The technique of blood glucose determination must be considered because different methods vary in specificity and sensitivity to glucose. The blood specimen itself is important; according to several reports (and my own experience), during each hour of standing at room temperature, whole blood glucose values decrease about 10 mg/100 ml unless a preservative is added. A high hematocrit value accentuates glucose decrease due to RBC metabolic activity. Fluoride is still the most recommended preservative. Plasma and serum are more stable than whole blood. If serum can be removed from the cells before 2 hours, serum glucose values remain stable for up to 24 hours at room temperature (although some authors report occasional decreases). Refrigeration assists this preservation. Serum or plasma values are generally considered to be 10%-15% higher than those of whole blood. However, several studies have reported considerable variation, ranging from 3% to 47%, in this difference over periods of time. Most current automated equipment use serum. Some small whole-blood office-type or portable analyzers are available, either single-test dedicated instruments (e.g., Yellow Springs glucose analyzer), reagent cartridge type (e.g., Abbott Vision or HemoCue-BG), or reagent strip types (Kodak Ektachem or Bohringer Reflotron). Venous blood is customarily used for glucose measurement. Capillary (arterial) blood values are about the same as those for venous blood when the patient is fasting. Nonfasting capillary values, however, average about 30 mg/100 ml (1.6 mmol/L) higher than venous blood, and this difference may sometimes be as great as 100 mg/100 ml (5.55 mmol/L).

Biochemical methods. There are a considerable number of methods for blood glucose determination. These may be conveniently categorized as nonspecific reducing substance methods, which yield values significantly above true glucose values (Folin-Wu manual method and neocuproine SMA 12/60 automated method); methods that are not entirely specific for glucose but that yield results fairly close to true glucose values (Somogyi-Nelson, orthotoluidine, ferricyanide); and methods using enzymes that are specific for true glucose (glucose oxidase and hexokinase). There are certain technical differences and interference by certain medications or metabolic substances that account for nonuniformity of laboratory methodology and that in some instances may affect interpretation. Reference values mentioned in this chapter are for serum and for true glucose unless otherwise specified.

“Bedside” paper strip methods. Another test for glucose consists of rapid quantitative paper strip methods (Dextrostix, Visidex, Chemstrip-BG, and others) available from several manufacturers. A portion of the paper strip is impregnated with glucose oxidase, an enzyme specific for glucose, plus a color reagent. One drop of whole blood, plasma, or serum is placed on the reagent area, and the color that develops is compared with a reference color chart. Visidex has two reagent areas that correspond to low- and high-glucose value areas. Small electronic readout meters are available for several of the manufacturer’s paper strips. The meters have generally been reported to make a substantial improvement in accuracy. Evaluations of the various paper strip methods provide a consensus that, with experienced personnel and with the use of a readout meter, experiments using quality control material or glucose solutions generally agree with standard laboratory methods within about ±5%. Using actual patient fingerstick capillary blood specimens, values between 40 and 130 mg/100 ml (2.2-7.2 mmol/L) usually agree within about ±15% (range, 8%-40%) with values obtained by standard laboratory methods. Persons without much familiarity with the technique may obtain more erratic results. These paper strip methods have been used with venous whole blood or finger puncture blood as a fast way to diagnose hypoglycemia and hyperglycemia in comatose or seriously ill persons and to provide guidance for patient self-adjustment of insulin dosage at home.

Some cautions include possible differences between capillary (finger puncture) blood and venous blood values, alluded to previously; and effects of hematocrit value on results, since blood with a low hematocrit value (<35%) produces a higher result (by about 10%-15%), whereas blood with a high hematocrit value (>55%) produces a lower result. This creates a special problem with newborns, who normally have a high hematocrit value compared to adults. Also, quality control or evaluation of different manufacturer’s products by using glucose solutions may not accurately predict results using patient blood specimens. Very high serum levels of ascorbic acid (vitamin C) or gross lipemia may interfere. Patients with hyperosmolar hyperglycemia, with or without ketosis, may show test strip results that are lower than true values. Capillary specimens from cyanotic areas or from patients in shock may produce falsely low results. In one study of patients in shock, 64% of patients had fingerstick levels over 20% less than venous ones, and 32% of patients had fingerstick levels over 50% less than venous ones.