Glucose Concentration Difference Between Arterial, Capillary, and Venous Blood

Introduction

Three of the major factors that influence glucose test results are the type of chemical analysis used for the test, the type of sample analyzed (whole blood verses plasma), and the source of the blood (venous, capillary, or arterial)^[1]. Home glucose monitoring has traditionally relied on a drop of capillary blood from the finger, but off-finger capillary sites are now being used and questions have arisen about their comparability.

Used only with express written permission

Until recently, capillary blood from a fingerstick was the standard sample used in home glucose monitoring. Occasionally, a blood sample from the earlobe or heel (infant monitoring) was also used. Capillary samples from the finger or ear lobe have been closely associated with arterial blood values, i.e., their glucose and oxygen properties are more similar to arterial blood values than venous blood values^[2,3]. However, even with fingerstick blood, concerns have been expressed about the variation in finger sampling technique and changes in peripheral blood flow as these may alter the composition of capillary blood. The main worry that has been expressed is contamination of the test sample, i.e., too much squeezing or 'milking' of the fingertip to produce a drop of blood may cause inaccuracies from either excess tissue fluid or hemolysis.

With the newest self-monitoring of blood glucose (SMBG) systems, capillary blood samples from sites other than the fingertips (forearm, upper arm, palm of the hand, calf or thigh) are used to measure glucose. These different locations must deal with the blood variation concerns of the finger plus address the spatial and temporal heterogeneity of the local cutaneous blood flow. It has been claimed that forearm capillary blood samples are more similar to venous blood values than arterial blood values. Specifically, a TheraSense™ FreeStyle™ test strip package insert states:

"Blood glucose in forearms and fingertips is not always the same…. FreeStyle arm measurements, on average, are slightly lower than FreeStyle finger measurements. The difference is similar in magnitude to the difference generally observed between capillary finger measurements and venous measurement [4]…Venous whole blood results are about 7% lower than a capillary sample from the same person with normal glucose levels."

However, an empirical conversion factor between forearm capillary and venous blood glucose levels has neither been supported nor disproved in the literature.

This paper reviews three physiological areas that can alter a blood glucose reading from one blood compartment to another. In doing so, it attempts to clarify how glucose concentration might differ in various blood samples. It also increases familiarity with physiological parameters that affect glucose levels and possible problems with the analytical method used to measure glucose. This should lead to better use of the available technology and help supply the most clinically accurate glucose reading for the patient.

1. Glucose test values may not match with different blood samples because glucose is being consumed by the body

Glucose diffuses through the capillaries and is consumed by the cells, so arterial glucose concentration (the capillaries' source) should be higher than venous glucose concentration (the capillaries' drain) unless capillary diffusion or muscle glucose consumption has been stopped. It has been shown that in fasting subjects the glucose levels in arterial, capillary, and venous samples are practically the same (venous glucose is generally 2-5 mg/dL lower than fingerstick capillary or arterial blood glucose)^[5,6]. It is only after meals, when glucose uptake in the periphery is rapid, that glucose levels in fingerstick capillary blood samples can exceed those in concurrently drawn venous samples. A typically quoted value is up to 80 mg/dL difference between venous and fingerstick capillary blood glucose values one hour after ingestion of 100 grams of glucose^[2].

Current literature has attempted to determine exactly how glucose levels in venous, arterial and fingerstick capillary blood vary so comparisons can be made. Venous blood is usually employed for laboratory analysis and is preferable in diabetes testing^[6]. However, because of the widespread use of SMBG instruments, fingerstick capillary blood samples have also become a standard. Fingerstick capillary blood has been shown to be predominantly arterial^[7] and so approximates the concentration of arterial blood. Somogyi compared the glucose content of blood samples simultaneously drawn from the femoral artery and the fingertip of non-diabetics one-hour after ingestion of 50 grams of glucose. The ingested glucose would produce a substantial difference between the arterial and venous glucose levels, and so should indicate whether fingerstick capillary blood was predominantly arterial, venous, or a combination of the two. The discrepancies between arterial and fingerstick capillary blood were less than 1 mg/dL for all three subjects studied and seemed to justify the substitution of fingerstick capillary for arterial blood glucose.

Somogyi also studied the difference between fingerstick capillary and venous glucose levels during the fasting state on 100 healthy individuals (fasting for 10-14 hours). The average fasting glucose value in fingerstick capillary blood samples was 89 mg/dL (78 - 97 mg/dL) with the average venous blood glucose value 5 mg/dL lower (84 mg/dL). Averages can be misleading as the capillary-venous blood glucose difference did vary widely among subjects. Differences between the fasting fingerstick capillary and venous blood glucose levels ranged from 1 mg/dL to 7 mg/dL in 93% of the patients studied; however, the fasting fingerstick capillary glucose values were 10 to 13 mg/dL higher than venous glucose in 3% of the healthy individuals studied. For all 100 individuals, the fingerstick capillary (paper assumed this to be arterial) blood glucose was always higher than the venous blood glucose in line with the fact that cells continually assimilate blood sugar. Hence the concentration of glucose inevitably decreases during the passage of blood through the tissue. The study noted that the magnitude of the fingerstick capillary (paper assumed this to be arterial) to venous blood glucose difference was not a characteristic of the individual as the value would vary from day to day in a healthy person, and it did not correlate to the fasting blood glucose value over the concentration range studied.

In the same study, both venous and fingerstick capillary blood glucose values were followed for a period of 4 hours in 44 healthy individuals that had ingested a 100-gram glucose load. As shown in Figure 1, a substantial increase in the fingerstick capillary to venous blood glucose difference were measured after an oral administration of glucose, and this difference remained consistently higher than the initial fasting level until the blood glucose returned to the fasting blood glucose level. The paper also found that the larger the glucose load ingested, the higher the glucose peaks, and the greater the maximal difference between the fingerstick capillary (paper assumed this to be arterial) blood glucose level and the venous blood glucose level.

Used only with express written permission

Figure 1: The average arterial and venous blood glucose levels from 44 healthy individuals after a 100-gram glucose load was ingested at time t=0. Fingerstick capillary blood glucose was measured in place of arterial blood glucose after the two were shown to be equivalent.

A study by Liu measured arterial, fingerstick capillary, and venous blood samples from six healthy males for oxygen saturation and glucose^[8]. Each subject's right hand was placed in a warm air box at 55-60 degrees C to determine if warm air would arterialize the venous blood obtained from a cannula inserted into the dorsal right-hand vein. The oxygen saturation measured in the arterial blood was 97%. The oxygen saturation measured in venous blood on a nonheated hand was 80%. The oxygen saturation measured in the heated 'arterialized' venous blood was 94% or approximately 3% below the average arterial value. Glucose levels also showed equilibration between the two blood compartments with heating. The difference between fasting arterial glucose levels and venous glucose levels with no heating of the hand ranged between 4-9 mg/dL (6% - 9%), and this glucose difference significantly correlated with the differences in oxygen saturation between the two blood supplies. The difference between the arterial glucose levels and 'arterialized' venous glucose levels obtained by heating the hand averaged less than 2 mg/dL difference, and this glucose difference had a low correlation with the differences in oxygen saturation between the two blood supplies.

The difference between capillary and venous blood in the postprandial state is due to muscles removing more glucose from the blood than the liver in the presence of adequate insulin action^[6]. Absolute values of glucose uptake into body organs should follow the organs metabolism and, in general, the higher an organ’s metabolism, the greater the blood flow. Table 1 shows the blood flow to different organs and tissues under basal conditions and gives an indication of their glucose needs. While inactive muscle constitutes between 30-40 percent of the total body mass it requires only 15% of the blood flow; however, during heavy exercise, muscle blood flow can increase as much as 20-fold to handle the increased metabolic activity^[9]. As more blood is shifted to the muscle, less blood goes to the tissues where it is not needed at the moment. During exercise the flow to skin is initially reduced but is later increased to get rid of excess heat. This action confirms a fundamental principle of circulatory function: controlling local blood flow allows the workload on the heart to be minimized while controlling the body’s temperature and maintaining sufficient nutrients at critical tissue sites.

Used only with express written permission

Table 1: Blood flow and blood flow by weight to different organs and issues under basal conditions [9].

Although key organs such as the liver, kidney and muscle during exercise consume most of the available glucose, the epidermis layer of the skin also has a very high metabolic activity and thus must have a high rate of glucose assimilation. The entire epidermis completely renews itself in a period varying from 45 to 75 days^[10].

It has been shown that a lack of insulin (in the de-pancreatized animal) shows an arterial to venous glucose difference that is extremely small and that injection of insulin produces an increase in this difference^[3]. As such, glucose uptake by the tissue is dependent on the sensitivity of the tissue to insulin, the circulating insulin level and the local blood flow. Diabetics may have various degrees of peripheral insulin resistance or various blood insulin levels or both, so a single patient’s nonfasting difference may not be seen in other patients. The nonfasting difference will depend on meal size, meal content, time of sample collection, and individual patient variability.

In summary, glucose levels in arterial and fingerstick capillary blood have been so closely correlated that most studies refer to arterial glucose measurements even if they measure fingerstick capillary samples. When studies are performed with the patient under fasting conditions, glucose levels in fingerstick capillary blood gives reliable quantitative estimates of the venous glucose concentration as determined in the laboratory for most patients. However, when the patients are under a glucose load the venous and fingerstick capillary glucose levels diverge in a similar but unpredictable manner where the venous value may be anywhere from 2% lower during fasting to 26% lower within one hour after a glucose load.

Unfortunately, empirical conversion factors have been applied to generate equivalent glucose values for different blood sample compartments without adequate data to show equivalence. One such conversion is that fingerstick capillary blood has a glucose concentration that is 7-8% higher than the concurrently drawn venous concentration^[11]. Others have presented charts showing the equivalence of venous and capillary glucose levels that differ between 0% to 13% depending on the glucose level^[12]. The validity of these conversion factors has been called into question since individual differences between capillary and venous blood glucose values are too great to allow for a meaningful transformation to be applied^[13,14]. It can be reasonably concluded that there is no simple conversion factor available to explain differences between glucose values in the various blood compartments.

2. Glucose test values may not match because the body is consuming oxygen

Like glucose concentration, the oxygenation of venous blood is dependent on three main factors: the oxygen saturation of arterial blood, the oxygen consumption of the tissue drained by the vein concerned, and the rate of blood flow through the tissue. Oxygen sensors measure the partial pressure or tension (pO2) of oxygen, and this is simply the saturated density of free oxygen in blood.

The analytical methods that measure for glucose must be capable of dealing with oxygen variation in the blood sample. However, some SMBG meters have been shown to be sensitive to the large oxygen variation seen between fingerstick capillary (arterial) and venous blood samples, and there are warnings in the package inserts against venous blood use. Many analytical procedures are used to measure blood glucose but the most common techniques are enzymatic. Enzymes commonly used in commercial test strips are glucose oxidase, glucose dehydrogenase, or hexokinase combined with glucose-6-phosphate dehydrogenase.

Glucose oxidase has historically been the preferred enzyme because of its excellent specificity for glucose, good room temperature stability, and relatively low cost. However, the reaction requires an adequate oxygen supply, and this leads to an oxygen dependence problem in certain measurement systems. Electrochemical measurement combined with glucose oxidase involves a mediator to transfer electrons between the electrodes. The mediator attempts to replace oxygen in the reaction sequence. This makes oxygen in the blood sample a competitor in the reaction and produces varying results with varying oxygen concentrations (oxygen dependence). A Glucometer™ Elite™ test strip labeling stated: "A venous whole blood sample usually reads higher than a (fingerstick) capillary sample from the same person (approximately 7% higher on average with normal glucose samples) due to the unique electrochemical properties of the test strip." Electrochemical test strips that are calibrated using fingerstick capillary blood can read up to 30% higher when tested with venous blood because of its 50-60% lower pO2 values^[15]. A similar situation exists with some optical reflectance methods. Generally, atmospheric oxygen is sufficient to meet the glucose oxidase reaction requirements, but different test strip design can block the diffusion of oxygen to the reaction site. To get around poor oxygen diffusion, a dye system has been utilized that essentially takes the place of oxygen in the reaction. This replacement gives very fast color development, but the oxygen content in the sample competes with the intended reactant in the oxidation reaction creating oxygen dependence. Commercial analyzers attempt to circumvent oxygen effects by pre-dilution of the sample into an oxygenated buffer. Instruments that use a glucose oxidase reaction include optical measurement devices OneTouch™ SureStep™, AccuChek™ Easy™ system, AccuChek Instant™ system and electrochemical measurement devices Glucometer Elite, and the laboratory systems Beckman™ Glucose Analyzer and YSI™ Glucose Analyzer^[16].

Glucose dehydrogenase can be made oxygen independent when it is combined with a cofactor called pyrroloquiniline quinone (PQQ). Using this enzyme combination effectively eliminates oxygen competition and enables the use of venous or arterial samples where extremes of pO2 may occur. The trade-off is reduced specificity for glucose in that it also detects maltose, galactose, and metabolites of maltodextrins. There is also reduced operational stability when compared to glucose oxidase. The electrochemical measurements by the AccuChek Advantage™ system and TheraSense FreeStyle previously used this reaction mechanism^[16], but due to maltose reactions they have been changed.

Hexokinase combined with glucose-6-phosphate dehydrogenase also avoids oxygen dependence, but the test strip is inherently more sensitive to heat and moisture, and therefore special attention is paid to packaging. The Bayer™ Encore™ product uses this mechanism^[16].

Glucose comparison studies between arterial, capillary, and venous blood must consider the significant differences in oxygen tension between the blood compartments when using analytical systems that are oxygen dependent. Ideally, the effect of pO2 needs to be examined by monitoring oxygen concentrations and determining if a correlation exists for glucose. Only Liu's paper discussed earlier has been found to adequately perform this task^[8].

3a. Glucose test values may not match because of low blood flow in the forearm

The first two sections in this paper (glucose consumption and oxygen variation) concern physiological parameters that would lead to a bias between glucose test results taken simultaneously from two different blood compartments during either fasting or the meal cycle.

A third physiological parameter that would cause glucose in one blood compartment to lag or lead another is flow or circulation problems in a capillary bed. Many medical and physical conditions can affect capillary blood flow with the problem being either systemic or localized. Localized variations in blood flow associated with the capillary beds would be a major contributing factor to erroneous comparison data between two capillary blood supplies such as within the finger and forearm. A localized variation in blood flow would also be a contributing factor in glucose differences measured within capillary, arterial, and venous blood.

Blood flow to skin capillary beds is controlled by two major mechanisms: autonomic nerve control of metarteriole and muscle control of capillaries through a precapillary sphincter. The metarteriole is a preferential shunt around the capillary bed that directly connects the arteriole to the venule and is under the control of the nervous system. In the skin, opening or closing of these shunts is important in heat regulation of the body, and the blood flow in these shunts does not participate in transfer of gases, nutrients, or wastes. The precapillary sphincter is a band of smooth muscle at the junction of each capillary vessel and arteriole. These sphincters regulate the amount of blood that enters into the capillary bed, and as a result, blood does not flow continuously through the capillaries, but intermittently in a series of pulses. This alteration of blood flow through the capillaries is termed vasomotion. Vasomotion is a subtle and esoteric concept that can globally result in lower blood flow. The frequency of vasomotion translates into more or less flow. With these phenomena in mind, only an average rate of blood flow, capillary pressure, and transfer of substances can be discussed. These average functions are in reality the functions of literally billions of individual capillaries, each operating intermittently in response to the local conditions of the tissue. This physiological temporal variation in flow has also been described as regular rhythmic changes in flux that occur with periods that range from approximately one second to several minutes^[17].

Two basic theories for the regulation of local blood flow involve either 1) vasodilators regulated by the rate of tissue metabolism or 2) lack of nutrient availability^[9]. As an example, a local drop in pO2 is the most important factor in the lack of nutrient theory because oxygen is usually the rate-limiting metabolite delivered by the blood. As explained by Guyton and Hall^[9]:

“Because smooth muscle requires oxygen to remain contracted, one might assume that the strength of contraction of the sphincters would increase with an increase in oxygen concentration. Consequently, when the oxygen concentration in the tissue rises above a certain level, the pre-capillary and metarteriole sphincters presumably would close until the tissue cells consume the excess oxygen. But when the excess oxygen is gone and the oxygen concentration then falls low enough, the sphincters would open once more to begin the cycle again.”

Closed capillaries provide a reserve flow capacity and can open quickly in response to local conditions such as higher metabolic rates, a fall in pO2 or a fall in glucose when additional flow is required. Additionally, the amplitude of blood flow can also be sensitive to external stimuli such as ambient temperature and pain, and internal stimuli such as exercise and psychological stress.

Lower flow in the capillaries will lead to greater exchange of nutrients and metabolites. Simplistically, a drop of blood moving slowly will have more time to lose glucose to the consuming tissue compared to a drop of blood moving quickly. In tissues like the heart, all capillaries are normally open to perfusion, but in skeletal muscle and intestine only 20% - 30% of capillaries are normally open^[18]. As an example, it is possible that only 70% of the forearm capillaries are flowing normally at any one time, and 30% have slower-moving blood that is being depleted of glucose and oxygen by diffusion into the cellular space. Lancing into such a location would produce glucose readings lower than both arterial and venous blood glucose since more glucose consumption would occur in areas with no flow. If the measurement technique were oxygen sensitive, then the measurement would also be lower because of oxygen consumption by the surrounding tissue. Ideally, blood collection from sites such as the forearm and thigh should target a highly perfused capillary bed, and either compensate for or be independent of temporal changes in blood flow. Research papers have not been found that investigate how glucose levels vary under these situations. Two published studies focusing on other measurement parameters noted that the capillary glucose level lags behind the venous level in returning to normal^[3,19]. In both these papers, a discernible lag was noted but no explanation was attempted.

Amira Medical™ conducted a time-based study measuring venous, fingerstick capillary, and forearm capillary blood glucose to determine if forearm capillary blood glucose values more closely follow fingerstick or venous blood. Ten individuals (5 type-1 and 5 type-2) were tested first under fasting conditions and then after ingestion of a 75-gram glucose load. Venous and capillary glucose values were monitored for a period of up to five hours. A 75-gram glucose load was chosen because it is standard in routine screening for diabetes and is more sensitive than blood glucose determinations after a meal high in carbohydrate or a mixed meal of carbohydrate and protein. After a mixed meal, postprandial insulin levels are higher and blood glucose levels are lower than after a glucose load since glucose and amino acids potentiate each other with respect to insulin release^[6]. It is possible that the amount of glucose time lag between blood compartments is dependent on the glucose load, but it was not studied in this experiment.

A set of blood samples consisting of duplicate venous blood samples, a single forearm capillary blood sample, and a single fingerstick capillary sample were collected within a 10-minute window and measured with the AtLast blood glucose system. A Glucometer Elite blood glucose system was also used in the study to measure venous and fingerstick capillary blood in duplicate but so closely matches the AtLast™ data that it is not graphed in the data sets to avoid clutter. Samples were drawn after an overnight fast at 10-minute intervals for 30 minutes under the fasting condition. After the ingestion of glucose, blood samples were again drawn at 10-minute intervals for up to five hours. At the beginning, middle, and end of the five-hour experiment, a venous and a fingerstick capillary sample were also collected and measured in duplicate on a laboratory YSI blood glucose analyzer. All blood samples were obtained from the subject while they were seated. A lag in glucose values over time were calculated using a peak-to-peak method. This was accomplished by fitting each blood compartment’s glucose values to a 6th order polynomial, determining the polynomials peak and comparing the peak times for venous, fingerstick, and forearm. The fingerstick capillary blood was collected from each subject an average of 0.8 minutes after the venous blood and the forearm capillary blood was collected from each subject an average of 3 minutes after the venous blood. This should be kept in mind when reviewing the data because these time lags were not subtracted from the peak lag time data presented in this paper.

Figures 2 shows the blood glucose variation achieved with three of the 10 subjects (Subject #1, a type-1 diabetic; Subject #5, a type-2 diabetic and Subject #7, a type-2 diabetic). These three graphs represent the range of time lags seen in the data. Using fingerstick blood glucose as the marker for the peak time lag, venous blood glucose lagged fingerstick blood glucose by 0 minutes, 6 minutes and -4 minutes for subjects #1, #5, and #7 respectively (a negative number means the peak fit shows venous blood leading fingerstick blood). Also, forearm blood glucose lagged fingerstick blood glucose by 28 minutes, 43 minutes, and 8 minutes for subjects #1, #5, and #7 respectively. YSI blood glucose measurements confirmed the accuracy of the AtLast blood glucose measurements at the beginning, middle and end of the tests. When the peak time lags for all 10 subjects were averaged, venous blood glucose lagged fingerstick blood glucose by 4.9 minutes on average and forearm blood glucose lagged fingerstick blood glucose by 16.2 minutes on average.

Used only with express written permission

Used only with express written permission

Used only with express written permission

Figure 2: Glucose values over time from venous, fingerstick and forearm blood compartments. Each graph shows data from a single subject.

Figure 3 shows three correlation graphs for rapidly changing glucose values using data from all 10 subjects. Figure 3a shows the correlation between venous and forearm capillary blood glucose measured with the AtLast blood glucose system that has 81.1% of values in the A-region utilizing Clarke Error Grid. Figure 3b shows a correlation between fingerstick capillary and forearm capillary blood glucose measured with the AtLast blood glucose system that has 73.2% of values in the A-region. For comparison, Figure 3c shows a correlation between venous and fingerstick capillary blood glucose measured with the AtLast blood glucose system that has 89.0% of values in the A-region. Data of steady-state glucose values (not shown) taken under fasting conditions gave >97.0% of values in the A-region for the three graphs.

Used only with express written permission

Used only with express written permission

Used only with express written permission

Figure 3: Glucose correlation between venous, fingerstick, and forearm blood compartments during rapidly changing glucose levels. Each graph is compiled using data from all 10 subjects in the study.

The study concluded that there is a lag between forearm capillary blood glucose and both venous and fingerstick capillary blood glucose. There is also significant variability in the length of lag between people in the study group. It is hypothesized that low blood flow in the forearm capillary bed is a contributing factor to the observed lag. The rate of glucose consumption, oxygen consumption and insulin concentration could also be playing a role in the effect, but their concentration variation will also correlate with low blood flow as well as diffusion resistance in the capillaries. The correlation graphs indicate that venous blood glucose is a better reference for forearm capillary glucose measurements in studies that use nonfasting subjects. However, this is not because forearm capillary blood is more like venous. Venous blood is 'down stream' from the capillaries and so better reflects any temporal decreases in the flow from the multitude of capillaries that feed the vein even as it is affected by the artery shunts and muscle glucose consumption.

Because local blood flow can be increased by either heat or injury, it may be possible to minimize or eliminate delay times by enhancing circulation. Friction rubbing, resistive heating, chemical heating, vibration, or repetitive lancing could increase local blood flow and so minimize glucose variation between blood compartments. Unfortunately, published data is not available on how any of these variables affect the lags between blood compartments and what minimal stimulus would be needed for improvement.

3b. Glucose test values may not match because of low blood flow in the fingers

Another contributing parameter that combines flow restriction and diffusion can be found in finger capillary blood, but in this physiological effect the flow stops almost completely. To restate, glucose consumption in the capillary system goes up as the flow decreases so low flow capillary blood samples can be depleted of their glucose supply. A familiar example would be hypothermia where long exposure to cold weather shuts down blood flow to the peripheral tissue. In published cases of restricted blood flow to the fingers, fingerstick capillary blood was not the clinically appropriate sample for monitoring blood glucose.

Many papers have been found in this category that indicates glucose measurement problems do occur. Shock or severe hypotension (systolic blood pressure of 80 mm Hg or less) are examples of clinical conditions that adversely affect the measurement of glucose in fingerstick capillary blood [20-22]. It is generally accepted that shock results from inadequate blood flow through the body resulting in limited delivery of oxygen and nutrients to the tissue cells. In the most complete study, only 36% of hypotensive patients had a fingerstick capillary glucose within +/- 20 % of the laboratory value and almost one third of patients were misidentified as hypoglycemic by the fingerstick method; two of these patients were actually hyperglycemic^[20]. All studies noted that the test strip measurement was accurate when using venous blood and compared to a venous laboratory measurement. Also, all studies recommended use of venous blood when a glucose test strip was used to determine glucose in hypotensive patients.

The administration of vasoactive drugs can influence capillary flow independent of the shock state. Not all patients in the two studies were on vasoactive drugs but up to 72% were. It has been documented that dopamine, a common vasopressor drug used in the intensive care unit, inhibits the glucose oxidase reaction on a test strip^[23]. However, since the above studies monitored glucose using the same instrument with both fingerstick capillary and venous blood, the dopamine effect would only come into play if there were a large dopamine concentration difference between the two blood compartments. One study^[20] was unable to show any relationship between the degree of fingerstick capillary glucose reduction and the use of intravenous dopamine. Unfortunately, dopamine blood concentration was not measured so this could still be a factor in the study results although it would not affect their conclusions.

Fingerstick capillary blood may also not be the clinically appropriate sample for patients in cardiac arrest, as a study showed it to be relatively nonspecific for identification of hypoglycemia in this patient population^[24]. In this study, the sensitivity and specificity of fingerstick capillary blood for detection of hypoglycemia were 75% and 38% respectively; whereas, test strip analysis of venous blood correctly identified all hypoglycemic patients (sensitivity of 100%), with no patients incorrectly categorized as hypoglycemic (specificity 100%). An explanation for the low fingerstick glucose readings was not found, but a combination of increased glucose use and decreased peripheral blood flow was attributed to be the most likely contributor. This study also concluded that test strip determination of blood glucose is reliable for cardiac arrest patients only if done on a sample of venous blood.

The major discrepancy in these studies between venous and capillary blood glucose measurements probably reflects continued glucose utilization by peripheral tissues in the presence of vascular stasis. This is likely caused by peripheral vasoconstriction with shunting of blood from the periphery and continued tissue glucose consumption. Neural regulation of skin blood flow includes the presence of arteriovenous anastomoses, which are highly innervated structures involved in thermoregulatory processes. These shunts provide a low-resistance pathway for blood flow where large volumes of blood can be partitioned to a superficial venous plexus, largely bypassing the nutritive capillaries of the skin. An attractive hypothesis is that diabetes may result in the loss of neural control of these vessels such that there is increased shunt flow creating a deficit in skin blood flow at the nutritive capillary level^[25].

Peripheral vascular disease or poor peripheral flow is likely to occur in patients when dehydrated, hypovolaemic, hypotensive or suffer from small vessel disease^[5]. Hyperosmolar hyperglycemia is another example of clinical conditions that adversely affect the measurement of glucose in fingerstick capillary blood. Circulation may also be compromised due to vasoconstriction from drug therapy, hypothermia, edema, diabetes, peripheral vascular disease, cardiovascular disease, or even hemodilution from cardiopulmonary bypass.

Conclusions

Simultaneous measurements of arterial and venous blood samples should produce different glucose values in healthy people due to glucose utilization by peripheral tissues. Unfortunately, the magnitude of this glucose difference cannot be predicted due to the large number of variables that affect it. Since capillary blood has been expanded to refer to blood collected from the finger, forearm, ear, heel, calf, and stomach, questions have arisen if each of these is predominantly arterial or venous. Published studies have justified equating arterial and fingerstick capillary glucose levels under most conditions but no other capillary blood source has been equally studied.

Local, rhythmic changes of blood flux within capillary beds play a larger role in the variation of forearm capillary blood glucose vs. fingerstick capillary blood glucose than the differences between arterial and venous values. It is not to say that forearm capillary is more like venous, but that the independent temporal changes in select capillary beds affect the venous value because it is upstream. An attempt should be made for blood analysis from sites such as the forearm to be either compensated for or to be made independent of temporal changes in blood flow. That said, venous blood glucose is a better reference for arm blood glucose than fingerstick capillary measurements. In the time course data sets (Figure 3) where glucose excursions were induced by Glucola™ (75 grams of glucose), a reference to venous blood glucose produced 7.9% more values in the A-region for the AtLast forearm capillary measurement than when compared with fingerstick capillary samples.

Although it appears that blood flow problems may be linked to the variation between forearm capillary glucose measurements and either arterial or venous blood glucose measurements, there are a number of other physiological parameters that can affect a glucose measurement from capillary blood sources. Ideally, it would be advisable to standardize the analytical chemistry method used to measure glucose and the blood compartment from which the sample is drawn, and to adopt a uniform method of blood collection. Unfortunately, such a worldwide standardization would stifle research and development into new, less painful glucose instruments since it would limit market acceptance of any technology that did not meet these standards. Therefore it is necessary to better understand the glucose variation in any biological fluid used to measure glucose and how they compare to more traditional glucose measurements.

The three physiological parameters presented in this paper could all occur simultaneously or one at a time. Glucose data collected from a single individual could show a bias only on the first blood measurement and a bias with lag on the second only a short time later. In comprehensive studies, other parameters will need to be measured (oxygen, blood flow, and others) to separate these factors and better understand glucose physiology. Additional studies will help clarify when the current glucose measurement technology will be most accurate and when it might be clinically unacceptable. It was noted in the Liu paper that heating the skin 'arterialized' the venous blood. Both heat and vacuum may stimulate the skin so that some of these physiological parameters are minimized, but currently this hypothesis has not been proven. Published studies have narrowed the areas needing further studies, but additional research is needed. It should remain an exciting area of research for years to come.

Acknowledgments

I wish to thank Uwe Kraemer for substantive discussions and Phil Stout, Michelle Delli-Santi, Gina Moss, and Anne Callahan with help in collecting the data sets at Amira Medical.

References

1. Eriksson KF, Fex G, and Trell E: “Capillary-venous differences in blood glucose values during the oral glucose tolerance test”, Clin Chem 1983;29(5):993.
2. Rasaiah B: Self-monitoring of the blood glucose level: potential sources of inaccuracy, Can Med Assoc J 1985;132:1357-1361.
3. Somogyi M: “Studies of arteriovenous differences in blood sugar: effect of alimentary hyperglycemia on rate of extranepatic glucose assimilation”, J Biol Chem. 1948;174:189-200.
4. Burtis CA, Ashwood ER (editors): Tietz Textbook of Clinical Chemistry, 2nd Editon, W.B. Saunders Company, 1994;959.
5. Wickham NWR, Achar KN, Cove DH: “Unreliability of capillary blood glucose in peripheral vascular disease”, Practical Diabetes 1986;3(2):100.
6. Fajan SS: “What is diabetes?”, Med Clin North Am 1971;55:793-805.
7. Chaisson KM: “Comparison of arterial and capillary blood glucose with the use of the Accu-Chek III”, Progress in Cardiovascular Nursing 1995;10(4):27-30.
8. Liu D, Moberg E, Kollind M, Lins PE, Adamson U and MacDonald IA: “Arterial, arterialized venous, venous and capillary blood glucose measurements in normal man during hyperinsulinaemic euglycaemia and hypoglycaemia”, Diabetologia 1992;35:287-290.
9. Guyton AC and Hall JE: “Local control of blood flow by the tissues; and humoral regulation”, In Medical Physiology, 10th edition, Philadelphia, W.B. Saunders Company 2000:175-182.
10. Ryan TJ: “Cutaneous Circulation”, In Goldsmith LA, ed. Biochemistry and Physiology of the Skin. Vol.II. New York/Oxford: Oxford University Press, 1983:817-877.
11. “Diabetics go home”, Lancet 1980;2:217.
12. Alberti KGMM and Skrabalo Z: “Standardization of biochemical methods in the diagnosis and management of diabetes”, IDF Bull 1982;27:17-45.
13. Larsson-Cohn U: “Differences between capillary and venous blood glucose during oral glucose tolerance tests”, Scand. J. Clin Lab Invest 1976;36:805-808.
14. Neely RDJ, Kiwanuka JB, Hadden DR: “Influence of sample type on the interpretation of the oral glucose tolerance test for gestational diabetes mellitus”, Diabetic Medicine 1991;8:129-134.
15. Gutman S: “Review criteria for Assessment of portable invasive glucose monitoring in vitro diagnostic devices which use glucose oxidase, dehydrogenase or hexokinase methodology”, FDA document, January 21, 1998.
16. Ervin, KR and Kiser EJ: “Issues and implications in selection of blood glucose monitoring technologies”, Diabetes Technology & Therapeutics 1999;1(1):3-11.
17. Schechner JS: “Blood glucose monitoring in the cutaneous micro-environment”, Diabetes Technology & Therapeutics 1999;1(1):39-40.
18. Granger DN: “Capillary Exchange”, In Johnson LR, ed. Essential Medical Physiology, 2nd edition, Philadelphia/New York: Lippincott-Raven Publishers, 1997:217-235.
19. Heght A, Weisenfeld S and Goldner MG: “Factors influencing oral glucose tolerance: experience with chronically ill patients”, Metabolism 1961;10:712-723.
20. Atkin SH, Dasmahapara A, Jaker MA et al: “Fingerstick glucose determination in shock”, Ann Intern Med 1991;114:1020-1024.
21. Sylvain HF, Pokorny ME, English SM et al: “Accuracy of fingerstick glucose values in shock patients”, Am J Critical Care 1995;4(1):44-48.
22. Sandler M and Low-Beer T: “Misleading capillary glucose measurements”, Practical Diabetes 1990;7(5):210.
23. Keeling AB, Schmidt P: “Dopamine influence on whole-blood glucose reagent strips”, Diabetes Care 1987;10:532
24. Thomas SH, Gough JE, Benson N, Austin PE, and Stone CK: “Accuracy of fingerstick glucose determination in patients receiving CPR”, Southern Med J 1994;87(11):1072-1075.
25. Stansberry KB, Hill MA, Shapiro SA et al: “Impairment of peripheral blood flow responses in diabetes resembles an enhanced aging effect”, Diabetes Care 1997;20(11):1711-1716.

All trademarks are the property of their respective holders.