Errors in clinical measurement: What and why?

Errors in clinical measurement: What and why?

Mohamed Mahmoud and Oliver Pratt highlight the most common clinical measurement errors encountered in daily practice and advise as to how they may be avoided

 

Abstract
The recording of accurate clinical data and physiological parameters is a key aspect of modern anaesthetic and critical care practice. It is a tool which help clinicians in diseases diagnosis, identification of abnormalities, and to guide clinical interventions. Such a tool, if used appropriately, will prevent any potential patient harm or unexpected event. Errors in clinical measurement can lead to either incorrect or missed diagnoses or even failed interventions. A clinician must therefore be aware of the most common errors which affect the accuracy of recording clinical data. Such error should be prevented and the source should be identified and corrected. Errors can be multifactorial in their origin; causes are generally classified as device, operator, or patient related. This article will highlight the most common errors encountered in our daily practice and advise as to how they may be avoided.

 

Introduction
Measurements of clinical parameters are used to assist medical professionals to identify normal patterns as well as any abnormalities or deviation from the standard. The reliance on such measurements has grown rapidly in the last few years. With the aid of clinical measurement, clinicians can be more certain that their diagnoses, based on objective data, are more likely to be correct.
In 2015, the Association of Anaesthetists of Great Britain & Ireland (AAGBI) published its most recent minimum monitoring standards for patients undergoing general anaesthesia. These minimum monitoring standards include use of a pulse oximeter, non-invasive blood pressure monitor, Electrocardiograph (ECG) and measurement of inspired and expired oxygen, carbon dioxide, nitrous oxide and volatile anaesthetic concentrations (if used). Ventilator parameters such as airway pressures, minute volume and respiratory rate should be recorded. A peripheral nerve stimulator should also be used if neuromuscular blocking drugs are administered. According to the GMC, failure to adhere to such guidance is considered malpractice. National Patient Safety Agency has published many alerts in relation to inappropriate use of monitoring system. For example, in 2009 NPSA highlighted that patients may be harmed if a wrong infusion containing dextrose is attached to keep the arterial line open. This might cause a false high blood sugar reading which, if treated, can cause sever hypoglycaemia and even cardiac arrest.

 

Components of a typical measurement system
The basic function of any monitoring system is to collect data from a patient or subject and process it to produce a meaningful and reproducible display. Figure 1 shows a schematic representation of a basic clinical measurement system. The first stage involves collection of specific signals from the patient – for example in an ECG this would be electrical impulses caused by cardiac action potentials. The
collected signals are amplified, filtered, and processed by a microcomputer in order to produce a readable display.

Clinical measurement systems must produce data that is both accurate, and precise. These concepts are defined as follows:
Accuracy is defined as the degree of correctness of the measurement when compared to the true or an absolute value. Calibration is an essential step to ensure an accurate reading. It is used to test and optimally adjust measuring instruments. Most devices are calibrated during the manufacturing process, however certain devices require additional calibration before or during usage. One-point calibration is required for a liner relationship. This involves measuring a system output against a known real value, e.g. zeroing of the invasive arterial blood pressure system. Multiple-point calibration involves using three or more know values or concentration, e.g. blood gas machine calibration using different pH solutions.
Precision relates to the reproducibility of the repeated measurement – a monitor which is precise, will give the same reading repeatedly when measuring the same signal. Precision therefore describes the “scatteredness” of data recordings.

We can illustrate the concepts of accuracy and precision by considering a single pin prick test to assess blood Haemoglobin concentration. The test will be will be accurate if the measured readings reflect the true haemoglobin level in blood. The test will be precise if it gives the same result each time a measurement is obtained.

 

Input-output relationship
Monitoring systems produce an output signal, in response to an input. In an ideal monitor, the magnitude of the output signal is related to the magnitude of the input in a constant, linear fashion.
Filters are incorporated into measurement systems to prevent artefacts or unnecessary interference of any other signal. Type of filter includes high-frequency and low-frequency filters which allow only the passage of the high and low frequencies respectively. A band pass filter blocks the signals above and below a certain bandwidth allowing only certain frequencies to pass.
Amplifiers used to process the biological signals should have a high signal-to-noise ratio and a high common mode rejection ratio. Both characteristics ensure the preferential amplification of signals being measured in comparison to any noises. The degree of amplification is known as the Gain. Gain stability will ensure an accurate reading when dealing with repeated or continuous measurements. Amplifiers contain semiconductors, with variant conductivity, which if heated will cause a drift and an inaccurate output.
Drift is a form of error which is expressed as an inaccurate output value which does not reflect the input change. The relationship between the input and the output is judged by linearity. A liner system uses the same amplification factor of an input to gain an accurate output. If a different amplification, values must be applied to obtain an accurate out, the relationship will be expressed as a non-liner.
Hysteresis is a form on non-linearity and is caused by losing the energy during the measurement process. The lost energy is spent to overcome either the measurement resistance such as temperature measurement. Allowing a degree of variation in the amplification is an essential requirement during the manufacturing of monitors which express hysteresis. In such case, the amplifiers will respond differently
to a decreasing or an increasing input.

 

Measurement errors
Measurement errors are classified into systematic or random error. Table 1 illustrates the main differences between both types. Random errors are caused by an external unexpected noise which affect the measuring system. A movement artefact or using the diathermy will affect pulse oximetry reading. To correct this error, averages of repeated measurements are taken to minimise the artefact effect.

Screen Shot 2017-08-02 at 10.16.47
A more clinically relevant classification of measurement errors divides those into three main categories: Firstly, device-related errors which are attributed to faulty device or inappropriate calibration. Using a faulty transducer for an estimation of the invasive arterial blood pressure will lead to an inaccurate measured value which will be either positive or negative in relation to the true value.
The operator factor is the second cause of errors. Proper understanding of physical principle of each monitor as well as anatomical consideration is essential for obtaining an accurate measurement. A wrong attachment of the pulse oximeter will lead to an abnormal low result or failure to obtain a measurement.
Patient-related factors are another source of measurement errors. Certain pathology would render measurement difficult or even impossible. Patient shivering during the ECG recording will distort the ECG details.

 

Pulse oximetry
Devices are unreliable at low saturation as the displayed values are extrapolated rather than measured. During the manufacturing stage, it is impossible to truly calibrate the transducer against a low saturation as it would be unethical to allow volunteer oxygen saturation to drop to a low level.
Patient-related errors relates to the peripheral circulation and the vascular tone. The increased pulsation at venous circulation (tricuspid regurgitation) will lead to an underestimation of the oxygen saturation. Similarly, using a high airway pressure during mechanical ventilation will create a pulsatile venous flow which will be misinterpreted by a monitor as an arterial pulsation and hence an inaccurate
lower reading will be displayed. Contrarily, detecting the pulse wave will be difficult in the presence of minimal blood pulsatile component. Vascular tone changes, hypotension, hypothermia and peripheral vasoconstriction will cause difficulty in eliciting oximetry reading.
The presence of abnormal haemoglobin will affect the reading accuracy. Different chemical combination of haemoglobin within the red blood cells will have a different absorption spectra of red and infrared light. Carboxyhaemoglobin is formed due to a
combination of carbon monoxide and haemoglobin. This combination compromise less than 2 per cent of total haemoglobin in normal individuals. Higher percentage of carboxyhaemoglobin, during carbon-monoxide poisoning, will lead to a false high oximetry reading. If, for example, a patient has a carbon monoxide saturation of 20 per cent and an oxygen saturation of 75 per cent, the displayed saturation on the monitor will be 95 per cent, which is the sum of both. Monitors recognise carboxyhaemoglobin in a similar way to an oxyhaemoglobin as both have the
same wave length of absorption at 660 nm wave length (red).
Contrarily, methaemoglobin will cause a false low reading of oxygen saturation as it does not combine with an oxygen molecule. Error is precipitated by excess nitrates and local anaesthetic toxicity. Bilirubin and skin pigmentation have minimal effect on pulse oximetry reading, while dark nail polish will affect pulse oximetry reading.
The pulse oximetry reading does not reflect a real-time reading but represents an average reading over 10-20 seconds. Monitor algorithm use such methodology to avoid the effect of movement artefacts. Movement, vibration (patient-related) and electromagnetic interference will affect the reading and distort the wave morphology.

 

Capnography
End-tidal carbon dioxide (EtCO2) measurement is used to convey information about respiratory and cardiac systems. In comparison to the pulse oximetry, EtCO 2 value represents a real-time change in the alveolar gases. Noticeably, when using side-stream analyser, there will be a slight time delay before the sample being displayed on monitor. This delay will depend upon sampling tube length and sampling rate.
Patient-related errors represent the majority of EtCO2 inaccurate reading. Genuinely, measured values of the EtCO2 , which are diluted by physiological dead space, will be much less than the alveolar CO2. In healthy adults, the EtCO2 value is about 0.5 kPa less than arterial partial pressure of CO2. Any factors which aggravate dead space such as hypotension, or placing the ECO2 sensor in distal proximity
to the alveoli will result in a lower EtCO2 value. In the presence of hypotension, displayed reading does not represent arterial partial pressure CO2. Importantly, a sudden drop of EtCO2 value might indicate a pulmonary embolism which will convert the affected alveolar space into a dead space.
Operator-related error will arise if specific ventilatory settings, which allow a low tidal volume and a high respiratory rate, have been used. This will cause an inaccurate low EtCO2 value. Similarly, some patient factors will cause the same error. Airway obstruction will lead to a slow rise of EtCO2 slope curve. In both situations, the value of EtCO2 will not represent the actual EtCO2. Common operator errors involve a disconnection error, either in breathing circuit or sampling line, which will be identified as an absent waveform. Other causes include kinking or blockage of the sampling line or loose connection and system leak, which will lead to dilution of the CO2 sample and hence a lower EtCO 2 reading.
Device-related factors are less common in modern devices. Errors might arise upon using nitrous oxide which has an infrared light absorption spectrum overlapping that of CO2. This could lead to a falsely high reading. A faulty sensor or blocked sample line are not uncommon source of inaccuracy.

 

Electrocardiogram (ECG)
ECG is used to monitor the electrical activity of the heart. Signal obtained is manipulated by different factors. Device-related errors include improper skin electrode and cable placement which should have the same length to minimise the effect of electromagnetic interference. Ideal positions should be over a bony prominence rather than over soft tissue.
Noises either from the body itself (muscles) or from an external source (diathermy apparatus) will disturb the ECG signal quality. ECG-filters are incorporated into modern ECG devices to remove any noise. High-frequency filters eliminate interference caused by main current or muscle movement while low-frequency filters are used to reduce the effect of respiration artefacts. Differential amplifiers are
also used as a method to avoid interference from external sources. Capacitive or inductive interference will cause ECG wave distortion.
Intraoperative use of diathermy will cause a distortion of ECG wave especially if the patient diathermy plate was inadequately positioned.

 

Non-Invasive Blood pressure (NIBP)
Ideal cuff inflation devices should inflate rapidly and deflate slowly. Any rapidly deflated device could lead to a difficulty in detecting arterial pulsation. The cuff size will affect the blood pressure measurement. Ideal cuff width should be 80 per cent of the midarm circumference and its length should cover two-thirds of the arm.
Patient-related errors include obesity, arrhythmias especially atrial fibrillation and any systolic blood pressure less than 60mmHg.
Operated-related errors will lead to an overestimation of the blood pressure if a narrow cuff is used. A wider cuff will cause a falsely low blood pressure. Any external pressure on cuff or conducting tube will also affect the reading accuracy.

 

Invasive blood pressure measurement
Positioning of the transducers is the most common cause of operator errors. A transducer should be ideally placed at the level of right atrium (at the level of midaxillary line). For each 1cm change in the transducer position, the arterial pressure will change by approximately 0.75mmHg.
Calibration and zeroing eliminates device errors and ensure the absence of any drift from the base line by subtracting the effect of atmospheric pressure on a measuring system. Resonance and damping affect the accuracy of measured value. Resonance will occur if the resonant frequency lies within the blood pressure wave form frequencies. Generally, any non-stiff long saline-filled catheters increase the resonance, while using a wider and stiff catheter decrease it.
Damping is associated with any restriction to the transmission of an arterial blood wave from the artery up to the sensing diaphragm which will cause an underestimation of the measured value. Common causes of damping are either a clot in the arterial line (patient-related) or an air bubble in catheter (device-related). Both damping and resonance will not reflect true systolic and diastolic values, however mean blood pressure values may still be accurate.

 

Blood gas analysis
Errors in ABG will lead to inappropriate diagnosis. Common error includes:
Operator Errors

  • Air bubbles in sample: affect Oxygen and carbon dioxide tension
  • Excess heparin: leads to inaccurate low pH value.
  • Delay in processing the sample: associated metabolism leads to a lower PO2, pH, sodium but higher potassium and PCO2 values. It is advisable to store blood gas sample at low temperature if the immediate analysis of
    blood gases is not available. This will ensure minimising the metabolic effect

Patient errors

  • Temperature of blood gas sample will affect the PO2, and PaCO2 results.

Device errors

  • Faulty electrode
  • Un-calibrated device

 

Cardiac output monitoring

  • Pulmonary artery (PA) catheters: If the PA catheter tip is inappropriately positioned, an inaccurate measurement will be displayed. PA catheter requires frequent calibration with a known volume which should be injected quickly. Any changes in the volume of injectate or the speed of injection might affect the calibration. Three cardiac out readings are essential for obtaining
    an acceptable calibration value. The recorded reading will be in-accurate in the presence of any intracardiac shunting.
  • Oesophageal doppler: Inappropriate probe position causes inaccuracy. Repeated adjustment of probe position, to be in close proximity to descending aorta, is essential to ensure an accurate reading. Accurate measurement relays on normal flow in the descending aortaic so any pathophysiological abnormalities of descending aorta will lead to inaccurate reading. The used algorithm in such devices to estimate the aortic cross-sectional area is a source of inaccuracies.
  • LiDCO rapid: Reading will not be accurate in the presence of aortic valve regurgitation or patient with pronounced peripheral arterial vasoconstriction.

 

Nerve stimulators
Wrong placement of neuromuscular monitoring electrodes might result in direct muscle stimulation and contraction. Most commonly used method is ulnar nerve stimulation to observe Adductor Pollicis contraction. Usually current needed for transcutaneous stimulation is about 60mA while the electric current used for nerve identification during regional blocks is much less (1-2 mA). Successful block is likely
to obtained if a response is still elicited when the current is ≤0.3mA. Accuracy of peripheral nerve stimulators decreases with time. Such deterioration will like to cause less current delivery which will increase the risk of nerve damage. On the other hand, a failed block is likely to happen if a device is inappropriately delivering a higher current.

 

Bispectral Index (BIS)
NICE has recommended the use of EEG-based depth of anaesthesia monitors in patients receiving total intravenous anaesthesia. A range of BIS value, during general anaesthesia, between 40–60 will indicates a low probability of awareness with recall. Values above 60 would be of a less significant clinical value. Inappropriate position of the 4-electrodes and even muscle relaxants can lead to an inaccurate result.

 

Temperature Monitoring
Infrared thermometers are not the gold standard equipment for measuring the temperature. When using tympanic thermometer, the clinician should ensure there is no wax obstructing ear canal. A temperature probe should be ideally positioned in a close proximity to the tympanic membrane but not to the sides of ear canal.

 

Ultrasound
The use of the ultrasound has grown up significantly over the last few years. Poor understanding of the anatomy or inadequate ultrasound skills contribute to difficulties in image generation and interpretation. Errors associated with using the ultrasound doppler for measurement of blood velocity and prediction of pressure gradient occur if the wrong measuring point has been chosen. The operator should use the continuous wave doppler for a high velocity estimation and the pulsed wave doppler for a low velocity.
Device-related error will arise if wrong probe was chosen. Curvilinear and liner ultrasound transducer can be used for visualisation of deep and superficial structure respectively, while cardiac probe is used for transthoracic echocardiography.
Patient-related errors include inappropriate patient position which will cause inadequate or poor-quality images. Abnormal anatomy or obesity will cause observational error.

 

FURTHER READING
Smith & Aitkenhead’s. Textbook of Anaesthesia. 6th edn. Elsevier Ltd, 2013.
Davey A, Diba A. Ward’s anaesthetic equipment. 6th edn. Elsevier Ltd, 2012.
Al-Shaikh B, Stacey S. Essentials of anaesthetic equipment. 4th edn. Elsevier Ltd, 2013.
Ehrenwerth J, Eisenkraft JB, Berry JM. Anaesthesia equipment: principles and applications. 2nd edn. Elsevier Saunders, 2013.
Magee P, Tooley M. The physics, clinical measurement and equipment of Anaesthetics practice. 2 nd edn. Oxford University Press, 2011.

 

Keywords: Measurement error; Accuracy; precision; artefact; calibration; monitoring, measurement system, linear relationship
Royal College of Anaesthetists CPD matrix: 1A03
Learning Objectives:
By reading this article you should be able to:

  • Able to identify sources of error and correct it.
  • Describe mostly frequent errors in our daily practice.
  • Avoid common errors during for commonly performed clinical measurement.

 

Authors

Mohamed Mahmoud, MBChB, FRCA, MSc, MD, Specialist Registrar of Anaesthesia at North West School of Anaesthesia, UK.
Oliver Pratt, MBChB, FRCA, Consultant of Anaesthesia at Salford Royal Hospitals NHS Foundation Trust, UK.
Conflicts of interest: none declared.

Categories: ARTICLES

About Author