never comes. The model predicts that a rise in right atrial or intrathoracic pressure, or a fall in relaxation-phase aortic pressure, will tend to reduce coronary flow, unless something else widens the gradient at the same time.

The things that can be acted on are the ones the model names: complete recoil, short and infrequent pauses, ventilation that does not keep the chest pressurized, and, when invasive lines are already in place, reading the arterial and right atrial pressures together and in the same phase rather than the arterial pressure alone. The 2025 AHA guidance puts the arterial number in the same light: when an arterial line is in place, arterial diastolic pressure "can approximate coronary perfusion pressure and myocardial perfusion during CPR," but "there are inadequate adult data to suggest any specific pressure in adults." Nothing in this section is a target.

The brain depends on the same compressions, through the arterial pressure they generate. Its own gradient is opposed by intracranial and venous pressure rather than right atrial pressure, but the practical lesson is shared: poor compressions tend to fail both organs together.

Two mechanisms do this, and 3.1 introduced them. In the cardiac-pump description, the sternum squeezes the ventricles against the spine, the mitral and tricuspid valves typically close, and blood is pushed out through the aortic and pulmonary valves much as in a heartbeat. In the thoracic-pump description, the compression raises pressure throughout the chest, the heart behaves more like a passive conduit, and blood is pushed out of the lungs, heart, and great vessels together; valves in the large veins at the thoracic inlet are thought to keep that pressure from driving blood backward into the head and arms, so the forward path toward the arteries is favored. Echocardiographic and pressure studies during CPR have been interpreted as supporting each pattern in different patients and at different times, which is why this text treats both as contributing and does not name one as the mechanism. The balance probably varies with the patient; in one human echo study, the ventricles were compressed more in narrower chests, and heart size, hand position, and arrest duration are thought to matter as well.

Release

Release is the other half of the cycle, and it is not suction in the sense of the heart pulling blood in. When the hands come off the sternum, the elastic chest wall springs back, intrathoracic pressure falls, sometimes slightly below atmospheric, and the pressures inside the heart and great veins fall with it. Blood returns because mean systemic filling pressure in the veins outside the chest, the pressure that the stretched systemic veins exert on the blood they hold, is now higher than right atrial pressure, the same gradient described in 5.1, and the coronary arteries fill because relaxation-phase aortic pressure is held above right atrial pressure, as described in 3.1. The aortic valve typically closes under the higher pressure above it and keeps the aorta from emptying backward into the ventricle during this phase. Nothing pulls the blood; the release removes the pressure that was opposing its return.

Five variables, one cycle

The compression variables are not independent dials. Each cycle has a push phase and a release phase; duty cycle is the fraction spent pushing. Changing one variable moves the others.

Rate shortens the cycle. Above the guideline range of 100 to 120 per minute, there is less time per cycle to push to depth and to release fully, so depth and recoil tend to fall. Below it, the gradient decays between compressions. Depth raises aortic pressure, but excessive depth, above about 6 cm in observational data, has been associated with lower survival. Recoil lets the chest and heart re-expand in the relaxation phase; in a pig study in which a device held compression depth constant, incomplete recoil lowered relaxation-phase aortic pressure and narrowed the coronary gradient while right atrial pressure changed little. Duty cycle trades the two phases against each other: a longer push gives more time for forward flow, a longer release gives more time for refilling and coronary filling. The optimal duty cycle is not known; the 2025 guideline says it may be reasonable to make compression and recoil times approximately equal (COR 2b, LOE C-EO), and calls the optimum a knowledge gap.

An illustrative calculation, from the guideline numbers rather than from measurement, shows why rate and duty cycle are tied together. At 100 compressions per minute each cycle lasts 0.6 seconds; at 120 per minute, 0.5 seconds. With compression and recoil times approximately equal, the release phase available for refilling and coronary filling is about 0.3 seconds at the lower rate and about 0.25 seconds at the upper. Beyond the range the release gets shorter still, and any leaning eats further into it. These figures show the arithmetic of the cycle, not targets, and they do not say how much flow occurs in each phase.

Ventilation acts on the same chest. A positive-pressure breath raises intrathoracic pressure, and if breaths come too often or are held too long, the chest spends more of each minute at a higher pressure, which tends to oppose venous return during release in the way 5.1 describes. The animal and human evidence on ventilation rate during CPR is gathered in 5.1. The guideline rates are 30 compressions to 2 breaths without an advanced airway and 1 breath every 6 seconds with one.

Fatigue degrades all of them together. Depth begins to fall after about 90 to 120 seconds, while rate holds, which is why changing compressors at about 2 minutes is built into the cycle. The practical model is that a compression is good only if all four variables are within range at once, and that a change aimed at one, such as pushing faster, should be judged by what it does to the others. Feedback on depth and rate is one way teams see what the eye misses. End-tidal CO2 gives an indirect signal of the output those compressions produce, with the limits set out in Part VI. During arrest, real-time audiovisual feedback on CPR quality is reasonable (COR 2a, LOE B-R), though its association is with ROSC and survival to discharge rather than survival with favorable neurological outcome; in training, feedback devices are recommended (COR 1, LOE A).

Over the first several compressions after a pause, aortic pressure climbs toward a plateau. When compressions stop, it falls almost at once. A pause therefore costs its own seconds plus the run of compressions needed to rebuild the gradient afterward. Several short pauses can cost more perfusion than their total time suggests, because each one restarts the climb.

Chest compression fraction, the share of the resuscitation during which compressions are actually happening, is the bedside measure of this. The 2025 adult BLS guidance says a fraction of at least 60% may be reasonable, based on observational data that are mixed, and high-performing teams may exceed 80%.

Some pauses are necessary: two breaths in 30:2 CPR without an advanced airway, a rhythm or pulse check, and delivery of a shock. The job is to make them few and short. Professionals often take longer than 10 seconds to decide whether a pulse is present, so the instruction is to restart compressions if a pulse is not definitely felt within about 10 seconds.

The perishock pause, the gap just before and after a shock, is especially costly because it comes when a successful shock most needs perfused myocardium. Two randomized trials found no survival benefit from pausing after a shock to inspect the rhythm before restarting CPR, and failing to restart promptly is associated with less return of spontaneous circulation.

How the airway changes the plan: without an advanced airway, the compression cycle already includes a pause for two breaths, so a compressor swap or a quick look can be folded into a pause that would happen anyway. Once an advanced airway is in, breaths are given continuously without pausing compressions, so there is no breath pause to share. The only planned pauses left are rhythm checks and shocks, and compressor changes belong there.

The limitation: the build-up of pressure after a pause comes from animal and limited human measurements, and the exact number of compressions needed varies. The direction is consistent, but the curve is not something you can see at the bedside.

It can have organized electrical activity. It can contract. And those contractions can produce enough flow to perfuse the body. A monitor shows only the first. VF and pulseless VT have electrical activity that does not produce effective contraction. PEA has organized electrical activity with too little mechanical output to make a pulse. Asystole has neither.

Each of the three has a location and a physical form. The electrical event begins in specialized tissue: normally the sinus node in the wall of the right atrium, then the atrioventricular node, the His bundle, the bundle branches, and the Purkinje fibers that spread the impulse through the ventricular muscle. Depolarization is a wave of change in the voltage across cell membranes, carried by ions moving through channels, and the surface ECG records the sum of those voltage changes from outside the body. Contraction is a separate, chemical and mechanical event inside each muscle cell: depolarization lets calcium enter and be released from internal stores, calcium allows the contractile proteins to bind and shorten, and the process consumes ATP. Flow is a third step again: shortening of the ventricular wall raises pressure in the chamber, and only when that pressure exceeds aortic pressure does the aortic valve open and blood leave. Each link can fail while the one before it survives. An ischemic, acidotic, or energy-depleted cell can depolarize without contracting usefully, and a heart that contracts can still eject little if it has nothing to fill it or something blocking its outflow.

VF and pulseless VT are shockable because their failure is in the first link. In VF, wavelets of depolarization circulate chaotically through the ventricles, usually by re-entry, so different regions are activated at different times and no coordinated contraction occurs. In pulseless VT, the rhythm is organized but too fast, or the myocardium too damaged, for the ventricles to fill and eject a pulse. A shock passes current through the heart and depolarizes enough of the myocardium at once that the circulating wavefronts have no excitable tissue left to enter, and they stop. What follows depends on whether the conduction system and myocardium can then generate and respond to an organized impulse. A shock does not make the heart contract, and a shock is not indicated for PEA or asystole, where there is no re-entrant rhythm to terminate.

PEA covers a range of mechanical states that the monitor cannot separate. At one end, organized depolarization produces no detectable contraction at all, true electromechanical dissociation. At the other, the ventricles contract, sometimes vigorously, but eject too little to make a palpable pulse, a state often called pseudo-PEA. The difference matters because it points to different causes. A heart contracting around little blood suggests a filling or obstruction problem such as hypovolemia, tamponade, tension pneumothorax, or massive pulmonary embolism; a heart that is barely moving suggests advanced ischemia, acidosis, or a metabolic or toxic cause. Point-of-care ultrasound can show cardiac motion, and an arterial line can show small pressure pulses that a finger cannot feel. Point-of-care ultrasound can be used by an experienced operator when it does not interrupt CPR, though its usefulness during arrest is not well established, and seeing motion on ultrasound is not the same as measuring flow. Asystole is the absence of detectable electrical activity, and with it the absence of contraction; before accepting it, it is standard practice to check that leads and gain are not hiding a fine VF.

For resuscitation, the practical rule is that each kind of observation answers only its own question. The monitor answers whether the heart is depolarizing in an organized way and whether a shock is indicated. Ultrasound answers whether it is moving, and sometimes why it is not filling. A pulse, an arterial waveform, or a sustained rise in end-tidal CO2 answers whether it is producing flow. Using one to answer another's question is where errors come from: an organized rhythm read as a perfusing heart, or visible cardiac motion read as adequate output. This is reasoning from physiology; the guideline treatment of PEA and asystole, compressions, epinephrine, and the search for a cause, does not change with the subtype, and no trial has tested whether treating PEA by subtype improves survival.

Shockable rhythms are electrical problems with an electrical treatment. In dog experiments, fibrillation stopped when a shock depolarized a critical mass of myocardium at once; that gives an organized rhythm a chance to take over. It works best early. In the ARREST registry of witnessed out-of-hospital arrest with an initial rhythm of VF, the first shock terminated VF about 93% of the time when it came within 6 minutes and about 75% of the time when it was delayed past 16 minutes. Each minute of delay to the first shock was associated with a relative reduction of about 6% in survival to discharge.

Time changes the myocardium as well as the odds. As VF continues, the myocardium uses up its energy stores, and a shock is less likely to produce an organized rhythm that can contract. CPR before the shock can restore some of that substrate, which is why prolonged VF and freshly witnessed VF are not the same problem.

The 2025 ALS guidelines prefer a single shock to stacked shocks, followed by immediate CPR for about 2 minutes. Whether stacked shocks are better in a monitored, witnessed arrest is unknown. Use a biphasic defibrillator when one is available, at the manufacturer's energy. If that energy is unknown, the maximal dose on that device may be used; the algorithm gives a typical biphasic initial range of 120 to 200 joules as an example. A monophasic device, if that is all you have, is set at 360 joules. Later shocks are at least that same energy, and a higher energy may be considered when the device can escalate and a perfusing rhythm has not returned.

The limit of the model: an organized rhythm after a shock does not prove that the heart is perfusing. That question is answered by a pulse, an arterial line, or a sustained rise in end-tidal CO2, not by the monitor. The usefulness of pad vector change and double sequential defibrillation for VF still present after three or more shocks has not been established, so fix pad contact, pad position, and CPR quality before adding an unproven technique.

After defibrillation the myocardium has been ischemic for the duration of VF and has just been depolarized en masse. It often produces an organized rhythm with weak or no effective contraction for a period afterward, so the first post-shock rhythm is commonly asystole or PEA before any pulse returns. This is the mechanical reason to resume compressions immediately after a shock without pausing to inspect the rhythm or feel for a pulse: compressions supply coronary flow to a stunned myocardium during the interval when it most needs it, and the rhythm and pulse are checked at the end of the 2-minute cycle. Randomized trials of pausing to analyze after the shock found no survival benefit.

Two mechanisms look the same on a single rhythm check. In incessant VF, the shock did not terminate VF. In recurrent VF, the shock terminated VF and it returned, usually hidden by the CPR that followed; AHA 2025 states this accounts for about 80% of VF seen at post-shock rhythm checks. Because the mechanism usually cannot be known during CPR, the guideline uses 'persisting VF/pVT' for VF/pVT that remains after 3 or more consecutive shocks. Telling them apart requires seeing the rhythm in the seconds after the shock, before CPR artifact hides it. AHA notes that recurrence is usually obscured by the CPR that follows, and that reliable ways to distinguish the two during ongoing CPR still need investigation. Incessant VF points to the shock: pad position, contact, energy, and the state of the myocardium. Recurrent VF points to a substrate that keeps re-triggering it, and AHA notes it may be better treated with post-shock rhythm stabilization than by changing a shock that worked. Antiarrhythmics may be considered for VF unresponsive to defibrillation. The usefulness of vector change and double sequential defibrillation for persisting VF/pVT after 3 or more consecutive shocks has not been established (COR 2b, LOE B-R). The incessant/recurrent distinction is defined in the AHA Part 9 synopsis and carries no COR/LOE of its own.

Arterial oxygen content falls if ventilation is left out for long, especially in asphyxial arrest. For health care professionals, the 2025 BLS guidance says it is reasonable to give both compressions and ventilations, with enough tidal volume to see the chest rise, while avoiding both hypoventilation and hyperventilation.

Breathing also changes the circulation, and to see how, it helps to start with where the returning blood is and what pushes it. Most of the blood in the systemic circulation sits in the veins and venules, a large, thin-walled, distensible reservoir spread through the abdomen, the limbs, and the skin. If the heart stopped and flow ceased everywhere, pressures throughout the systemic vessels would move within seconds toward a common value, and that pressure is the mean systemic filling pressure. It is the pressure that the volume of blood exerts on the stretched walls of the systemic vessels, most of all the veins, measured in mm Hg against the same atmospheric zero as every other pressure in this text. It is not produced by the heart. It exists because the vessels hold more blood than they can contain without being stretched.

That distinction between holding and stretching is the core of the idea. Part of the blood in the veins only fills them to their resting shape and exerts almost no pressure; this is the unstressed volume. The blood beyond that stretches the walls, and only this stressed volume produces filling pressure. How much pressure a given stressed volume produces depends on venous compliance, the change in volume per change in pressure (mL per mm Hg): compliant veins accept a lot of blood with a small rise in pressure, stiff or constricted veins much less. Mean systemic filling pressure therefore rises when blood or fluid is added, and when sympathetic tone or a vasopressor constricts the veins, which is thought to shift blood from unstressed to stressed volume. In ICU patients, raising or lowering norepinephrine moved estimated filling pressure in the same direction, but it also changed the resistance to venous return, so cardiac output did not always rise. These are estimates outside of CPR, and the volume shift itself is inferred rather than measured. Pressure from outside the veins can act in either direction: in dogs, raised abdominal pressure pushed blood from the abdominal veins toward the chest, but it impeded return when the abdominal veins collapsed. It falls with blood or fluid loss and when venous tone is withdrawn, as after induction drugs. Pressure, volume, and distension are three different things here: the same total blood volume can produce different filling pressures depending on tone, and a vein can be full without being under much pressure.

Venous return is a flow, measured in liters per minute, and it needs a gradient like any other flow. Blood moves from the peripheral veins, at roughly mean systemic filling pressure, toward the right atrium, so the pressure it flows against is right atrial pressure. This is the same right atrial pressure that sits on the outlet side of the coronary bed in 3.1, now acting as the back pressure for the whole systemic venous return. A catheter reads it as intracavitary pressure relative to atmosphere, and that is the relevant value here, because most peripheral veins, outside the chest, sit at roughly atmospheric surrounding pressure; abdominal veins sit at abdominal pressure. When intrathoracic pressure rises, intracavitary right atrial pressure rises with it, even though the pressure actually distending the atrium may not change and the atrium may hold no more blood. For venous return, the rise in the measured value is what matters, because it narrows the difference that drives blood into the chest.

With those pieces in place, the relation can be written. Venous return is driven by the difference between mean systemic filling pressure and right atrial pressure. Written out, venous return equals mean systemic filling pressure minus right atrial pressure, divided by the resistance to venous return, with the pressures in mm Hg, venous return in liters per minute, and resistance in mm Hg per liter per minute. The resistance to venous return does not change that gradient; it sets how much flow a given gradient produces. An illustrative example, with numbers chosen only to show the arithmetic and not as measured or target values: if filling pressure is 12 mm Hg and right atrial pressure is 4, the gradient is 8 mm Hg; if a breath raises right atrial pressure to 9 while filling pressure stays at 12, the gradient falls to 3. In a hypovolemic patient whose filling pressure starts at 7, the same rise in right atrial pressure from 4 to 7 closes the gradient entirely. The arithmetic does not give the flow in any real patient, because filling pressure can move with the breath and resistance is not known.

Every positive-pressure breath is also a circulatory event. A positive-pressure breath raises intrathoracic pressure, and with it right atrial pressure. If right atrial pressure rises more than filling pressure over that breath, the gradient narrows, and with venous resistance unchanged, venous return falls. More breaths, or breaths held longer, keep that gradient narrowed for more of each minute. With less blood entering the heart, each compression has less to eject, and coronary perfusion pressure tends to fall. The prediction weakens when filling pressure rises with the breath, and the same rise in right atrial pressure costs a smaller fraction of venous return when filling pressure starts high: a volume-loaded or venoconstricted patient tolerates the same breath better, while a hypovolemic patient loses venous return sooner. In a study of 10 anesthetized adults undergoing defibrillator implantation, holding airway pressure at 15 rather than 0 cmH2O during apnea raised right atrial pressure and mean systemic filling pressure (measured during brief induced VF) by similar amounts, so the gradient was unchanged, yet left ventricular stroke volume fell by a mean of 23% (with wide variation); that is one bounded observation, not a description of the arrested circulation during CPR. The guideline names the harms of excessive ventilation directly: gastric inflation, regurgitation, aspiration, and decreased cardiac output.

During CPR this reasoning applies with two differences. The heart is not lowering right atrial pressure by ejecting, so the release phase of each compression, when the chest recoils and intrathoracic pressure falls, is the main opportunity for the right atrial pressure to drop below filling pressure and for blood to return. And the same right atrial pressure now sits downstream of two gradients at once: the venous return that refills the heart and the coronary perfusion that nourishes it. A rise in intrathoracic pressure that raises right atrial pressure therefore tends to narrow both. This is physiological reasoning built from the relations above; mean systemic filling pressure is not measured during clinical resuscitation, and the recommendation to avoid excessive ventilation rests on the guideline's statement of harms and on animal and observational data rather than on trials that titrated ventilation to a venous-return target.

The evidence for the circulatory cost of extra breaths during CPR is mostly animal. In pigs, faster ventilation rates during CPR raised mean intrathoracic pressure and lowered coronary perfusion pressure and survival; in people, very fast bagging has been observed during resuscitation, but the hemodynamic and survival effects come from the animal data. The pig result fits the mechanism above, since each extra breath adds time at raised intrathoracic pressure, which works against the venous-return gradient and the coronary gradient during the same release phase. This is the home of that finding; the sections on coronary perfusion and CPR mechanics point here rather than restating it.

Breaths also have to be judged by what they do for oxygen delivery, which is flow multiplied by arterial oxygen content (Part II). A breath raises content only by refreshing alveolar oxygen so that hemoglobin passing the lungs can load. During CPR pulmonary blood flow is low, so the blood that has to be oxygenated each minute is small, and once the hemoglobin leaving the lungs is nearly saturated, further breaths add almost no content. Their effect on intrathoracic pressure, however, acts on the flow term. Because content and flow multiply, a small gain in one cannot make up for a larger loss in the other. An illustrative calculation, with numbers chosen only to show the arithmetic and not measured in any patient: if faster bagging raised arterial content from 19 to 19.5 mL O2/dL, about 3%, but lowered the output of compressions from 1.5 to 1.2 L/min, 20%, delivery would fall from about 285 to about 234 mL O2 per minute (L/min times mL/dL times 10). The opposite error is real too. Ventilation that is too sparse lets content fall, especially in asphyxial arrest, which is why the guidance asks for neither hypoventilation nor hyperventilation.

More is therefore not better for reasons that are easy to run together. Beyond the point where the blood is saturated, extra breaths add little oxygen. They lengthen the time the chest spends at raised pressure, which tends to oppose venous return and narrow the coronary gradient. And they carry the harms the guideline names, gastric inflation, regurgitation, and aspiration. After ROSC the same excess acquires one more cost, because a falling PaCO2 constricts the cerebral arterioles, as set out in the post-ROSC targets.

The airway state determines the rhythm of breathing. Without an advanced airway the ratio is 30 compressions to 2 breaths. With an advanced airway in place, give 1 breath every 6 seconds (10 breaths per minute) without pausing compressions. Begin with a bag-mask and oxygen. If placing an advanced airway would interrupt compressions, defer it until initial CPR and defibrillation have failed, or until there is ROSC.

Induction drugs remove sympathetic tone, lowering mean systemic filling pressure. Positive-pressure ventilation raises intrathoracic and right atrial pressure. And the transition can remove a patient's own respiratory compensation for a metabolic acidosis if the ventilator delivers less minute ventilation than the patient was generating. A patient who was already short of volume or acidemic before the tube has the least reserve.

Gas trapping adds a fourth mechanism. When expiration is too short for the lungs to empty, as in severe asthma and, by the same physiology, in COPD or rapid bagging, each breath starts before the last has left. Volume and pressure accumulate in the alveoli (intrinsic PEEP, or auto-PEEP), intrathoracic pressure rises, right atrial pressure rises, and venous return falls by the same mechanism as a tension pneumothorax, but on both sides. The prediction is falling blood pressure, then PEA, with high airway pressures and quiet but equal breath sounds. Brief disconnection from the bag or ventilator lets the chest empty and may relieve hyperinflation; a rise in pressure soon after is consistent with gas trapping. The graded scope is therapeutic and limited to life-threatening asthma refractory to standard therapies, where active exhalation maneuvers may be considered (COR 2b, LOE C-EO); extending it to COPD or rapid bagging is physiological reasoning, not a graded recommendation. A sudden rise in peak pressure or difficulty ventilating in severe asthma should prompt evaluation for tension pneumothorax, which disconnection does not relieve. In life-threatening asthma, a low respiratory rate and tidal volume is reasonable (COR 2a, LOE C-LD).

What remains uncertain: in the arrested patient, the relative weight of these mechanisms usually cannot be measured, and more than one is often present. The timing, arrest within minutes of the tube, is the strongest clue that the procedure changed the physiology.

Asking what that variable is, and whether it is the one limiting the resuscitation, keeps four questions separate: what the drug does, whether it changes the variable, whether that helps the patient, and whether this patient resembles those studied.

Epinephrine. Its alpha-adrenergic effect constricts peripheral arteries, raising relaxation-phase aortic pressure during CPR, which raises coronary perfusion pressure if right atrial pressure does not rise with it. Its beta-adrenergic effects raise myocardial contractility, heart rate, and oxygen demand, which can increase myocardial oxygen consumption in a heart receiving little flow and may favor arrhythmia. The intended target in arrest is the alpha effect. Giving epinephrine is recommended (COR 1, LOE B-R), at 1 mg every 3 to 5 minutes: as soon as feasible in non-shockable rhythms, and after initial defibrillation attempts in shockable rhythms. Neither high-dose epinephrine nor vasopressin as a substitute offers an advantage. The surrogate and the outcome do not move together for this drug: epinephrine increases ROSC and survival to admission, but there is limited evidence that it improves survival to discharge or functional neurological outcome.

Amiodarone and lidocaine. These target the electrical substrate that lets VF restart or persist, not the circulation. Amiodarone blocks several channels and prolongs repolarization; lidocaine blocks sodium channels in depolarized, ischemic tissue. Either may be considered for VF or pulseless VT unresponsive to defibrillation, and the evidence does not distinguish between them. Doses are on the algorithm: amiodarone 300 mg, then 150 mg; lidocaine 1 to 1.5 mg/kg, then 0.5 to 0.75 mg/kg. They are recommended only for VF or pVT unresponsive to defibrillation; there is no fibrillating substrate to treat in PEA or asystole.

Calcium and sodium bicarbonate. Targeted drugs, not general ones: calcium for calcium-channel blockade and selected situations, bicarbonate for sodium-channel blockade. Routine use is not recommended; see the reversible-causes part for each indication and its strength.

Compression quality depends on who is compressing and for how long, so compressor changes are scheduled at rhythm checks rather than left to fatigue. Pauses are shortest when the next action is prepared before the current one ends: the defibrillator charged during compressions, the next compressor ready, the airway operator waiting for a planned pause instead of creating one. A team leader who is not doing a task can watch the clock, the compression fraction, and the reversible-cause list, and can notice when the hypothesis should change. Closed-loop communication is intended to reduce missed or duplicated drugs. After the event, debriefing is how the team learns what the monitor recorded and the room did not see. Debriefing after the event, using the recorded CPR data, is reasonable (COR 2a, LOE B-NR).

Here the term means the standard cycle is not working. The response has three steps, in order of cost.

First, recheck the basics, because they fail silently. Is compression depth still adequate after compressor changes? Is recoil complete? Is the compression fraction high, or have pauses crept in? Is the tube still in the trachea, and is ventilation slow enough? Are the pads making good contact, in a sensible position, at the right energy?

Second, change the hypothesis. A failing resuscitation is evidence that the working diagnosis may be wrong. Revisit the reversible causes from the history and the course so far: a fast, narrow PEA that never responds may be an obstruction or an empty circulation; recurrent VF may be ongoing ischemia; a wide, slow rhythm may be metabolic or toxic. Treat on suspicion where the guideline supports it, knowing which treatments are recommended, which are of uncertain benefit, and which are not established.

Third, escalate where the system allows. Extracorporeal CPR (ECPR) replaces the failing circulation with a circuit and buys time to treat a cause such as coronary occlusion, pulmonary embolism, hypothermia, or poisoning. Its benefit depends on selection and on a system that can start it quickly; it is not a rescue for every arrest. Rapid intra-arrest transport for ECPR may be considered for limited, highly selected adults with OHCA (COR 2b, LOE B-R), within a system that sets and reassesses selection criteria (COR 2a, LOE C-LD). It is reasonable to rewarm adults in hypothermic cardiac arrest using ECLS (COR 2a, LOE B-NR). Routine mechanical CPR is not recommended (COR 3: No Benefit, LOE B-R); it may be considered where high-quality manual compressions are challenging or dangerous, such as during transport, if interruptions during deployment are strictly limited (COR 2b, LOE C-LD).

The problem changes. During CPR the aim was to deliver any flow at all. After ROSC the aim is to keep oxygen, carbon dioxide, pressure, and temperature in ranges that limit further injury to a brain and heart that have just been through ischemia. The myocardium is often stunned, vascular tone is abnormal, and the brain may not autoregulate, so values that a healthy patient would tolerate can still cause harm.

The patient can still die of the arrest because the returning pulse ends only the no-flow part of it. Three threats remain, and they run on different clocks. The first is that the circulation fails again: the cause that stopped the heart may still be present, the stunned myocardium may not sustain its output, and a vasodilated circulation may not hold pressure, so shock and re-arrest are the dangers of the first minutes to hours. The second is the brain, which tolerates ischemia among the least of any organ and which, during reperfusion, depends on a pressure gradient it may no longer regulate. The third is every other tissue that was ischemic, where the injury tends to show itself later as organ dysfunction. None of these is undone by the pulse, and each can be made worse by what the team does next.

What is being protected has a location and a mechanism. During arrest, the brain, the myocardium, and every other tissue ran short of oxygen, used up their energy stores, and accumulated the products of anaerobic metabolism. When flow returns, oxygen arrives at cells whose ion pumps, mitochondria, and membranes have been damaged, and the restored supply is thought to generate reactive oxygen species and trigger inflammation in the vessel walls. The result is a second injury that unfolds over hours to days after the pulse returns: in the brain as neuronal death and swelling, in the heart as stunned muscle that contracts weakly even when its arteries are open, and in the circulation as vasodilation and leaky capillaries. Each of the targets below acts on one link in that chain, by limiting how much oxygen, carbon dioxide, pressure, or heat the injured tissue is exposed to.

What you are protecting in the next hour can therefore be stated as four exposures and one search. Oxygen, carbon dioxide, arterial pressure, and temperature are each set by the team, and each acts on one link in the chain just described: oxygen on the supply to reperfused cells and on oxidative injury, carbon dioxide on the tone of the cerebral arterioles, arterial pressure on the gradient into a brain that may follow pressure passively, and temperature on how much oxygen that brain consumes. The search is for the cause, because a cause left in place can stop the heart again however well the four exposures are handled; the 12-lead ECG, the decision about angiography, and imaging belong to that search, and persistent shock and its causes are taken up in the next section. The paragraphs that follow take the exposures in turn, with the ranges the guideline gives and the reasoning behind each.

The 2025 post-cardiac arrest recommendations stay inside ranges that trials have actually compared, and each target guards against a specific way the injured brain and heart can be hurt again. Give 100% oxygen until SpO2 or PaO2 can be measured reliably, because turning the oxygen down before you have a number risks hypoxemia. Once saturation is trustworthy, titrate the inspired oxygen fraction to an SpO2 of 90% to 98% (PaO2 about 60 to 105 mm Hg) and avoid both hypoxemia and hyperoxemia. The two ends fail in different ways: too little oxygen extends the ischemic injury, while too much can add oxidative injury to tissue that has just been reperfused.

The oxygen target is easier to read with the distinction drawn in Part II. SpO2 and PaO2 describe how fully hemoglobin is loaded and the partial pressure of dissolved oxygen, not how much oxygen reaches the tissue, which also depends on hemoglobin and cardiac output. Once hemoglobin is close to fully saturated, a higher PaO2 adds very little content, so hyperoxemia offers almost no gain in delivery while raising the oxygen tension in tissue that is vulnerable to oxidative injury. The target range therefore sits where content is near its maximum and the excess partial pressure is avoided. A patient with severe anemia or low cardiac output can be hypoxic at the tissue level even within the saturation range, so a normal SpO2 is necessary but not sufficient evidence that oxygen delivery is adequate.

In adults who remain comatose after ROSC, keep PaCO2 in a normal physiologic range, generally 35 to 45 mm Hg. Carbon dioxide is one of the strongest controls on cerebral vessel tone, so a low PaCO2 constricts cerebral arterioles and reduces flow to a brain that is already short of it. Mild hypercapnia, the opposite move, did not improve six-month neurologic outcome in the TAME trial.

Carbon dioxide acts on the cerebral vessels from the inside. It diffuses freely from blood into the smooth muscle of the cerebral arterioles, where it changes the local pH; a fall in PaCO2 raises that pH and the arterioles constrict, and a rise lowers it and they dilate. Cerebral blood flow therefore moves with PaCO2 while the vessels can still respond. After arrest, end-tidal CO2 can underestimate PaCO2 by a variable amount, for the dead-space reasons set out in Part VI, so a ventilator adjusted to a normal-looking end-tidal value can leave the arterial value higher or lower than intended. The blood gas, not the capnograph, is the measurement that answers this question.

Avoid hypotension by holding mean arterial pressure at least 65 mm Hg. When autoregulation is impaired, cerebral blood flow follows arterial pressure more directly, so a pressure the healthy brain would tolerate can leave the injured brain underperfused. Randomized comparisons of higher versus lower MAP targets have not shown better survival or neurologic outcome, and aiming below 65 mm Hg has not been shown to be safe in this group.

Arterial pressure matters to the brain through a gradient, like every other flow in this text. Cerebral perfusion pressure in the beating heart is mean arterial pressure minus the pressure opposing inflow, intracranial or venous pressure, whichever is higher. In a healthy brain, autoregulation adjusts arteriolar tone so that flow stays nearly constant across a range of perfusion pressures. After arrest that range may be shifted or lost, so flow follows pressure more passively: a fall in mean arterial pressure lowers flow directly, and swelling that raises intracranial pressure narrows the gradient from the other side. The individual patient's autoregulatory range cannot usually be measured at the bedside, which is why the guideline sets a floor rather than an individual target. Temperature acts on the demand side of the same balance. Cerebral metabolic rate rises with temperature and falls with cooling, so fever increases the oxygen the injured brain needs while its supply is uncertain; the trials compared specific temperature targets, and the guideline recommends keeping temperature within 32 °C to 37.5 °C without naming one value within that range as best.

Record a 12-lead ECG as soon as feasible. An immediate post-ROSC tracing can overcall ST elevation, and the absence of ST elevation does not rule out an intervenable lesion. Persistent ST elevation with a suspected cardiac cause calls for emergency coronary angiography, whether or not the patient is comatose. Without ST elevation, emergency angiography is reasonable for selected patients with a suspected cardiac cause who have cardiogenic shock, recurrent ventricular arrhythmias, or evidence of significant ongoing myocardial ischemia, again regardless of coma. In a comatose patient with none of those features, no ST elevation, shock, electrical instability, or significant ongoing ischemia, emergency angiography is not recommended over a delayed or selective strategy, because randomized trials found no benefit from going early. For adults who remain unresponsive to verbal commands, a deliberate temperature-control strategy is recommended, holding a temperature between 32 °C and 37.5 °C. In that same group, it is reasonable to continue temperature control for at least 36 hours. In adults with spontaneous hypothermia after ROSC, or who are being rewarmed after hypothermic temperature control, it may be reasonable to avoid rewarming faster than 0.5 °C per hour. Routine rapid infusion of cold intravenous fluid for prehospital cooling is not recommended.

Each of these reasons explains why an extreme is harmful; none, by itself, shows that a particular value is best. The targets are population ranges tested against alternatives, and an individual patient may sit at a different point on each curve. Severe anemia, chronic hypertension with a shifted autoregulatory range, raised intracranial pressure, or ongoing shock can all change what a given number means. During care, the practical stance is to meet the guideline ranges while asking, for this patient, what each number does and does not reveal about oxygen delivery, cerebral flow, and the heart.

The limit of the model: these are ranges for a population, and the evidence for several of them is that a different target did not help, not that this exact range is optimal. The reasoning explains why each extreme is dangerous. It does not identify the best value for a particular patient.

It is usually multi-mechanism: post-arrest myocardial stunning with low output, vasodilation from systemic inflammation, the original cause if it persists (ischemia, embolism, toxin, hemorrhage), and the effects of drugs given during arrest. The mean arterial pressure floor and the 12-lead ECG are set out in the post-ROSC targets; that target is a floor, not an optimum, and the 2025 guideline removed the previous systolic blood pressure target because the trials used MAP. There is insufficient evidence to recommend a specific vasopressor. Routine temporary mechanical circulatory support should not be used for cardiogenic shock after ROSC, though it may be considered in highly selected patients with refractory shock. Finding the cause matters as much as supporting the pressure: beyond the 12-lead ECG, it may be reasonable to use echocardiography or point-of-care cardiac ultrasound to identify clinically significant diagnoses requiring intervention, and head-to-pelvis CT to investigate the cause of arrest and complications of resuscitation.

Temperature control is set out in the post-ROSC targets. Seizures occur in about 10% to 35% of patients who do not follow commands, and many are visible only on EEG, so EEG should be performed and interpreted promptly in patients who do not follow commands; clinically apparent seizures should be treated, treating nonconvulsive seizures is reasonable, and routine seizure prophylaxis is not recommended. Sedatives and neuromuscular blockers can be cleared slowly after arrest and confound the examination. Prognosis should be multimodal, never based on a single finding, and delayed until enough time has passed to avoid confounding; it is reasonable to consolidate prognostic assessment at a minimum of 72 hours after normothermia and discontinuation of sedatives. The uncertainty to name is specific: a poor early examination may reflect drugs, temperature, or organ failure rather than brain injury.