hypoglycemia was observed, occurring asymptomatically at night, and was followed by a ... sessing diabetic control (e.g., daily urine tests for glucose.
s
tandard Parameters of Diabetic Control: Are They Reliable?
ROBERT J. WINTER, NEIL J. STONE, JOYCE E. WISE, HOWARD S. TRAISMAN, AND ORVILLE C. GREEN
To evaluate the reliability of the traditional methods to assess short-term control of diabetes, 25 children with insulin-dependent diabetes were studied with a 24-h glucose profile in addition to the traditional assessment techniques. Patient compliance was eliminated as much as possible from the experimental design. The correlation of the routine methods with the 24-h glucose profile was excellent, and a scoring system for control was empirically derived. The single method of assessment that correlated best with the overall control score was the traditional daily urine test. In 6 of the 25 subjects studied, relative hypoglycemia was observed, occurring asymptomatically at night, and was followed by a hyperglycemic rebound. Traditional assessment techniques did not detect this event. Five additional patients had symptomatic daytime hypoglycemia. We conclude that the traditional daily urine tests are adequate indicators of day-to-day control in most diabetic patients, given adequate compliance. Our data also suggest that asymptomatic nocturnal hypoglycemia occurs frequently in children with diabetes, although clinical proof is difficult in the absence of a 24-h glucose profile, DIABETES CARE2.336-341, JULY-AUGUST 1979.
H
ow does the clinician assess control in the diabetic patient on a day-to-day basis? This question is fundamental to patient management and yet there is no close agreement on definitions of'good" control using traditional methods of assessment.1"5 The measurement of glycosylated hemoglobin (hemoglobin A lc ) may prove to be a better indicator of long-term control, but it provides no information about day-to-day variation in blood glucose. The value of conventional parameters for assessing diabetic control (e.g., daily urine tests for glucose and ketones, plasma"glucose determinations, and 24-h quantitative urinary glucose excretion) has been questioned.6 If such conventional parameters are indeed unreliable, this could reflect a limitation of the methods, poor patient compliance, or a combination of the two. In an attempt to answer these questions, we evaluated 25 children with insulin-dependent diabetes in a manner that eliminated patient compliance as a major determinant of overall metabolic control. Using a portable, constant blood withdrawal pump permitting an accurate 24-h profile of plasma glucose as a standard of control, we compared routine parameters of day-to-day control with the profile and with an empiri-
336
cally derived scoring system. In this setting, in which patient compliance is not a major factor, there is excellent correlation between the standard indices of control and the 24-h profile of plasma glucose. Moreover, we have shown that hyperglycemia rebound following asymptomatic hypoglycemia was common and usually unrecognized by the routine indicators of control. PATIENTS STUDIED
Twenty-five children with insulin-dependent diabetes (duration greater than 1 yr) were selected at random from the Diabetes Clinic at Children's Memorial Hospital. All were in good general health; subjective assessment by the physicians attending these children suggested an even distribution among good-fair-poor control categories. Each subject was admitted to the Clinical Research Center (CRC) of Children's Memorial Hospital after signed informed consent was obtained from the subjects and their parents. The usual insulin and dietary regimen of each child was maintained during the hospitalization; a 48-h period prior to the 24-h blood withdrawal study allowed each subject to become acclimated to the hospital
DIABETES CARE, VOL. 2 NO. 4, JULY-AUGUST 1979
ASSESSING STANDARD PARAMETERS OF DIABETIC CONTROL / ROBERT J. WINTER AND ASSOCIATES
routine. General descriptive information on the subjects is presented in Table 1. Table 2 shows the daily schedule followed by each subject.
TABLE 2 Daily schedule of patients studied Clock time
Schedule
Urine block
CLINICAL METHODS
O
n the third day of hospitalization a constant blood withdrawal procedure was begun for each subject, using a nonthrombogenic intracath connected to a portable, constant withdrawal pump (Cormed), as previously described.7 Blood was withdrawn at a constant rate of 8.0 ml/h; collection vessels, kept on ice, were changed every 30 min and the plasma immediately separated and frozen. Subjects were fully ambulatory on the CRC and in general maintained a normal routine of daytime activities and sleep. Urine was collected in "blocks" for quantitative glucose measurement simultaneously with the blood withdrawal. Block collections corresponded to the (1) breakfast-lunch, (2) lunch-dinner, (3) dinner-bedtime, and (4) bedtime-breakfast time periods, as shown in Table 2. In addition, urine glucose and ketones were determined in second-voided urine specimens obtained prior to meals and at bedtime. Continuous supervision was maintained throughout the study by nurses especially trained in the care of diabetes. METHODS
Glucose was measured in each 30-mih aliquot of plasma and in the block urine collections using a glucose oxidase method. Glucose and ketones were estimated in the secondvoided urines by the 2-Drop Clinitest and Acetest methods, respectively. The terminology used for this study is summarized in Table 3. The integrated concentration of glucose (ICG) for the 24-h study period was determined for each subject as a mean of the 48 half-hourly values. In addition, each profile was examined
0700 0730 0800 0830
First urine void Second urine void Insulin injection and breakfast Finish breakfast
Morning 0730-1100
1100 1130 1200 1230
First urine void Second urine void Lunch Finish lunch
Afternoon 1130-1600
1500 1600 1630 1700 1730
Afternoon snack First urine void Second urine void Insulin (if on split dose) and dinner Complete dinner
Evening 1630-2000
2000 2030 2100 2200
First urine void Second urine void Bedtime snack Retire for night
Nighttime 2030-0700
in detail fo evaluate the glucose excursions throughout the study period. The fasting blood glucose (FBS) was estimated as the glucose concentration in the plasma collected in the 30-min interval immediately preceding the morning insulin injection prior to breakfast. The maximum (G-max) and niinimum (G-min) glucose values were similarly determined from the appropriate half-hourly value. Glucose brackets (GB), originally described by Koenig et al., 8 were estimated by the sum of the three preprandial 30-min glucose values and the three 30-min glucose values from the interval 60-90 min
TABLE 3 Terminology for parameters of diabetic control
TABLE 1 Patients studied with 24-h glucose profiles
ICG N = 25 Male Female Age Duration of diabetes
GB
10 15 13.5 ± 2.4 yr 4.8 ± 3.0 yr
Insulin therapy Single dose Split dose
19 6
Clinical assessment Good control Fair control Poor control
8 10 7
(mean ± 1 SD)
24 UG G-max G-min FBS £U
£K
Integrated concentration of glucose, calculated as the mean of the 48 half-hourly values (mg/dl). Glucose brackets, calculated as the sum of the three preprandial and three postprandial glucose values (mg/dl). 24-h urinary glucose (g/24 h). Maximum glucose value during the 24-h study period (mg/dl). Minimum glucose value during the 24-h study period (mg/dl). Fasting blood glucose (mg/dl). Sum of the four, second-voided urine tests for glucose (g/dl). Each urine test has potential maximum of 5 g/dl; hence maximum ^ U is 20 g/dl. Sum of the four, second-voided urine tests for ketones, in arbitrary units. Each test was scored 0-1-2-3, corresponding to negative-small-moderate-large. Potential maximum of 2)K is thus 12.
DIABETES CARE, VOL. 2 NO. 4, JULY-AUGUST 1979
337
ASSESSING STANDARD PARAMETERS OF DIABETIC CONTROL / ROBERT J. WINTER AND ASSOCIATES
TABLE 4 Diabetic control—quartile data (mean ± 1 SD) Insulin-dependent diabetic patients Parameter
Normals
No. of patients ICG
8 117 ± 13 697 ± 72 1 153 ± 21 87 ± 11 101 ± 17 0
GB 24 UG G-max G-min FBS
2U
Nos . 1-7
169 1209 28 335 45 135 6
8-14
7
7 ± 49 ± 406 ± 18 ± 62 ± 17 ± 50 ± 5
230 1439 54 364 112 226 10
± ± ± ± ± ± ±
12 187 32 51 44 59
5
15-20
21-27
6 281 ± 15 1826 ± 206 * 468 ± 53 139 ± 55 235 ± 90 14 ± 4
361 ± 60 2090 ± 530 * 563 ± 76 219 ± 67 269 ± 66 14 ± 6
7
Urine collections were not performed on enough of these patients to permit meaningful analysis.
As some variables (ICG, 24 UG) reflect control over prolonged intervals, these may obscure glucose fluctuations. Thus, a scoring system for diabetic control was empirically and arbitrarily derived. As illustrated in Table 6, the scoring system uses seven variables, which include observations at a discrete point in time (FBS, G-max, G-min) as well as those assessing control over a longer period of time (2U, GB, 24 UG, ICG). A point system for good, fair, and poor categories was used for each variable, and the total control score thus determined (maximum score = 21 points). This scoring RESULTS system was applied to data from each subject; the correlation ata for all subjects studied are shown in Table 4, coefficients for the relationship of the overall score to each of separated into quartiles on the basis of rank the seven individual parameters evaluated are shown in Table order of the ICG. These data are compared with 7. The best correlation with the overall control score was for those from eight normal children previously re- XU, the traditional urine tests performed four times daily. The application of this scoring system is illustrated in Table ported.7 Considerable hyperglycemia was present in most of the diabetic children evaluated. Euglycemia for the entire 8. Patients 7 and 8 had identical ICGs of 219 mg/dl. Patient 24-h period was not achieved in any insulin-dependent subject; 7 achieved a high control score of 18, and the degree of glucose two diabetic subjects did achieve normal ICGs (80 and excursion was not great. By contrast, patient 8 exhibited wide 121 mg/dl) but only at the expense of rather prolonged excursions of glucose (G-max = 380 mg/dl, G-min = 64 mg/dl) and received a lower score of 9. hypoglycemia. Implicit in the scoring system is a "penalty" for low plasma Linear regression analysis was performed between ICG and all other parameters. The correlation coefficients from these glucose. In 6 of the 25 (24%) subjects studied, the plasma analyses are shown in Table 5. after the meal was consumed. The 24-h urinary glucose (24 UG) was determined from the sum of the four block urine collections during the 24-h study period. Glucosuria on spot urines was quantitated as the sum of the four second-voided urine tests for glucose (2U, as g/dl, with a maximum of 20 g/dl). Ketonuria was estimated as 0-1-2-3 (equivalent to negative-small-moderate-large), with the sum of the four tests for ketones (2K) calculated for each subject.
D
TABLE 6 Diabetic control—scoring system
TABLE 5 Diabetic control—parameter evaluation, linear regression analysis ICG vs. Parameter GB G-max G-min 24 UG
2U FBS
IK
338
r
P
N
0.89 0.89 0.86 0.84 0.74 0.71 0.32
