We have in our study added data on capillary and venous sampling with a device that seems practical to use prehospitally. Our study of CAP and VEN sample analysis showed generally higher values (mean bias, 0.7 and 0.9 and upper level of limits of agreement, 2.4 and 1.9 mmol/L) than the reference method GEM (Fig.
2). This agrees with results from Baldari et al. [
6] that also demonstrated higher levels of lactate in healthy athletes when using handheld analyzers. The Bland-Altman plot shows high bias in both VEN and CAP, although somewhat higher in VEN. The precision is superior with VEN than CAP which may be more important than the bias. High precision is associated with low risk for underestimation, which is, from a clinical point of view, safer. The Bland-Altman plot also illustrates if the variation of differences is constant through different values on the x-axis. It seems to be the case for values under 4, but it’s difficult to assess if the same is true for values > 4 as only a few values are > 4 in our study [
10]. Moreover, no sign of systematic change in bias in the Bland-Altman plot. In Table
3 only small changes in mean bias was found when GEM values were 0–2 mmol/L and 2–3.5 mmol/L (Fig.
2). Our results show a low accuracy with CAP, as well as with VEN analyzed with the handheld AP compared to the reference method GEM. This is in discrepancy to Baldari [
6] and Pyne et al. [
7] who investigated the reliability of the AP in young healthy athletes and where the results showed a good linear relationship with the reference method using capillary blood samples. Pattharanitima et al. [
8] also showed a high correlation between arterial and capillary lactate values, as well as high correlation between arterial and central venous lactate values in 30 septic inhospital patients using AP. Our results concur with Datta et al. [
11] who demonstrated a poor agreement of capillary lactate analyzes in comparison to peripheral venous lactate measurements.
Table 3
Mean, standard deviation and bias of Accutrend Plus and GEM
Accutrend Plus |
Capillary | 2.18 | 0.86 | 0.77 | 3.29 | 0.93 | 0.47 | 2.55 | 1.51 | 0.72 |
Venous | 2.27 | 0.60 | 0.86 | 3.70 | 0.74 | 0.88 | 2.71 | 1.48 | 0.88 |
GEM |
Venous | 1.41 | 0.34 | – | 2.82 | 0.72 | – | 1.83 | 1.37 | – |
The limits of agreement show an existing bias in both lactate measurements but slightly larger for VEN. The limit is wider for CAP compared to VEN. The scatter of the differences is constant in increasing values on the x-axis, both for CAP and VEN-values. The bias in CAP is lower than in VEN, but for values below 2 mmol/L and between 2 and 2.5 mmol/L bias is more stable for VEN (Table
3).
Limitations
Time between sample and analysis was seconds for the handheld analyzer. The blood samples were collected upon arrival by two nurses (authors 1 and 4) using the same procedure for all patients included. No local injuries or issues in the limbs where the samples were collected where found for any of the patients included in the study during the stay at the emergency ward. All samples tested with GEM were analyzed within 20 min (30 min stipulated by manufacturer). We consider our bias concerning differences in sampling procedures minute.
As shown in Table
1 the majority (91,2%) of the included patients were triaged orange/yellow and only a few red patients could be included. This may have selected a category of patients with similar lactate levels and minimized the spread of our measurements. The discrepancy to the results found by Baldari [
6] and Pyne et al. [
7] who investigated capillary samples in young healthy athletes and Pattharanitima et al. [
8] who analyzed capillary samples from septic patients could perhaps be explained by the difference in the selection of patients. Where our study had a relatively large number of patients (
n = 160 triaged mainly as orange/yellow priority) with a similar level of lactate the Baldari study [
6] showed a wider range of lactate levels. The red/orange patients included with possible low blood-pressure may have had peripheral vasoconstriction. This may account for differences between VEN and CAP levels.
We found that lactate levels measured with CAP, as well as VEN, were higher than the values measured with GEM (Fig.
2). The spread of 2 mmol/L above values analyzed by GEM was more frequent for CAP than for VEN. Baldari et al. [
6] found that the deviation with the handheld compared to stationary blood-gas analyzers escalated with higher lactate values. The measured levels of lactate in our study were generally lower than 6 mmol/L which gives a small spread of the results and may account for our small deviation from reference. Further studies with patients in critical conditions with a wider spread of lactate levels are needed to be able to fully evaluate the potential use of CAP samples.