Oxygen saturation and haemodynamic changes prior to circulatory arrest: Implications for transplantation and resuscitation

Abstract

Aims

To describe the progression of oxygen saturations and blood pressure observations prior to death.

Introduction

The progression of physiological changes around death is unknown. This has important implications in organ donation and resuscitation. Donated organs have a maximal warm ischaemic threshold. In hypoxic cardiac arrest, an understanding of pre-cardiac arrest physiology is important in prognosticating and will allow earlier identification of terminal states.

Methods

Data were examined for all regional patients over a two-year period offering organ donation after circulatory death. Frequent observations were taken contemporaneously by the organ donation nurse at the time of and after withdrawal of intensive care.

Results

In all, 82 case notes were examined of patients aged 0 to 76 (median 52, 4 < 18 years). From withdrawal of intensive care to death took a mean of 28.5 min (range 4 to 185). A terminal deterioration in saturations (from an already low baseline) commenced 14 min prior to circulatory arrest, followed by a blood pressure fall commencing 8 min prior to circulatory arrest, and finally a rapid fall in heart rate commencing 4 min prior to circulatory arrest. Two patients had a warm ischaemic time of greater than 30 min; 15 patients had a warm ischaemia time of 10 min or greater; and 53 patients had a warm ischaemia time of 5 min or less. It was observed that 0/82 patients had saturations of less than 40% for more than 3 min prior to cardiac arrest and 74/82 for more than 2 min.

Conclusions

There is a perimortem sequence of hypoxia, then hypotension, and then bradycardia. The heart is extremely resistant to hypoxia. A warm ischaemic time of over 30 min is rare.

Introduction

The perimortem physiology of the human body is poorly understood. Several animal studies have shown the physiology of some species as they approach circulatory arrest, and such data have been extrapolated to gain a greater understanding of what may happen in humans. Human data may improve our understanding of the physiological progression towards death. It may also have a significant impact on improving rates of successful organ donation after circulatory death (DCD) and optimising appropriate allocation of medical resources.¹ Further to this it would help us to understand the hemodynamics and oxygenation in the peri-arrest period, which would provide important information to help guide resuscitation and its prognostic aftermath.

The insults that lead to death can be considered as primary brain events (e.g. traumatic brain injury or intracerebral vascular events) or events commencing external to the brain that lead to brain hypoxia. The latter is most often the result of circulatory failure. Circulatory failure can be further subdivided into a cardiac cause, where at the point of cardiac arrest the rest of the body is physiologically relatively normal, or a non-cardiac cause, where the cardiac arrest is as a result of a catastrophic failure of the heart secondary to external factors.² Overwhelmingly the most common non-cardiac cause is severe and sustained hypoxia and in some patient groups, such as paediatrics, this is almost universally the cause of any cardiac arrest.^3,4

In humans with severe hypoxia, each organ has a different susceptibility to permanent cell damage. It is known that acute severe hypoxia can cause loss of consciousness in 10 to 20 s and permanent brain damage in 3 min.⁵ Animal work suggests that the point of complete and total irreversible brain failure may be up to 60 min after circulatory arrest,⁶ but restoration of circulation must be within 11 min for restoration of normal cerebral function in laboratory animals.⁷ Other organs, for instance the kidneys, have a much longer viable hypoxic time. During transplant operations, when the kidney is cooled, it can remain viable whilst hypoxic for up to 36 h.⁸

It is unknown however what amount, both in time or depth, of hypoxia will cause cardiac standstill. Such data would be very important for three reasons. Firstly, it would enable treating physicians to gain an understanding of how close their patient may be to cardiac arrest, thereby quantifying the urgency of the situation. Secondly, for the patient who has suffered a hypoxic cardiac arrest, it would give an indication of the potential damage which has already occurred to vital organs, specifically the brain. And thirdly, it would give more understanding for the patient who is in the process of donating their organs following circulatory arrest, to have more knowledge of the likely warm ischaemic time potential organs have undergone. Due to the obvious ethical problems this is a difficult area to research though some preliminary work has commenced.⁹

Many cardiac arrests are unexpected events that occur without monitoring in the preceding minutes and hours, meaning that robust data collection is usually impossible. However, all patients undergoing controlled DCD, as practiced in the UK, have observations taken at frequent intervals from the moment of withdrawal of life-sustaining treatments (WLSTs) until circulatory arrest. These data are recorded and stored on the UK National Health Service Blood and Transplant database. This data would allow us to analyse the sequence of changes leading to a hypoxic cardiac arrest.

Method

We reviewed all DCD donations over the calendar years of 2014 and 2015 in one region in the UK. These intensive care patients have both a family agreement that the WLST is appropriate and a consent for organ donation, which follows once circulatory arrest has occurred. To ensure a timely transfer to the operating room for organ recovery, full intensive care monitoring is continued until death is confirmed by a doctor.

Consent was obtained from the NHS Blood and transplant service, UK, to access the already collected data.

The notes and observations of all potential DCD patients were reviewed. Those with incomplete observations, or who did not die, were not analysed.

Following this, observations from the point of withdrawal of life-sustaining treatment (WLST) to circulatory arrest were recorded. Observations recorded were heart rate (HR) using three cardiac chest leads, percutaneous oxygen saturations, systolic blood pressure (SBP), diastolic blood pressure (DBP) and mean blood pressure (MBP) using an invasive arterial line. Every observation recorded by the attending specialist nurse for organ donation was recorded. The time of circulatory cessation was taken as the loss of cardiac output (not the loss of electrical activity) in keeping with the Academy of Medical Royal Colleges Code of Practice for the Diagnosis and Confirmation of Death, which requires a minimum of 5 min observed cessation of the circulation before death can be diagnosed following cardio-respiratory arrest.¹⁰

Data were analysed using Microsoft Excel and ‘R’. The time of loss of cardiac output was taken as time zero, or the moment of circulatory arrest. Data were then analysed at time points prior to the moment of circulatory arrest. Loewe’s linear regression curve was produced for all of the observations and Kaplan–Meier plots were constructed for different oxygen saturation points.

Results

Complete observation data were recorded from 82 patients (Figure 1). The age of the patients varied from 4 to 76, with a median age of 52 and mean age of 50. The majority (49, 60%) of patients were male.

Figure 1.

CONSORT flow diagram.

Time from WLST to circulatory arrest varied from 4 to 185 min, with a mean time of 28.5.

Figure 2 shows the mean, and 95% confidence range, of observations at times prior to circulatory arrest. This shows a terminal deterioration in saturations (from an already low baseline) commencing 14 min prior to circulatory arrest, followed by a blood pressure fall commencing 8 min prior to circulatory arrest, and finally a rapid fall in heart rate commencing 4 min prior to circulatory arrest. The terminal deterioration in saturations may represent reflex peripheral vasoconstriction as relative tissue hypotension begins,¹¹ or the exhaustion of the pulmonary-circulatory oxygen reservoirs.

Figure 2.

Changes in SBP, MBP, DBP, SpO₂, HR in 20 min preceding circulatory arrest. HR: heart rate; SpO₂: oxygen saturations; SBP: systolic blood pressure; DBP: diastolic blood pressure; MBP: mean blood pressure.

The timing of circulatory arrest as predicted by a patient’s saturations is shown in Figure 3. Each Kaplan-Meier plot shows the probability of circulatory arrest at a time elapsed once the patient’s saturations has crossed a threshold of 90%, 80%, 70%, and 60%. In general, the survival plots are reasonably close together, implying that once the saturations have started to drop, then circulatory arrest is imminent. The probability of circulatory arrest at different saturation points is shown in Table 1.

Figure 3.

Kaplan–Meier plots showing probability of survival with increasing time once the saturations have crossed a threshold of 90%, 80%, 70%, and 60.

Table 1.

Time to cardiac standstill at various probabilities.

Time until cardiac standstill once a saturation threshold has been passed.
	Saturation percentage threshold
	90%	80%	70%	60%
50% Probability of circulatory arrest	12 min (11–15)	11 min (7–14)	10.5 min (8–13)	9 min (7–12)
75% Probability circulatory arrest	22 min (17–31)	19.5 min (14–26)	16 min (14–25)	16 min (13–22)
95% Probability circulatory arrest	56 min (37–153)	52 min (53–153)	45 min (32–123)	34 min (28–117)

Patients who had inotropic support withdrawn at the time of WLST had a higher probability of circulatory arrest at any given time point when compared to those who did not/were not on inotropes. This makes physiological sense as these patients will have had a mixture of hypoxic and primary cardiac arrest due to a reduction in contractility.

Two patients had a warm ischaemic time (defined as saturations below 70% and SBP (in adults) below 50 mmHg) of greater than 30 min, and three greater than 20 min; 15 patients had a warm ischaemia of 10 min or greater. In all, 53 patients had a warm ischaemia time of 5 min or less.

Figure 4 shows the proportion of patients with oxygen saturations of <40% for the given time durations prior to circulatory arrest. The 40% saturation threshold was chosen as some experimental data shows that brain lactate increases below saturations of 40%, implying critical hypoxia. These timings are of the first documented measurement under 40%, so the true proportions will be higher than stated. A total of 85% of patients had saturations of less than 40% for more than 3 min prior to circulatory arrest (Figure 5).

Figure 4.

Kaplan–Meier plot showing probability of survival. Blue line shows patients who did not have inotropic support stopped and red line those that had inotropes stopped.

Figure 5.

Proportion of patients with hypoxia <40% at given time points.

Discussion

Several prediction models have been developed for early circulatory arrest after treatment withdrawal;^12–15 however, the high level of variability in patient groups and healthcare environments may limit their general applicability. We believe our data builds on this work and helps to explore the physiological sequencing of circulatory arrest in humans. The sequence of saturation fall, blood pressure fall, and then a terminal bradycardia demonstrates the sequence of premortem physiological collapse.

It is ethically challenging to investigate the process of death. This retrospective study has weaknesses in that our patients were undergoing withdrawal of life-sustaining treatment, were predominately brain-injured, had been accepted for DCD, and were not standardised or randomised. The data were not collected for the purpose of this study. There were many different pathologies, which may have different effects on the body. We were also not able to differentiate patients who showed some evidence of respiratory effort after WLST and those who did not.

We are also aware that pulse oximeters can only be as accurate as their empirical calibration curves,¹⁶ therefore the accuracy of pulse oximetry at low oxygen saturations (SpO₂ below 75%) is unknown and is based on extrapolation from data at higher oxygen saturations. Studies continue to show significant bias, increasing as oxygen saturation decreases,¹⁷ which is likely to be associated both with a fall in pulsatile flow, peripheral vasoconstriction, and reduced arterial oxygen levels.^18,19

The implications of our data are manifold. Our data creates a strong inference that good oxygenation can prevent cardiovascular collapse or cardiac standstill: likewise maintaining blood pressure can help defer cardiac standstill in the face of hypoxia. The earlier in the sequence of collapse that an intervention can be made, the more likely it is that the sequence can be arrested. Clearly, we were unable to intervene in our patients (and it is inconceivable that a randomised trial of interventions in perimortem patients would ever be possible or ethical), so although extrapolating from our data is not completely scientifically robust, it is likely to be the closest we can get to reliable data. Our results also seem entirely biologically plausible. Resuscitation experts often say that the best resuscitations are those which never occur: our data reinforce the importance of maintenance of oxygenation, and later blood pressure, for the maintenance of life.

The heart is a muscle with a very high metabolic demand. Sufficient fuel, in the terms of oxygen delivery (DO₂), is necessary for continued function. Saturated blood is a key ingredient: once the saturations are below 35%, the blood pressure begins to decline. Sufficient blood pressure to maintain perfusion through the coronary arteries is required. Our data imply that once the MBP (for this predominately adult population) is below 50, a rapid decline towards cardiac arrest occurs.

There does not seem to be a clear calculation which will inform physicians of an amount of hypoxia which will cause cardiac standstill. This is likely to be hugely multifactorial, with the patient’s age, fitness, and pathology all playing important parts. However, in 50% of our patients, a saturation of less than 60% for 9 min led to circulatory arrest.

We were naturally unable to define the amount of hypoxia which will cause permanent brain damage. It is known that paediatric cardiac patients, who may live with hypoxia, usually around 75%, for many years, have a reduced IQ when compared to the general population. Some animal studies show that neuronal lactate increases once the brain saturations are below 42%, implying anaerobic metabolism and imminent cell death. Studies on experimental animals reveal that at PaO₂ 50 mmHg an increase in the [lactate]/[pyruvate] ratio and a decrease in brain tissue pH are observed.²⁰

Normal human brain consumes 3.3 mL of oxygen per 100 g of brain per minute, which represents ∼20% of total body resting oxygen consumption despite the fact that the brain represents only ∼2% of body weight. Anaerobic glycolysis cannot sustain the energy requirements of adult brain for more than a few minutes.²¹

Overall, current medical knowledge points to a high probability of severe and permanent brain damage with the level and duration of hypoxia we have seen necessary to cause a hypoxic cardiac arrest. This is very different to cases of primary cardiac arrest. In such cases, where the brain is well perfused when the cardiac arrest occurs, instant and effective cardio-pulmonary resuscitation can maintain adequate perfusion to the brain for a long time, whilst a reversible cause is tackled.

Outcomes from hypoxic cardiac arrest are almost universally extremely poor.²² Cases of excellent post-cardiac arrest outcomes are almost universally in patients with a primary cardiac arrhythmia, who are given excellent and immediate bystander CPR.²³ Our data underline the reasons for poor hypoxic cardiac arrest outcomes, and should inform resuscitation guidelines where a distinction between cardiac arrest, and hypoxic cardiac arrest should be made. In general, it is likely that the overwhelming majority of hypoxic cardiac arrest patients have already suffered very severe hypoxic brain damage by the time of cardiac standstill. This should inform decisions on how long it is appropriate to continue resuscitation, and to consider the escalation of care and duration of intensive care in these cases.

Given that organ DCD is likely to be the most effective way to increase organ donation numbers,²⁴ these data provide important information on the generally short warm ischaemic time. In our patients, only two had a warm ischaemic time above the threshold of 30 min and only 1 more above 20 min. That 53/82 patients had a warm ischaemic time of less than 5 min shows that most organs available for transplant with donation after circulatory arrest have a minimal time where organ damage is occurring. This underlines the generally excellent outcomes of organs donated after circulatory death.

Conclusion

Prior to hypoxic cardiac arrest, we have demonstrated that there appears to be step-wise and concurrent terminal physiological deterioration lasting a total of 14 min, comprising: 6 min of severe hypoxia, then 4 min of hypotension with hypoxia, and a final terminal bradycardic, hypotensive, and hypoxic phase lasting 4 min. For our predominately adult population, blood pressure seems to fall after the saturations go below 35%, and the terminal heart rate decline begins once the mean blood pressure falls below 50 mmHg.

Significant hypoxia is required to cause cardiac standstill. It is likely that by the time of cardiac arrest caused by hypoxia, irreversible severe brain damage has already occurred. This should inform decisions on resuscitation. There is likely to be a window of intervention where the terminal decline can be arrested if suitable interventions to maintain oxygen saturations and blood pressure are implemented when the situation is recognised. In general, the warm ischaemic times in these patients prior to circulatory arrest is short.

We aim to continue data collection, ideally with prospective research, to further differentiate between patient groups and pathologies.

Ethics committee approval

Approval was gained from NHS Blood and Transplant, UK to allow access to the data.

Authors' contributions

Colin Gilhooley – data collection and analysis. Written manuscript. Geoff Burnhill – data analysis. Dale Gardiner – review of manuscript and advice of reviewing literature. Harish Vyas – review of manuscript and advice of reviewing literature. Patrick Davies – idea conception and review of manuscript.

Declaration of conflicting interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dale Gardiner is Deputy National Clinical Lead for Organ Donation, NHS Blood and Transplant, UK.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.