The Variability of Blood Glucose in Type 1 Diabetes

Stefan Du Rietz  (e-mail: sdr (at) this domain)

Probability distribution of all blood glucose (BG) readings

Read more about the statistical method in my Introduction.

These three examples are from extremely well-controlled individuals with HbA_1c within the normal range and no serious hypoglycaemias.

SDR is a man, 55 years old, with type 1 diabetes for 49 years. After such a long time there is probably no remaining endogenous insulin supply. The only serious complication occurred around 1990: proliferative retinopathy which – after laser treatment and much tightened control with normalization of HbA_1c – disappeared and did not recur.

Instead, recurrent serious hypoglycaemias became a problem. Eventually, however, a method was developed to avoid them.

SDR makes eight regular BG measurements per day: before and after breakfast, lunch and dinner, before sleep and in the middle of the night. He takes four regular injections of insulin per day: rapid acting insulin (lispro) before breakfast, fast acting (regular) insulin or lispro (depending on the time of day) before lunch, regular insulin before dinner and NPH insulin before sleep.

If BG is too high he takes an additional injection of lispro and, if before a meal, he postpones and reduces the meal. After an unusually high BG reading (>10 mmol/l) he makes an additional measurement after one hour in order to estimate the BG rate of decrease and the optimal time of the postponed meal.

If BG it is too low he immediately takes additional carbohydrates of high glycaemic index. The same is true in case of neuroglycopenic symptoms between the BG measurements. If there is any uncertainty regarding the neuroglycopenia he makes an additional BG measurement.

SDR uses the ONE TOUCH Profile (LifeScan Inc.) that has proved to be the most accurate of portable BG meters, although the PubMed is somewhat misleading [1, fig. 1]. The accuracy is also maintained at low BG readings [2, fig. 2], which is very important for training of neuroglycopenia recognition (usually referred to as "hypoglycaemia awareness"). Computer download of BG data in mg/dl yields the highest resolution.

Fig. 1 shows the statistical distribution of all his regular BG meter readings during 1¾ year, 1998-03-15 to 1999-12-14 (640 days). Only 1 in 53.9 (less than 2%) of the regular measurements is missing. Evidently there is a great variability, despite the frequent BG measurements and subsequent control actions.

How to read the graphs.

Figure 1. Cumulative probability distribution of all BG readings (n = 5025):
GM = 5.12 mmol/l  FV = 1.47  FV⁴ = 4.62

AKE is a woman, 27 years old, with type 1 diabetes for 10 years. She also tests eight regular times per day, including in the middle of most nights (157 out of 252).

Fig. 2 shows the statistical distribution of all her regular BG meter readings during her second pregnancy, 1997-07-22 to 1998-03-30 (252 days). Only 1 in 14.6 (less than 7%) of these numbers is missing, most of them at night.

How to read the graphs.

Figure 2. Cumulative probability distribution of all BG readings (n = 1878):
GM = 5.96 mmol/l  FV = 1.36  FV⁴ = 3.45

JH is a woman, 27 years old, with type 1 diabetes for one year, i.e. still in the "honeymoon" phase. She tests four regular times per day when she takes insulin: before breakfast, lunch, dinner and sleep.

Fig. 3 shows the statistical distribution of all her BG meter readings during this first year, 1998-07-01 to 1999-06-29 (364 days) after the diagnosis. Only 1 in 45.5 (less than 2%) of these numbers are missing.

How to read the graphs.

Figure 3. Cumulative probability distribution of all BG readings (n = 1424):
GM = 4.55 mmol/l  FV = 1.33  FV⁴ = 3.15

Discussion

The variations in all BG readings of SDR, AKE and JH are quite large. The varying insulin requirement is not possible to optimally supply with injections of constant (or even adapted) insulin doses and the small remaining endogenous insulin production of AKE and JH is completely insufficient for meals. Only after one night fast (fasting BG) does it diminish their BG variability to any particular extent.

References:

1. Brunner GA, Ellmerer M, Sendlhofer G et al: "Validation of home blood glucose meters with respect to clinical and analytical approaches." Diabetes Care 21: 585-90, 1998. PubMed  Reprint (PDF)
2. Trajanoski Z, Brunner GA, Gfrerer RJ et al: "Accuracy of home blood glucose meters during hypoglycemia." Diabetes Care 19: 1412-5, 1996. PubMed

© 1999-2024 Stefan Du Rietz.  Last modified: 2022-06-21 00:41 (MEST).