Hemolysis Rates in Whole Blood Samples for Blood Gas/Electrolyte Analysis by Point-of-Care Testing

To the Editor:

Arterial blood gas/electrolyte analysis is an essential investigation in critically ill patients. The collection of arterial heparinized whole blood samples can be challenging, with a need to minimize in vitro hemolysis. Hemolysis is not readily detected in whole blood samples, and it may give rise to spurious results with significant increases in PCO₂, bicarbonate, and potassium (1). The spurious increase in potassium is important, as abnormalities in potassium concentration are common in critically ill patients and may be life threatening. Unrecognized hemolysis may result in an incorrect diagnosis of hyperkalemia or mask true hypokalemia. The prevalence of significant hemolysis in blood gas samples has been reported as 1.2% in samples from an emergency department (ED) (1–3).

We compared hemolysis rates in blood gas samples taken in an intensive care unit (ICU) with those taken in an ED in Altnagelvin Hospital. Samples from 100 consecutive patients in the ED were taken by arterial puncture and analyzed by trained clinical staff on the Cobas b123 POC analyzer (Roche Diagnostics International Ltd.); arterial line samples from 100 consecutive patients in the ICU were analyzed by trained clinical staff also on a Cobas b123 POC analyzer. In both ED and ICU, blood samples were collected in heparinized syringes [heparinized 1-mL blood gas sampling syringes, (Becton Dickinson)] and analyzed immediately. Following the analysis, the surplus blood was immediately transferred by a laboratory technologist using gentle evacuation from the unneedled syringe into an open-top nongel lithium heparin blood tube [BD™ Vacutainer™ lithium heparin plasma tubes (Becton Dickinson)]. Samples were centrifuged Centrifuge Hettich EB270 (Hettich Laboratory Technology) for 10 min at 1630 g and analyzed on the Cobas 8000 analyzer (Roche Diagnostics) for hemolysis index (HI). [Studies on a separate series of 25 whole blood samples showed that the whole blood transfer and centrifugation process was associated with only a negligible increase in the HI of 3.0 mg/dL (median), 4.2 mg/dL (mean).] The laboratory participates in the UK National External Quality Assurance Scheme serum indices and demonstrated satisfactory performance during the study period.

Samples were categorized as “no hemolysis” (HI < 10 mg/dL), “slight hemolysis” (HI, 10–99 mg/dL), “hemolysis” (HI, 100–199 mg/dL), “severe hemolysis” (HI, 200–599 mg/dL), and “gross hemolysis” (HI > 600 mg/dL). Hemolysis was more frequent and of greater severity in samples from ED than in those from ICU(χ² test, P < 0.0001; Table 1).

For serum samples received by the laboratory from ICU (4342 samples) and ED (32709 samples), the hemolysis rates were also higher in samples from ED than in those from ICU (χ² test, P < 0.0001; Table 1). The ICU samples were collected by nursing and medical staff, generally from indwelling lines, while the ED samples were collected by medical and nursing staff, generally by peripheral venipuncture.

Hemolysis rates are higher in whole blood than serum samples from both ICU and ED and higher in ED than in ICU for both sample types.

For serum samples analyzed in the laboratory, potassium results are labeled as “unreliable” if the HI > 99 mg/dL and not reported if HI > 200 mg/dL. This means that 13% of ED potassium results and 4% of ICU potassium results obtained from whole blood samples by point-of-care analysis would have been considered as potentially inaccurate.

The percentage of whole blood samples with significant hemolysis is higher than that reported recently by Lippi (1.2%) on the basis of visual inspection but comparable to the older findings of Hawkins (18%) (1, 4). It is unclear why there should be a significant difference in hemolysis rates between ICU and ED; it is unlikely to be because of differences in in vivo hemolysis rates. In vitro hemolysis may result from mechanical factors such as inappropriate needle size, excess blood flow rate, and overly vigorous sample mixing. An important difference between the ED and ICU whole blood samples is that the latter were taken from arterial lines and the former by arterial puncture. A further difference may be that the ICU samples were collected by a team of highly specialized clinicians with extensive experience in arterial sampling. In contrast, the ED samples were collected by a broader group of clinical staff, who, although trained in blood gas sampling, may have had less cumulative experience than ICU staff. The fact that hemolysis rates were also higher for serum samples collected in ED is also noteworthy.

In conclusion, this study shows that hemolysis rates in whole blood samples remains an important problem and may differ between clinical units. Instrument manufacturers should consider the introduction of hemolysis checks in whole blood analyzers.

• Received February 27, 2018.
• Accepted April 13, 2018.

• © 2018 American Association for Clinical Chemistry