Sage Journals: Discover world-class research

Abstract

Objective:

We evaluated effects of unmasking of continuous display of continuous glucose monitoring (CGM) on quality of glycemic control and glycemic variability.

Methods:

We reanalyzed CGM data from 85 patients using a 7-day glucose sensor. Glucose values were “masked” during the first week but “unmasked” during the next 2 weeks. We evaluated 48 criteria for quality of glycemic control, including mean glucose, SD, percentage of values within-, above- or below- specified ranges, Schlichtkrull's M ₁₀₀ index, mean amplitude of glycemic excursion (MAGE), mean of daily differences (MODD), the J index, “Index of Glycemic Control” (IGC), Hyperglycemia Index, Hypoglycemia Index, High Blood Glucose Index (HBGI), Low Blood Glucose Index (LBGI), average daily risk range (ADRR), GRADE scores, and CONGA_n. We calculated SD values between daily means, between days-within time points, within days, between time points (for the average glucose profile for several days), and within series for time segments of arbitrary length.

Results:

Unmasking CGM displays resulted in rapid, highly statistically significant improvement in 29 indices, including percentage within, percentage above, and percentage below target range, mean glucose, SD, SD of daily means, MODD, M ₁₀₀, IGC, GRADE, HBGI, and J index. Both hyperglycemia and hypoglycemia improved during the first week after unmasking; further improvement in hypoglycemia was seen during the following week. Results obtained using multiple criteria were consistent and highly correlated.

Conclusions:

Continuous access to display of CGM sensors dramatically improved 29 indices of glycemic control and glycemic variability.

Introduction

Continuous glucose monitoring (CGM) with real-time display of interstitial fluid glucose to the subject has been shown to significantly reduce hemoglobin A1C^1

–4 in individuals with both type 1 and type 2 diabetes, in both well- and poorly controlled subjects, and in patients using multiple daily injections or continuous subcutaneous insulin infusion. Garg and co-workers have previously shown that CGM with real-time display resulted in statistically significant changes in the glucose distribution, with reduced frequencies at extremely high and low values combined and increased frequency at the center of the distribution.^2,3 In view of concerns that glycemic variability might play a role in the pathogenesis of complications of diabetes^5

–11 and due to progressively increasing usage of CGM, there is need for improved methods to characterize variability in CGM data.

Methods

Data

Data were obtained from 85 patients who utilized a 7-day CGM sensor (SEVEN™ System, DexCom, Inc., San Diego, CA) for three consecutive 1-week periods.² Display of CGM data was masked during Period 1 but unmasked during Periods 2 and 3 and continuously available to the subject. Characteristics of the patients and details of the study protocol have been reported previously.²

Measures of quality of glycemic control and glycemic variability

We computed 20 measures of quality of glycemic control and 28 measures of glycemic variability. Indices of glycemic control included the grand mean of glucose values,^12,13 the percentages of values above various thresholds (170, 200, or 250 mg/dL), below various thresholds (80 or 50 mg/dL), or within specified ranges (80–200, 70–180, or 80–140 mg/dL),² Schlichtkrull's M using a reference value of 100 mg/dL, M ₁₀₀,¹⁴ J ,¹⁵ Index of Glycemic Control (IGC) (previously termed “Figure of Merit”), which combines the results from a Hyperglycemia Index and a Hypoglycemia Index,^16
–18 the average daily risk range (ADRR), which combines information from an High Blood Glucose Index (HBGI) and a Low Blood Glucose Index (LBGI),^19,20 the GRADE score,²¹ and the percentages of the GRADE score attributed to glucose values within the hyperglycemic, euglycemic, and hypoglycemic ranges.²¹ The IGC¹⁶ was calculated with the following parameters: lower limit of target range (LLTR) 80 mg/dL; upper limit of target range (ULTR) 140 mg/dL; exponent for deviations of glucose below the LLTR, b = 2; exponent for deviations of glucose above the ULTR, a = 1.1.

We also computed 28 measures of glycemic variability, including the overall SD (SD or SD_T),^12,13 mean amplitude of glycemic excursions (MAGE),^22,23 mean of daily differences (MODD),^23,24 and CONGA_n for n = 1–4 and 24.²⁵ We examined an alternative method for calculating the MODD^23,24 when dealing with data from multiple days: MODD_d measures variability of glucose values separated by exactly 24 × d h. The details, advantages, limitations, and interrelationships of these methods are discussed elsewhere.^16,17 We also examined six subtypes of SD,^16,17 described further below.

New measures of glycemic variability

Several types of SD values were calculated using a new approach based on the principles underlying analysis of variance.^16,17 We calculate: (1) an overall SD or SD_T—the SD of all glucose values from all days for a specified period of time; (2) the average SD of glucose values within each day, SD_w; (3) the SD of the mean glucose value for any specified time of day (abbreviated hh:mm ) based on the average glucose profile by time of day from several days, SD_hh:mm; (4) the SD for a series of glucose values for a time segment of any length ( h , in hours), SD_{ws h}; (5) the SD of daily means, SD_dm; (6) the SD between days-within specified time points averaged over all times of day, SD_{b hh:mm}; and finally (7) the SD between days, within time points, after correction for changes in the daily mean, SD_{b hh:mm//dm}.

Statistical analysis

Results were available from 85, 84, and 83 subjects in Periods 1, 2, and 3, respectively. We performed an analysis of variance of the data to obtain estimates of several new types of SD.^16,17 We compared the results from the three periods using a general linear model to establish overall statistical significance, followed by Student's t tests (paired within individuals) comparing Periods 2 and Period 3 with Period 1. We used Wilcoxon two-sample tests to evaluate the robustness of results with respect to the need for assumptions regarding gaussianness of distributions. We used correlation coefficients and a correlation matrix to examine the interrelationships among the measures of quality of glycemic control and of glycemic variability (see Supplementary Appendix at www.liebertonline.com/dia). We ranked the criteria in terms of their statistical significance (P values) when comparing Period 3 with the control period, Period 1, reflecting their ability or sensitivity to detect the changes introduced by unmasking of CGM data.

Results

A total of 365,533 glucose values were obtained of the maximally expected 495,936, or 74%.

Euglycemia

Figure 1A shows the change in percentage of glucose values within the ranges 80–200, 70–180, and 80–140 mg/dL after unmasking of the CGM sensor display in Periods 2 and 3 relative to the corresponding percentages during the control period, Period 1, when the sensor was masked. There was a rapid and highly statistically significant improvement: all three P values were <0.0001. The percentage improvement was slightly higher when using a relatively stringent criterion (80–140 mg/dL) rather than a relaxed criterion such as 80–200 mg/dL. There was a corresponding statistically significant increase (improvement) in the %GRADE_EUGLYCEMIA (P < 0.05).

FIG. 1.

Response to unmasking of CGM sensor output during Weeks 2 and 3, expressed as percentage of corresponding value during the control period (Week 1). P values for paired comparisons (within individuals) between Period 1 and Period 3 are as indicated. (A) “Euglycemia”: % of glucose values within the specified target ranges 80–200 mg/dL (black triangles) (P < 0.0001), 70–180 mg/dL (red squares) (P < 0.0001), and 80–140 mg/dL (blue circles) (P < 0.0001), and percentage GRADE_EUGLYCEMIA (green diamonds) (P < 0.01). These four curves appear above the baseline value of 100%. Hyperglycemia: percentage of glucose values >180 mg/dL (dotted curve, light orange squares with superimposed plus sign) (P < 0.0001), >250 mg/dL (dashed curve, orange squares with superimposed asterisk) (P < 0.01), Hyperglycemia Index, and HBGI (dot-dash curve; pink triangles) (P < 0.01 and <0.0001, respectively). These four curves fall below the initial (baseline) level of 100%. (B) Hypoglycemia: percentage of glucose values <80 mg/dL (black circles) (not significant), <50 mg/dL (red squares with superimposed asterisk) (P < 0.01), LBGI (blue squares) (not significant), percentage GRADE_HYPOGLYCEMIA (green diamonds) (P < 0.01), and Hypoglycemia Index (pink triangles) (P < 0.01). (C) Quality of glycemic control: percentage of glucose values 80–200 mg/dL (black circles), Mean_T (red squares), M ₁₀₀ (blue triangles), J (green diamonds), IGC (turquoise squares with superimposed asterisk), and GRADE (pink squares with superimposed multiplication sign). (All six criteria showed a statistically significant change between Period 1 vs. Period 3: P < 0.0001.) (D) Glycemic variability: SD_T (green circles) (P < 0.01), SD_dm (pink squares with superimposed plus sign) (P < 0.001), MAGE (black circles) (P < 0.01), MODD (red diamonds) (P < 0.001), CONGA₂₄ (blue triangles) (P < 0.05), SD_w (turquoise squares with superimposed multiplication sign) (not significant), and SD_{b hh:mm} (light orange squares with superimposed asterisk) (not significant).

Hyperglycemia

Figure 1A shows the reduction in percentage of hyperglycemic values using criteria of 180 and 250 mg/dL (P < 0.0001 and ∼0.01, respectively). There were corresponding decreases in the Hyperglycemia Index (P ∼ 0.001) and HBGI (P < 0.0001).

Hypoglycemia

Figure 1B shows the changes in measures of hypoglycemia in response to unmasking of the CGM display. We show results for percentage of glucose values <80 and <50 mg/dL. As was the case for hyperglycemia, using a more stringent criterion (<50 mg/dL) results in a larger percentage change relative to control (Period 1). We also examined the Hypoglycemia Index, LBGI, and %GRADE_HYPOGLYCEMIA. All of these criteria showed a statistically significant improvement following unmasking of the real-time displays of CGM data. The improvement of hypoglycemia continued in Period 3, whereas the improvement in hyperglycemia was largely completed in Period 2.

Quality of glycemic control

Figure 1C shows the trends in several additional criteria for quality of glycemic control: M ₁₀₀, J , IGC, GRADE, and ADRR. The response for the percentage of glucose values in the range 80–200 mg/dL is shown for reference. These criteria showed a statistically significant improvement after unmasking of CGM data.

Glycemic variability

Figure 1D shows the response of several criteria for glycemic variability, including the “classical” measures MAGE and MODD, a recently introduced measure, CONGA₂₄, the overall SD or SD_T, and the SD between daily means (SD_dm). All of these measures improved significantly (P < 0.05). Additionally, the SD_w and the SD_{b hh:mm} decreased, reaching statistically significance when Period 2 is compared with Period 1, but not when Period 3 was compared with Period 1 (see Supplementary Appendix, Table A2, at www.liebertonline.com/dia). A more complicated pattern of changes was observed for CONGA_n and SD_{ws h} for values of n and h between 1 and 24. Variability over a time span of 24 h was decreased (MODD, SD_dm, CONGA₂₄), but relatively high frequency/short periodicity variability increased (see Supplementary Appendix, Table A2 and Figs. A4 and A5, at www.liebertonline.com/dia).

Figure 1 shows the results grouped by functional category (quality of glycemic control, glycemic variability). Results for all criteria examined, ranked in order of statistical significance, are summarized in the Supplementary Appendix, Tables A1 and A2, at www.liebertonline.com/dia. These results make it possible to identify the criteria that have the greatest sensitivity for the detection of a response to a therapeutic intervention such as the one used in the present study.

The results for 20 criteria for quality of glycemic control are summarized in the Supplementary Appendix, Table A1, at www.liebertonline.com/dia. Sixteen of these parameters were statistically significantly different (P < 0.05) when comparing Period 3 (“unmasked”) with Period 1 (“masked”). The criteria that showed the greatest improvement in terms of magnitude of t tests and P values were the IGC and the percentage of values within the target range (80–200 mg/dL or 70–180 mg/dL). Eight criteria were significant at the level of P < 0.0001 for Period 3 compared with Period 1. Four criteria were not statistically significantly different between Periods 1 and 3 (ADRR, LBGI, percentage of glucose values below 80 mg/dL, and %GRADE_{HYPERGLYCEMIA}).

There was a statistically significant improvement in 13 of the 28 criteria for glycemic variability when comparing Period 3 with Period 1 (see Supplementary Appendix, Table A2, at www.liebertonline.com/dia). Four more criteria were statistically significantly improved in Period 2 compared with Period 1 (see Supplementary Appendix, Table A2, at www.liebertonline.com/dia). The parameters that improved with the highest statistical significance were MODD, the SD_dm, the overall SD or SD_T, and SD_{b hh:mm}. The SD_hh:mm was reduced (P < 0.05). CONGA₂₄, a measure of between-day variability²⁵ that is highly correlated¹⁷ with MODD, was also reduced (P < 0.05).

Fifteen measures of variability did not show a statistically significant change between Period 1 and Period 3 (see Supplementary Appendix, Table A2, at www.liebertonline.com/dia). However, changes in four of these parameters were significant when comparing Period 2 with Period 1, including MODD₄ (the differences between glucose values exactly 96 h apart), the SD_{ws

24}, SD_w, and SD_{b hh:mm//dm}. Thus, some changes following unmasking of the CGM sensor display may be transitory.

Interrelationships among criteria

We identified a number of relationships among the criteria being used for glycemic control and for glycemic variability. There was a linear relationship between percentage of values within the range 80–200 mg/dL and the five parameters: M ₁₀₀, J , IGC, GRADE, and ADRR (see Supplementary Appendix, Fig. A1 and Tables A3–A5, at www.liebertonline.com/dia). All six measures are very highly and significantly correlated despite the fact that each effectively assigns different weights and importance to hyperglycemic and hypoglycemic values. Correlation coefficients for the percentage of measurements in the range 80–200 mg/dL with M ₁₀₀, GRADE, J , IGC, and ADRR were r = −0.886, −0.875, −0.848, −0.761, and −0.642, respectively.

The measures of glycemic variability within days are highly correlated among themselves and with the overall variability: there are high correlations among MAGE, CONGA₂, SD_w, SD_hh:mm, and SD_{ws h} for h = 1–24 hours and SD_T (see Supplementary Appendix, Tables A3–A5, at www.liebertonline.com/dia).^16,17 Similarly, measures of glycemic variability between days are also highly correlated among themselves and with overall variability: there are high correlations between MODD, CONGA₂₄, SD_{b hh:mm}, SD_{b hh:mm//dm}, and SD_T (see Supplementary Appendix, Table A4, at www.liebertonline.com/dia).^16,17 MODD and MODD_d for d = 1–6 showed high correlations with each other, with CONGA₂₄, and with SD_{b hh:mm}. Based on theoretical considerations, both MODD and CONGA₂₄ should be directly proportional to SD_{b hh:mm}.¹⁷

Discussion

Effects of continuous display of CGM

The present analysis demonstrates that the unmasking of continuous monitoring of glucose results in highly statistically significant improvement in many parameters of glycemic control (Fig. 1A–C) (see Supplementary Appendix, Table A1, at www.liebertonline.com/dia) and a corresponding reduction of glycemic variability (Fig. 1D) (see Supplementary Appendix, Table A2, at www.liebertonline.com/dia). This includes improvement in percentage of glucose values within a specified target range (Fig. 1A) and the corresponding reduced frequency of values in the hyperglycemic (Fig. 1A) and hypoglycemic ranges (Fig. 1B). Improvement in the relative frequency of “euglycemic” values and reduction of hyperglycemia occurred within the first week, whereas improvement in hypoglycemia continued during the second week after unmasking (Period 3) (Fig. 1A and B). These findings are consistent with previous studies that showed a reduction in A1C over a period of months and with a dose–response relationship to the extent of use of the CGM.^1,3,4 We confirm previous findings based on the same data, that CGM leads to increased frequency of glucose values in the range 80–200 mg/dL and reduced likelihood of glucose values above or below this range.²

The IGC and “percentage between 80 and 200 mg/dL” were the most sensitive criteria. The latter criterion is robust with respect to selection of target range despite its theoretical limitation that it does not consider the severity of departures of the glucose values outside the target range. In contrast, several other criteria ( M ₁₀₀, Hyperglycemic Index, Hypoglycemic Index, IGC, HBGI, LBGI, ADRR, and GRADE) take into account the extent to which the observed or measured glucose values depart from the target range.^17

–21 Some of these indices perform quite well (IGC, GRADE, J ), whereas others are relatively insensitive to the unmasking of CGM (ADRR). There is remarkable uniformity of results among most of the criteria tested. The M ₁₀₀, J , IGC, GRADE, and ADRR are all closely related to percentage within 80–200 mg/dL (see Supplementary Appendix, Fig. A1, at www.liebertonline.com/dia).

Variability

Use of CGM with real-time display resulted in a reduction in total variability, SD_T, by 6.5%. The reduction in total variability was associated with a relative reduction in the variability of daily means (SD_dm) of 17.7%, a reduction in SD_{b hh:mm} of 7.1%, and a corresponding reduction in MODD of 8.2%. Within day variability, SD_w, and the closely related SD_{ws

24} showed significant changes in Period 2 but not in Period 3 compared with Period 1.

Differences in “sensitivity” of alternative criteria to detect changes

There were statistically significant changes in 29 of 48 parameters of glycemic control and glycemic variability when the display of glucose values was unmasked (see Supplementary Appendix, Tables A1 and A2, at www.liebertonline.com/dia). The ability to detect this effect varied considerably among the several criteria. The most sensitive were the IGC and the percentage of glucose measurements falling within the target range 80–200 mg/dL, but the results were fairly insensitive to the exact choice of target range. A series of closely related criteria (see Supplementary Appendix, Tables A1 and A2, at www.liebertonline.com/dia) also detected the effects of unmasking of CGM with a very high degree of sensitivity. For this patient population, CGM had its predominant effect on reducing hyperglycemia (e.g., Hyperglycemia Index, HBGI, “percentage >180 mg/dL,” or “percentage >250 mg/dL”) (Fig. 1A), while also reducing the frequency of hypoglycemia (Fig. 1B). Criteria such as the Hypoglycemia Index and “percentage of observations below 50 mg/dL” showed relatively large percentage change but were accompanied by larger standard errors for the difference between periods, making them less sensitive to the effect of unmasking of real-time displays of CGM.

It would be advantageous if a criterion can combine information from both hypo- and hyperglycemic events. Criteria such as “percentage below 50 mg/dL” or “percentage above 180 mg/dL,” Hypoglycemia Index, Hyperglycemia Index, LBGI, and HBGI only consider one aspect of the glucose values. When these can be combined with the information reflecting the percentage of values in the target range and on the other side of the target range (as in M ₁₀₀, IGC, GRADE), one should obtain a more sensitive measure. It is also desirable, a priori, to utilize a weighting scheme so that values that fall progressively farther outside the target range are given progressively larger penalties. This property is incorporated into M ₁₀₀, IGC, Hypoglycemia Index, Hyperglycemia Index, GRADE, LBGI, HBGI, and ADRR. Measures such as these generally provided better sensitivity, as reflected in their relatively high rank in terms of statistical significance in (see Supplementary Appendix, Table A1, at www.liebertonline.com).

Comparison with previous methods of analysis

The present data set had been analyzed previously² using the frequency distribution of glucose values, examining the changes in the frequency of glucose values in the tails of the distribution. That analysis pooled results from multiple individuals. In the present approach, we use each individual subject as his or her own control. Many of the observed comparisons have Student's t values of 3, 4 or 5, corresponding to P values of P < 0.001 to P < 0.00001—truly impressive levels of statistical significance (see Supplementary Appendix, Tables A1 and A2, at www.liebertonline.com/dia). Several of the previously developed criteria such as “percentage of glucose values within the range 80–200 mg/dL” (or 80–140, or 70–180 mg/dL), M ₁₀₀, J , GRADE, MAGE, MODD, SD_T, CONGA₄, CONGA₂₄, and HBGI also performed extremely well in terms of detecting highly significant change.

Explanations for the observed changes in variability

Using feedback from real-time CGM data, the patient can avoid excessively long or severe excursions into the hyper- or hypoglycemic regions, thus “trimming the tails” of the distribution. This will reduce variability between days (MODD, CONGA_1–24, SD_dm, and SD_{b hh:mm}) and reduce the overall variability, SD_T. When using CGM, the patient has greater freedom to adjust several lifestyle factors, e.g., postponing a meal or reducing food intake if experiencing hyperglycemia, consuming a snack or the next meal earlier if experiencing hypoglycemia. Thus, timing of meals might be expected to show more variability than in the absence of CGM feedback. This would be expected to generate an increase in SD_{b hh:mm}, SD_{b hh:mm//dm}, and MODD. Changes in the shape of the glucose profile by time of day on different days will reduce SD_hh:mm and the ratio of SD_hh:mm to SD_w.

The rank order of the t and P values for different criteria in the Supplementary Appendix, Tables A1 and A2, at www.liebertpub.com/dia varies somewhat depending on whether one compares Period 2 or Period 3 with Period 1. This is in part due to the fact that there were temporal trends (Fig. 1). Periods 2 and 3 are not a single homogeneous and stable period of time in terms of glycemic control, variability, and patterns.

Clinical application of new measures of glycemic variability

The present study illustrates several concepts and introduces new terminology to help physicians and patients describe and utilize the information in complex glycemic patterns. These concepts can potentially help improve communication between physician and patient.^16,17 These approaches to measurement of glycemic variability are expected to be useful when evaluating therapeutic interventions, e.g., medical therapy, islet cell or pancreatic transplantation, or closed loop feedback control. One can monitor changes in individual patients using these criteria relative to their previous or baseline values, using each patient as his or her own control to monitor changes with time and responses to therapeutic interventions. It will be important to obtain normative data on multiple patient populations (e.g., type 1 and type 2 diabetes, patients using multiple daily injections or continuous subcutaneous insulin infusion, patients using various therapeutic regimens, and for patients with various levels of glycemic control as reflected in their hemoglobin A1C). One could then categorize patients as being in excellent, good, fair, or poor control in terms of a subset of the parameters shown in Figure 1 and in Supplementary Appendix, Tables A1 and A2, at www.liebertonline.com/dia, and calculate a combined or weighted average of these scores.

Selection of criteria to characterize glycemic variability

In view of the plethora of criteria that are currently available as augmented by the new methods described here^16,17 and the high degree of correlation among them (see Supplementary Appendix, Tables A3–A5 and Figs. A1–A4, at www.liebertonline.com/dia), which criteria for glycemic variability are most appropriate for routine purposes—both clinically and in clinical research? Each has their advantages and disadvantages.^16,17 We believe that the most useful criteria are likely to be the SD_T, the SD_w, and the SD_{b hh:mm}. The SD_w provides essentially the same information as MAGE but is easier to calculate and will generally be more precise and reproducible. The SD_{b hh:mm} provides essentially the same information as MODD or CONGA₂₄ but facilitates calculation for time series from multiple days. The criteria (SD_T, SD_w, and SD_{b hh:mm}) are SD values and thus have a shared conceptual, mathematical, and statistical foundation. Other criteria are likely to be important in special circumstances.^16,17

Conclusions

Use of CGM with continuous display improves 29 indices of quality of glycemic control and glycemic variability, including improvements in mean glucose (Mean_T), the overall SD or SD_T, the SD_dm, M ₁₀₀, J , IGC, GRADE, percentage of values in a specified range (80–200 mg/dL), percentage of values in the hyperglycemic range (>180 or >250 mg/dL), Hyperglycemia Index, HBGI, percentage of observations in the hypoglycemic range (<50 mg/dL), and Hypoglycemia Index (P < 0.05). The IGC and percentage of glucose values within 80–200 mg/dL had the greatest sensitivity to detect the response to this therapeutic intervention. The present study provides multiple insights regarding the methods to characterize CGM data.

Footnotes

Acknowledgments

The authors thank Bradley S. Matsubara, M.D. for many helpful suggestions. We thank Katherine H. Nakamura, Ph.D. for her excellence in computing the extensive and meticulous statistical analyses. Mr. Lucas Bohnett developed software for calculation of MAGE.

Author Disclosure Statement

David Rodbard, M.D. has served as a consultant to DexCom, Inc. Tim Bailey, M.D. has conducted clinical research studies and served as a speaker for DexCom, Inc. The original² and the present studies were supported by DexCom, Inc., San Diego, CA.

References

Juvenile Diabetes Research Foundation Continuous Glucose Monitoring Study Group. Tamborlane

, Beck

, Bode

, Buckingham

, Chase

, Clemons

, Fiallo-Scharer

, Fox

, Gilliam

, Hirsch

, Huang

, Kollman

, Kowalski

, Laffel

, Lawrence

, Lee

, Mauras

, O'Grady

, Ruedy

, Tansey

, Tsalikian

, Weinzimer

, Wilson

, Wolpert

, Wysocki

, Xing

. Continuous glucose monitoring and intensive treatment of type 1 diabetes. N Engl J Med, 2008; 359:1464–1476.

Garg

, Jovanovic

. Relationship of fasting and hourly blood glucose levels to HbA1C values: safety, accuracy, and improvements in glucose profiles obtained using a 7-day continuous glucose sensor. Diabetes Care, 2006; 29:2644–2649.

Bailey

, Zisser

, Garg

. Reduction in hemoglobin A1C with real-time continuous glucose monitoring: results from a 12-week observational study. Diabetes Technol Ther, 2007; 9:203–210.

Deiss

, Bolinder

, Riveline

, Battelino

, Bosi

, Tubiana-Rufi

, Kerr

, Phillip

. Improved glycemic control in poorly controlled patients with type 1 diabetes using real-time continuous glucose monitoring. Diabetes Care, 2006; 29:2730–2732.

Hirsch

, Brownlee

. The effect of glucose variability on the risk of microvascular complications in type 1 diabetes. Diabetes Care, 2007; 30:186–187.

Monnier

, Mas

, Ginet

, Michel

, Villon

, Cristol

, Colette

. Activation of oxidative stress by acute glucose fluctuations compared with sustained chronic hyperglycemia in patients with type 2 diabetes. JAMA, 2006; 295:1681–1687.

Wentholt

, Kulik

, Michels

, Hoekstra

, DeVries

. Glucose fluctuations and activation of oxidative stress in patients with type 1 diabetes. Diabetologia, 2008; 51:183–190.

Ceriello

, Esposito

, Piconi

, Ihnat

, Thorpe

, Testa

, Boemi

, Giugliano

. Oscillating glucose is more deleterious to endothelial function and oxidative stress than mean glucose in normal and type 2 diabetic patients. Diabetes, 2008; 57:1349–1354.

Diabetes Research in Children Network (DirecNet) Study Group. The effect of glucose variability on the risk of microvascular complications in type 1 diabetes. Diabetes Care, 2007; 30:185.

10.

Kilpatrick

, Rigby

, Atkin

. The effect of glucose variability on the risk of microvascular complications in type 1 diabetes. Diabetes Care, 2006; 29:1486–1490.

11.

Kilpatrick

, Rigby

, Atkin

. HbA1C Variability and the risk of microvascular complications in type 1 diabetes: data from the DCCT ([abstract 817-P] Diabetes, 2008; 57,Suppl 1:A236.

12.

Hirsch

. Glycemic variability: it's not just about A1C anymore! Diabetes Technol Ther, 2005; 7:780–783.

13.

Rodbard

. Optimizing display, analysis, interpretation and utility of self-monitoring of blood glucose (SMBG) data for management of patients with diabetes. J Diabetes Sci Technol, 2007; 1:62–71.

14.

Schlichtkrull

, Munck

, Jersild

. The M-value, an index of blood-sugar control in diabetics. Acta Med Scand, 1965; 177:95–102.

15.

Wójcicki

. “J”-index. A new proposition of the assessment of current glucose control in diabetic patients. Horm Metab Res, 1995; 27:41–42.

16.

Rodbard

. Interpretation of continuous glucose monitoring data: glycemic variability and quality of control. Diabetes Technol Ther, 2009; 11,Suppl 1:S-55–S-67.

17.

Rodbard

. New and improved methods to characterize glycemic variability using continuous glucose monitoring. Diabetes Technol Ther, 2009; 11:551–565.

18.

Rodbard

, Jovanovic

, Garg

, Matsubara

, Nakamura

, Bailey

, Zisser

, Kaplan

. Reduced glycemic variability with the use of real-time continuous glucose monitoring [abstract OR-208] http://professional.diabetes.org/Abstracts_Display.aspx?TYP=1&CID=74519. 2009 August 31.

19.

Kovatchev

, Otto

, Cox

, Gonder-Frederick

, Clarke

. Evaluation of a new measure of blood glucose variability in diabetes. Diabetes Care, 2006; 29:2433–2438.

20.

Kovatchev

, Cox

, Gonder-Frederick

, Young-Hyman

, Schlundt

, Clarke

. Assessment of risk for severe hypoglycemia among adults with IDDM: validation of the low blood glucose index. Diabetes Care, 1998; 21:1870–1875.

21.

Hill

, Hindmarsh

, Stevens

, Stratton

, Levy

, Matthews

. A method for assessing quality of control from glucose profiles. Diabet Med, 2007; 24:753–758.

22.

Service

, Molnar

, Rosevear

, Ackerman

, Gatewood

, Taylor

. Mean amplitude of glycemic excursions, a measure of diabetic instability. Diabetes, 1970; 19:644–655.

23.

Service

, O'Brien

, Rizza

. Measurements of glucose control. Diabetes Care, 1987; 10:225–237.

24.

Service

, Nelson

. Characteristics of glycemic stability. Diabetes Care, 1980; 3:58–62.

25.

McDonnell

, Donath

, Vidmar

, Werther

, Cameron

. A novel approach to continuous glucose analysis utilizing glycemic variation. Diabetes Technol Ther, 2005; 7:253–263.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.48 MB

Improved Quality of Glycemic Control and Reduced Glycemic Variability with Use of Continuous Glucose Monitoring

Abstract

Objective:

Methods:

Results:

Conclusions:

Introduction

Methods

Data

Measures of quality of glycemic control and glycemic variability

New measures of glycemic variability

Statistical analysis

Results

Euglycemia

Hyperglycemia

Hypoglycemia

Quality of glycemic control

Glycemic variability

Interrelationships among criteria

Discussion

Effects of continuous display of CGM

Variability

Differences in “sensitivity” of alternative criteria to detect changes

Comparison with previous methods of analysis

Explanations for the observed changes in variability

Clinical application of new measures of glycemic variability

Selection of criteria to characterize glycemic variability

Conclusions

Footnotes

Acknowledgments

Author Disclosure Statement

References

Supplementary Material