Estimation of blood haemoglobin concentration using the HemoCue® during caesarean section: the effect of sampling site☆

Introduction 

The HemoCue® (HemoCue® Ltd, Sheffield, UK) is a portable device that allows rapid assessment of haemoglobin concentration by the bedside or during surgery.1 Indeed, the value of such testing is highlighted as a specific recommendation in the latest report of the Confidential Enquiries into Maternal and Child Health.2 The HemoCue® is widely used in our maternity unit, both intra-operatively and in the diagnosis and management of ante- and postpartum haemorrhage, and we have found it useful in this setting as have others.3

The HemoCue® has been shown to be both convenient and accurate, although analysis of capillary blood (using a skin prick) may be less accurate than that of venous or arterial samples unless the recommended sampling method is strictly followed.4, 5 We are unaware of any studies comparing capillary sampling with formal venous sampling during obstetric neuraxial anaesthesia, although the accuracy of analysing venous samples with the HemoCue® per se during caesarean section has been confirmed.3

During neuraxial analgesia and anaesthesia, an obvious and convenient place for taking a capillary sample is the toe, since with an effective block the sampling will be painless. However, neuraxial blockade causes vasodilatation and increases blood flow in the legs, whilst reducing flow to the arms and upper body.6, 7, 8 In addition, even with lateral tilting the gravid uterus may cause vena caval obstruction,8, 9 which may increase venous pressure in the legs. Due to these concerns and the requirement of reaching under the sterile drapes to access the toes, some anaesthetists prefer to take samples from the fingers or thumb. The combined effect of the above circulatory changes on the haematocrit of capillary samples obtained from the toes and the fingers/thumb, and therefore on the accuracy of the HemoCue® when used with skin pricks at these sites during caesarean section, is unknown. We compared haemoglobin concentrations derived from capillary samples taken from the toe and thumb and venous samples, analysed with the HemoCue®, with those obtained from venous samples analysed in the laboratory in women undergoing caesarean section under spinal anaesthesia.

Methods 

After local research ethics committee approval, written consent was obtained from 50 healthy women undergoing elective caesarean section at >37 weeks of gestation under spinal or combined spinal-epidural (CSE) anaesthesia. Women with preeclampsia, multiple pregnancy or cardiovascular disease were excluded. Monitoring included standard electrocardiogram, pulse oximetry and non-invasive blood pressure. A 16-gauge cannula was placed in the arm or hand, and Hartmann’s solution at room temperature given both as a preload (400-600mL) and as intra-operative maintenance up to 2000mL in total, with Gelofusine given thereafter. A combined spinal-epidural or spinal anaesthetic was performed in the sitting position according to the preferences of the anaesthetist in charge of the case, using 2.0-2.5mL of 0.5% hyperbaric bupivacaine with diamorphine 300-400μg intrathecally. Women were then placed supine with left lateral tilt using a wedge. The blood pressure was maintained within 25% of baseline with boluses of either ephedrine 3-6mg or phenylephrine 25-50μg, and 200-mL boluses of i.v. fluid.

At the end of surgery each patient had three samples of blood taken: one venous and two capillary. The venous sample was from the antecubital fossa of the non-cannulated arm, and was sent in an EDTA tube for formal laboratory haemoglobin measurement. A droplet of this sample was also analysed for haemoglobin concentration using the HemoCue®. The other two samples were taken at the same time as the first using standardised (automated) skin prick sampling at (i) the side of the thumb of the non-cannulated arm, 5min after a blood pressure reading and (ii) the side of the ipsilateral big toe.

The person taking the capillary samples was the same for all patients and had received training from the manufacturers of the HemoCue® on the correct method of use of the instrument. The capillary samples were taken according to the manufacturer’s recommendations, which include choosing a suitable digit for sampling, making a skin puncture that will allow a ladybird size drop of blood to be expelled, drawing up the blood sample with the cuvette ensuring that the entire chamber is filled (not only the circular section) and wiping off the excess on absorbent material taking care not to siphon out the contents. This sample is then placed into the HemoCue® device. This is a portable device that measures haemoglobin concentration using modified azidemethaemoglobin reaction. The erythrocytes are lysed and the released haemoglobin is converted by sodium nitrite from the ferrous to the ferric state to form methaemoglobin, which then combines with azide to form azidmethaemoglobin. The latter is then measured by a photometer.

Other information recorded included details of the neuraxial block, total volume and type of i.v. fluids given, amount of vasoconstrictors required, estimated blood loss, duration of surgery and the patient’s temperature at the time of sampling.

The values of haemoglobin concentration were compared using Bland-Altman analysis.10 This statistical test compares two methods of clinical measurement by plotting the difference between the values obtained by each method against the “actual” value, estimated by calculating the mean of the two values. This analysis was done separately for laboratory vs. toe capillary, laboratory vs. thumb capillary, and laboratory vs. venous samples (Altman, personal communication regarding evaluation of three variables instead of the two originally described). A consistent difference between the two methods (mean difference) represents “bias” of one method over the other, while a wide spread of differences about the mean represents poor precision (indicated by a large value for 2 SD).10

Results 

All the women underwent successful neuraxial anaesthesia with a volume of 0.5% bupivacaine of 2.3±0.1mL and dose of diamorphine of 337±43μg. Total volume of intravenous fluids was 1873±544mL while median (IQR [range]) total dose of phenylephrine was 300 (150-600 [0-1300])μg and estimated blood loss was 600 (500-700 [200-1400])mL. No blood products were given intra-operatively. Duration of surgery was 47±16min and the patients’ temperature at the time of sampling was 36.0±0.5°C.

Bland-Altman plots for the capillary HemoCue® measurements of haemoglobin concentration against laboratory measurements are shown in Fig. 1. Bias (mean difference) and precision (± 2 SD) were respectively −0.1g/dL and ±1.8g/dL for the thumb samples and −0.2g/dL and ±1.6g/dL for the toe samples. Bias and precision for the HemoCue® against laboratory measurements for the venous samples were −0.2g/dL and ±1.6g/dL respectively (data not shown).

View Large Image Download to PowerPoint

Fig. 1. Bland Altman plots of haemoglobin concentration measured in the laboratory and with the HemoCue® using capillary samples: (a) thumb; (b) toe. The mean difference (bias)±2 SD (precision) are shown with horizontal dotted lines.

Discussion 

We were concerned that the commonly used sites for taking a drop of blood for a HemoCue® measurement (namely the toe and thumb or finger) under neuraxial anaesthesia for caesarean section might be inaccurate, first because of venous obstruction secondary to aortocaval compression (even with lateral tilt), and second, because of vasomotor changes secondary to neuraxial blockade. Our results suggest that, in terms of accuracy, both sites are equally suitable for use during caesarean section under neuraxial anaesthesia.

Our values for bias and precision for venous samples (−0.2g/dL and ± 1.6g/dL respectively) are greater than those of Yau et al.,3 who reported values of −0.04g/dL and ± 0.64g/dL respectively when analysing venous samples taken during caesarean section. Other investigators using the HemoCue® device for whole blood in a broad range of clinical settings, not only pregnancy, have reported values for bias ranging from −0.038 to 0.63g/dL and precision (2 SD) ranging from 0.4 to 1.74g/dL;3, 11, 12, 13, 14, 15; our value for bias is on the lower side of the above range but our precision lies at the upper end (Fig. 2). Since the precision in our study was similar for the capillary samples (thumb ± 1.8g/dL; toe ± 1.6g/dL) and the venous blood samples (± 1.6g/dL) that were analysed in the laboratory, we conclude that operator error was not the cause of the relatively poor precision since if it were, 2 SD would have been greater for the capillary sampling method. The value therefore probably reflects the HemoCue® device itself.

View Large Image Download to PowerPoint

Fig. 2. Mean difference (bias) ± 2 SD (precision) reported by studies of the HemoCue® for analysing whole blood, compared with laboratory analysis. The horizontal dotted line indicates a bias of zero.

The HemoCue® has been subject to varying opinions as to its value. There are those who expressly advocate its use1, 3, 16 and those who do not4, 5 questioning precision, operator experience and repeatability as areas of weakness. In a study of term women, Yau et al.3 found a very close agreement between the HemoCue® and laboratory testing (bias [precision] = −0.04 ±0.64) which was not matched in our study. We would agree with Van de Louw et al.17 that the results of HemoCue® testing should be interpreted with some caution reflecting the high range of 2 SD in the literature. There is no other point of care device that is comparable to the HemoCue®, so it remains useful as a clinical guide in the acute setting, though it should not replace formal laboratory venous sampling, which remains the gold standard.

It has been suggested that operator experience can be a cause of inaccuracies that would limit use of the HemoCue®.4 However, our study suggests that operator error was not a factor in the bias and precision values. We ensured that the single operator received one-to-one training from a company representative, and their recommended technique was employed for all the sampling. Users of the device should be adequately trained1 as with any medical device.

We did not standardise the anaesthetic technique or fluid regimen since we wished to study the HemoCue® in the pragmatic setting of routine clinical practice. Scrutiny of the anaesthetic details of the cases included in our study reveals them to be typical of practice in our unit.

We cannot extrapolate our findings to patients in the lithotomy position or under general anaesthesia. It is possible that prolonged procedures in lithotomy may introduce inaccuracy in HemoCue® performance when the foot is used for sampling, especially after a prolonged period during which the foot is raised above the body.

For caesarean section under neuraxial anaesthesia, our results suggest that either toe or thumb is suitable as a site for capillary sampling; since the foot is numb under these circumstances we would suggest a toe rather than a thumb, unless the toes are inaccessible.