Diet is a cornerstone of diabet

Document Sample
 Diet is a cornerstone of diabet Powered By Docstoc
					Clinical Care/Education/Nutrition
 O R I G I N A L                A R T I C L E

Effects of Meal Carbohydrate
Content on Insulin Requirements in
Type 1 Diabetic Patients Treated
Intensively With the Basal-Bolus
(Ultralente-Regular) Insulin Regimen
RÉMI RABASA-LHORET, MD                                 DANIELLE POISSON, RN
                                                                                                                    iet is a cornerstone of diabetes man-

JEAN GARON, MD                                         JEAN-LOUIS CHIASSON, MD
HÉLÈNE LANGELIER, RDT                                                                                               agement and yet the most challenging
                                                                                                                    aspect of diabetes care (1,2). The dif-
                                                                                                              ficulty with diet prescription is that most of
                                                                                                              the time, it does not take into account indi-
                                                                                                              vidual food choices and lifestyle. In fact,
OBJECTIVE — In this study, we evaluated the effects of high- (55%) and low- (40%) car-                        one of the main recommendations of the
bohydrate diets on insulin requirements in nine type 1 diabetic subjects treated intensively with             Diabetes Control and Complication Trial
ultralente as basal insulin and regular insulin as premeal insulin adjusted to the carbohydrate               (DCCT) is that nutrition should be based
content of meals.                                                                                             on flexibility in making food choices to fit
                                                                                                              individual lifestyles while focusing on
RESEARCH DESIGN AND METHODS — Nine subjects were randomized in a
                                                                                                              improved metabolic control (1,3–5). One
crossover design to follow two diets consecutively for a period of 14 days each. A 3-day food
diary was completed for each diet with the amount of carbohydrate in the mixed meals rang-                    way to implement this recommendation is
ing from 21 to 188 g. Preprandial (5.9 vs. 6.1 mmol/l) and postprandial (8 vs. 8.9 mmol/l) cap-               to teach patients to count the carbohydrate
illary glucose and fructosamine (310 vs. 316 µmol/l) were comparable on both the low- and                     content of their meals and to adjust their
high-carbohydrate diets.                                                                                      insulin doses accordingly. Thus, carbohy-
                                                                                                              drate counting provides increased flexibil-
RESULTS — The assessment of meal carbohydrate content by the patients was excellent, with                     ity of food choices and facilitates better
  85% of cases falling within 15% of computer-assisted evaluation. When premeal regular                       glycemic control; such a dietary approach
insulin was prescribed in U/10 g of carbohydrate, the postprandial glycemic rise remained con-                is better suited to intensified insulin ther-
stant (2.4 ± 2.8 mmol/l) over a wide range of carbohydrate ingested (21–188 g) and was not                    apy. This is based on the fact that after a
affected by the glycemic index, fiber, and caloric and lipidic content of the meals. This tight
                                                                                                              mixed meal, the postprandial blood glu-
control was maintained during the low- and high-carbohydrate diet without any change in
insulin requirements (breakfast, 1.5 vs. 1.5 U/10 g of carbohydrate; lunch, 1.0 vs. 1.0; supper,              cose rise is essentially determined by the
1.1 vs. 1.2) and in basal ultralente insulin requirements (22.5 vs. 21.4 U/day).                              amount of carbohydrate in the meal, and
                                                                                                              insulin requirement should be proportional
CONCLUSIONS — These results indicate that in type 1 diabetic subjects 1) increasing the                       to the carbohydrate content of the meal
amount of carbohydrate intake does not influence glycemic control if premeal regular insulin is                (6–10). Compared with the traditional
adjusted to the carbohydrate content of the meals; 2) algorithms based on U/10 g of carbohydrate              meal plan based on food exchanges, carbo-
are effective and safe, whatever the amount of carbohydrate in the meal; 3) the glycemic index,               hydrate counting is more precise, easier to
fiber, and lipidic and caloric content of the meals do not affect premeal regular insulin require-             teach, and much more flexible and theo-
ments; 4) wide variations in carbohydrate intake do not modify basal (ultralente) insulin require-            retically should facilitate better glycemic
ments; and, finally 5) the ultralente-regular insulin regimen allows dissection between basal and
                                                                                                              control (2).
prandial insulin requirements, so that each can be adjusted accurately and independently.
                                                                                                                   However, in the basal-bolus insulin
                                                                    Diabetes Care 22:667–673, 1999            regimen, algorithms based on units of
                                                                                                              insulin per gram of carbohydrate (11,12)
                                                                                                              have not been validated over a wide range
                                                                                                              of carbohydrate content within a given
                                                                                                              mixed meal on an ambulatory basis. Fur-
                                                                                                              thermore, no one has assessed the impact
                                                                                                              of low- and high-carbohydrate diets on
From the Research Center, Centre Hospitalier de l’Université de Montréal, Campus Hôtel-Dieu and Depart-
ment of Medicine, University of Montreal, Montreal, Ontario, Canada.                                          basal insulin requirement. Therefore, the
   Address correspondence and reprint requests to Dr. Jean-Louis Chiasson, Director, Research Group on Dia-   aim of this study was to validate the algo-
betes and Metabolic Regulation, Research Center, CHUM, Campus Hôtel-Dieu, 3850 St. Urbain St., Mon-           rithms used for premeal insulin dose
tréal, Québec, H2W 1T8. E-mail:                                                    adjustment over a wide range of carbohy-
   Received for publication 17 April 1998 and accepted in revised form 1 February 1999.
   Abbreviations: DCCT, Diabetes Control and Complications Trial; GI, glycemic index.                         drate meal content and to assess the effects
   A table elsewhere in this issue shows conventional and Système International (SI) units and conversion     of low- and high-carbohydrate diets on
factors for many substances.                                                                                  basal insulin requirement.

DIABETES CARE, VOLUME 22, NUMBER 5, MAY 1999                                                                                                           667
Carbohydrate counting for type 1 diabetic subjects

RESEARCH DESIGN AND                                  insulin therapy using the basal-bolus          given in the same region (arms, legs, but-
METHODS                                              approach. They were all familiar with car-     tocks) throughout the study.
                                                     bohydrate counting and with adjustment of           Patients were asked to follow, in random
Subjects                                             insulin doses according to specific algo-       order and in a crossover design with no
Nine type 1 diabetic subjects (7 men, 2              rithms.                                        washout period, a low-carbohydrate diet
women) well controlled with ultralente                   Target blood glucose was between 4         (40% carbohydrate, 40% lipids, 20% pro-
(beef-pork) as basal insulin and regular             and 7 mmol/l for the premeal period and at     teins) and a high-carbohydrate diet (60%
insulin (Humulin R) as premeal insulin               bedtime and between 7 and 10 mmol/l for        carbohydrate, 25% lipids, 15% proteins) for
participated in the study. Their mean age            1 h after the meal. Basal insulin was given    14 days each, while maintaining their usual
was 33.4 ± 2.9 years (range 21–46), BMI              as ultralente insulin at bedtime and was       caloric intake. All patients met with a regis-
was 22.6 ± 1.0 kg/m2, the duration of dia-           adjusted by 2 U at a time to achieve fasting   tered dietitian at the screening visit. A dietary
betes was 12.4 ± 2.6 years, and HbA1c was            capillary glucose between 4 and 7 mmol/l.      evaluation was performed based on a 24-h
6.0 ± 0.3%. Two patients had mild                    Premeal regular insulin, given 15–30 min       recall and a food consumption frequency
retinopathy, one had peripheral neuropa-             before the meal, was expressed in U/10 g of    questionnaire to establish their current
thy, and one was treated for hypertension.           carbohydrate to be ingested, as described      caloric intake. At the second visit, they were
All patients were C-peptide–negative after a         previously (11,12). The dose was adjusted      instructed to follow one of the two experi-
glucose challenge. Gastroparesis was                 by 0.2 U/10 g of carbohydrate to achieve       mental diets using a food groups list and a
excluded by a radioisotopic test meal using          and maintain 1-h postprandial capillary        typical diet. They were instructed by the
digestible solid food labeled with 99mTc             blood glucose between 7 and 10 mmol/l.         dietitian on how to fill out the 3-day food
(13). All subjects were highly motivated             Regular insulin was injected in the            diary at the time of consumption. They were
and already well controlled by intensified            abdomen, while ultralente injection was        also instructed to measure food with meas-

Table 1—Diet composition based on the 3-day dietary record

Composition                                        Low-carbohydrate diet                  High-carbohydrate diet                      P values
Energy (kcal)                                           2,244 ± 106.8                        2,374 ± 214.4                               NS
  Total (%)                                              38.8 ± 1.32                          52.8 ± 1.6                               0.001
  Total (g)                                             215.9 ± 7.8                          312.2 ± 26.2                              0.002
  Simple carbohydrate
    Total intake(%)                                      18.9 ± 1.1                           24.4 ± 1.4                               0.004
    Carbohydrate intake (%)                              48.8 ± 2.37                          46.1 ± 1.7                                NS
    Grams                                               104.4 ± 4                            143.6 ± 12.9                              0.03
  Complex carbohydrate
    Total intake (%)                                     19.8 ± 1.1                           28.3 ± 1                                 0.001
    Carbohydrate intake (%)                              51.0 ± 2.37                          53.7 ± 1.7                                NS
    Grams                                               111.8 ± 8.32                         168.1 ± 15.6                              0.001
Glycemic index                                            82.6 ± 1.2                           83.8 ± 1.76                               NS
Fibers (g)                                                 9.3 ± 0.68                          10.9 ± 0.65                               NS
  Total (%)                                               38.1 ± 1.67                          28.7 ± 1.36                             0.001
  Total (g)                                               95.6 ± 6.8                           77.0 ± 9.4                              0.06
  Saturated fatty acids
    Total intake (%)                                      13.3 ± 0.48                           9.6 ± 0.63                             0.001
    Lipid intake (%)                                      35.3 ± 1.1                           33.5 ± 1.3                               NS
  Polyunsaturated fatty acids
    Total intake (%)                                       6.6 ± 0.65                           5.2 ± 0.42                              0.05
    Lipid intake (%)                                      17.0 ± 1.0                           18.1 ± 1.4                                NS
  Cholesterol (mg)                                      347.7 ± 40.5                         254.6 ± 43.0                                NS
Protein (%)                                               19.3 ± 7.9                           15.6 ± 0.92                              0.02
Protein (g)                                             108.4 ± 7.3                            89.9 ± 6.22                              0.02
Protein (g/kg)                                             1.7 ± 0.1                            1.4 ± 0.1                               0.01
Data are means ± SEM. In this table, % always represents percentage of energy (kcal).

668                                                                                                 DIABETES CARE, VOLUME 22, NUMBER 5, MAY 1999
                                                                                                                   Rabasa-Lhoret and Associates

                                                                                                        lary glucose were available for analysis in
                                                                                                        89.5 and 88.3% of cases, respectively. Values
                                                                                                        that were not available were either missing
                                                                                                        (n = 3) or, more frequently, taken outside the
                                                                                                        acceptable time range (n = 31). Whether
                                                                                                        these were included for analysis or not, they
                                                                                                        did not change the results; however, they
                                                                                                        were omitted for figures and final analysis.
                                                                                                             Overall compliance with the two diets
                                                                                                        was good, as shown in Table 1. The pro-
                                                                                                        portion of carbohydrate in the high-carbo-
                                                                                                        hydrate diet was significantly higher than
                                                                                                        in the low-carbohydrate diet, whether it
                                                                                                        was calculated as a percentage of total calo-
Figure 1—Pre- (AC) and postprandial (PC) capillary blood glucose under high- and low-carbohydrate
diets. Data are means ± SEM.
                                                                                                        ries (52.8 ± 1.6 vs. 38.8 ± 1.3%; P 0.004)
                                                                                                        or absolute amount (312.2 ± 26.2 vs. 215.9
                                                                                                        ± 7.8 g; P 0.03) of carbohydrate. The per-
uring cups or spoons and to weigh foods            (1 h) capillary blood glucose measured               centages of simple and complex carbohy-
they could not measure. At the subsequent          with a memory glucose meter (One Touch;              drate, the glycemic index, and fiber intake
visit 14 days later, they met with the dietitian   Lifescan Canada, Burnaby, British Colum-             were similar in the two diets. The low-fiber
again to review the diaries and to clarify any     bia). These dietary records provided infor-          intake reflects the average amount of fibers
missing elements, including food prepara-          mation on 162 meals with a very wide                 normally ingested by our population (19).
tion methods, amount of fat added, etc.            range of carbohydrate intake (21–188 g).             Because both diets were isocaloric, the pro-
Dietary data from the diaries were analyzed        Glycosylated hemoglobin was measured at              portion of lipids in the high-carbohydrate
for carbohydrates, proteins, lipids, fibers and     the beginning of the study by high-perfor-           diet was lower than that in the low-carbo-
glycemic index (GI), using “Nutritionist 3”        mance liquid chromatography (normal                  hydrate diet (28.7 ± 1.4 vs. 38.1 ± 1.7%; P
computer software (version 7.0 1991 from           range 3.5–5.7%) (17). At the end of each                0.001); however, the distribution of dif-
N-Squared Computing, Salem, OR) and the            diet, glycemic control was assessed by fruc-         ferent sources of lipids, saturated, mono-
Bayer Database as modified by T.M.S.                tosamine measurement (18). Patients were             unsaturated, and polyunsaturated fatty acid
Wolever (1988, unpublished database).              instructed to maintain their usual physical          and cholesterol was constant. Protein
These modifications were based on the Con-          activities throughout the study and were             intake was slightly higher in the low-car-
densed Canadian Nutrient File (14), which          asked to report any unusual physical activ-          bohydrate diet (19.3 ± 0.8 vs. 15.6 ± 0.9%;
was updated for carbohydrates and fiber             ities; none were reported. The study was             P 0.02). During the high-carbohydrate
according to the manufacturers’ information        approved by the ethics committee with all            diet, capillary glucose values tended to be
(McCance and Widdowson [15]), and by               participating subjects giving signed volun-          slightly higher (Fig. 1) but never reached
estimation of some unlisted food items. We         tary informed consent.                               statistical significance. Maintenance of good
did not have any problem regarding missing                                                              glycemic control, as shown by capillary
values in the database. The GI values, using       Statistical analysis                                 glucose, was confirmed by the measure-
white bread as a reference, were included in       Statistical analysis was done using analysis         ment of fructosamine, which was 296 ±
the Bayer Database based on published data         of variance repeated measures with the               14.5 µmol/l at baseline and 310 ± 12.4
where possible (16) and on estimation for          paired t test when applicable (SigmaStat,            µmol/l at the end of the low-carbohydrate
other food items by Wolever. When coding           version 2, Jandel, San Rafael, CA).                  diet compared with 316 ± 15.4 µmol/l at
the food diaries, if a particular food was                                                              the end of the high-carbohydrate diet.
missing in the database, the dietitian would       RESULTS — Data on meal composition,                       Correlation between patient and nutri-
find a food or a combination of food with           carbohydrate evaluation and insulin doses            tional software-assisted evaluation of car-
the same value. Recipes were entered as sin-       were compiled by the patients in 100% of             bohydrate content of the meals was excellent
gle ingredients dividing for the amount actu-      cases. Data on pre- and postprandial capil-          (r = 0.89) (Fig. 2). In 85.2% of cases, patient
ally eaten by the patient.
      The patients were advised on a bed-
time snack ( 20 g carbohydrate), without           Table 2—Insulin requirements
insulin to prevent early nocturnal hypogly-
cemia. No specific instructions were given          Insulin                                             High-carbohydrate diet   Low-carbohydrate diet
regarding fiber intake in this study. In the
last week of each diet, they were asked to         Premeal insulin, regular (U/10 g of carbohydrate)
fill out the 3-day dietary journal, recording         Breakfast                                                1.5 ± 0.1                1.5 ± 0.1
at the time of consumption (2 weekdays               Lunch                                                   1.0 ± 0.03                1.0 ± 0.1
and 1 weekend day), for each meal and                Dinner                                                   1.2 ± 0.1               1.1 ± 0.04
snack, the amount of food consumed, the            Basal insulin, ultralente (U/day)
carbohydrate content, the amount of                  Bedtime                                                 21.4 ± 3.8               22.5 ± 4.0
insulin injected, and pre- and postprandial        Data are means ± SEM.

DIABETES CARE, VOLUME 22, NUMBER 5, MAY 1999                                                                                                       669
Carbohydrate counting for type 1 diabetic subjects

                                                                                                         No severe hypoglycemia occurred
                                                                                                    throughout the study. The frequency of
                                                                                                    minor hypoglycemia ( 3.5 mmol/l) was
                                                                                                    not different between the two diets (low-car-
                                                                                                    bohydrate diet, n = 15; high-carbohydrate
                                                                                                    diet, n = 14). One minor hypoglycemic
                                                                                                    episode was attributed to an overevaluation
                                                                                                    of carbohydrate content in the low-carbo-
                                                                                                    hydrate diet, and another was due to inap-
                                                                                                    propriate insulin adaptation for an unusual
                                                                                                    physical exercise on the high-carbohydrate
                                                                                                    diet; all other episodes of hypoglycemia
                                                                                                    could be explained by classic risk factors.
                                                                                                    The high regular insulin doses related to the
                                                                                                    high-carbohydrate content of meals ( 100
                                                                                                    g/meal) were not associated with episodes of
                                                                                                    late postprandial hypoglycemia. Body
                                                                                                    weight was stable throughout the study.

                                                                                                    CONCLUSIONS — For type 1 dia-
                                                                                                    betic patients treated with multiple
                                                                                                    insulin injections, the use of carbohydrate
Figure 2—Correlation between patient and computer-assisted assessment of carbohydrate content of    counting and premeal insulin prescrip-
mixed meals.                                                                                        tion per grams of carbohydrate offers
                                                                                                    numerous advantages, including greater
                                                                                                    flexibility in choosing food, portion size,
evaluation was within 15% of dietetic eval-      had no influence on the postprandial                and timing of meals as well as in physical
uation. For the remaining 14.8%, most            increase in capillary glucose (Fig. 4).            activity (4,7,20). The usefulness of the
cases were an underevaluation, and no                For a comparable carbohydrate con-             system has been confirmed in the DCCT
major adverse consequence occurred, as           tent of a meal, there was a wide variation in      with better control for patients using car-
judged by capillary glucose results or the       glycemic response, whether it was evalu-           bohydrate counting in the intensively
frequency of hypoglycemia. Only one              ated as the 1-h capillary blood glucose            treated group resulting in a supplemen-
minor hypoglycemia and one hypergly-             absolute value (8.4 ± 3.7 mmol/l) or as the        tary 0.56% reduction in HbA1c, compared
cemia episode ( 15 mmol/l) could be              postprandial increase in blood glucose (2.4        with those not counting carbohydrate (5).
directly related to over- or underevaluation     ± 2.8) with a slight trend toward decreased        However, the validity of the algorithms
of the carbohydrate content of meals.            fluctuations for high-carbohydrate content          has never been shown over a wide range
     Using the aforementioned premeal            ( 100 g) (Fig. 2).                                 of carbohydrate content in mixed meals
algorithms for insulin adjustment, glycemic
control was well maintained, despite a wide
range of carbohydrate content in individual
mixed meals (Fig. 1). Furthermore, such
tight glycemic control did not necessitate
any change in the dosage of premeal insulin
requirement based on 10 g of carbohy-
drate or in the dosage of basal ultralente
insulin (Table 2).
     There was no correlation between 1-h
postprandial capillary glucose, the post-
prandial increase in blood glucose, and the
total carbohydrate content of the meal. The
mean postprandial increase in blood glu-
cose was linear and horizontal (2.4 ± 2.8
mmol/l) over the entire range of carbohy-
drate content explored, indicating the
validity and the safety of the premeal algo-
rithms over a wide range of carbohydrate
(21–188 g of carbohydrate) (Fig. 3). This
relationship was true for each meal under
both diets. The fiber content of the meal,        Figure 3—Postprandial increase in capillary blood glucose over a wide range of carbohydrate intakes
glycemic index, and caloric or lipid intake      in mixed meals for type 1 diabetic subjects using carbohydrate counting.

670                                                                                                DIABETES CARE, VOLUME 22, NUMBER 5, MAY 1999
                                                                                                                       Rabasa-Lhoret and Associates

                                                                                                            These studies were done, however, in
                                                                                                            poorly controlled, conventionally treated
                                                                                                            type 1 diabetic subjects without specific
                                                                                                            algorithms for insulin adjustment and/or
                                                                                                            without capillary glucose measurement. It
                                                                                                            is clear that, to be effective, intensified
                                                                                                            insulin therapy requires a teaching and
                                                                                                            training program, not only on carbohy-
                                                                                                            drate counting but also on various aspects
                                                                                                            of diabetes management, particularly spe-
                                                                                                            cific algorithms to adjust insulin doses to
                                                                                                            achieve good glycemic control (12). It is
                                                                                                            interesting to note that the morning pre-
                                                                                                            meal insulin dose was 50% higher than the
                                                                                                            noon and evening premeal dose (1.5 vs. 1.0
                                                                                                            U/10 g carbohydrate). This is consistent
                                                                                                            with other studies that we have done in the
                                                                                                            past (11,12,19). It is suggested that there is
                                                                                                            an increased insulin resistance in the morn-
                                                                                                            ing compared with the rest of the day. It is
                                                                                                            most likely explained, at least in part, by
Figure 4—Postprandial increase in capillary blood glucose over a wide range of fiber, total lipid, caloric   the dawn phenomenon (30), because the
content, and glycemic index of mixed meals for type 1 diabetic subjects using carbohydrate counting.        fasting capillary blood glucose was fairly
                                                                                                            well controlled by the evening basal ultra-
                                                                                                            lente insulin (Fig. 1).
with subcutaneous insulin injections. In              week after week. But the best argument in                  The mean increase in postprandial
the present study, we have demonstrated               support of the feasibility of carbohydrate            plasma glucose was 2.4 ± 2.8 mmol/l over
that intensive insulin therapy in type 1              counting by patients is the tight glycemic            the entire range of carbohydrate content
diabetic patients with the basal-bolus                control achieved in our study when adjust-            studied (Fig. 3). The high standard error
(ultralente-regular insulin) approach                 ing their premeal insulin dose based on the           attests to the wide fluctuation in postpran-
based on carbohydrate counting is feasi-              carbohydrate content of meals (Fig. 1).               dial glycemic excursion for comparable car-
ble, effective, and safe.                                  That premeal insulin requirement                 bohydrate intake, despite the use of
     It has been claimed that the estimation          should be proportional to the carbohydrate            algorithms based on carbohydrate count-
of dietary carbohydrate is difficult to use in        content of meals has been shown by a                  ing. A number of factors can explain this
the treatment of diabetes, even for health            number of investigators (6–8,10,20,23,24).            wide variation: 1) the poor reproducibility
professionals (21). The present study                 Using artificial pancreas, Slama et al. (8,10)         of insulin absorption after subcutaneous
shows, on the contrary, that motivated                have reported in type 1 diabetic subjects             injections (31–33); 2) the timing of insulin
patients are able to learn to count carbohy-          that there is a linear relationship between           injections before meals (34); 3) variation in
drate in meals to be ingested. On the basis           the amount of insulin delivered intra-                the accuracy and precision of capillary glu-
of the 3-day nutritional diary, the validity of       venously to maintain normal postprandial              cose measurements (22); 4) premeal plasma
which has been established for dietary inter-         blood glucose profile and the amount of                glucose (35); and finally, 5) meal composi-
vention studies, we report that 85% of                carbohydrate (from 20 to 140 g) in the                tion that can affect gastric emptying (36,37)
carbohydrate estimation by patients was               mixed meals. Others have shown that these             or insulin action (38,39). Despite all these
within 15% of computer-assisted nutritional           patients cannot vary the carbohydrate con-            potential factors that could affect the post-
evaluation (Fig. 2). This is more than                tent of their diet without changing their             prandial rise in plasma glucose, they do not
acceptable if one considers that there is at          insulin doses to avoid hyper- or hypogly-             preclude the achievement of tight glycemic
least a 10% variation between the different           cemia (25,26). In the present study, using            control. In fact, the present study demon-
software used in assessing the carbohydrate           specific algorithms based on the carbohy-              strates that in home situations on a day-to-
content of food (our experience), and that a          drate content of meals, patients were able            day basis, the glycemic index of the diet, the
20% variation in capillary glucose meas-              to maintain good postprandial plasma glu-             fiber and lipid content, as well as the caloric
urement is deemed to be acceptable (22).              cose during both the low- and high-carbo-             value of meals do not have any significant
Such accuracy requires teaching by a dieti-           hydrate diets (Fig. 1). Furthermore, our              impact on premeal insulin requirement to
tian familiar with carbohydrate counting              study shows that when premeal insulin is              achieve and maintain good postprandial
and the provision of a pamphlet or booklet            given per 10 g of carbohydrate, the post-             glycemic control (Fig. 4). We have already
with the carbohydrate content of most com-            prandial plasma glucose rise is not affected          shown in the basal-bolus approach with
mon food items. In our experience, diabetic           by the amount of carbohydrate over a wide             constant carbohydrate intake that premeal
patients learn very rapidly to recognize the          range (21–188 g) in mixed meals (Fig. 3).             or basal insulin requirements are not mod-
carbohydrate content of food, because                 Some investigators have reported no bene-             ified by large variations in the glycemic
   90% of food items usually eaten by a               fit of carbohydrate counting compared                  index or fiber content of mixed meals (19).
given subject recur within 1 week as well as          with simple dietary advice (21,27–29).                Therefore, it is suggested that the carbohy-

DIABETES CARE, VOLUME 22, NUMBER 5, MAY 1999                                                                                                          671
Carbohydrate counting for type 1 diabetic subjects

drate content of meals is the major deter-        ultralente is small (47) and most probably          potentially help to improve glycemic con-
minant of premeal insulin requirement; tak-       of no clinical significance (46,48–50). Other        trol and quality of life.
ing this factor only into account is sufficient   studies have shown that human ultralente
to achieve relatively tight postprandial          could be used as basal insulin in intensified
glycemic control. It is very likely that these    insulin therapy (47).                               Acknowledgments — We are grateful to
results also apply to lispro as premeal                 Finally, our study reveals that adjusting     Susanne Bordeleau-Chénier for preparing this
insulin. Lispro insulin has been shown to         the premeal insulin dose with algorithms            manuscript and illustrations and to Ovid Da
be associated with better postprandial            based on the carbohydrate content of meals          Silva for editing the text.
glycemic control and to be more repro-            is valid and safe. Even in situations of
ducible than regular insulin (40,41). Lispro      extreme carbohydrate meal content, the
has also been reported to be a suitable pre-      algorithms remain safe with no increased            References
meal insulin for intensified treatment of          risk of hypoglycemia on the low-carbohy-             1. American Diabetes Association: Nutritional
type 1 diabetic patients when used with           drate diet or of hyperglycemia on the high-             recommendation and principles for people
                                                                                                          with diabetes mellitus (Positional State-
human ultralente as basal insulin (42,43).        carbohydrate diet. In fact, they are probably           ment). Diabetes Care 17:519–522, 1994
Furthermore, we have demonstrated that            safer than conventional therapy, where               2. Franz MJ, Horton ES, Bantle JP, Beebe CA,
lispro given as premeal boluses based on          patients often vary the carbohydrate content            Brunzell JD, Coulston AM, Henry RR,
carbohydrate control of meals results in          of meals due to lack of compliance (51). Our            Hoogwerf B, Stacpoole PW: Nutritional
better postprandial glycemic control than         results do not mean, however, that type 1               principles for the management of diabetes
regular insulin when administered by con-         diabetic subjects do not have to follow rec-            and related complications. Diabetes Care
tinuous subcutaneous insulin infusion (44).       ommended dietetic prescriptions, including              17:490–518, 1994
     The present study also demonstrates          qualitative and quantitative carbohydrate or         3. DCCT Research Group: The effect of inten-
that basal ultralente insulin is not affected     fat intake (1,2). Carbohydrate counting is a            sive treatment of diabetes on the develop-
by the amount of carbohydrate in the meal         useful tool that increases flexibility and thus          ment and progression of long-term
                                                                                                          complications in insulin-dependent diabetes
(Table 2), suggesting that our algorithms for     allows individualized dietetic approaches to            mellitus. N Engl J Med 329:977–986, 1993
insulin adjustment allow us to dissect            help reach better glycemic control.                  4. Daly A: Carbohydrate counting: new teach-
between basal insulin requirement from                  Unfortunately, the impact of the low-             ing resources. Pract Diabetol 15:19–23, 1996
that required for meals. This is strongly         and high-carbohydrate diet on lipid profile           5. Delahanty LM, Halford BN: The role of diet
supported by the observation that a major         was not evaluated in this study. Low-carbo-             behaviors in achieving improved glycemic
increase in dietary carbohydrate does not         hydrate diets are frequently high-fat diets             control in intensively treated patients in
have any significant effect on basal insulin       with potential adverse effect on plasma lipids          the diabetes control and complication trial.
requirement. It is also supported by other        and are not recommended (52). High-car-                 Diabetes Care 16:1453–1458, 1993
studies using artificial pancreas (6,9). We        bohydrate diets, particularly in type 2 dia-         6. Vlachokosta FV, Piper CM, Gleason R,
have also made similar observation in             betic subjects, have been shown to be                   Kinzel L, Kahn CR: Dietary carbohydrate, a
                                                                                                          Big Mac, and insulin requirements in type 1
another study where we have varied the            associated with prolonged increase in VLDL              diabetes. Diabetes Care 11:330–336, 1988
glycemic index and the fiber content of the        triglycerides (53). In type 1 diabetic subjects      7. Irsigler K, Kritz H: Long-term continuous
diet (19). This does not mean, however, that      on intensive insulin therapy, Stone and Con-            intravenous insulin therapy with a portable
the premeal insulin (regular) does not, at        nor (54) reported no significant increase in             insulin dosage-regulating apparatus. Dia -
least partially, provide some basal insulin       plasma triglycerides. It may be that high-car-          betes 28:196–203, 1979
need. In fact, that premeal regular insulin       bohydrate diet does not adversely affect the         8. Halfon P, Belkhadir J, Slama G: Correlation
did provide some basal need is suggested by       lipid profile if optimal glycemic control is             between amount of carbohydrate in mixed
Anderson and Brunelle (45), who have              preserved, such as in the present study.                meals and insulin delivery by artificial pan-
shown that when patients are transferred                The present study indicates that 1) a             creas in seven IDDM subjects. Diabetes Care
from premeal regular to premeal lispro, the       large increase in carbohydrate intake does              12:427–429, 1989
                                                                                                       9. Mirouze J, Selam J-L, Pham TC, Cavadore D:
basal insulin requirement had to be               not modify glycemic control if regular insulin          Evaluation of exogenous insulin homeostasis
increased to maintain similar glycemic con-       dose is adjusted in proportion to the carbo-            by the artificial pancreas in insulin-dependent
trol. Finally, it is possible that the bedtime    hydrate content of the meal; 2) premeal algo-           diabetes. Diabetologia 13:273–278, 1977
snack did influence the ultralente require-        rithms in units of insulin per 10 g of              10. Slama G, Klein J-C, Delage A, Ardila E,
ment. However, this was not reflected by a         carbohydrate are valid and safe over a wide             Lemaignen H, Papoz L, Tchobroutsky G:
difference in the basal insulin requirement       range of carbohydrate content in mixed                  Correlation between the nature and
between the low- and high-carbohydrate            meals; 3) variations in the glycemic index,             amount of carbohydrate in meal intake and
diet. The patients participating in the pres-     fiber, lipid, or caloric intake in daily life have       insulin delivery by the artificial pancreas in
ent study were using beef-pork ultralente         little influence on premeal insulin require-             24 insulin-dependent diabetics. Diabetes
insulin as basal insulin. Since the introduc-     ments; 4) wide variations in carbohydrate               30:101–105, 1981
                                                                                                      11. Chiasson J-L, Ducros F Poliquin-Hamet M,
tion of human ultralente insulin, much con-       intake do not modify basal ultralente insulin           Lopez D, Lecavalier L, Hamet P: Continuous
troversy has arisen over differences in           requirements; and 5) the ultralente-regular             subcutaneous insulin infusion (Mill-Hill
pharmacokinetics. It is very likely, however,     insulin regimen allows dissection of basal              Infuser) vs. multiple injections (Medi-Jector)
that our observations with animal ultralente      from prandial insulin requirements.                     in the treatment of insulin-dependent dia-
also apply to human ultralente because their            Carbohydrate counting can be used                 betes mellitus and the effect of metabolic
time of action is not very different (46), and    safely as part of the intensification of treat-          control on microangiopathy. Diabetes Care
the spike of action described for human           ment of type 1 diabetic patients and can                7:331–337, 1984

672                                                                                                   DIABETES CARE, VOLUME 22, NUMBER 5, MAY 1999
                                                                                                                        Rabasa-Lhoret and Associates

12. Rabasa-Lhoret R, Chiasson J-L: Insulin                 composition is a determinant of lispro-               role in insulin sensitivity and the metabolic
    therapy: decreasing diabetes complications.            induced hypoglycemia in IDDM. Diabetes                disturbances of diabetes mellitus. Lancet
    JCME 9:75–88, 1997                                     Care 20:152–155, 1997                                 i:785–789, 1963
13. Feldman M, Smith HJ, Simon TR: Gastric           27.   Waldron S: Current controversies in the         40.   Selam J-L: Du concept des analogues rapi-
    emptying of solid radiopaque markers: stud-            dietary management of diabetes in child-              des aux propriétés de l’insuline lispro. Dia -
    ies in healthy subjects and diabetic patients.         hood and adolescence. Br J Hosp Med                   betes Metab 23:45–49, 1997
    Gastroenterology 87:895–902, 1984                      56:450–455, 1996                                41.   Vignati L, Anderson JH, Szwast S,
14. Bureau of Nutritional Sciences, Department       28.   Walker L: Counted diet versus non-                    Symanowski S: Twice-daily lispro results in
    of National Health and Welfare Canada:                 counted diet in the management of insulin-            less glucose variability compared to human
    Nutrient value of some common foods.                   dependent diabetes in children. Diabetes              regular insulin. Diabetologia 38 (Suppl.
    Computer file. Ottawa, Canada, Dept. of                 Gen Pract Winter:7–8, 1992                            1):A191, 1995
    National Health and Welfare Canada               29.   Mitchell RD, Nowakowska JA, Hurst AJ:           42.   Burge MR, Waters DL, Holcombe JH,
    (COP.CA.2.1989-1200)                                   Comparison of official 10 g carbohydrate              Schade DS: Prolonged efficacy of short-act-
15. McCance RA: In The Composition of Food. 4th            exchange system with simplified dietary                ing insulin lispro in combination with
    ed. Paul AA, Southgate DAT, Eds. London,               advice in insulin-dependent diabetics. J              human ultralente in insulin-dependent dia-
    Her Majesty’s Stationery Office, 1978, p. 417          Hum Nutr Diet 3:19–26, 1990                           betes mellitus. J Clin Endocrinol Metab
16. Jenkins DJA, Wolever TMS, Jenkins AL:            30.   Bolli GB, Gerich JE: The “dawn phenome-               82:920–924, 1997
    Starchy foods and glycemic index. Diabetes             non”: a common occurrence in both non-          43.          Bastyr EJ III, Holcombe JH, Ander-
    Care 11:149–159, 1988                                  insulin-dependent and insulin-dependent               son JH, Clore JN: Mixing insulin lispro
17. Cole RA, Soeldner JS, Dunn PJ, Bunn HF:                diabetes mellitus. N Engl J Med 310:                  and ultralente insulin. Diabetes Care
    A rapid method for the determination of                746–750, 1984                                         20:1047–1048, 1997
    glycosylated hemoglobin using high pres-         31.   Antisferov M, Woodworth JR, Mayorov A,          44.   Zinman B, Tildesley H, Chiasson J-L, Tsui
    sure liquid chromatography. Metabolism                 Ristic S, Dedov I: Within patient variability         E, Strack T: Insulin Lispro in CSII: results of
    27:289–301, 1978                                       in postprandial glucose excursion with lispro         a double-blind crossover study. Diabetes
18. Johnson RN, Metcalf PA, Baker JR: Fruc-                insulin analog compared with regular                  46:440–443, 1997
    tosamine: a new approach to the estimation             insulin. Diabetologia 38 (Suppl. 1):A190,       45.   Anderson JH, Brunelle RL: Reduction of
    of serum glycosylprotein: an index of dia-             1995                                                  postprandial hyperglycemia and frequency
    betic control. Clin Chim Acta 127:87–95,         32.   Galloway JA, Spradlin CT, Nelson RL,                  of hypoglycemia in IDDM patients on
    1983                                                   Wentworth SM, Davidson JA, Swarner JL:                insulin-analog       treatment.       Diabetes
19. Lafrance L, Rabasa-Lhoret R, Poisson D,                Factors influencing the absorption, serum              46:265–270, 1997
    Ducros F Chiasson J-L: The effects of dif-             insulin concentrations, and blood glucose       46.   Seigler DE, Olsson GM, Agramonte RF,
    ferent glycaemic index foods and dietary               responses after injections of regular insulin         Lohman VL, Ashby MH, Reeves ML, Skyler
    fibre intake on glycaemic control in type 1             and various insulin mixtures. Diabetes Care           JS: Pharmacokinetics of long acting (ultra-
    diabetic patients on intensive insulin ther-           4:366–376, 1981                                       lente) insulin preparations. Diabetes Nutr
    apy. Diabet Med 15:972–978, 1998                 33.   Wredling R, Liu D, Lins P-E, Adamson U:               Metab 4:267–273, 1991
20. Chantelau E, Sonnenberg GE, Stanitzek-                 Variation of insulin absorption during sub-     47.   Freeman SL, O’Brien PC, Rizza RA: Use of
    Schmidt I, Best F Altenähr H, Berger M:                cutaneous and peritoneal infusion in insulin-         human ultralente as the basal insulin com-
    Diet liberalization and metabolic control in           dependent diabetic patients with                      ponent in treatment of patients with IDDM.
    type 1 diabetic outpatients treated by con-            unsatisfactory long-term glycaemic response           Diabetes Res Clin Pract 12:187–192, 1991
    tinuous subcutaneous insulin infusion.                 to continuous subcutaneous insulin infu-        48.   Selam J-L, Turner D, Woertz L, Eichner
    Diabetes Care 5:612–616, 1982                          sion. Diabetes Metab 17:456–459, 1991                 HL, Lauritano A, Charles MA: Comparison
21. Waldron S: Childhood diabetes-current            34.   Dimitriadis G, Hatziagellaki E, Alex-                 of safety and effectiveness of human and
    dietary management. Current Paediatrics                opoulous E, Kordonouri O, Komesidou V,                bovine long-acting insulins. Diabetes Res
    3:138–141, 1993                                        Ganotakis M, Raptis S: Effects of -glu-               12:131–134, 1989
22. Jones HE, Cleave B, Zinman B, Szalai JP      ,         cosidase inhibition on meal glucose toler-      49.   Hildebrandt P, Berger A, Volund A, Kühl C:
    Nichol HL, Hoffman BR: Efficacy of feed-               ance and timing of insulin administration             The subcutaneous absorption of human
    back from quarterly laboratory comparison              in patients with type I diabetes mellitus.            and bovine ultralente insulin formulations.
    in maintaining quality of a hospital capil-            Diabetes Care 14:393–398, 1991                        Diabet Med 2:355–359, 1985
    lary blood glucose monitoring program.           35.   Nielsen PH, Nielsen GL: Preprandial blood       50.   Holman RR, Steemson J, Darling P, Reeves
    Diabetes Care 19:168–170, 1996                         glucose values: influence on glycemic                  WG, Turner RC: Human ultralente insulin.
23. Grant AM, Christie MR, Ashcroft SJH:                   response studies. Am J Clin Nutr                      Br Med J 288:665–668, 1984
    Insulin release from human pancreatic islets           49:1243–1246, 1989                              51.   Toeller M: Diet and diabetes. Diabetes Metab
    in vitro. Diabetologia 19:114–117, 1980          36.   Maerz LL, Sankaran H, Scharpf SJ, Deveney             Rev 9:93–108, 1993
24. Rizza RA, Gerich JE, Haymond MW, West-                 CW: Effect of caloric content and composi-      52.   American Diabetes Association: Nutrition
    land RE, Hall LD, Clemens AH, Service FJ:              tion of a liquid meal on gastric emptying in          recommendations and principles for people
    Control of blood sugar in insulin-depen-               the rat. Am J Physiol 267:R1163–R1167,                with diabetes mellitus. Diabetes Care 21
    dent diabetes: comparison of an artificial              1994                                                  (Suppl. 1):S32–S35, 1998
    pancreas, continuous subcutaneous insulin        37.   Vist GE, Maughan RJ: Gastric emptying of        53.   Coulston AM, Hollenbeck CB, Swislocki
    infusion and intensified conventional ther-             ingested solutions in man: effect of bever-           LM, Reaven GM: Persistence of hyper-
    apy. N Engl J Med 303:1313–1318, 1980                  age glucose concentration. Med Sci Sports             triglyceridemic effect of low-fat high-car-
25. Levy-Marchal C, Perlman K, Zinman B,                   Exerc 26:1269–1273, 1994                              bohydrate diets in NIDDM patients.
    Albissier M: Preprogrammed intravenous           38.   Reaven GM: Role of insulin resistance in              Diabetes Care 12:94–100, 1989
    insulin infusion in diabetic humans: meta-             human disease. Diabetes 37:1595–1607,           54.   Stone D, Connor WE: The prolonged
    bolic consequences of altering meal size.              1988                                                  effects of a low-cholesterol, high-carbohy-
    Diabetes Care 9:283–290, 1986                    39.   Randle PJ, Garland PB, Hales CN, New-                 drate diet upon serum lipids in diabetic
26. Burge MR, Castillo KR, Schade DS: Meal                 sholm EA: The glucose-fatty acid cycle: its           patients. Diabetes 12:127–132, 1963

DIABETES CARE, VOLUME 22, NUMBER 5, MAY 1999                                                                                                              673

Shared By: