- Click here for explanation of classification scheme.

The purpose of this study was to compare the effectiveness and cost of self-monitoring of blood glucose (SMBG) to routine urine glucose testing as part of a standardized treatment program for patients with type II (2) diabetes who do not use insulin, and assess the respective costs of the 2 monitoring techniques.

1. Type 2 diabetes
2. Fasting plasma glucose (FPG) >8.8 and <22 mmol/L (approx. 158-396 mg/dl)
3. No history of ketoacidosis
4. Current treatment with diet or diet + 1 oral antidiabetes agent 
5. No active infection or serious concurrent illness
6. No physical or mental handicap precluding participation as determined by the cognitive-capacity screening exam and physical abilities questionnaire
7. Willingness to follow study protocol

1. Had SMBG devices previously
2. Serum creatinine >177 µmol/L (approx. 2 mg/dl)

Recruitment:

Subjects were recruited from general medical and triage clinics of the Durham Veterans Administation Medical Center, Durham North Carolina.

Design:  Randomized Controlled Trial

Blinding:

Blinding of the subjects and/or physician was not possible due to the nature of study interventions.  The treatment algorithm was used to reduce group bias when determining treatment changes.

Intervention:

1. After obtaining informed consent subjects were randomized in groups of 10 to a urine-testing group or SMBG group with the use of a computer-generated table of random numbers.
2. All attended monthly clinic visits for 6 months.
3. Nutrition counseling done at initial visit.
4. Instruction on glucose checking:
   • Subjects in both the urine and blood check groups were individually instructed in techniques and after 7-10 days of use were asked to demonstrate the technique to clinic staff at the initial clinic visit.
   • Subjects were required to demonstrate proficiency before being allowed to proceed in the study.
5. Treatment algorithm:
   • Physician initiated treatment was to follow an algorithm based on the results of the urine and blood glucose monitoring, food and exercise records and monthly fasting plasma glucose to modify oral diabetes medications or to initiate insulin.  Treatment changes were made monthly as needed.  Subjects received no other diabetes treatment advice from non-study physicians.

Statistical Analysis:

Mean changes in outcome variables were compared between groups using a two-sample t-test.  Changes within the groups were compared using a one-sample t-test.   Necessary sample size was calculated to detect a 25% improvement in any major outcome variable at the 5% significance level.

Timing of Measurements

All attended monthly clinic visits for 6 months.

Dependent Variables:

1. Fasting plasma glucose: monthly.   
   • Accu-Chek I meter with Chemstrips bG was used because of its relative simplicity and acceptable accuracy.
   • Subjects expected to complete 36-blood checks/month.
   • Subjects instructed to monitor blood glucose level before each meal every other day (goal: <7.7 mmol fasting and <8.8 mmol before lunch or dinner).
2. A1C test: 0, 3 and 6 months
3. Total cholesterol and HDL-cholesterol
4. Urine glucose
   • Tes-Tape was used to monitor glucose control because of its simplicity and ability to detect <0.05% glucose concentration.
   • Subjects were expected to complete 36 urine tests/month (positive or negative reading).
   • Subjects instructed to test a single-voided urine specimen before each meal every other day (goal: negative urine check).

5. All subjects were weighed monthly.

Independent Variables:

1.  Nutrition intervention

• RD instructed all subjects on diet based on weight and physical activity.
• All instructed to consume 30-40 g fiber/d. 
• All subjects instructed to complete food diary 1 d/wk and a daily exercise diary.

2.  Physician algorithm

• Based on the percentage of negative urine glucose,  percentage of glucose checks within targe (<8.8mM) and the level of FPG, subjects were assigned treatment of none (diet instruction all subjects received), daily exercise diary/food intake dairy, exercise/food diary and oral agent, or exercise/food diary plus insulin. 
• This algorithm was applied monthy at the visits.  Oral agents and insulin doses were increased as needed per algorithm.

3  In addition to the study measures, compliance was assessed at monthly visits by the demonstration of urine/blood testing proficiency, the completeness of testing records, and attendance at required monthly visits.

Control Variables

Initial N:  61 male patients 

Attrition (final N):  54 completed the study, 5 of the subjects did not meet inclusion criteria, 2 others dropped out of the study for unknown reasons.

Age, Ethnicity, Other Relevant Demographics:

Characteristics	Urine-Testing	SMBG
N	27	27
Age, yr	57.9±10.7	58.2±9.7
White, %	67	59
Weight, kg	96.0±22	90.0±11
Duration of Diabetes, yr	9.0±10.3	6.8±6.5
Oral meds, %	85	85
Fasting glucose, mmol/L (mg/dl)	12±2.6 (216±47)	12±2.4 (216±43)
A1C, %	11.7±3.0	12.4±3.3
T-chol, mmol  (mg/dl)	5.6±1.2 (201±46)	5.8±1.5 (224±58)
HDL-chol, mmol (mg/dl)	1.0±0.3 (39±12)	1.1±0.3 (43±12)

Anthropometrics:  no differences between groups at baseline

Location:  North Carolina

During the 6-month study, both the urine testing group and the SMBG group showed similar improvement in glycemic control.  The improvements within each group of FBG and A1C were statistically significant however comparisons between groups showed no statistically significant differences in mean FBG, A1C or weight.

Of interest, 17 (31%) of 54 patients normalized A1C values, 9 in the urine testing group and 8 in the SMBG group during the course of this study.

Comparison by Group

	Urine-Testing	SMBG
Fasting Glucose, Mmol (mg/dl)         Baseline	12.0±2.6 (216±47)	12.0±2.4  (216±43)
6-months	10.5±3.0 (189±54)	10.6±3.6 (191±65)	p>0.86
Change	-1.5±2.8 (-27)	-1.4±3.2 (-25)
p	<0.01	<0.03
A1C, %         Baseline	11.7±3.0	12.4±3.3
6 months	9.7±2.6	10.4±2.9	p>0.95
Change	-2.0±2.4	-2.0±3.4
p	<0.001	<0.01

There was no statistically significant weight change between the urine-testing group and the SMBG group (p>0.19). There was no weight gain in the urine-testing group ad 2.0±4.5 kg. weight gain in the SMBG group.

The estimated cost of glucose monitoring of the blood was $481/patient/year compared with $40/patient/year for urine testing.

SMBG is no more effective and is 8-12 times more expensive than urine glucose testing in facilitating improved blood glucose control.

When considering the use of public or insurance funds to pay for the diabetes testing supplies, the results do not support widespread use of SMBG (over urine glucose testing) in patients with type 2 diabetes not treated with insulin.

Each subject was treated by a team including physician, physician associate, diabetes teaching nurse and dietitian.  Each team member provided reinforcement of treatment goals using a behaviorally oriented approach.  Subject's attendance at the monthly visits was >98%.

Government:	VA HSR&D
University/Hospital:	Durham VA Medical Center, Duke University Medical Center

Points to consider:

Subjects did show improved glycemic control during the 6 months of the study, equally in both testing groups.  This is a short amount of time.  Could these improvements in glycemic control be sustained over time with each of the testing techniques?  These authors could have planned a followup assessment of the same subjects at some time interval following the study period but this was not mentioned in this article.

The behaviorally focused treatment program was highly structured with monthly visits.  This intensity of caregiver attention would likely not continue longterm.

The subjects' diabetes control was quite poor at the beginning of the study.  There was no discussion of hypoglycemia events.   Even though glycemic control improved equally in both groups, many subjects were still above target at 6 months.  I would expect a difference in glucose testing techniques to be shown if attempts were made to get all subjects to near normal goals.

The authors' gave no credit to the algorithm for which they used to make advances in treatment levels.  Was this type of algorighm standard care in 1990?

Additionally, all medications and monitoring supplies were provided at no cost.   Would these subjects have shown similar improvements in glycemic control outside of the research setting?

Limitation of study included population specific to type 2 non-insulin using.  Subjects were all men.  Subjects were mostly the same age and same education level and few had newly diagnosed diabetes. To determine the influence of the study visits alone, future studies could have a third subject group that does no testing at all.