Fun with a glucometer

Table of contents

The graphs on this page are somewhat interactive; you can mouse over different items for more detail. They're drawn with d3.js (initially based on its "line" example) and require a modern web browser that supports SVG.


Glucose is a simple sugar used by cells for energy. If the level of glucose in your bloodstream (also known as your blood sugar) gets too low, you're hypoglycemic; too high, and you're hyperglycemic. Your body (ideally) regulates your level via insulin, a hormone produced and released by beta cells in your pancreas that instructs your liver and your fat and muscle tissues to remove glucose from the bloodstream and store it as glycogen (a process called glycogenesis).

Very bad things can be caused by or are correlated with chronic hyperglycemia:

  • nerve damage (neuropathy), leading to desensitization, pain, erectile dysfunction, and gangrene
  • retina damage (retinopathy), leading to blurred vision and eventual blindness
  • kidney damage (nephropathy)
  • glucose toxicity, where the same beta cells that are needed to produce insulin to lower blood sugar are damaged by high blood sugar
  • increased risk of heart disease and cancer

Two common causes of hyperglycemia are type 1 diabetes, where the body is unable to produce sufficient quantities of insulin to lower its glucose level, and type 2 diabetes, where insulin is being produced but cells aren't responding properly to it.

Several methods exist to measure how well someone's body deals with glucose. The most common is the fasting plasma glucose test, where the subject avoids eating for a period of time (I've seen eight to twelve hours) before having their level measured. The American Diabetes Association currently classifies levels below 100 milligrams per deciliter as "normal", 100-125 mg/dL as suggesting "pre-diabetes", and 126 mg/dL and up as indicating diabetes. Before 1998, the guidelines given by the ADA were more relaxed; "normal" extended up to 110 mg/dL and diabetes didn't begin until 140 mg/dL.

Glucose tolerance tests are administered less frequently since they're more of a pain to do. After fasting, the subject drinks a glass of water containing 75 grams of dissolved glucose and their glucose level is repeatedly measured over the next few hours. The ADA says that a level lower than 140 mg/dL at the two-hour mark is "normal", 140-199 indicates pre-diabetes, and 200 and up indicates diabetes.

In Blood Sugar 101 and on her website, Jenny Ruhl gives convincing arguments that the fasting test is a significantly worse indicator of diabetes risk than the tolerance test (citing data from the Baltimore Longitudinal Study of Aging) and that the ADA's guidelines are far too conservative. She argues that damage can begin to occur at levels between 140 and 160 mg/dL, and that the process can get pretty far along before it's severe enough that it becomes apparent through the fasting test.

I was curious about my glucose levels, since:

  • My fasting level has been near the high end of the ADA's "normal" range (although falling, most recently) the last few times that it's been measured by a lab.
  • I sometimes have headaches in the morning and was wondering if they could be caused by low blood sugar (see this list of hypoglycemia symptoms).
  • I've never had my glucose tolerance (as opposed to my fasting level) measured.
  • I like to measure things.


Bayer Contour glucometer
The glucometer itself

Glucometers are devices that measure glucose levels, usually via enzymes that react to the glucose levels in small blood samples. I ordered Bayer's "Contour Blood Glucose Monitoring System" from Amazon for $13.57 (which, given the device's list price of $80, qualifies as a "eh, it must've just fallen off the back of a truck" discount — I suspect that someone's getting ripped off here, but I'm not sure who). In 2016, the part of Bayer that made Contour devices was apparently purchased by PHC Holdings (formerly Panasonic Healthcare Holdings) and rebranded as Ascensia.

For my thirteen bucks, I got an electronic device that analyzes blood samples and stores the results, a Bayer Microlet 2 lancing device, ten non-reusable test strips, ten non-reusable (but if you're doing repeated testing over a short period, using a new one each time seems a bit wasteful...) lancets (pins for the lancing device), a small bottle of control solution containing a known amount of glucose for testing the device's accuracy, a black carrying case, and a couple of manuals. I paid $25 for fifty more test strips and $17.77 for two hundred more lancets ( I dunno, I just like needlesthe smallest quantity that I could find).

Bayer Microlet 2 lancing device
Needle comes out of the top

The lancing device is a small, plastic, spring-loaded gun that you use to shoot a pin into your finger to produce a drop of blood. There's a dial on top for adjusting the pin's depth. I found that the shallowest setting didn't pierce my skin, but somewhere around the middle worked fine. In the interest of science, I also bravely tried the deepest setting once, but I didn't notice any difference in terms of pain or blood — the lancets are really, really tiny. The basic idea is to stick a new test strip in the glucometer, load up the lancing device with a fresh lancet and use it to prick a finger to produce a drop of blood, and then stick the business end of the test strip into the blood so it can suck it up.

Boring paragraph involving metric units: Before drawing conclusions from the readings given by a glucometer, note that labs in the U.S. measure glucose levels in plasma, while home-testing devices analyze whole blood (unless you own a centrifuge). Levels in plasma are 10-15% higher than those in whole blood; the rule of thumb that I've seen is to multiply a whole blood measurement by 1.12 to get a corresponding plasma measurement. However, most recent glucometers (including my Contour) already do this adjustment for you. Check the specifications in the manual before comparing your own numbers with those from a lab. (Note also that much of the rest of the world uses mmol/L (thousandths of a mole per liter) instead of mg/dL. To convert mmol/L of glucose to mg/dL, multiply by 18.)

Foods' effects on glucose levels

Glucose typically comes from carbohydrates in the diet, although it can also be produced in the liver from other sources (gluconeogenesis). People on very-low-carbohydrate diets like the Atkins diet are often trying to enter ketosis, a state where the body effectively uses stored fat as an energy source. Glycemic index describes the speed with which a food's carbohydrates break down and release glucose into the bloodstream (higher means faster), while glycemic load multiplies a food's GI by its quantity of carbohydrate to describe the actual effect on glucose levels.

My day-to-day diet is usually pretty low-carbohydrate (although not ketogenic). For my first test, I figured I'd check my fasting level, then see what a glass of whole (raw!) milk does to me, and then eat a normal (for me) breakfast of eggs and bacon:

The glucometer reported a drop to 55 mg/dL soon after I drank the milk, which I wasn't expecting at all: milk contains lactose, a sugar that's made up of galactose and glucose — it should've made my blood sugar go up, not down. After breakfast, my level jumped up between 100 and 110 (likely due to the milk) before rebounding a bit below the initial level and stabilizing.

A brief aside on accuracy: apart from the post-milk reading listed above, the device produced plausible numbers. I tested my results after drinking milk a second time on a different day and got the results more in line with what I expected: a small initial spike, then a drop, and then a second residual increase:

I was also surprised that my fasting level was so much lower than what I'd seen in previous lab tests. When I used the control solution to test the glucometer's accuracy, I saw 123 mg/dL (expected range: 105-145) with the first set of test strips and 124 mg/dL (expected range: 106-147) with the second. The next time that I have a blood sample drawn, I'll probably take my own readings before and after to see how the numbers compare.


Since my fasting level as reported by the device was quite low, I was getting really curious about the hypoglycemia-headaches thing I mentioned earlier. I decided that I'd measure my level just before I went to bed, halfway through the night, and after waking up:

Sure enough, it fell steadily throughout the night. Ideally, my liver should have enough glycogen stored up to keep my blood sugar at a stable level while I sleep. Since then, I've experimented with consuming various snacks before bed or upon waking up in the middle of the night, but with limited success:

Eating food containing carbohydrates but with a low glycemic index (raspberries, with a glycemic index between 30 and 40, and milk, around 30) just before bed appears to have propped up my level for a few hours, but it dropped back to my normal morning level even when I got up a bit after 4 a.m. to eat more. In addition, I found that eating just before bed made it even harder for me fall asleep.

plain bagel on a scale

Bagels' effects on glucose levels

Finally, I wanted to see what happened when I ate food with a fairly high glycemic index and glycemic load. Random people on the Web suggest that the glucose solution typically used for testing is disgusting and that you can use a plain bagel (with no toppings, as fat may slow absorption) for home-testing.

I bought a 60-cent plain bagel from the corner store. Per the USDA's national nutrient database (plug: I've created an easy-to-use-site for viewing nutrition data), a 115-gram plain, enriched bagel contains 39 grams of water, 60 of carbohydrate (43 starch, 10 sugars, 2 fiber), 12 of protein, and 1.5 of fat. With a glycemic index above 70 and a glycemic load around 40, this bagel is considered a high-GI, high-GL food.

This particular experiment was carried out with less rigor than I would've liked; in the interest of time, I ate the bagel after the second milk test described earlier rather than starting from a fasting state (my level had fallen close to its initial fasting level, for what it's worth).

The increase was pretty dramatic, although less steep than I was expecting. I would've liked to continue testing (well, as much as one can like shooting pins into one's fingers) to see how long the level took to fall back to its initial state, but I had a prior obligation.


My body appears to be able to keep its blood sugar at safe levels given the types of food that I eat day-to-day. The difference after eating a high-calorie, high-fat, low-carbohydrate breakfast (the eggs and bacon) versus a low-calorie, low-fat, high-carbohydrate one (the bagel) was striking, and the ADA's advice to diabetics to eat a low-fat, high-carbohydrate diet leaves me puzzled although looking at their site again in 2020, it seems like it's been updated to be more reasonable). Seeing that much of the bagel's glucose still hadn't been dealt with after two hours gave me pause, but I suspect that the previously-drunk milk slowed digestion of it. If not, well... I don't eat many bagels outside of glucose tests.


I repeated the bagel test from a fasting state and saw a much lower peak. Assuming that the light sprinking of sesame seeds on top of the second bagel didn't affect things too much, it seems likely that the milk that I consumed three-and-a-half hours before the previous test had an effect after all.