Diabetlgia 10, 139--143 (1974) 9 by Springer-Verlag 1974 On Measuring the Adequacy f Diabetes Regulatin: Cmparisn f Cntinuusly Mnitred Bld Glucse Patterns with Values at Selected Time Pints* G.D. Mlnar, W.F. Taylr and A. Langwrthy May Clinic and I~{ay Fundatin, l~chester, l~{innesta Received: August 11, 1972, and in revised frm: December 12, 1973 Summary. Individual bld glucse (BG) measurements at selected time pints were cmpared with cntinuusly recrded BG data as criteria f the adequacy f diabetes regulatin. Indices reflecting the adequacy f diabetes regulatin have previusly been develped frm cntinuusly mnitred BG measurements during studies under standardized near-nrmal living cnditins. These indices are: (1) mean amplitude f glycemie excursins (M2~GE), (2) diurnal mean bld glucse (MBG), (3) mean f daily differences f paired BG values (MODD). Because f the intensive studies necessary t btain these indices, apprximatins based n individual BG measurements which might easily be btained in practice were sught. The BG value 80 rain after breakfast crrelated best with the MAGE. The average f the fasting BG value and the value at 80 rain after breakfast crrelated well with the MBG. These individual BG measurements distinguished the grups f subjects. The difference between fasting BG values n successive days (AFBG) crrelated well with the ]~{ODD. Hwever, unlike MODD itself, AFBG did nt distinguish the grups f subjects. Sme ther selected BG values with different timing were nearly equally highly crrelated with these three criteria f BG behavir. Thus, relatively few but critically timed BG measurements n successive days, with suitable urinary glucse measurements, during standardized therapeutic prgrams may serve as an index f the efficacy f the therapy. By these same means, the characteristics f the patient's diabetes might als be assessed. Key wrds: Bld glucse, cntinuus bld glucse analysis, diabetes therapy criteria, unstable diabetes. This reprt initiates a search fr practical applicatins f the knmedge acquired by cntinuus bld glucse (BG) mnitring [1--3]. In this article we cmpare individual measurements f BG with cntinuusly recrded BG data as criteria f the adequacy f diabetes regulatin. The individual measurements were selected at time pints fund t be mst useful fr gauging the BG variatins. In a previus reprt [4] we analyzed the relatinships between amunts f urinary glucse and crrespnding BG values determined by a cntinuus mnitring prcedure. Bld and urinary glucse determinatins are the mst widely used labratry measures f the adequacy f diabetes regulatin [5--10] despite the recgnitin that disturbances f metablism f carbhydrates ther than glucse, and f prtein and fat as well, are part f the bichemical derangements in diabetes mellitus. Glucse measurements are mre generally available as well as mre readily interpretable than ther measurements. T identify and quantify several aspects f a patient's BG regulatry stability, we have previusly analyzed the results f ur cntinuus BG mnitrkngs [1] in terms f (1) the extent f within-day BG fluctuatins (the mean amplitude f BG excursins: MAGE) [2], (2) between-day BG variability (the mean f daily differences in matched BG values : MODD) [3], and (3) verall glycemia (the mean BG level: MBG). These * This investigatin was supprted in part by Research Grant AM-10152 frm the Natinal Institutes f Health, Public Health Service, and by grants frm the Endictt-Bhn Fundatin and Eli Lilly and Cmpany. three indices -- MAGE, MODD, and MBG -- tgether help characterize the behavir f the bld glucse in that each measures smething the thers d nt measure. Each f these indices als distinguishes amng nrmals, stable diabetics, and unstable diabetics in ur small but intensively studied samples. These same indices als serve as criteria f diabetes regulatin. Hwever, they can be calculated nly frm cntinuusly btained BG data. Therefre, we further analyzed ur data t seek simpler and mre cnvenient methds fr similar characterizatin f patients, utilizing intermittently btained (discrete) BG values. Rdsumd f Recent Wrk Leading t This Paper Clinical data cncerning the subjects in this investigatin and descriptins f the experimental design, bichemical methds, and data-prcessing prcedures have been published [1, 11, 12]. The mst relevant aspects f the methds and findings will be briefly summarized belw. In these studies the emphasis was n investigatin f unstable diabetics wh were cmpared with stable diabetics and nrmal subjects during standardized near-nrmal living cnditins. Tw prgrams f individually ptimized therapy with insulin were used, while diet and exercise regimens remained unchanged. One prgram was nce- r twice-daffy injectin f intermediate-acting insulin, with r withut shrtacting insulin frm the same syringe. The ther prgram
140 G.D. MLnar et al. : Adequacy f Diabetes Regulatin was fur-times-daily injectin f shrt-acting insulin. The time schedule f insulin injectins, meals, exercise, and sleep was cnsistent and precisely adhered t thrughut the studies. The diet f all subjects prvided fur majr meals f equal calric cmpsitin and carbhydrate, prtein, and fat cntents. The exercise was a standardized walk three times daily during the third pstprandial hur. Each investigatin culminated in a 6-day metablic balance study after the insulin, diet, and exercise regimens had been individually standardized and ptimized. Fr 48 h during this 6-day perid, cntinuus BG mnitring was perfrmed. Fig. 1 shws the cntrasting BG patterns f an unstable diabetic, a stable diabetic, and a nrmal subject. The grss qualitative differences between the diabetics and the nrmal as well as between the unstable and the stable diabetic are apparent. The diabetics lack the base-line stability and the rhythmicity de- 0 4 400 3 300 ~2 ~200 -: 1 cr~ E 100 200 1 100 0 1 viage 154 VIAGE 67 100 viage 44 0 0.7 11 15 t9 23 03 07 11 Clck time Fig. 1. Cntinuusly sampled and analyzed bld glucse patterns ver tw successive 24 h perids in a nrmal subject (Bttm), a stable diabetic (Middle), and an unstable diabetic (Tp). The first 24 h is pltted as a slid line and the secnd 24 h, as a brken line. The 48 h mean bld glucse (MBG) is shwn by the hrizntal line transeeting each panel; its numerical value is at the left. Mean f daily differences f paired bld glucse values (MODD) is shwn by the hatched area (abslute difference between the first and secnd 24 h) ; its numerical value is belw. Mean amplitude f glycemic excursins (MAGE) is shwn by interrupted lines bracketing the MBG; its numerical value is at the right. Symbls: B = breakfast; L = lunch; SK = snack; D = dinner; Su = supper ; E = 1 h f exercise; I = insulin mnstrated by the nrmal subject. The unstable diabetic has the highest, the nrmal has the lwest, and the stable diabetic has intermediate values fr the M_AGE (the measure f the extent f BG excursins), the I~ODD (the measure f day-t-day BG variatin), and the MBG (the measure f mean BG). Fig. 1 als illustrates the experimental cnditins. The daily events in the mrning (between breakfast and lunch) and in the evening (between dinner and supper) were similar and, accrdingly, the majrity f analyses presented belw are based n BG results during these times. Cnventinally utilized BG values -- fasting and 1, 2, and 3 h after meals (breakfast and dinner) -- are shwn in Fig. 2 as means fr grups f unstable diabetics, stable diabetics, and nrmals. The values were selected frm the cntinuus BG recrds. There are differences between after-breakfast and after-dinner BG levels and patterns, especially amng diabetics, as well as differences in BG levels and patterns between diabetics and nrmals. Fig. 2 als indicates that the mean BG values at the cnventinally sampled pstprandial times indicate fairly well the magnitude and timing f the mean maximal BG level. Minimal BG levels, hwever, ccur at quite different times. \ 3 B 3 f \ F Nrmals x Diu" ti s f-~sl~ De c '[Unstable --H0urly values..... Range D X 2 it t 200 ' -4%\ 100 ;. %6 007 09 1t 13 15 17 19 l 21 ~ 213 01 Clck time Fig. 2. Mean bld glucse values fr three nrmal subjects, fur stable diabetics, and six unstable diabetics at varius times, taken frm the cntinuusly sampled and analyzed bld recrds. The 7 a.m. value is the fasting bld glucse cncentratin; the ther values are at 1, 2, and 3 h after breakfast and after dinner. The maximal and minimal bld glucse cncentratins after these tw meals are als indicated Methds Out f the large number f BG measurements available frm the cntinuus mnitring data, we first chse t study thse which were custmary r had been used in ther studies (fasting; 1, 2, and 3 h after
G.D. Mlnar et al. : Adequacy f Diabetes Regulatin 141 breakfast; 1, 2, and 3 h after dinner). Als, we btmned frm the cntinuus data the average times f ccurrence f varius daily maxima and minima and used BG measured at thse times. This led us t add BG measured at 80 and 280 min after breakfast and after dinner (as the respective apprpriate times f daily maxima and minima in diabetics). 1 Frm this set f measurements we lked fr the mst prmising nes t be used in place f (r as clues t the phenmena characterized by) MAGE, MBG, and MODD. In this wrk we chse the simple prduct mment cefficient f crrelatin as an index f assciatin. It is easy t use and serves as a reasnable device fr deciding whether ne measurement was "clser" t a criterin than anther. We als examined these individual measurements fr their usefulness in distinguishing amng the grups -- nrmals, stable diabetics, and unstable diabetics. Results Prtins f the abve crrelatin studies gave the results summarized in Table 1. M_AGE was mst clsely assciated with BG at 80 min after breakfast (r-----0.88). Als, high assciatin was fund between MAGE and the difference, BG value at 80 minus BG value at 280 rain after breakfast, and the fasting BG. ~ MBG was highly assciated with mst f the BG values at the selected time pints. We chse ]3G, the mean f fasting BG and BG at 80 min after breakfast, fr which r=0.89, recgnizing that several thers were nearly as highly crrelated. MODD was studied in cnjunctin with ABG, the day-t-day differences in the varius BG values. The best crrelatin appeared t be with the fasting BG day-t-day differences (r ---- 0.89). We call this AFBG. The freging results are based n analysis f the data frm stable and unstable diabetics treated with intermediate-acting insulin. The use f data frm m~stable diabetics treated with shrt-acting insulin gave 1 In diabetics, the majrity f the daily maxima ccurred in the mrning. In nrmals, the daily maxima were evenly distributed during the mrning, evening, and night. In all subjects, a higher prprtin f daily BG minima ccurred during the mrning than during any ther segment f the day r night. The timing f pstcibal ]3G maxima and minima after the start f the preceding meal was different between diabetics and nrmals. The mean elapsed time t BG maxima after the majr daytime meals was 80 rain (range, 77 t 91 rain) in diabetics and 40 rain (range, 34 t 44 rain) in nrmals. The mean elapsed time t BG minima was 280 rain (range, 188 t 319 rain) in diabetics and 120 rnin (range, 104 t 144 rain) in nrmals. 2 With bth prgrams f treatment in unstable diabetics, IVIAGE was clsely assciated with ]3G at 80 rain after breakfast and with the difference f the BG values at 80 and 280 rain after breakfast. Hwever, with shrtacting insulin treatment in unstable diabetics, MAGE was less clsely assciated With the fasting BG (r -~ 0.40). similar results. MAGE was mst clsely assciated with the BG at 80 min after breakfast (r----0.75). Fr MBG with FBG, r ----- 0.83 ; fr MODD with AFBG, r = 0.83. Thus, by measuring the fasting BG value and the value at 80 min after breakfast, we fund highly psitively crrelated measurements with the cntinuus-mnitring indices. Table 1. Crrelatin cefficients f cntinuus-mnitring indices with selected discrete BG values Selected BG value at Crrelatin cefficient s with MAGE b MBG c MODD a After breakfast (rain) : 60 0.66 0.82 0.83 80 0.88 0.83 0.71 120 0.48 0.55 0.03 180 0.26 0.33 --0.07 280 0. 0.71 0.71 80-- 280 e 0.83 0.64 -- 0.10 After dinner (min): 60 0.56 0.76 0.77 80 0.51 0.81 0.84 120 0. 0.82 0.77 180 0.47 0.77 0.68 280 0.44 0.87 0.91 Fasting 0.80 0.85f 0.89 Crrelatin cefficients cmputed frm data n stable and unstable diabetics treated with intermediateacting insulin [1--3]. b MAGE = mean amplitude f glycemic excursins, a within-day BG variability index [2]. c ivibg = mean (diurnal) BG, an index f verall glycemia [2]. d 1VIODD = mean f daily differences (f paired BG values n successive days), an index f between-day BG variability [3]. e Value at 80 rain minus value at 280 rain. Fr mean f fasting BG and value 80 rain after breakfast, 0.89. T evaluate the ptential usefulness f selecting the BG values at the chsen times, we used graphic examinatin and regressin analysis f the relatinships f the selected individual BG measurements with the indices frm cntinuus BG measurements. Fig. 3 indicates an impressive assciatin between the BG at 80 rain after breakfast and MAGE. The spreading ut f the three grups f patients is n less impressive fr the individual BG measurement than fr M_AGE. There is n verlap. The use f the regressin line shwn in Fig. 3 as an estimate f MAGE is gd. The standard errr abut the regressin line is 24.6 rag/100 ml and the relative errr (relative t the mean MAGE) is 21.3%. Similarly BG prvides an adequate estimate f MBG (Fig. 4). The standard errr is 23.5 rag/100 ml and the relative errr is 16.5~. The spread f the three grups f patients indicates n verlap. Fr MODD the situatin is less satisfactry (Fig. 5). The AFBG des nt distinguish nrmals frm diabetics nr stable frm unstable diabetics, althugh MODD Diabetlgia,Vl. 10 10
142 G.D. MInar et al. : Adequacy f Diabetes Regulatin itself separates the three grups withut verlap. The high crrelatin cefficient, r=0.89, is partly explained by a single aberrant bservatin. Deleting that pint reduces the cefficient t r = 0.62. It evidently is nt safe r useful t estimate MODD frm the day-tday difference f fasting BG. Mrever, we have nt fund help in estimating MODD by attempting t use day-t-day differences in ther discrete BG measurements. MAGE 2 200 1 100 x ~,D x-x l [ I 1 ~ I I I I 0 0 100 200 300 400 0 BG 80 minutes after breakfast Fig. 3. Plt f BG values 80 min after breakfast against mean amplitude f glycemic excursins (MAGE). Nte gd separatin amng nrmals ( ), stable diabetics (D), and unstable diabetics (O). Fr details, see text 300 2 200 MBG 1 100 0 I I 1 I ~ 1-2. I00 200 300 400 Fig. 4. Plt f BG (average f fasting BG value and BG value at 80 rain after breakfast) against diurnal mean BG (MBG). Separatin f grups f subjects and symbls as in Fig. 3. Fr details, see text BG Discussin During treatment f diabetics, BG measurements may serve as an index f the efficacy f a particular therapeutic regimen. The mst efficacius therapeutic regimen wuld result in minimal within-day and between-day BG variability at a near-nrmal mean diurnal glycemia. It wuld als minimize bth hyperglycemic and hypglyeemie extremes. Our results n estimating majr indices f the efficacy f a therapeutic regimen fit smewhere between the rutine measurements f diabetes regulatin and experimental cntinuus BG mnitring. We believe that ur results prvide a beginning with which scientifically riented clinicians can apprach the estimatin f BG variability in their patients. Hithert, nly during frmal investigatinal cnditins have such estimates been btained [2, 3, 13--19]. Nw this might be accmplished under cnditins feasible in the setting f at least sme utpatient clinics and many hspitals and at what wuld be t mst patients an acceptable cst ill the number f daily bld samplings. Of curse, this apprach needs critical evaluatin in larger grups f patients studied ver lnger perids. Such further studies will, hwever, f necessity, be less intensive than ur studies. 180 t60 140 120 t00 MODD 80 60 40 20 0 I T I i I I 40 80 120 160 200 240 280 h FBG Fig. 5. Plt f AFBG (day-t-day differences in fasting BG) against mean f daily differences f paired BG values (MOOD). Symbls as in Fig. 3. Fr details, see text It is likely that, with therapeutic regimens similar t ur experimental cnditins, ur findings may be applicable t a larger sample f all diabetics. This is suggested by ur preliminary clinical experience t date with the same and additinal subjects. It is pssible that, even n different albeit standardized regimens, sme applicatin f ur findings may be feasible. The daily maximal BG and the daily BG range are lgical crrelates f MAGE. Others have als fund that, in diabetics n a variety f therapeutic regimens, the maximal BG level ccurs mst frequently after breakfast [13, 18, 20, 21]. The essential cnditins fr btaining interpretable data are that as many as pssible f the variables affecting diabetic patients (such as diet, insulin, and exercise) be standardized and that the measurements be made repeatedly n successive days with identical timing. Shuld unavidable alteratins f timing f therapeutic variables ccur, it is prbably best t alter the timing f BG measurements crrespndingly. Fr instance, if a meal is delayed by 10 min, it is best t delay the pstprandial BG measurements by the same interval, a The actual time interval between the meal and the BG measurement (fr example, 60 r 80 min 3 We have timed pstprandial BG measurements frm the start f meals fr cnsistency and because the timing frm the start f meals is easier t standardize.
G.D. Mlnar et al.: Adequacy f Diabetes Regulatin 143 after breakfast) is prbably less imprtant than the day-t-day cnsistency f diet, insulin, and exercise. The practice f btaining BG measurements when the lwest cncentratins are expected -- namely, befre meals and when clinical indicatins suggest excessive insulin actin -- shuld als be brne in mind. BG values befre lunch, befre dinner, and at ther times when peak insulin actin is suspected may prvide useful infrmatin cncerning pssible hypglycemia. Furthermre, with suitable quantitative urinary glucse measurements, additinal useful althugh less direct infrmatin abut BG variability and mean glycemia [4] may be btained. It is encuraging t nte that a few well-timed BG values appear t prvide useful clues (althugh with much lss f precisin and detail) t the quantitative character f BG behavir. The lss f precisin and detail was t be expected [19]. The disappinting lack f satisfactry clues t between-day BG variability btainable frm day-t-day changes in discrete BG measurements further emphasizes the wealth f iefrmarin prvided by cntinuus BG mnitring [1--3, 19, 20]. Yet it is with reference t between-day variability that cntinuus BG mnitring has its principal drawback: it is limited, fr varius reasns, t 48 t 72 h [1, 12, 19]. ]~eeause f the imprtance f betweenday BG variability [13, 14, 18], better means f assessing this aspect f BG variability need t be iund. There is clearly a need fr systematic assessment f the adequacy f diabetes regulatin. The lack f satisfactry BG criteria applicable t the assessment f diabetes regulatin under lng-term, real-life cnditins has hampered bjective appraisal f the results f therapy. This in turn has made crrelatin between adequacy f diabetic regulatin and ther phenmena, including the incidence and prgressin f diabetic cmplicatins, virtually impssible [22]. References i. Mlnar, G.D., Ackerman, E., Rsevear, J.W., Gatewd, L.C., Mxncss, K. E. : Cntinuus bld glucse analysis in ambulatry fed subjects. I. General methdlgy. May Clin. Prc. 43, 833--851 (1968) 2. Service, F.J., Mlnar, G.D., Rscvcar, J.W., Ackerman, E., Gatewd, L.C., Taylr, W.F. : Mean amplitude f glyccmic excursins, a measure f diabetic instability. Diabetes 19, 644--655 (1970) 3. Mlnar, G.D., Taylr, W.F., It, M.M.: Day-t-day variatin f cntinuusly mnitred glycemia: a further measure f diabetic instability. Diabetlgia 8, 342--348 (1972) 4. Service, F.J., Mlnar, G.D., Taylr, W.F.: Urine glucse analyses during cntinuus bld glucse mnitring. J. Amer. reed. Ass. 222, 294--298 (1972) 5. Krall, L.P., Jslin, A.P.: General plan f treatment and diet regulatin. In: Jslin's Diabetes Mellitus, p. 258, 11 th editin. (Ed. Marble, A., White, P., Bradley, R. F., Krall, L. P.). Philadelphia: Lea and Febigcr 1971 6. Marble, A. : InsMin in the treatment f diabetes. In: Jslin's Diabetes Mellitus, p. 290, llth editin. (Ed. Marble, A., White, P., Bradley, l%.f., Krall, L.P.). Philadelphia: Lea and Febiger 1971 7. White, P., Graham, C. A. : The child with diabetes. In : Jslin's Diabetes Mellit.us, p. 341--345, llth editin. (Ed. Marble, A., White, P., Bradley, R.F., Krall, L. P.). Philadelphia: Lea and Fcbiger 1971 8. Clwell, A.R., Sr. : Clinical use f insulin. In : Diabetes Mellitus : Thery and Practice, p. 624-- 637. (Ed. Ellenberg, Rifkin, H.). New Yrk: McGraw-Hill Bk Cmpany 1970 9. Knwlcs, tt.c., Jr. : Cntrl f diabetes and the prgressin f vascular disease. In: Diabetes Mellitus: Thery and Practice, p. 666-- 673. (Ed. Ellenberg, M., Rifkin, It.). New Yrk: McGraw-Hill Bk Cmpany 1970 10. Sussman, K.E.: Juvenile-type diabetes and its cmplicatins: theretical and practical cnsideratins, p. 68--70. Springfield, Illinis: Charles C. Thmas, Publisher 1971 11. Service, F.J., Mlnar, G.D., Rsevcar, J.W., Ackcrman, E., Taylr, W.F., Crmer, G.M., Mxncss, K.E. : Cntinuus bld glucse analysis in ambulatry fed subjects. II. Effects f anticagulatin with heparin. May Clin. Free. 44, 466--477 (1969) 12. Rsevear, J.W., l~faff, K.J., Service, F.J., Mlnar, G.D., Ackcrman, E.: Glucse xidase methd fr cntinuus autmated bld glucse determinatin. Clin. Chem. 15, 680--698 (1969) 13. Czyzyk, A., Pnikwska, I.: La significatin des valeurs particuli@res de la glicdmie dans le cycle glycemique jurnalicr. Acta diabct, lat. 4, 374--385 (1967) 14. Izz, J.L., Crump, S.L.: A clinical cmparisn f mdified insulins. J. clin. Invest. 29, 1514--1527 (19) 15. SchSne, G., Zimmer, ~.: Uber Blutzuckcrtageskuryen bci Gesunden und Zuckerkranken. Klin. Wschr. 14, 1672--1677 (1935) 16. Hallas-Mller, K.: The lent insulins. Diabetes 5, 7-- 12 (1956) 17. Brcssler, 1~., Gallway, J.A.: Insulin treatment f diabetes mellitns. Med. Clin. N. Amer. 55, 861--876 (1971) 18. Schlichtkrull, J., Munck, 0., Jcrsild, M. : The M-value, an index f bld-sugar cntrl in diabetics. Acta reed. scand. 177, 95--102 (1965) 19. Miruze, J., Satingher, A., Sany, C., Jaffil, C.: Cefficient d'efficacitg insulinique: cefficient M de Schlichtkrull crrig@ ct simplifid par la technique de l'crffegistrement glyc@miquc cntinu. Diabete 11, 267--273 (1963) 20. Jersild, M. : La glyc6mic pst-prandiale des diabdtiques. Diabete 14, 225--227 (1966) 21. Sindni, A., Jr. : Fasting bld sugar vs. pstprandial bld sugar as bserved in nrmal individuals, medical (nn-diabetic) patients, and patients with diabetes ; special references t : plain, prtamine zinc and glbin insulins, cmpatible hyperglycemia and arterisclersis. Amer. J. dig. Dis. 13, 178--192 (1946) 22. Kaplan, M.tt., Feinstein, A.R. : A critique f methds in reprted studies f lng-term vascular cmplicatins in patients with diabetes mellitus. Diabetes 22, 160-- 174 (1973) Dr. G.D. Mlnar Dept. f Endcrinlgy and Internal Medicine May Clinic ~chestcr Minnesta 55901 USA 10"