Exposure to extreme cold results in a series of predictable changes in human physiology, which can be broadly described as "everything stops working". From the coagulation cascade to the the nephron to the interactions of haemoglobin with oxygen, all things become slower and more disorganised. Ultimately, the borders of survival can be found somewhere around 13.7°C, which was the temperature of the coldest survivor (unreliable accounts suggest it may be as low as 9°C for children). Beyond this point, the defrosted organism tends to be permanently damaged by their experience.

Question 1 from the second paper of 2000 asked the candidates to list the clinical effects of severe accidental hypothermia. The defining feature of this question was the accidental nature of the injury: the college were not interested in the pathophysiology of carefully maintained therapeutic hypothermia. Then, extreme hypothermia was ignored by the college for almost twenty years, ultimately reappearing in Question 7 from the second paper of 2019, where the examiners were mainly interested in the consequences of hypothermia for the management of cardiac arrest.

Among the prim,ary exam revision sections, there was already a chapter on the physiology of hypothermia which was probably too extensive for a last minute cram. This did not stop the author from copying that apocryphal content into these revision notes. That fringe material can be found trailing below the table for physiological consequences of hypothermia. The time-poor candidate may safely stop reading there.

A summary is all that is required for answering Question 1, and it is offered in a table below.  If one were to read only one peer-reviewed article, it should be the 2012 NEJM review by Brown et al. The hypothermia enthusiast may also derive some amusement from reading K.C. Wong's 1983 article,  or these two thorough reviews by Kees H. Polderman.

Definition and staging of hypothermia

Primary hypothermia is the process an otherwise healthy person being heat-depleted by exposure to some extreme cold environment. In contrast, secondary hypothermia can occur accidentally at a relatively normal temperature, and reuires the person to be in some way susceptible. Table 1 from Brown et al (2012) lists about thirty different conditions which may predispose a person to hypothermia. The notables include alcohol intoxication, hypoadrenalism, hypothyroidism, malnutrition and impaired shivering (e.g by quadriplegia or stroke).

A definition of the stages of hypothermia offered in the college model answer is as follows:

• Mild hypothermia: 32-35 degrees
• Moderate hypothermia: 28-32 degrees
• Severe hypothermia: under 28 degrees

This staging system is well known, but requires the measurement of core temperature, which may not be possible in the prehospital environment. There, the Swiss system is probably going to be more useful (Durrer et al, 2003). The Swiss system classifies hypothermia according to clinical features rather than temeperature:

• HT 1: Clear consciousness with shivering
• HT 2: Impaired consciousness without shivering
• HT 3: Unconsciousness
• HT 4: Apparent death
• HT 5: Irreversible hypothermia

The first three HT grades correlate roughly with the mild-moderate-severe temperature range; HT 4 spans the interval of 24-28 degrees and is defined as "apparent" death because historically people retrieved from frozen lakes with this sort of temperature have been asystolic, and yet went on to make reasonable neurological recoveries. The "real" death in HT 5 is associated with a temperature below 13.7 degrees, which is the lowest temperature ever survived by anybody (perhaps referring to this case report of a seven year old Swedish girl who fell into the winter sea).

Physiological consequence of hypothermia

Hypothermia results in a series of predictable physiological derangements in the human organism.The nature and severity of these derangements depends on the depth of hypothermia. A number of excellent free resources are available for this topic. K.C. Wong's 1983 article on this issue is one such resource, and I will refer to it constantly. Additionally, Kees H. Polderman has written two brilliant reviews on the topic of low temperature physiology; I am not sure whether he is the all-knowing guru of hypothermia, but he certainly has his articles available as free full-text, which earns massive brownie points with me.

The Physiological Consequences of Hypothermia

Endocrine and metabolic consequences Decreased metabolism and oxygen consumption Decreased carbohydrate metabolism and hyperglycaemia Decreased drug metabolism and clearance Essentially unchanged electrolytes Haematological consequences Increased hematocrit and blood viscosity Neutropenia and thrombocytopenia Coagulopathy and platelet dysfunction Respiratory consequences Decreased respiratory rate and medullary sensitivity to CO2   Acid-base changes: alkalosis and hypocapnea Rise of pH with falling body temperature Fall of PCO2 with falling body temperature Increased oxygen solubility and O2-haemoglobin affinity

Cardiovascular consequences Decreased cardiac output and bradycardia QT prolongation and the J wave Arrhythmias - classically AF and VF Resistance to defibrillation Vasoconstriction Renal consequences "Cold diuresis" due to decreased vasopressin synthesis Central nervous system effects Confusion and decreased level of consciousness Shivering Increased seizure threshold Immunological consequences Decreased granulocyte and monocyte activity

These effects can be represented on a large confusing diagram.

Clinical features of hypothermia

In their comments to Question 7 from the second paper of 2019, the cranky CICM examiners complained that many of the candidates "listed things that were not clinical signs e.g. ECG changes and ETCO₂". These days of course the intensivist has an absolute orgy of data available at the bedside, including potentially core temperature, mixed venous oxygen saturation, cardiac output and EEG; but the examiners wanted to reach back to a simpler time. For this, an excellent review by Rosin et al (1964) was highly informative. A slightly more modern take by Aslam et al (2006) is also available for free.

Clinical examination findings

• Observation
  • Mottled appearance
  • Cyanosis
  • Pallor
  • No shivering
  • Cold oedematous skin
• Respiratory findings
  • Apnoea, hypoventilation or Cheyne-Stokes respiration
  • Creps on auscultation, suggestive of pulmonary oedema
• Cardiovascular findings
  • Sinus bradycardia, AF or asystole
  • Hypotension
  • Weak thready pulse, or altogther impalpable peripheral pulses
  • Poor capillary refill
• Neurological findings
  • Coma (unresponsiveness below 32°C)
  • Increased muscle tone
  • Sluggish deep tendon reflexes
  • Extensor plantar responses
  • Loss of cranial nerve reflexes (with fixed mid-dilated pupils)
• Gastrointestinal findings
  • Absent bowel sounds
  • Abdominal tenderness (hypothermia-induced pancreatitis)
• Renal findings
  • Decreased urine output

The effect of hypothermia on metabolic rate

It is known that the rate at which chemical reactions proceed changes roughly in parallel with temperature; this relationship is represented by the Arrhenius equation. So one might surmise that the metabolic processes which are responsible for the ongoing function of our bodies will also obey this law. And indeed they do. In dogs, the relationship of metabolic activity and temperature has been studied in a rather broad range, down to 20° C. These dogs were attached to a closed anaesthetic circuit, and their whole-body oxygen consumption could be measured very precisely. The O2 consumption decreased by 50% at 28 degrees, and by 75% at 20°. One could extrapolate this into a graph, where one assumes that at 0°, the frozen dog metabolises nothing, and consumes no oxygen.

From this crude graph, one can imagine that the oxygen demands of a mammal decrease to about 60-70% when they are cooled to the normal range of therapeutic hypothermia, around 32-33°.

The oft-quoted figure (that for every 1 degree below baseline the metabolic rate decreases by 6%) is therefore probably not entirely accurate, as the relationship of metabolic rate to temperature does not appear to be linear.

Otis and Jude arrived at this figure in 1957, on the basis of more frozen dog experiments; I am unable to query their methodology because the article is behind a paywall.

The whole-body oxygen consumption is an average; obviously different tissues will handle the same temperature differently. For instance, at a normal temperature the kidney is the most metabolically active organ, consuming 8% of the total body oxygen while occupying only 0.5% of its mass. However, renal oxygen requirements (and renal blood flow) decrease at the fastest rate of all organ systems. Quoting a graph from Emil Blaire's 1964 masterpiece "Clinical Hypothermia", K.C. Wong reports that renal blood flow decreases by 40% at 32°, whereas cerebral blood flow only decreases by 25%. Cerebral blood flow could actually decrease significantly more, as can be demonstrated from some simple calcuations.

If we extrapolate these percentages to what we know about the metabolic demands of the human brain, we can calculate the values for cerebral oxygen and glucose demand. With a metabolic rate down to 60%, the brain will use only 1.8-3.1ml O₂ per 100g of brain tissue every minute. Whole-brain O₂ delivery can thus theoretically decrease to 27ml/min, which is the content of only about 135ml of well-oxygenated blood. Thus, cerebral blood flow could potentially decrease not by 25%, but by a staggering 80%. Of course, the blood coming out of the jugular will be totally anoxic in this scenario. One would not want to cut it so close, or keep going like this for long.

The effect of hypothermia on nutrient metabolism

Carbohydrate metabolism slows down, as does everything else. The glucose in the bloodstream remains underutilised, leading to hyperglycaemia. This thought to be the consequence of decreased insulin sensitivity.

Exogenous insulin is required, as hyperglycaemia is bad for the injured brain.

Gastrointestinal motility is suppressed in hypothermic individuals. If you were going to feed them, it would be better to start feeds after you have rewarmed them to a reasonably normal temperature.

The effect of hypothermia on drug metabolism and pharmacokinetics of clearance

Drug metabolism is decreased in hypothermia for two reasons. Firstly, the rate of enzyme-drug reaction (particularly among the CYP450 enzymes) slows down in a predictably Arrhenian fashion; secondly, the hepatic blood flow decreases, which theoretically should mean that the most rapidly cleared drugs are the most affected.

Indeed, it seems the circulating levels of propofol increase by 30% for every 3° below normal; and by 15% for fentanyl. To extrapolate - at 32° C, propofol clearance is halved, and fentanyl clearance reduced by 25%.

An excellent article from 2007 presents the results of an exhaustive Medline search, asking what happens to the ubiquitous ICU drugs inside hypothermic patients. I will present the salient features of this article in this table.

Effects of Therapeutic Hypothermia (~32°C)
on Drug Clearance

Drug	Alteration in clearance	notes
Propofol	Reduced to 50%	Hypothermia causes a decrease in inter-compartmental clearance of propofol, resulting in much more drug becoming "trapped" in the brain tissue. This may have something to do with the decreased cerebral blood flow.
Fentanyl	Reduced to 75%	Some of the "offset" of the effects of fentanyl is actually due to redistribution, and this will not be dramatically affected by hypothermia.
Midazolam	Reduced to 1%	Midazolam clearance decreases 100-fold in patients cooled to below 35° - which is a massive impairment to clearance, not to mention the effects of renal failure
Remifentanil	Reduced to 70%	Even remifentil is not immune - for every 5° decrease in body temperature, certain authors recommend a 30% downward dose adjustment. A remifentanil overdose has grave implications for the already bradycardic hypothermia patient. Of course, its clearance rate is still ridiculously quick, and once rewarming is commenced, remifentanil will wash out without problems.
Thiopentone	Reduced to 50%	Barbiturate metabolism will slow, but thankfully some of them are excreted unchanged in the urine, which ameliorates this effect somewhat.
Rocuronium	Reduced to 50%	Duration of effect increased by 5 minutes per every 1° below 37°. Thus, at 32°, a good dose of rocuronium may have useful effects for up to 60 minutes.
Atracurium (thus, also cisatracutium)	Reduced to 50%	This molecule is mainly degraded by non-enzymatic decomposition, which is temperature dependent (that's why the ampules are stored in the fridge). It stands to reason that the rate of this decomposition should be decreased at lower temperatures.
Phenytoin	Reduced to 25-50%	Its difficult to say with precision what happens at 32°, because the studies were done at 34° and 36°. That 2-degree difference resulted in an almost twofold decrease in phenytoin clearance.
Gentamicin	Reduced to 75-80%	Here, gentamicin is used as a model of a drug the clearance of which is not related to metabolism, because it is excreted unchanged in the urine. However, in a hypothermic patient glomerular blood flow is greatly reduced, and this decreases the filtration of such drugs.

Additionally, it should be mentioned that antiarrhythmic drugs will accumulate to toxic levels, and should be used sparingly. And vasopressors (such as adrenaline, particularly in the sort of doses one might use for cardiac arrest) should also be expected to have their clearance rates halved.

The effect of hypothermia on serum electrolytes

Weirdly, not much happens to these. At least not in the range of temperatures to which the ICU patients are routinely subjected. Sure, "cold diuresis" results in the increased possibility of potassium loss, but renal potassium excretion also slows, so these typically don't result in a clinically significant hypokalemia. The anaesthetised dogs in the 1950s experiments were cooled to 19 degrees and only experienced a potassium decrease from 3.5 to 3.0 mmol/L. Sodium calcium and magnesium all remained stable within a mild to moderate range of hypothermia.

However, below 25° this begins to change. Sodium and potassium decrease, and chloride increases-

we think. The greatest change in electrolyte concentration was observed in measurements of myocardial cells rather than serum samples - with the aim of determining which electrolyte shifts may be responsible for the frequent arrhythmias which derailed early cooling efforts. However, serum electrolytes changed as well. Chloride increased from 107-108 to 112mmol/L; similarly sodium decreased by 4-5mmol/L, and potassium decreased by about 0.5-0.6mmol/L. The significance of these electrolyte shifts is difficult to elaborate upon. Certainly, in the carefully monitored ICU environment, they appear to be trivial - as an entire team of medical staff armed with replacement solutions hovers constantly at the bedside.

The effect of hypothermia on fluid compartments, hematocrit and blood viscosity

Viscosity and hematocrit both increase during deep cooling, and the mechanisms of these changes have been explored. 

Firstly, there is a loss of plasma volume.

Where this plasma goes, nobody knows. Because serum protein levels remains largely unchanged and there is no change in red cell volume, investigators have suggested that this is a loss of whole plasma, rather than the "dehydration" of the bloodstream. There is decreased lymphatic flow, suggesting that some vascular beds somewhere have lots of low-hematocrit blood sequestered inside them.

There is also decreased interstitial fluid space, and decreased capacity for transcapillary ultrafiltration, suggesting that a lot of the microvasculature no longer participates in the normal balancing act of intravascular-extravascular fluid. This means that blood loss will not be readily compensated for by the migration of fluid out of the interstitium. Also, this means your fluid boluses will not redistribute as readily, and will remain in the intravascular space for a longer period.

Not only is there plasma volume loss, but the physicochemical properties of blood plasma change at low temperature. This is a very complicated topic; an excellent book has been written about it by L.Dintenfass, which summons some exotic references. Fascinating though it may be, I caution the casual reader from jumping into that rabbit hole. I will summarise: ultimately, blood viscosity depends on the interaction of blood protein elements, the aggregation of blood cells, the internal viscosity of blood cells, and the hematocrit. Overall, it has been found that the ratio of plasma viscosity to whole blood viscosity remains unchanged over the temperature range of 10°-37°. However, the aggregation of red cells and their internal viscosity increase, and so the overall viscosity increases - in fact it almost triples from 37° to 10°. Furthermore, this is exaggerated by low shear states - that is to say, the faster blood is flowing, the less viscosity it appears to have.

That brings us to the next reason for hypothermic hyperviscosity- the low flow state. In conditions of decreased cardiac output and increased systemic vascular resistance, blood will flow more slowly, and therefore exhibit a greater tendency towards hyperviscosity. This has implications for organ perfusion. However, at present there are no guidelines urging us to hemodilute our hypothermic patients. To the contrary, the patients post cardiogenic cardiac arrest would probably benefit from an increased hematocrit, and the transfusion guidelines suggest that we should aim for a hemoglobin level of around 100 g/L.

In fact, many patients will arrive to the ICU post-cardiac arrest in a state of post-resuscitation hemodilution, having had litres of fluid prior to their admission; and we never cool them as deeply as would be required to produce some of the above-discussed effects. However, if one were to suffer their cardiac arrest while trapped under a moose in the Siberian tundra, these mechanisms would be acutely relevant to their retrieval team.

Neutropenia and thrombocytopenia as a consequence of hypothermia

There is an observed neutropenia and thrombocytopenia which occurs during deep hypothermia, and which is thought to be due to the sequestration of neutrophils and thrombocytes in the slow-flowing capillary beds of the liver and spleen. Certainly, hypothermic dogs seem to drop their platelet counts from 208,000 to 58,000 when cooled to 20°; something similar happens to their neutrophils. These cells al reappear readily upon rewarming, and selective organ biopsies and splanchnic vessel catheterization has revealed the spleen and liver as the dominant sites of sequestration.

In short, though thrombocytopenia in the hypothermic patient may be a barrier to invasive procedures, it reverses readily and one should not spent too much time stressing about it.

Coagulopathy of hypothermia

There is an excellent article in the Journal of Trauma which discusses this topic alongside acidosis. Naturally, hypothermic trauma patients (trapped under a moose in the tundra, waiting for retrieval, etc.) are a population in whom coagulopathy is a serious threat, and it approaches the issue from this viewpoint; however the exploration of hypothermic coagulopathy is still relevant to the intentionally hypothermic post-arrest patient, particularly as these patients frequently arrive to the ICU in a post-resuscitation state, acidotic and haemodiluted.

The major contributing factor to hypothermic coagulopathy are a failure of fibrinogen synthesis and

A swine model of hypothermia (to 32°) has revealed that the rate of fibrinogen synthesis halves at this temperature. Fortunately, fibrinolysis is not affected; clot strength remains the same at this temperature, and hemostasis in post-operative patients - once achieved- should not be adversely affected.

However, you just try to form a new clot, and see what happens. Another 32° swine model (by the same author, Martini) has investigated the independent contributions from hypothermia and acidosis on clot formation. Leaving acidosis aside for now, Martini has determined that hypothermia decreases the rate of thrombin formation, suggesting that it is the Factor VIIa-Tissue factor pathway which gets affected.

Additionally, the hypothermic platelets just don't want to aggregate. This platelet dysfunction occurs by multiple mechanisms, mainly related to the dysfunction of platelet surface receptors, some sort of loss of affinity for their ligands. This is rapidly reversed by rewarming.

This may prompt one to say "Aha! The platelets too few and wont aggregate, fibrinogen is lacking, and the extrinsic pathway is impaired. Can I simply infuse some extra platelets FFP and cryoprecipitate?" Indeed, how can one reverse the coagulopathy of hypothermia, and realistically does one ever need to? In his 2012 review article, Polderman the hypothermia guru addresses these concerns. In even profound hypothermia, the greatest contribution to coagulopathy seems to be acidosis; the contribution of hypothermia is minor, and can be safely ignored even if the patient has been thrombolysed!

However, Polderman acknowledges that in some cases where there is cardiac arrest AND uncontrolled bleeding, one may need to come to a compromise with the surgeons. He presents data that suggests one can cool such a patient to 35° instead of 32°, and rewarm them of the bleeding becomes a lifethreatening problem. He provides a sample of personal practice for such circumstances; to transfuse the patient with platelets and to administer DDAVP.

The effect of hypothermia on respiratory function

Again we return to the early animal studies. The initiation of hypothermia has the effect of increasing ventilation - but we never see this, as the patient is mechanically ventilated. Anatomical dead space increases due to "cold bronchodilation", decreasing peak airway pressures (which is favourable, but typically lost in the quagmire of pulmonary oedema and aspiration pneumonia). As hypothermia becomes established and the metabolic activity begins to decrease, respiration becomes more shallow and the respiratory rate decreases, attending to the body's decreased CO2 clearance requirements. The sensitivity of the medulla to CO₂ also begins to decrease below 34°, which complements the decreased respiratory drive. At 24°, spontaneous respiration ceases; below 20° there is no response to either hypoxia nor hypercapnea.

The effect of hypothermia on arterial blood gas analysis

• Interpret the pH at 37° without correction
• Interpret the PaCO₂ at 37° without correction
• Interpret the PaO₂ at 37° without correction.

In short, do everything as you normally would.

Why, you ask?

The lower the temperature, the higher the corrected pH.

pH changes with temperature, and this relationship is almost linear. The linked article addresses this concept with diligence, brevity and a heaping serving of complicated maths. For the calculus-averse, Kerry Brandis has gone through this with patience and passion. For those who do not appreciate brief and lucid explanations, a rambling series of digressions is also available as a part of the ABG Interpretation chapter.

Briefly, the reason for the change in pH associated with changes in temperature is explained by the fact that dissociation is an endothermic reaction - i.e. for a given acid HA, with more energy available in the system more HA molecules will dissociate into their components, H⁺ and A^-. The balance of HA and its dissociation products will therefore trend left (towards HA) if energy is subtracted from the system, eg. by a big cooling blanket. With less H⁺ activity in the system, the pH will increase (because it is a negative logarithm of H⁺ activity). Thus, the apparent alkalosis of hypothermia is observed.

Now, blood gases are measured at 37°, which is usually close enough to our patients actual temperature. However, if our patient is not normothermic, we have the option of correcting the ABG results for their actual body temperature. This will give very different results.

Say you correct the pH for 20° or 30°, whatever the patients temperature is. You get a number. But you have no normal reference ranges for pH at that temperature - only for 37°. If you start using the normal values for pH and pCO₂ at 37° to interpret the gas at 20°, you will run into all sorts of trouble. For one, the patient will appear to have a respiratory alkalosis, and you will be tempted to play with the ventilator.

But in actual fact, at all temperatures intracellular pH remains at pN- the normal pH of neutrality, required for cellular function. The reason for this is that protein buffering of intracellular pH (via imidazole histidine residues) is also temperature dependent, and changes in parallel with body temperature. Not only that, but as intracellular pH increases (obeying the linear relationship) so must the extracellular pH increase, in order to maintain the normal pH gradient (from pH 7.4 to pH 6.8, at 37°). Thus, if the cellular machinery is already functioning under ideal conditions for that temperature, the introduction of a respiratory acidosis (by changing ventilator settings) will result in a deterioration of this function, by decreasing the transmembrane pH gradient.

Lastly, and most importantly, behold the definition of neutrality (going back to Arrhenius)- it is not "a pH of 7.0", but rather the presence of equal numbers of H⁺ and OH^- ions. Because temperature has an equal effects on the concentration of each of them, neutrality is preserved no matter the temperature.

The lower the temperature, the lower the corrected pCO₂.

Solubility of CO₂ is increased in hypothermia; it is a temperature-dependent property. The PCO₂ of your hypothermic patient, if interpreted at the actual body temperature, will appear abnormally low (even if you already expect it to be low, given the decreased metabolic rate).

The lower the temperature, the lower the corrected PaO₂.

But: changes in temperature do not significantly alter the oxygen content of blood.

Solubility of O₂ is also increased in hypothermia. In fact, total blood oxygen content is increased in hypothermia. Now, the solubility of oxygen in water increases only by 100% from 37° to 0°, but the affinity of hemoglobin for O₂ increases by a whopping factor of 22. The hemoglobin hungrily absorbs the dissolved oxygen, decreasing the PO₂ - thus increasing the concentration gradient and enticing more O₂ into the blood from the alveolar gas. (this doesn't mean that the oxygen-carrying capacity of the blood is dramatically increase, of course- there is still 1.37ml of oxygen per 1g of hemoglobin).

Now, you put this cooled sample into the blood gas machine, and the electrode will measure its PO₂ at 37°. The increase in temperature at the electrode will force the hemoglobin to give up a lot of its oxygen, and some of the dissolved oxygen will also come out of solution. The result is a slightly raised PaO₂.

However, this is not a realistic way of looking at the sample. Who cares what the available oxygen content of blood is at 37°? At a low temperature, all that oxygen is trapped in high-affinity bonds with hemoglobin, and is unavailable to the mitochondria.

How much oxygen is available to these cold mitochondria? There are numerous complex formulae to arrive at the corrected PO₂ value. However, a simple method is to subtract 5mmHg for every 1°C from the measured PaO₂ - thus, at 32° with a measured PaO₂ of 100mmHg the actual PaO2 is 70mmHg.

So what do you do? Does PaO₂ need to be corrected at the bedside?

In short, no.

Ashwood and colleagues present us with an excellent table (see Table 5 in their article) which demonstrates that for any given hemoglobin saturation, oxygen content of blood (in mls of gas per ml of blood) is not significantly altered, even though the calculation of PaO2 gives vastly different values.

For example, at 95% SaO₂ the hypothermic patient (at 22°) would have a PaO2 of 95mmHg when measured at 37°, and 27mmHg when measured at 22° - that difference is massive. But the calculated oxygen content of blood varies very little - from 0.1920 to 0.1933 mls/ml.

Thus, there is little point in correcting PaO₂for temperature. The American Association for Respiratory Care recommended against the routine correction of blood gas samples for temperature.

A more indepth discussion of the influence of partial pressure and temperature on gas solubility is carried out somewhere among the ridiculous digressions of the Arterial Blood Gas Analysis chapter.

Decreased cardiac output associated with hypothermia

Like with respiratory function, we the ICU people never get to see the normal physiology in action. The normal sympathetic overdrive which increases heart rate and cardiac output in awake mammals tends to fail them with anaesthesia, and what we see instead is a temperature-proportional decrease in cardiac output.

K.C. Wong quotes some inaccessible early papers as reporting that the cardiac output decrease is predominantly due to a decrease in heart rate, which decreases by 50% at 28° and by a massive 80% at 20°. Apparently, these decreases in cardiac output closely mirror the decreases in whole-body oxygen demand. The bradycardia is mainly due to the decrease in conduction through the chilled conduction system.

However, again these heart rates of 10-15 is something we never see. Rather, we see bradycardia to 40 or so, as well as a host of ECG abnormalities and arrhythmias. Below 28°, the risk of cardiac arrhythmias becomes very high. These can take any form, be it SVT, AF or VF.

QT prolongation and the J wave

The QT interval becomes prolonged. Additionally, a J wave or "Osborne wave" appears - I refer the reader to a page from Life In The Fast Lane, from whose excellent collection I have shamelessly misappropriated the image below:

Infrequently acknowledged complications of hypothermia which are relevant to the heart are coronary vasoconstriction and increased blood viscosity. These are counterproductive to the stunned ischaemic myocardium. Certainly, its metabolic rate is reduced, but we must be mindful of the fact that many cardiac arrest survivors had some sort of massive coronary arterial event, and even if they didn't their coronary arteries are probably not in a very good condition. Anything which decreases flow through these diseased vessels is going to contribute to the already high risk of post-cardiac arrest arrhythmias.

Arrhythmias in deep hypothermia

In brief, one can expect the slowing of the rate, the development of AF, the increased risk of VF, and ultimately everything trending towards asystole.

A 1960s article brings us early experiences with hypothermia and arrhythmias, derived from a case series of cardiac surgical patients managed intraoperatively at low temperature. In this case series, all rhythms slowed, AV nodal conduction slowed, and in a couple of chronic fibrillators sinus rhythm was suddenly restored for the duration of the cooling.

At extremely deep hypothermia, it appears the myocardium develops VF or asystole at between 26° and 19°. If you were lucky to be wandering around in the tundra with a supply of Class I antiarrhythmic drugs, you could push this temperature even further - the abovelinked study mentions that quinidine-treated dogs were cooled to as low as 12° before cardiac events took them out. Though excited by this advancement, Frank Gollan writes soberly:

"The euphoria of investigators who have taken this hurdle soon changes to despondency when they try to rewarm the animals."

Indeed. Most of those dogs perished. However, others were able to rapidly cool some mice to about 2°, maintained them without oxygen or heartbeat for up to 1 hour, and then defrosted them ...reasonably safely.

And this brings up a valid question. If these patients are going to fibrillate all over the place, how do we stop that?

Certainly, studies on the early antiarrhythmic pharmacopoeia had confirmed that Class I agents (specifically, quinidine) were the most effective, whereas other agents were either useless or actually pro-arrhythmic. The pro-arrhythmic effect was found to be related to the QT prolongation effect of some antiarrhythmic agents, which is enhanced by the cold; particularly sotalol was found to be strongly pro-arrhythmic in hypothermia, and should not be used. Even good old amiodarone- the workhorse of the ICU- can prolong QTc by 100msec or so, and the Irish have recommended that you watch those patients carefully.

So which anti-arrhythmics are safe? Presently, there is no agreement. Lignocaine does not prolong the QT interval, and may be one of the answers, but its use in hypothermic patients is not well researched.

An article on accidental hypothermia from 2002 writes about hypothermic VF arrest, and recommends bretylium as "the only anti‐arrhythmic agent of any use in this situation" on the basis of ancient yellowed scrolls and dog studies.

So, what do if they obstinately refuse to remain in sinus rhythm, and arrest again? The AHA recommendations for cardiac arrest in "special circumstances" make some adjustments to the normal ALS algorithm to accommodate the unusual cardiac electrophysiology associated with hypothermia. Though the use of adrenaline in hypothermic arrest is supported by evidence, it should be given every 4th cycle of CPR rather than every 2nd cycle - because of the diminished clearance.

Resistance to defibrillation

Though some theoretical models have suggested that direct current may be less effective in hypothermia, swine models have instead demonstrated that hypothermia improves the chances of successful defibrillation. Most guidelines recommend that attempts at defibrillation are going to be most successful at temperatures above 28-30°; however, case reports exist of successful defibrillation in patients with core temperatures as low as 25.9°.

Pragmatically, it is probably useless to defibrillate somebody in whom the body temperature will not allow an adequate cardiac output even while in sinus rhythm. It would probably be better to perform CPR on these people until sufficient rewarming has been accomplished - say, above 28°. Above this temperature, the risk of arrhythmia diminishes. If you defibrillate them at this stage, they are more likely to stay defibrillated, and cardiac output may be sufficient to support their diminished metabolic requirements.

Vasoconstriction

Systemic vascular resistance increases in hypothermia, and this is to be expected, as it is a stereotypical sympathetic response to falling body temperature. From the intensivists' point of view, it is slightly inconvenient - firstly, it increases the afterload on the diseased ventricle, and secondly, it decreases the blood flow to the patients surface, making them resistant to cooling. Interestingly, the resistance of the pulmonary vasculature does not change across a broad range of low temperatures.

The effect of hypothermia on renal function: mechanism of cold diuresis

There is the well known phenomenon of "cold diuresis", which is the observed(and occasionally torrential) increase in the output of dilute urine associated with hypothermia. The senior staff at our unit have explained this away as a dysfunction of the tubule, resulting from decreased ion channel function associated with lower temperature. This sounded plausible. A study of chilled rats paints a different picture - it seems the cold diuresis is due to a non-osmotic suppression of vasopressin synthesis which occurs at low temperatures. Certainly, a vasopressin infusion readily reverses cold diuresis in the abovementioned rats.

Again in rats, hypothermia protected ischaemic kidneys from the adverse effects of ischaemia-reperfusion. The investigators report also that after 60 minutes, the effect was lost - so this is the window of opportunity. Cooling the rats after 60 minutes did nothing to protect their kidneys.

The effect of hypothermia on consciousness

K.C. Wong reports 31-30 degrees as the threshold beyond which consciousness becomes "clouded", and sedation requirements decrease. This was observed firsthand by the participants in the great Heroic Age of polar exploration; in his "History of accidental hypothermia", Henry Guly quotes from Sir Walter Scott:

"There can be no doubt that in a blizzard a man has not only to safeguard the circulation in his limbs, but must struggle with a sluggishness of brain and an absence of reasoning power which is far more likely to undo him… It is a rambling tale to-night and a half thawed brain."

For the pragmatic intensivist not interested in the rambling of frozen Scotsmen, the mental slowing associated with hypothermia is an important factor to consider in the choice of sedating agents. Not only are the pharmacokinetic properties of the drug affected, but the requirements for sedation are greatly decreased by this obtundation. I wont even mention the hypoxic brain injury, which is usually associated with post-arrest coma.

In short, you need much less propofol than you think you do, even adjusting for the decreased clearance.

The limits of cerebral hypothermia

How low can you go? There is good evidence that cerebral metabolism is aerobic even at low temperatures, which suggests that ongoing good oxygen supply is required. Indeed, if you continue to perfuse the brain with oxygenated blood, it may sustain little injury from hypothermia at temperatures as low as 20° - in fact the cardiothoracic anaesthetists do this routinely. (well, there may be some neurological hiccups, but these are typically blamed on microthrombi.) But can we go any lower?

A group in the carefree 1960s had twenty-six dogs whose aortas they cannulated, putting them on bypass and cooling their brains to between 0.2° and 14.1°. Temperatures above 10° seemed to be well tolerated - those canines had no deaths, and returned to something resembling normality. However, over half of the dogs with brains below 5° had died, predominantly from massive neurological disturbance and cardiovascular collapse. Thus, 10° core body temperature may be the safe threshold for hypothermia in the humans.

Thus far, case study evidence supports this theory. Anna Bågenholm, a Swedish radiologist, became a case study in hypothermia when she was rapidly submerged in water, and where she remained trapped beneath a layer of ice for 80 minutes. For 40 of those minutes she was able to breathe from a pocket of air; however, upon her extraction from the water, she had neither spontaneous respiration nor circulation. With CPR in progress, upon arrival to hospital, her core body temperature was recorded as 13.7°, the lowest on record for a surviving human victim of accidental hypothermia. ROSC was established around 65 minutes after arrival to ED (thus, after155 minutes of CPR in total, preceded by a hypothermic circulatory arrest for 40 minutes).

Her long term deficits were minimal, but she did end up spending well over 3 months in ICU, of which 35 days were ventilated. And, needless to say, the anaesthetist in charge of her resuscitation wrote her up as a case report and published in the Lancet. Since then, the ARC and ILCOR recommend that nobody be declared dead until they are "warm and dead", and that CPR be commenced on all hypothermic patients unless their chest wall is so rigidly frozen that it does not permit compressions.

Shivering

Shivering is a pert of the normal homeostatic response to hypothermia. K.C.Wong reports the findings of the hideous Nazi experiments, which found shivering an immediate response to any exposure to extreme cold. The ability of the organism to generate heat in this way is lost at around 27 degrees of core temperature.

Initially, as shivering begins, the whole body metabolic rate and oxygen consumption increase. Clearly, this is counterproductive, and must be stopped. Some authors advocate opiates, whereas others favour neuromuscular junction blockers.

Increased seizure threshold

It appears that the slowing of the cerebral metabolic rate decreases the likelihood of sezure activity, and protects the brain from seizure-associated excitotoxic damage. At least among experimental animal models of status epilepticus, benzodiazepines and hypothermia in synergy seem to have a profoundly-counter-epileptic effect.

Immune suppression

Not only is the release of proinflammatory cytokines decreased, but granulocytes and monocytes diminish in activity developing a difficulty migrating into tissues, with decreased rates of phagocytosis. This is beneficial to the injured brain, and limits the proinflammatory response to global hypoxia, but it certainly is useless if there is an active infection. The slow flow of blood in the peripheries doesn't help either. In short, the healing of wounds and and pressure sores will be markedly impaired, and post-operative infections become more likely.

Resuscitation of the severely hypothermic patient

The ARC ALS2 manual is perhaps the best reference point for resuscitation of hypothermia. They offer a list of important differences in the initial management of a hypothermic patient in cardiac arrest. In addition to this resource, Chris Nickson's LITFL page on hypothermia is an excellent overview of resuscitation from hypothermia, including all the relevant techniques.

Issues in prehospital management  and cardiopulmonary resuscitation

• When checking for pulse, one should be patient. Profound bradycardia and the decrease in cardiac contractility associated with severe hypothermia may result in a feeble slow pulse, which can be easily missed.
• Resuscitation should be attempted on all patients in whom there is no evidence of lethal traumatic injury, fatal illness, or prolonged asphyxia.
• Realistically, one should attempt CPR in all patients, unless the chest is actually incompressible because they are frozen solid.
• Use the same ventilation and chest compression rates as for a normothermic patient.
• CPR should continue until the patient has rewarmed to at least 35 degrees.
• A special low-reading thermometer is required to measure the correct core temperature.
• Defibrillation is unlikely to be effective below 28°C.
• Drug metabolism has been slowed significantly; one should not use any resuscitation drugs until the temperature is at least 30°C.
• Between 30°C and 35°C drug metabolism is still slowed, and adrenaline should be given at half the usual rate, i.e. every 4 cycles.
• Some attempts should be made to prevent the patient from getting any colder; for example, having just fished them out of a body of water, the rescuers should make some attempt to dry the patient, so that they do not lose any more heat by evaporation.

The phenomena of afterdrop and rescue collapse

• Afterdrop is the mythical phenomenon of continued core cooling after rescue. It is supposed to happen when circulation is restored following a prolonged period of poor or absent perfusion. Suddenly, cold blood from the peripheries returns to the heart, thereby dropping the core temperature. Apparently, it has been documented in artificial cooling experiments and inferred from discrepancies between rectal and core temperatures. Brown et al (2012) mention that it has never been reported in the era of accurate oesophageal temperature measurement and active rewarming.
• Rescue collapse is the very real non-mythical possibility that the previously stable hypothermic  patient will suddenly arrest when you fiddle with them. This tends to happen as they rewarm (and develop circulatory collapse from vasodilation). The other possibiluty is cardiac arrhythmia: pProcedures at this early stage (with a temperature around 24-28°C) could possibly trigger VF. The patient is more likely to be asystolic anyway. Fear of VF should not deter rescuers from intubating the patient.

Issues in emergency hospital management

• Rewarming is the priority.
• It is important to remember that the heart is irritable, and tickling it with CVC guidewires may precipitate "rescue collapse".
• With CPR in progress and a very low body temperature one may consider starting VA ECMO in the emergency department.
• The ideal gold standard method of measuring core temperature is with a PA catheter
• The next best measurement can be taken from the lower third of the oesophagus

Options for rewarming

• Passive rewarming:
  • The conscious patient should be rewarmed passively (i.e. allowed to shiver).
  • Removing wet clothes can prevent the loss of heat
  • A reflective blanket can help the retention of heat
• Active surface warming
  • Warm blankets
  • Chemical heat packs
  • Forced air surface warming (eg. Bair Hugger)
  • Arteriovenous anastomosis reheating (essentially, putting their hands into some warm water)
• Minimally invasive core warming
  • Heated humidified air in the ETT
  • Warm intravenous fluids (but this is limited by the volume you would be comfortable infusing)
• Invasive body cavity warming
  • Thoracic lavage: warm fluid administered via chest tubes into the pleural cavity; in via a midaxillary tube, out via a midclavicular tube.
  • Similarly, you can lavage the peritoneal cavity and the bladder
• Extracorporeal circuit rewarming
  • Could use a dialysis circuit, but .. why would you.
  • ECMO has the strongest evidence in this context; it is the preferred method.