Personalized Approach for the Management of Exercise-Related Glycemic Imbalances in Type 1 Diabetes: Comparison with Reference Method

Abstract

Background:

One of the most frequently adopted strategies to counterbalance the risk of exercise-induced hypoglycemia in patients with type 1 diabetes is carbohydrates supplement. Nevertheless, the estimation of its amount is still challenging. We investigated the efficacy of the personalized Exercise Carbohydrate Requirement Estimation System (ECRES) method compared to a tabular approach to estimate the glucose supplement needed for the prevention of exercise-related glycemic imbalances.

Method:

Twenty-six patients performed two one-hour constant intensity exercises one week apart; the amount of extra carbohydrates was estimated, in random order, by the personalized ECRES method or through the tabular approach; glycemia was determined every 30 minutes. Continuous glucose monitoring (CGM) metrics were calculated over the 48 hours preceding, and the afternoon and night following the trials.

Results:

Applying the personalized ECRES method, a significantly lower amount of carbohydrates was administered to the active patients compared to the tabular approach, median (interquartile range): 9.0 (0.5-21.0) g vs 23.0 (21.0-25.0) g; P < .01; the two methods were similar for the sedentary patients, 18 (13.5-36.0) g vs 23.0 (21.0-27.0) g; P = NS. After overlapping CGM metrics before the exercises, both methods avoided hypoglycemia and resulted in similar glucose levels throughout them. The ECRES method led to CGM metrics within the guidelines for either the afternoon and the night just following the trials, whereas the tabular approach resulted in a significantly greater time below range in the afternoon (11.8% ± 18.2%; P < .05) and time above range during the night (39.3% ± 29.8%; P < .05).

Conclusions:

The results support the validity of the personalized ECRES method: although the estimated amounts of carbohydrates were lower, patients’ glycemia was maintained within safe clinical limits.

Introduction

Physical activity and nutrition play an important role in the management of type 1 diabetes (T1D). Indeed, in subjects suffering from T1D, a healthy lifestyle including habitual physical activity contributes to enhance insulin sensitivity, reduces the risk of cardiovascular disease, may help to achieve and maintain better metabolic control, and improves psychological well-being.^1-4 Regular moderate-intensity physical activity should thus be encouraged in T1D subjects.^5-7 Nevertheless, glucose control is challenging and the risk of exercise-induced hypoglycemia represents one of the major barriers to physical activity in these patients.^8-10 On the other hand, excess carbohydrate intake may lead to an excessively high glycemia.¹¹

Existing guidelines for minimizing the risk of hypoglycemia during exercise are still inexplicit and require patients to determine his/her exercise-related self-management strategy by trial and error. To overcome this gap, Perkins and Riddell proposed a tabular approach that summarizes the amount of carbohydrates needed to avoid hypoglycemic hazard for a series of different types of activity and according to a few possible values of patient’s body mass.¹² To the best of our knowledge, this is one of the rare approaches accounting for a few important parameters. Nevertheless, other crucial factors such as patient’s insulin sensitivity and the state of the patient’s physical conditioning are additional factors that require consideration,^13-15 as well as the time distance from the last insulin administration.¹⁶

A personalized technology for patients affected by chronic conditions, such as T1D, might enhance self-management and improve patients’ lifestyle and well-being.^13,17 In this context, a patient-tailored and exercise-specific method able to estimate the amount of glucose supplement required to maintain safe blood glucose levels during physical activity was proposed and preliminarily tested during short^18,19 and prolonged exercise.^20,21 The Exercise Carbohydrate Requirement Estimation System (ECRES) algorithm is based on patient’s habitual therapy and diet and accounts for his/her insulin sensitivity. In addition, actual exercise intensity and duration, patient’s physical fitness level, and the timing of the activity relative to the last insulin bolus are parameters used by the algorithm, allowing to estimate the carbohydrates supplement for whatever exercise is performed and at any time of day.

We aimed at evaluating the efficacy of the proposed personalized ECRES method for the management of exercise-related glycemic imbalances in T1D people in comparison to the aforementioned tabular method of Perkins and Riddell,¹² assumed as reference method.

Methods

Participants

Patients were recruited at the Diabetes Center District 3, Azienda Sanitaria Universitaria Giuliano Isontina, Trieste, Italy. Inclusion criteria were individuals with T1D, 18 to 65 years old, both male and female, body mass index between 18 and 32 kg/m², diagnosed >24 months, on multiple daily injections with insulin Degludec (Novo Nordisk A/S, Bagsvaerd, Denmark) as basal insulin, glycated hemoglobin (HbA1c) in the range 42 to 75 mmol/mol (6%-9%). Eligible patients were thoroughly informed about the experimental protocol, and, subsequently, gave written informed consent prior to participation.

According to the self-reported physical activity habits, patients were classified as “active,” who exercised regularly for at least 30 minutes for three days/week, and “sedentary,” who exercised only occasionally.

The experimental procedures were approved by the Institutional Review Board of the Friuli-Venezia Giulia Region, Italy (No. 365 issued on September 3, 2015). All procedures were performed according to Good Clinical Practice and conformed to the standards set by the Declaration of Helsinki.

Experimental Protocol

In an outpatient setting, each participant performed two one-hour constant intensity treadmill walks/runs starting four hours after midday insulin treatment. Half the subjects were randomly allocated to first perform the exercise with the ECRES algorithm^19,22 to determine the amount of carbohydrates needed to avoid exercise-induced hypoglycemia, and seven days later, the same exercise applying the reference method,¹² while the other half performed the two exercises with opposite order (first exercise with reference, the second one with ECRES method). The amount of carbohydrates supplement calculated by one of the two methods was administered half an hour before the start of the exercise in the form of liquid dietary supplement (Glucosprint, Harmonium Pharma, Milan, Italy) containing d-glucose, vitamin B1, and manganese. In case of an excessive drop of glycemia during the exercise, the latter would be stopped and the same dietary supplement would be used to restore safe glucose levels.

Patients were instructed to maintain their usual ther-apy (ie, diet and insulin regimen) and to control their blood glucose levels according to the self-management procedures.

Reference Method

The carbohydrate requirements for a series of different activities and for a few possible patient’s body weights are reported in standardized tables.¹² According to our previous experience,¹⁹ a walking speed of about five kilometers per hour elicits a heart rate (HR) corresponding to about 65% of the maximal one. The values reported in the table of Perkins and Riddell¹² for a few values of patient’s body mass allowed us to define the following equation to estimate the required amount of carbohydrates (reqCHO) to avoid exercise-induced imbalances for one-hour walking at the selected speed (five kilometers per hour), according to the specific patient’s body mass:

$$\mathrm{reqCHO}=0.31\mathrm{g}/\mathrm{kg}·W+1\mathrm{g}$$ (1)

where W is the patient’s body mass expressed in kg.

The ECRES Algorithm

The ECRES algorithm, based on a series of models and relations described previously,^13,14,19,23 estimates in a personalized manner the amount of carbohydrates supplement (reqCHO) needed to avoid exercise-related glycemic imbalances. The algorithm is summarized in Figure 1.

Figure 1.

Schematic representation of the personalized ECRES algorithm. The following data have to be provided as inputs to obtain reqCHO (from top to bottom): (a) patient’s fitness level, average heart rate, and duration foreseen for the specific exercise; (b) patient’s usual therapy (ie, insulin types, doses, and time scheduling, together with the dietary carbohydrates) and time of day of exercise; (c) age, weight, and capillary blood glucose level measured just before the start. ECRES, Exercise Carbohydrate Requirement Estimation System.

reqCHO is calculated as a percentage (Pt%) of the total amount of carbohydrates burned during the exercise (CHOox) corrected by the excess or lack of glucose contained in the extracellular fluid compartment (Gb):

$$\mathrm{reqCHO}=\left(\mathrm{CHOox}·\mathrm{Pt}\%\right)\pm \mathrm{Gb}$$ (2)

CHOox is calculated as the product of exercise duration (exD) and the whole-body carbohydrate oxidation rate (CHOoxRate) estimated by models depending on the patient’s fitness levels (sedentary, active, or trained) and taking into account the expected exercise intensity expressed by the average expected HR^14,23:

$$\mathrm{CHOox}=\mathrm{CHOoxRate}(\mathrm{HR},\mathrm{fitness}\mathrm{level})\cdot \mathrm{exD}$$ (3)

Pt% depends mainly on and is almost proportional to the prevailing insulin concentration (IC) throughout the exercise. This is calculated over the specific exercise time period from IC(t) profile which, in turn, is estimated on the basis of the patient’s therapy data, accounting also for possible changes in insulin dose, and on standard pharmacokinetic profiles of the insulin analogs loaded in the system. The IC(t) profile is corrected for the patient’s insulin sensitivity S(j) of the specific day period (ie, j = morning, afternoon, or evening), estimated from the patient’s usual dietary carbohydrate-to-insulin ratio:

$$\mathrm{Pt}\%=f\left(\mathrm{IC}(t),S\left(j\right)\right)$$ (4)

Finally, Gb correction is obtained from the difference between the actual capillary glucose level measured before the start of the exercise (actual measured glycemia; aGL) and the theoretical glycemia, theoGL(t), the patient should have according to the time elapsed from the last meal and insulin bolus:

$$\mathrm{Gb}=\left(\mathrm{aGL}-\mathrm{theoGL}\left(t\right)\right)·\mathrm{ECF}$$ (5)

where ECF is the volume of the extracellular fluid compartment.

Data Acquisition and Analysis

One week before the first experimental session, patients attended the facility of the Diabetes Center District 3 (Trieste, Italy), where they were equipped with a continuous glucose monitoring (CGM) system (Dexcom G4, Dexcom Inc., San Diego, California, USA), which was removed 48 hours after the completion of the second experimental session. Patients were instructed to use self-monitoring of blood glucose measurements to calibrate the CGM devices, as per manufacturer’s specifications. The average blood glucose level and glycemic variability (expressed as the coefficient of variation) and percentage times metrics for CGM, namely, time within target glucose range (TIR; 3.9-10.0 mmol/L; 70-180 mg/dL), time below target glucose range (TBR; <3.9 mmol/L; <70 mg/dL), and time above target glucose range (TAR; >10.0 mmol/L; >180 mg/dL),²⁴ were calculated over the 48 consecutive hours before the exercises. In addition, the same parameters were calculated over the afternoon just after the exercise (from 18:00 to 22:59) and the subsequent night interval (from 23:00 to 07:00) as well as for the same time periods of the day before.

During the experimental sessions, HR was measured by a HR monitor (Polar, Kempele, Finland). Velocity and/or slope were measured by the treadmill and adjusted in order to maintain the HR on the target of 65% of patient’s theoretical maximal HR, in turn calculated as HRmax = 220 – age (expressed in years). All these variables were subsequently averaged over 15-minute intervals.

Venous and capillary blood withdrawals were performed 30 minutes before (−30′), just before the start (0′), at the middle (30′), at the end (60′) of the exercise, and one hour thereafter. Capillary glucose concentrations were measured immediately using a handheld glucometer (Accu-Chek Aviva, Roche Diagnostics, Indianapolis, IN, USA). Venous blood was collected into a two-milliliter Vacutainer tube (#368920) containing a glycolysis inhibitor (four milligrams of kalium oxalate + five milligrams of sodium fluoride); immediately after the collection, the tube was gently inverted and stored at plus four degree Celsius until the hospital laboratory centrifuged the samples and performed the measurements. Plasma glucose concentration was determined by applying a hexokinase-based methodology (Olympus Diagnostic Systems AU2700, Tokyo, Japan), with a coefficient of analytical variation determined in the laboratory of <2% in the range of 3.27 to 11.67 mmol·L⁻¹ (59-210 mg/dL). Glycemia in the range from 3.9 to 10.0 mmol·L⁻¹ (70-180 mg/dL) was defined as being on target.

Amounts of carbohydrates administered before the start and/or during the exercise were recorded in detail.

Statistical Analysis

Statistical analyses were performed using the Systat 13.2 software; if not specified otherwise, results were expressed as means ± SD and the significance level was set to P < .05.

Unpaired or paired t-tests (two tails) were appropriately used to assess differences between two groups of data (eg, active versus sedentary patients, CGM metrics before and after the exercise). One-sample t-test (one tail) was used to assess the difference between the CGM metrics and the corresponding guidelines.

Analysis of variance with repeated measures was used to detect differences in glucose levels throughout the trials, with time as within-subjects factors (Time effect) and the two different methods to estimate the required carbohydrates as between-subjects effect (Method effect). Mauchly’s test was used to check if the sphericity assumption appeared to be violated; where Mauchly’s test was significant, the Huynh-Feldt ε was used to adjust the degrees of freedom. Post hoc pairwise comparisons between levels of within-subjects factor were used to detect significant differences inside the within-subjects effects.

Nonparametric Mann-Whitney U test was used to detect significant differences in the amounts of required carbohydrates as estimated by the ECRES method and by the reference method.

Results

Participants

Overall, 26 patients participated in the study. All of them had no evidence of diabetes complications contraindicating physical activity. Anthropometric characteristics, together with diabetes and exercise-related parameters, are summarized in Table 1 for the active and sedentary patients and for the whole group. None of the reported parameters was statistically different between the active and sedentary patients.

Table 1.

Anthropometric, Clinical, and Exercise Characteristics for the Study Groups. Values Are Means ± SD.

	Active	Sedentary	All
Male/female	14/5	2/5	16/10
Age (years)	43.8 ± 12.4	38.0 ± 15.6	42.3 ± 13.3
Body mass (kg)	70.0 ± 12.9	73.5 ± 17.4	70.9 ± 14.0
Stature (m)	1.72 ± 0.11	1.71 ± 0.11	1.71 ± 0.11
Body mass index (kg/m2)	23.3 ± 2.2	25.3 ± 6.4	23.9 ± 3.8
Diabetes duration (years)	17.1 ± 14.4	17.4 ± 12.8	17.2 ± 13.7
HbA1c (mmol/mol)	54.0 ± 6.8	59.6 ± 6.6	55.5 ± 7.1
(%)	7.1 ± 0.6	7.6 ± 0.5	7.2 ± 0.6
Daily insulin dose (IU/day/kg)	0.47 ± 0.14	0.55 ± 0.17	0.49 ± 0.15
Heart rate at rest (bpm)	74.8 ± 11.5	76.5 ± 9.9	75.3 ± 10.9
Heart rate during exercise (bpm)	107.9 ± 11.8	117.0 ± 9.8	110.3 ± 12.0
Walking speed (km/h)	5.1 ± 0.9	5.3 ± 0.8	5.3 ± 0.8
Inclination (%)	4.2 ± 1.6	4.1 ± 2.6	4.1 ± 2.3

Experimental Conditions

The CGM metrics, ie, percentage times in range (TIR), above range (TAR), and below range (TBR), calculated for the afternoon and for the night just before each of the experimental sessions were not significantly different between the two sessions (Figure 2, left panel), as well as the metrics over the longer consecutive period of 48 hours before the trials. None of the average percentage times was significantly different from the corresponding guidelines (ie, TBR <4%, TIR >70%, and TAR <25%), although a great variability among patients was observed.

Figure 2.

Average percentage of times in range (TIR), above range (TAR), and below range (TBR) during the afternoon just after the exercise (from 18:00 to 22:59) and the subsequent night interval (from 23:00 to 07:00) as well as for the same time periods of the day before the experimental sessions carried out by using ECRES or the reference method to prevent exercise-induced hypoglycemia.

*significantly different from the guidelines for patients with type 1 diabetes. ECRES, Exercise Carbohydrate Requirement Estimation System.

The number of patients who experienced hypoglycemia during the 24 hours before the experimental sessions was 11 and 12 (out of 26) for the trials performed preventing exercise-induced hypoglycemia with ECRES and with the reference method, respectively, with the lowest observed glycemia amounting to 40 mg/dL.

No significant difference was observed in the exercise intensity between the two experimental sessions, either according to their sequence or according to the method used for the prevention of the possible exercise-induced hypoglycemia (P = NS for both comparisons). No significant difference was detected between active and sedentary patients (Table 1).

Exercise Tests

Cases of hypoglycemia were not observed during and in the one hour just after the exercises, either by applying the ECRES algorithm or by applying the reference method. On the other side, too high glucose levels (>10 mmol/L; >180 mg/dL) were observed at the end of the exercise in 4 out of 26 subjects (all active) during the trials managed with the reference method, with other 2 (1 active, 1 sedentary) showing an increase in glycemia above the same threshold in the following one hour; a too high glycemia occurred in three cases (two active, one sedentary) during the trials managed with ECRES, with one active patient showing an increase after the effort.

Mean (±SD) venous blood glucose levels measured before (ie, before carbohydrates administration), at the start, during, and after the trials are illustrated in Figure 3 together with the mean CGM values during the experimental periods. No significant difference in venous blood glucose levels was detected at any time point between the trials managed with the ECRES algorithm and with the reference method (Method effect, F = 0.244, P = NS). Conversely, the analysis of variance showed a significant effect of time (Time effect, F = 3.31, P < .05); post hoc pairwise comparisons highlighted a significant increase of glycemia from time −30′ (ie, the time point before the administration of the estimated carbohydrates) to the start of the exercise (P < .01); afterward, glycemia remained quite stable until the half of the exercise (P = NS between 0′ and 30′) and then decreased significantly by the end of the exercise (P < .01 between 30′ and 60′); glycemia again increased during the one-hour period after the end of the exercise (P < .01 between 60′ and 120′).

Figure 3.

Mean (±SD) venous glycemia before the administration of the carbohydrates supplement, at the start, middle, and end of the exercise and one hour thereafter. Average CGM traces obtained in the two experimental sessions are also illustrated.

Dots and continuous lines: ECRES method. Diamonds and dotted lines: reference method. Horizontal lines correspond to the safe glycemic limits. Shaded area represents the exercise period. CGM, continuous glucose monitoring. ECRES, Exercise Carbohydrate Requirement Estimation System.

Although blood glucose levels overall were not statistically different, Figure 3 shows that average glycemia was somewhat higher before the trial conducted with the ECRES method compared to the reference method. Nevertheless, by the end of the exercise, glycemia was very similar and within the safe clinical range when applying both approaches. As a consequence, the observed difference between glycemia before the exercise (time −30′) and its end (time 60′) suggests a decrease in glycemia toward the optimal values for the ECRES method (−0.78 mmol/L; −14 mg/dL), and a slight increase for the reference method (+0.22 mmol/L; +4 mg/dL).

Figure 4 illustrates the amounts of extra carbohydrates administered to prevent the possible exercise-induced hypoglycemia according to the two investigated methods. For the active patients, the amounts of extra carbohydrates calculated by the ECRES method differed significantly from those calculated with the reference method (9.0 g IQR 0.5-21.0 g vs 23.0 g IQR 21.0-25.0 g; P < .01), whereas the two quantities were similar for the sedentary patients (18 g IQR 13.5-36.0 g vs 23.0 g IQR 21.0-27.0 g; P = NS). Taking together the two groups, the ECRES algorithm estimated a significantly lower amount of carbohydrates as compared to that suggested by the reference method (median 13.5 g IQR 1.5-21.0 g vs 23.0 g IQR 21.0-25.0 g; P < .01).

Figure 4.

Box and whiskers plot of the amounts of carbohydrates administered to patients to prevent the possible exercise-induced hypoglycemia, calculated according to the ECRES method or the reference method. For the active patients, the amounts of extra carbohydrates calculated by means of the ECRES method were significantly lower than the corresponding amounts calculated with the reference method. This significant difference was not observed in the sedentary patients.

Whiskers are 1.5 × IQR. ECRES, Exercise Carbohydrate Requirement Estimation System.

CGM Metrics of the Period Following the Exercise

Figure 2, right panel, illustrates the CGM metrics obtained from the recordings of the afternoon hours just after the exercise and during the following night interval (23:00-07:00). No significant differences were observed for each of the calculated CGM parameters between the values obtained for the experimental sessions with the two methods to prevent exercise-induced hypoglycemia. In addition, no significant difference was found by comparing these CGM metrics with those obtained for the corresponding time periods in the 24 hours before the exercise. Nevertheless, following the trials performed using the reference method, the average percentage time below range (TBR) was significantly greater than the guidelines (P < .05) in the afternoon, whereas the time above range was significantly greater than the guidelines during the night (P < .05).

In the hours after the trials conducted with the ECRES method, 7 patients (out of 26) experienced hypoglycemia during the afternoon and 8 patients became hypoglycemic during the night, whereas the trials conducted with the reference method resulted in 10 and 4 patients experiencing hypoglycemia during the afternoon and during the night, respectively.

Discussion

The main finding of this study is that the proposed model based on a personalized approach resulted in overlapping glucose levels at the end of the exercise compared to the tabular approach,¹² assumed as reference method, despite lower amounts of carbohydrates required by T1D patients to avoid exercise-induced hypoglycemia were estimated.

An extra carbohydrate supplement alone might not always be the optimal strategy to prevent exercise-induced hypoglycemia, in particular when the physical activity is performed also with the aim of weight management. The ECRES algorithm estimating the required amount of carbohydrates likely minimizes the drawbacks of an unnecessary excessive carbohydrate snacking. To overcome this problem, insulin dose adjustments can be applied,^12,25 either as basal rate adjustment (applicable only in insulin pump users) or as mealtime bolus reduction (meaningful only for early-onset exercises, ie, within two hours from the meal). The latter is also allowed by the ECRES algorithm, which estimates the extra carbohydrates according to the reduced insulin dose.²²

The reference method presented on average a worse glycemic response compared to the ECRES method, although similar metabolic conditions were detected through CGM during the period before the experimental sessions. Indeed, the significantly higher amount of carbohydrates advised by the reference method resulted in a higher number of patients experiencing hyperglycemia in the one hour just after the exercise and, at the same time, in a percentage time below the hypoglycemic threshold significantly higher than the guideline during the afternoon after the exercise.

The results obtained in the present study, in particular the lower amounts of estimated carbohydrates for the active patients, are likely due to the tendency toward a personalized approach of the ECRES method, which also evaluates patient- and exercise-specific parameters crucial in determining the amount of required carbohydrates. Among the latter, ECRES takes into account the prevailing insulin concentration during the exercise at the specific daytime it is performed and changes in the patient’s specific insulin sensitivity throughout the day.^13,26 Furthermore, attention is paid to the patient’s training level, which implicates greater estimated amounts of glucose burned in sedentary patients compared to the active ones.^14,23 Finally, glycemia at the start of the activity and exercise intensity are considered in a more personalized manner than in the tabular approach^12,15 and/or in simple decision trees.²⁷ The amount of glucose already available in the body is estimated according to the measured starting plasma glucose level and to the extracellular fluid volume²⁸ and exercise intensity is evaluated by HR measurement. Although a tabular approach might be easily translatable in patients’ everyday life, we believe it would be very hard to maintain the simplicity of a tabular approach while accounting for all the above-mentioned factors which are mandatory to personalize the estimate. To boost the usability of the personalized ECRES algorithm, it was already implemented as an independent smartphone app,²⁹ but it can also be easily introduced in more general apps for T1D management.

CGM systems may serve as a tool to reduce the risk of exercise-induced hypoglycemia.³⁰ Nevertheless, it has been shown that their accuracy drops during physical activity when compared to reference glucose measurements,³¹ even so among the more recent devices,³² likely because of the rapidly changing blood glucose concentrations.^33,34

Insulin pumps and closed loop systems have the potential to further improve exercise management in patients with T1D^25,35 mainly by trying to mimic the physiological reduction of insulin release during exercise through a reduced basal insulin delivery. As a matter of fact, recent algorithms have reduced the exposure to exercise-induced hypoglycemia.^36,37 However, reduction of the basal insulin delivery rate can be efficacious only for late-onset exercises, namely, starting a few hours after the meal bolus; moreover, personalized reductions of the insulin delivery rate according to the intensity and daytime schedule of the exercise have never been proposed. Indeed, wide-ranging effects on insulin requirements and glucose turnover during physical activity make the development of an exercise smart artificial pancreas challenging,^38,39 and exercise in closed loop systems is a major hurdle.⁴⁰

Our study presents some limitations. Indeed, patients were not asked to provide an alimentary diary for the afternoon and night following the exercises, allowing us to better evaluate the behavior of glycemia during these time periods. Nevertheless, the same patients performed both trials and followed their own habitual post-exercise T1D self-management. Patients were asked to wear a CGM device to collect information about their glycemic imbalances around the trials. Although the alarms were set outside the optimal glycemic range to avoid interferences with patients’ usual self-management, it cannot be excluded that simply wearing the CGM device had an impact on the final obtained glycemic control. In addition, some patients experienced hypoglycemia in the 24 hours preceding the two experimental sessions. Nevertheless, the number of these patients was similar between the two experimental sessions, suggesting that no advantage/disadvantage was introduced for one of the tested approaches.

The main strength of this study is that, for the first time, the efficacy of the personalized ECRES method for the prevention of exercise-induced hypoglycemia in T1D patients, compared to a tabular approach, was investigated under real-life conditions. Although these do not always guarantee a good metabolic control, they likely are the most common conditions under which patients usually exercise.

Conclusion

The results of this study showed that the personalized ECRES algorithm estimated a significantly lower amount of carbohydrates required by T1D patients for specific physical activity as compared to the amount suggested by the chosen reference method. Patients’ blood glucose level, however, was maintained within safe clinical limits throughout and after the exercise. Consequently, the illustrated data support the validity of the estimates made by the personalized ECRES algorithm.