A colleague in Pennsylvania uses the HemoCue^® (fingerstick) for screening their blood donors for anemia. Their policy is to perform one hemoglobin determination per donor. However, they have found that occasionally they observe widely discrepant results if the hemoglobin determination is repeated for a donor who is expecting to have a normal hemoglobin level, but tests initially on the anemic side. For example, the first determination might be a hemoglobin of 11.2 g/dL, yet a second fingerstick performed within a few minutes (at the donor's request) might be 13.1 g/dL. The inquiring colleague is curious to learn what others use as their policy for when to repeat the HemoCue test when screening their donors. Are there policies that require a second (or third?) fingerstick determination in donors who expect to have a normal hemoglobin level, but are surprised at the initial result and "demand" another test, etc.?

The following comments have been received.

ADDENDA Aug. 4, 2003

1. A laboratory-based physician in Connecticut reports that from his experience as a blood donor, inconsistent hemoglobin test results with fingerstick samples is an extremely frustrating problem. He reports having been rejected twice now for low hemoglobin, but when he asks the lab to have his hemoglobin run on a venous sample using a GEN-S, the hematocrit is always 40 or more. He always had the opportunity to make a follow-up appointment to donate in a few weeks, but for most employees at company drives this is not always possible, nor are all donors so persistent. Furthermore, it makes the Connecticut physician tend to avoid drinking much for a few hours before donation to reduce the chance that he will fail the hematocrit cutoff, which he admits is contrary to usual donor recommendations. He wonders if the problem is pre-analytical and asks the following question "are the screeners squeezing fingers too hard and contaminating the specimens with extracellular fluid?" He concludes that in his opinion there is clearly room for improving this aspect of donor screening.

2. Editor's note: Colleagues may find this related issue of interest to the current discussion

ADDENDA Aug. 5, 2003

3. A colleague from New Zealand reports that if a donor's fingerstick hemoglobin result is below the acceptable level, a second fingerstick sample is obtained by another operator for repeat testing. If this second result is above the minimum level for hemoglobin, the donation is accepted (provided no other reason for deferral is discovered).

The reason for this approach is that in the New Zealander's experience, any error/problem with the fingerstick hemoglobin testing process usually gives a falsely low result due to contamination of the sample with tissue fluid or incomplete filling of the test cuvette. The New Zealander also comments that in his experience, capillary hemoglobin levels and venous hemoglobin levels are inherently different (with the venous levels consistently being lower than the capillary levels). He believes that this difference has nothing to do with the testing process or contamination. He adds that in his donor center, their acceptance range for hemoglobin is based on the capillary hemoglobin level, and not that obtained by venous sampling. Thus, they base their decision about donor acceptibility on the basis of fingerstick result alone. (See results of their study in #6 below.)

ADDENDA Aug. 6, 2003

4. A colleague in Canada reports that her hospital uses several HemoCue devices for screening patients for anemia. They have found the HemoCue test results on venous draws to be within 0.5 to 1 g/dl of the hemoglobin result obtained when they use their Cell Dyn 3200 on venous draws.

However, in their experience, when fingerstick specimens collected directly into the Hemocue cuvette were compared to venous draws, it was not uncommon for the venous draw (done on the Cell Dyn 3200) to give a substantially higher hemoglobin result than the fingerstick specimen using the Hemocue. They believe the problem was with the quality of the fingerstick and the amount of blood obtained freely (or with no squeezing). What seems to work better for them when testing fingerstick samples is to wipe away the first drop of blood and to test the second drop for the hemoglobin value.

Another observation they have made when testing fingerstick samples with the HemoCue was that the longer the test cuvette is allowed to sit (up to 10 minutes) before testing, the closer the Hemocue result came to that of the Cell Dyn 3200 for patients with 'resistant' red cells (eg. nucleated rbcs). The Canadian colleague believes this is due to the cyanide-based reagent used in the HemoCue cuvettes that causes the red cells to lyse, thus releasing the hemoglobin for measurement in the assay.

She wonders if some falsely HIGH hemoglobin results reported on fingerstick samples might be due to incomplete lysis of the red cells during testing. She cautions that the latter observation may not be applicable as they have observed it only in quality control material sent to them from their referral lab. They had noticed it with the rare newborn screened by public health. However they are not a large institution, so they do not have a wide variety of patients to evaluate. Their clinics are also unlikely to recheck a normal/high hemoglobin result from the HemoCue.

ADDENDA Aug. 8, 2003

5. A colleague in the UK (Dr. Frank Boulton, attribution used with permission) reports that in that country the HemoCue is reserved only for testing blood from donors who fail the copper sulfate gravimetric screen (SG 1.055 for men equiv. Hb 13.5g/dL; SG 1.053 for women equiv. Hb 12.5g/dL). He writes (verbatim) "Furthermore, only venous blood is used for the HemoCue, as we find fingerstick blood to be too variable. In two separate small studies of volunteers from whom blood was taken by fingerstick from each finger, (repeated when a new lancet came on the market) we found that in all subjects the readings from fingerstick HemoCue varied quite widely from the mean and from the HemoCue readings on venous blood taken at the same time. A man with a venous Hb of 15.7 g/dL had fingerstick HemoCue readings varying between 13.5 and 17 g/dL. All fingers were warm and the blood flowed freely. We also observed that the first fingerstick droplets to emerge had a mean Hb 0.16g/dL lower than; and the second droplets to emerge had a mean Hb 0.42g/dL higher than the venous sample. The current UK policy is to offer all donors failing the copper sulfate gravimetric test a HemoCue test on venous blood at session. If this reads 13.0g/dL or more from men and 12.0g/dL or more from women, the donor is accepted. This does mean that a second venipuncture is required (from the other arm); and is somewhat time-consuming especially if the session is busy, but on the whole more donors are retained than lost. I would not recommend using HemoCue on fingerstick samples. This is not the fault of the HemoCue - when performed properly they will assay accurately the sample of blood presented to the cuvette. (BUT NOTE THAT HEMOCUE ADVISES THAT THEIR CUVETTES SHOULD BE USED AT TEMPERATURES BETWEEN 15C AND 30C; SOME SESSIONS IN THE CURRENT EUROPEAN HEATWAVE MAY BE BEING HELD AT HIGHER AMBIENT TEMPERATURES.) The variable results from fingerstick samples derive from variable quality of the sample, not the HemoCue device. The gravimetric method seems to be less susceptible to variability, although this seems illogical; but the gravimetric method is remarkably rough and ready! James et al in 2003 (Transfusion 43; 400-404) found among UK donors that 18.7% of women failed copper sulfate (SG 1.053) but only 9.7% by HemoCue and 7.7% by automated counter (both set for Hb 12g/dL, also 5.3 - 6.0% were genuinely below 12g/dL). We look forward to non-invasive reliable pre-donation hemoglobinometry, but a system using pre-donation assessment clinics could also alleviate donor losses through inadequate methods of Hb screening as predonation venous HemoCue would not be practical for most services."

ADDENDA Aug. 11, 2003

6. The New Zealand colleague (#3 above) wishes to add to his previous comments. Recently, they compared fingerstick HemoCue results with venous HemoCue and venous Coulter hemoglobin results for 329 blood donors. All fingerstick tests were carried out at a temperature between 20C and 24C. The New Zealander offers the following results:

The New Zealand colleague concludes that their data show a significant difference between fingerstick and both the venous hemoglobin results, but no significant difference between the two venous sample results. Although the differences between fingerstick and venous samples were not consistent, in the vast majority of cases the fingerstick results were higher than the results on venous samples.

ADDENDA Aug. 12, 2003

7. Scott Montrose, Technical Director, Blood Bank of the Redwoods (attribution used with permission) reports that his facility compared fingerstick samples run on the HemataSTAT^® II (HSII) to venous samples drawn on the same individual that were run on the HSII and on the ABX Micros 60. A summary of  results from a correlation study of 50 donors in their center was provided by Scott, and appears below verbatim. The abbreviations are as follows:

• ABX: Venous sample run on ABX Micros 60
• HSII-fs: Finger stick sample run on the HemataSTAT II
• HSII-pct: Venous pre-procedure count run on the HemataSTAT II

Results  (ABX plotted on x axis, or HSII-pct plotted on x axis vs HSII-fs)

None of the comparative data met all of the established target criteria of a slope between 0.9 and 1.1, y-intercept of between +1 and –1, and an R2 value of >0.86. None of the correlation coefficient values were > 0.8000, even when comparing the HemataSTAT II venous values to the HemataSTAT II finger stick values; however that data set did yield an acceptable slope of 0.9074.

Discussion

The data collected suggests that the ABX Micros 60 and the HemataSTAT II methods do not meet the established correlation criteria, even if venous blood is used to eliminate the hemoconcentration or hemodilution phenomena that can be present when collecting a capillary blood sample by fingerstick. The x-y differences illustrate significant variation for the HSII-pct vs. HSII-fs, with differences ranging between +3 to –8.  The ABX vs. HSII-fs x-y differences show variation ranging from +2.3 to –4.3. The ABX vs. HSII-pct x-y differences show variation ranging from +8.3 to –3.3.  The data suggests a somewhat random variation in the values obtained, with no apparent bias to the pattern when comparing the ABX data to the HSII data.  When the x-y data for (HSII-pre) – (HSII-fs) is plotted in a histogram, there is a bias toward the negative side of the scale.  This would indicate that the preponderance of the fingerstick data has a higher value than the venous (pct) data.  This suggests that there may be a hemoconcentration phenomenon associated with collection of the capillary (fingerstick) samples.

Conclusion

The HemataSTAT II HCT% does not appear to correlate to the ABX HCT% values. The variation between the two methods appears to be random in nature. However, there does appear to be a bias toward a higher hematocrit with the fingerstick collection.

Please submit comments to the e-Network Forum.

Ira A. Shulman, MD
CBBS e-Network Forum Editor & Moderator