Pitfalls and challenges in the diagnosis of acid–base disorders by blood gas analysis
Blood gas analysis for the diagnosis of acid–base disorders is now mainly performed at the point-of-care and the results reported can be affected by a number of factors. Users should be aware of the possible pitfalls to prevent errors in diagnosis and to optimize patient management. CLI chatted to Dr Kate Shipman (Consultant Chemical Pathologist, University Hospitals Sussex NHS Foundation Trust, UK) and Dr Henry Carlton (Anaesthetic CT1 Doctor, University Hospitals Sussex NHS Foundation Trust, UK), who have recently published a narrative review on the subject, to find out more about the factors that can affect blood gas data.
What are acid–base disorders and why is blood gas analysis useful for their diagnosis and management?
Henry Carlton (HC) Acid–base disorders are the pathological absence of homeostatic pH. These disorders are mediated either through a respiratory or a metabolic etiology. They can be acute or chronic, vary in severity and they are often not sufficiently counteracted by the body’s own natural buffer systems.
Blood gas analysis is fundamental to the diagnosis and management of acid–base disorders. Now, because everywhere has a blood gas analysis instrument, it can be used at point-of-care (POC), only requires a very small amount of sample, and a multitude of different sample types can be used to obtain a result, most patients don’t come through A&E (Accident and Emergency Department) without at least a venous blood gas analysis having been done.
Kate Shipman (KS) pH is so unstable – it’s not a specimen that lasts well – so for anyone doing this analysis in a lab, the pH in the sample will have changed from when the sample was taken, so immediate access to analysis is necessary, hence the use of POC testing.
How is blood gas analysis normally done?
HC As mentioned, blood gas analysis is usually done at POC, so it’s an analysis that happens in the clinical area. Ideally, the analysis should use an arterial sample, so that’s usually a radial artery blood draw, preferably with some lidocaine as it can be slightly more painful than other types of blood draw. We are seeing a shift towards greater use of peripheral venous samples a lot more – in A&E if they pop a canula in and send off blood, that’s part of their routine protocol. However, capillary blood samples can be used, which is more often done in pediatric care where fingertip pinprick tests are used to get a result. Intraosseous (IO) blood samples can also be used, as when you are inserting an IO device you can get around 5 to 10 mL of blood on the first aspirate before you actually use the device. Additionally, central venous samples can be used, whether that’s upper limb or lower limb. The different samples vary slightly in their interpretation.
KS If you just want a pH measurement a venous blood sample would probably be sufficient, but if you want data on the patient’s oxygenation and ventilation then arterial blood is the best specimen. The method is an ion specific electrode, which provides a direct measurement. The POC device is usually set up for analysing arterial samples, so if you wanted to use venous or capillary specimens then the settings on the instrument would have to be changed.
HC Also, sometimes when you collect your sample, if you are not using ultrasound guidance or an established line it can be quite difficult to know what the sample actually is. For example, there are veins around the radial artery and if the patient is really quite unwell it can be difficult to match the pulse oximetry to the CO-oximetry [two different ways of assessing hemoglobin oxygen saturation, particularly important in patients with possible carbon monoxide (CO) poisoning].
Why is it important to understand the pitfalls in the interpretation of blood gas analysis?
HC Blood gas analysis is becoming a routine part of the diagnosis and management of acid–base disorders and many factors can affect how the results should be interpreted. These pitfalls need to be understood in order to provide appropriate patient care. The challenges can be broadly categorized as pre-analytical, analytical and post-analytical.
What are the common pitfalls in the pre-analytical stage?
HC The factors that affect this stage can themselves be broken down into three groups: issues with the sample type; patient preparation; and sampling method.
We have mentioned some of the issues with sample type above. Obviously, the blood circulates from lungs to heart and then round the body, so the arterial blood draw would be the most proximal and would give the best respiratory gas quotients. As you go more distal round the circulation you get more usage of the oxygen and more carbon dioxide build-up so that’s something to be aware of.
With regards to patient preparation, it’s important to make sure that the demographic details, such as the patient’s temperature, supplemental oxygen and their ventilation status are noted.
For the sampling method, there is a myth of pain provoking hyperventilation that will affect the results but we really didn’t find any evidence of that. There is also discussion about the historic use of glass syringes and the diffusion of respiratory gas in the blood going through the glass.
KS Also, the liquid anticoagulant can also cause problems but generally now people use dry bits of card with the heparin on, which is lower risk for interference. There are also factors that you can’t control, such as aberrant hemoglobin. However, there are things you can control, such as ensuring that samples are not taken directly after the use of intravenous contrast agent, for example. It is worth bearing in mind that there may be some factors that you will be unaware of until you realise your results don’t fit with what you expected.
It’s important to note that as the analysis is done at the POC, you don’t have analytical validation being performed by a trained laboratorian. If the person using the device hasn’t appreciated that there is a problem, then they may interpret the number as physiologically relevant, when it’s not.
HC And as we move towards electronic databases of all the results you can have distant clinicians looking at the results without any appreciation of an incorrect context.
KS Such as what you did to get that result, or what the patient actually looked like. It’s really important to know who your patient is and what your question is, otherwise you can become misled by the results.
What are the considerations in the analytical stage?
HC One factor to be aware of is the interference of pathology. For example, lipemia, icterus and hemolysis can affect blood results. However, the effects weren’t well described in the literature, the main message being that severe hemolysis can affect the respiratory quotients, so hemolysis should be reported on the blood gas results.
More interestingly, we found something called ‘leukocyte larceny’, where essentially a high concentration of blast cells – which are really metabolically active cells – actually use some of the oxygen and produce carbon dioxide and so distort the blood gas result in a time-dependent manner.
Another factor to be aware of is albumin, which is an acidic protein. Albumin concentrations can fall rapidly in critical illness, such as sepsis, which therefore affect the pH measurement and increase base excess.
KS The anion gap – as long as you have the right question – can be very informative about what is present. It is also useful in the setting of poisoning.
HC For example, in the analytical phase drug/toxin interference can cause anomalous blood gas readings. For example, salicylate overdose can cause spurious hyperchloremia (and therefore acidosis).
KS However, salicylate overdose can also cause alkalosis through respiratory stimulation. If you haven’t measured the salicylate, then the anion gap will be erroneously ‘normal’ or ‘negative’. In these cases, you might miss the poisoning as the anion gap has been affected by the poison (due to the hyperchloraemia masking the presence of the unmeasured anion).
HC This is actually more interesting in insidious salicylate overdoses – for example, preparations such as Pepto-Bismol and oil of wintergreen contain salicylate and if you didn’t know that patient had taken it you wouldn’t expect the negative anion gap.
KS This is the opposite of ethylene glycol poisoning, where you can use the lactate gap. If the lab uses a lactate dehydrogenase assay (in comparison to lactate oxidase which is the method used in the gas machine), the lactate gap between the gas results and the lab analysis tells you that you have ethylene glycol present and can be useful for indicating the severity of the overdose. This is very useful, particularly with unconscious patients, because an accurate ethylene glycol assay has to be done by specialist toxicology labs by headspace GC-MS and is not easy to get quickly out of hours. Additionally, these days, most antifreeze preparations contain fluorescein, so you can check for fluorescence in the patient’s urine, which is useful in pediatric patients, but not 100% reliable. It is only useful to identify a poison if there is an antidote – which there is for ethylene glycol – otherwise treatment is just supportive.
What are the post-analytical pitfalls?
HC The challenges of blood gas interpretation that occur at the post-analytical stage involve the values that the analyser calculates. For example, data is adjusted for temperature, but there are no normal reference ranges for hypo or hyperthermic patients – the data is all adjusted to a normal temperature based on the temperature that you tell the device for the patient. There are two methods that are typically used by a device to adjust for temperature: Alpha-stat and pH-stat. The Alpha-stat method can overestimate respiratory carbon dioxide (if the patient is hypothermic).
KS Sometimes people think that the numbers reported are what is ‘in’ the person and don’t realize that they have been adjusted by the device, so then they make an allowance in their head (that doesn’t need to be made) and so errors happen that way.
HC One situation where this is important is in communicating with cardiac anesthetists who cool their patients in certain circumstances.
KS It is important for users if they have a very hypothermic or hyperthermic patient that the results would be calculated and even to understand exactly which of the reported numbers are measured and which are calculated.
Blood gas analysis is a really powerful technique for adjusting ventilator settings for a patient who is at a relatively normal temperature, because you are looking at ‘big changes’ in respiratory failure. The challenges arise when the changes are subtle or the patient is, for example, a child with unusual congenital artery abnormalities, and if they arrive at the ITU (Intensive Treatment Unit) in sepsis and their blood gases were already abnormal, that’s when clinicians can struggle for example.
Regarding robustness, the blood gas analyser does do direct methods of analysis. For example, one problem we have in the lab is with pseudohyponatremia and pseudohypernatremia due to the solid phase in the specimens – the amount of fat or protein. If the patient is very lipemic, or they have myeloma, the sample will have a lot of fat or protein in it. This will significantly affect the sodium report, as we are using indirect electrodes then back-
calculating according to an assumed quantity of water that is not valid for that sample. The blood gas analyser is a good way of getting a much more robust direct measurement in the aqueous phase of the specimen, so its methodology is really good for those situations. If people see discrepancies between the POC blood gas analyser and the lab report, they may assume that the lab data is more robust, whereas actually in these circumstances the blood gas data is more accurate. In the UK, where measuring ionized calcium is uncommon, the blood gas analyser is also a way for users to access ionized calcium analysis, which is half the concen-
tration of the lab total calcium concentration, which confuses people also.
HC The reference ranges that we typically see and use were calculated for healthy individuals, so there are no reference ranges for pregnant or neonatal patients. Also, the ranges don’t account for age and we know that there are lots of differences in those demographics, so if you are using the same set of reference ranges you can end up making very misjudged interpretations.
KS There are some case reports of acid–base disorders being missed in pregnant women as the reported data were within the usual reference ranges, but the ranges were incorrect for pregnant women. Pregnancy affects the mother’s breathing as well as all sorts of other things. Also, reference ranges for neonates don’t really exist as it’s quite hard to get the samples for it, although the CALIPER (Canadian Laboratory Initiative on Pediatric Reference Intervals) Project in Canada is trying to do exactly that – to improve the diagnosis and monitoring of children and adolescents by establishing a comprehensive database of reference intervals for blood test results in a ‘well’ cohort.
If you could highlight just one point for users of blood gas analysis to be aware of, what would it be?
HC Mine might be quite basic, but I think never appreciated the distinction between the full blood test requests that get sent away have full quality control with a lab behind it and a biochemist like Kate, who actually oversees everything, whereas on the floor with the blood gas analysis POC device you are often a solo interpreter. Also, the device is sometimes a bit between ownership so is it being maintained by A&E or the lab? Blood gas data interpretation is deeper than you might think, so it’s good to be aware of your own limitations and to ask for senior help if you think something doesn’t seem quite right or the results are confusing.
KS I would say that for clinicians, you may be on your own and you are probably looking after very ill patients, you need to know who your patient is, what your question is, and how the test is going to help you. Because then if you understand the data, know what you are expecting, and how the results will change your management of the patient, then a blood gas analyser is a very powerful instrument. You have to be careful to input all the demographics of the patient and be aware of the limitations and what might go wrong. If you know what you are doing then you have a very robust device giving immediate results, which is very powerful.
Are there any improvements to blood gas analysis on the horizon?
HC Critically ill patients in ITU or HDU (High-Dependency Unit) tend to have blood samples analysed hourly, which along with the two sets of formal blood samples per day, can actually cause iatrogenic anemia. I have seen some literature about real-time blood gas analysis – a probe that sits in an artery that gives a constant readout of pH, pO2, pCO2. This would be a real benefit when we are adjusting ventilator settings – to immediately see what is helping and what is not.
KS Artificial intelligence (AI) might be helpful when a less experienced doctor is on the ward in the middle of the night to help with the data interpretation, but the crucial factor is to have the right demographic information there in the first place. In fact, the development of the AI might drive people to ensure that all the data is properly entered, as otherwise the AI won’t have the correct information to train on.
The interviewees
Dr Kate E. Shipman BMBCh MA
(Hons Oxon) FRCP FRCPATH EuSpLM, Consultant Chemical Pathologist
St Richard’s Hospital, Chichester,
West Sussex PO19 6SE, UK
Email: kate.shipman@nhs.net
Dr Henry C. Carlton MBChB BSc
(with Hons), Anaesthetic CT1 Doctor
St Richard’s Hospital, Chichester,
West Sussex PO19 6SE, UK
Email: henry.carlton@nhs.net
For further information see:
Carlton H, Shipman KE. Pitfalls in the diagnosis and management of acid–base disorders in humans: a laboratory medicine perspective. J Clin Pathol 2024:jcp-2024-209423 (https://jcp.bmj.com/content/early/2024/07/18/jcp-2024-209423.long).
