Thanks for those insights
@saz9961. I have often wondered about the science behind finger prick testing and your description goes a long way to satisfying that.
In terms of random errors I have also tended to think about sampling errors. A tiny drop of blood is taken from a large dynamic system and it is assumed that the value obtained from the drop represents the whole. Do you think that is reasonable and do you think that error is smaller or larger than the system measurement errors you have described.
Its not what I think, but established fact;
34 years ago;
In most hands, the glucose oxidase strip method is accurate and reliable. Since whole blood is used, the results tend to be slightly lower than simultaneous venous samples, but this is balanced by the fact that capillary blood has a higher glucose concentration than venous blood.
The concentration of glucose is highest in the arterial circulation. Laboratory determinations are usually done on venous samples. If the venous circulation is delayed, such as by leaving a tourniquet on for a prolonged period of time, the concentration falls even further. Thus, samples should be obtained after releasing the tourniquet. Studies have shown that blood glucose concentration may fall as much as 25 mg/dl when a tourniquet has been left in place for 6 minutes.
Glucose concentration may be determined in whole blood, plasma, or serum samples. If whole blood is used, the concentration will be lower than if plasma or serum is used. This is due to the greater water content of the cellular fraction. Under usual circumstances, the concentration of glucose in whole blood is about 15% lower than in plasma or serum
It is known that glucose levels in capillary blood in the fingertip after a liquid glucose load are constantly higher when compared with venous blood measurements (
1). Recently alternative sites for capillary blood drawing (e.g., forearm) have been proposed (
2) that are less painful compared with fingertip. Data have shown that there was no significant difference between the capillary blood drawn from forearm and fingertip in diabetic patients with glucose values in a wide range (
3). Nevertheless, some data have shown that glucose results from alternative sites and fingertip were not identical. This difference was more pronounced when there was a rapid increase or decrease of blood glucose values (
4).
These findings confirmed the already reported observation that up to 3 h after a liquid glucose load, capillary finger glucose levels are constantly higher (15–26%) than venous glucose levels. On the contrary, forearm glucose levels were closer to venous plasma glucose levels: There was no significant difference between them after 1 h, whereas a significant increase of 16% appeared at 2 and 3 h. These findings are in accordance with the concept of slower glucose kinetics at the forearm than the fingertip due to lesser arteriovenous anastomoses (
4). To be sure, this physiological difference needs to be taken into consideration in the detection of hypoglycemia in diabetic patients. However, it is precisely this physiological difference that supports the suggestion that capillary forearm glucose measurements using a portable glucose meter may be useful for the 50-g challenge test for gestational diabetes screening in an outpatient environment.
So there is wide variation. But its irrelevant. The ideal ranges are established based on finger prick testing, not on thigh testing, forearm testing etc. The CE mark on the meter (or UKCA) has specific meaning. Its a conformance marking. The glucose meters are medical devices, so that means conformance to ISO-15485, which is all about defining how a medical device is designed, made, and an adverse event process put in place. But home glucose meters also have an additional standard, to get that CE/UKCA mark; ISO-15197.
95% of results have to be within 0.83mmol/l for measurements below 5.5mmol/l and 15% for measurements above 5.5mmol/l.
So, not surprisingly, less accurate at low concentrations. For all intents and purposes, the readings might as well be +/- 1mmol/l. Round up/down
So @ a recorded 5.0mmol/l, the actual blood glucose might be 4.2-5.8mmol/l. At 7mmol/l, it might be 5.95-8.05mmol/l, but in practice, that inaccuracy will reduce with higher concentrations. 115% of 5.5 is 6.3mmol/mol.
So yesterdays waking concentration (fasting or whatever you want to call it) was 4.5 (5). A few hours after a 400kcal dinner of beef stew and green beans, it was 5.4 (5). Waking up this morning, 5.1 (5). I pigged out this morning. Instead of a sickly shake, and faced with the ridiculousness of a powdered bacon omelette, I decided to make a spanish omelette; a couple of eggs, pepper, onions, bacon, fried up in butter, slipped in some chillies and slivers of half a cooked new potato that was sitting in the fridge. 2 hours later, 5.7 (6). Oops. But then, 2h after a grim bowl of instant veg soup, it was 6.2 (6).
I'm no longer going to sweat it over 0.5mmol/l variations, as its meaningless.
