Q & A

October 2003

Q. Our laboratory routinely does fingerstick glucose tests on a glucose meter in the morning. Occasionally, the patient also has a basic profile ordered, in which case the meter sugar may be canceled or, if performed, will result in the nurses dosing the insulin from the meter value and then receiving a serum glucose as part of the profile. Although the meter and analyzer correlate on similar samples (fingerstick and capillary tube obtained for serum) within 15 percent, the values may be on opposite sides of an insulin dosage level, which causes concern. Should the dosage always be from the meter since that report is received quickly as a point of care, or should the glucose be reported from only one method and the other test canceled?

A. Your question about the consequences of biases between two different analyzers measuring the same analyte is important to laboratorians. Much has been written about practices to minimize this issue. As you suggest, these differences may lead to different insulin doses, which could possibly cause a patient to suffer an adverse effect.

The introduction of bedside glucose monitoring has led laboratorians to again focus on this subject. The issue potentially has major consequences since about one of every seven dollars spent on health care is spent on diabetes, and better control of diabetes results in delayed and decreased complications for those afflicted. Almost 20 percent of Americans will suffer from disorders of glucose metabolism during their lifetime, and the incidence of diabetes is increasing. Therefore, even small improvements in the care of diabetics may have significant effects.

Your question delves into many areas. First, you should be aware that the biases between glucose measurements made using reflectance meters and plasma glucose values made in the central laboratory are decreasing. This is in part because some manufacturers are recalibrating their analyzers to minimize these differences. We have seen this in some of the CAP Q Probes studies. In my opinion, there is no valid scientific reason for these differences to exist.¹ I would suggest that laboratorians add the phrase “minimal bias between bedside and central laboratory glucose measurements” to their list of desired features in a bedside glucose analyzer.

Second, even though there are what we consider to be large errors made in the measurement of glucose, far greater errors occur in the actions taken in response to a glucose result. For example, the typical response to elevated glucose values is to draw insulin into a syringe and inject it into the patient. Recent research has emphasized the imprecision of measuring the required amount of insulin, especially for neonates, where the insulin doses are small. Some older studies found that these errors may be as high as 100 percent and are markedly influenced by dead spaces in the syringe or air bubbles.² Even more recent studies indicate that measuring insulin in a syringe prior to injection may have a coefficient of variation of five to 33 percent.³ Errors are more frequent when a combination of insulins are prepared for an injection than when a single insulin is used.

Others have pointed out that the effect of injected insulin is dependent on the injection site. A common error is to inject insulin intramuscularly rather than subcutaneously, or into scar tissue where absorption is slow. Even when a single insulin was injected subcutaneously, the halftime was found to be 6.9 +/– 2.6 hours, with an intraindividual coefficient of variation of 26 percent and an interindividual coefficient of variation of 55 percent.⁴

Despite these large errors in the delivery of insulin compared with the relatively small errors in the measurement of glucose caused by biases and imprecision, there are those rare patients who are harmed by inaccurate glucose measurements. It is important that we continue to work to decrease the biases between these two glucose measurements, as others continue to work on the inaccuracies of insulin delivery.

As for your specific concern, I suspect that you will not see any complications caused by biases between your glucose methods, as the imprecision of insulin measurement and delivery is far greater and masks the error of glucose measurement. I recommend that you continue your practice of measuring glucose in the central laboratory when other tests are ordered and not measure glucose at the bedside at these times.

References

1. Howanitz PJ, Jones BA. Bedside glucose monitoring. Comparison of performance as studied by the College of American Pathologists’ Q-Probes program. Arch Pathol Lab Med. 1996; 120: 333–338.
2. Arnott RD, Cameron MA, Stepanas TV, et al. Insulin syringes: dangers of dead space. Med J Aust. 1982;10:39–40.
3. Smith CP, Sargent MA, Wilson BP, et al. Subcutaneous or intramuscular insulin injections. Arch Dis Child. 1991; 66: 879–882.
4. Kolendorf K, Aaby P, Westergaard S, et al. Absorption, effectiveness, and side effects of highly purified procine NPH-insulin preparations (Leo).
Eur J Clin Pharmacol. 1978;14:117–124.

Q. I am looking for a method to test body fluids (peritoneal, pleural) for specific gravity. We have a refractometer in the urinalysis laboratory, but the literature on urine specific gravity testing says the urine refractometer should not be used to test other body fluids. Ithink most labs use the urine refractometer for testing other fluids even though it is calibrated for urine specimens. What is a suitable method for handling lab requests for specific gravity on other body fluids?

A. The urine refractometer should only be used to determine urine specific gravity. It is optimized to urine where the solutes that contribute to specific gravity are urea, sodium, chloride, sulfate, and phosphate. The urine refractometer gives spurious results when pure sugar or salt solutions are measured, and conversion tables are required if simple salt solutions are used to check the refractometer’s calibration.

Because you ask about pleural and peritoneal fluids, I assume you plan to use specific gravity to estimate protein concentration since the electrolyte concentration of these fluids fairly closely mimics plasma in almost all cases. Estimating the specific gravity of a protein containing fluids classically requires a graded series of copper sulfate solutions of known specific gravity. A drop of fluid is pipetted into the solution and then a shell of protein-copper forms around the drop. If the drop has a specific gravity greater than that of the copper sulfate solution, it sinks to the bottom. If the specific gravity is less than that of the copper sulfate solution, it floats to the surface. If the specific gravity approximates that of the copper sulfate solution, then the drop remains suspended in the solution.

I’m not sure why you would want to use this relatively rough and labor-intensive method, however, since fluid protein levels can be measured directly by a number of colorimetric methods on chemistry analyzers. The primary reason to measure protein is to assist in differentiating a transudative or exudative fluid. (Cell counts are also quite useful in this regard.) The modified criteria of Light (Light RW, et al. Ann Intern Med. 1972; 77: 507–513) are a classic standard for establishing the exudative nature of a fluid. They require that the fluid-to-serum total protein ratio be greater than 0.5; the fluid-to-serum ratio of lactate dehydrogenase activity be greater than 0.6; and the LDH activity in the fluid be greater than two-thirds of the upper normal limit of lactate dehydrogenase in the serum.

Robert Novak, MD
Department of Pathology
Children’s Hospital
Medical Center of Akron (Ohio)
Chair, CAP Hematology/Clinical Microscopy Resource Committee