Self-monitoring of blood glucose in noninsulin-treated type 2 diabetes: an overview.

INTRODUCTION: The effectiveness of self-monitoring of blood glucose (SMBG) in noninsulin-treated patients with type 2 diabetes (T2DM) remains unclear. We aimed to review the trials investigating the effects of SMBG in this population.
METHODS: Medline was searched until June 29, 2009. Randomized controlled trials (RCTs) of at least 12 weeks' duration were included. Data on the following aspects were gathered: patient and study characteristics, effects on HbA(1c), quality of life and treatment satisfaction, and methodological quality.
RESULTS: The search revealed 9 original RCTs. These studies were very heterogeneous, and 5 were classified as of high quality. The studies with the best methodology did not show an effect of SMBG on HbA(1c), the studies with the worst methodological quality did. Two out of the 4 studies that assessed quality of life showed a significant change in favor of the control group, 1 study showed a significant change in favor of SMBG. DISCUSSION AND
CONCLUSION: We found an inverse relation between study quality and efficacy of SMBG. At this moment, there is no basis for general use of SMBG in noninsulin-treated T2DM patients.

Introduction

Self-monitoring of blood glucose (SMBG) is an important tool in the management of diabetes mellitus. Patients with type 1 diabetes frequently measure their blood glucose to evaluate and, if necessary, to adjust insulin treatment and lifestyle, and to detect hypoglycemia. It is recommended that patients should measure their blood glucose at least 3 times daily.1 Although effectiveness of SMBG has been established for insulin-treated patients with type 2 diabetes (T2DM), the optimal frequency and timing of SMBG have not yet been established.2,3 There is still much debate about the use and effectiveness of SMBG in noninsulin-treated T2DM.4 Many trials have investigated the effects of SMBG in this specific group and many (systematic) reviews have been written on this topic with conflicting conclusions.5

In insulin-treated patients SMBG can lead to (instant) insulin dose adjustments, thereby improving glycemic control and/or reducing episodes of hypoglycemia. This cascade is not applicable for patients not using insulin, because an instantaneously modifiable factor is lacking. Still, there are theoretically advantages of SMBG in this population as well. For instance, data of SMBG could be used during periodical consultation with the diabetes health care provider to start or titrate blood glucose-lowering agents. Furthermore, based on the measured blood glucose, patients could adjust physical activity or food intake, or both.

Many aspects need to be considered when evaluating the effects of SMBG in noninsulin-treated patients. Firstly, how SMBG is used depends on the patient’s level of diabetes education. Without any knowledge of the patient about the values to strive for it will be merely self-measurement and health care providers are needed to interpret the results and give advice on how to adjust physical activity and diet. Whereas, when the patient is fully informed about the treatment goals and is educated about how to achieve these goals, SMBG could function as a tool for self-regulation. Secondly, the frequency of measurements varies; SMBG can be performed infrequently or each day at fixed times, pre- or postprandial, or both. Thirdly, the population of noninsulin treated T2DM patients is heterogenic and may use SMBG differently. For example, newly diagnosed patients, who just started a specific diet versus patients on a maximum dosage of oral blood glucose-lowering agents, when insulin therapy will have to be the next step. Or patients in good glycemic versus patients in persistent poor glycemic control. Fourthly, in some situations the primary goal of the patient and the health care provider could not be to improve glycemic control with SMBG, but to improve the patient knowledge of diabetes and the effects of different behavior on glucose levels and thereby influencing nonclinical factors such as well-being and treatment satisfaction. Furthermore, performing SMBG includes the patient drawing blood from his or her finger tip regularly. This procedure can be painful or frightening and may also have its impact on well-being on its own.

We aimed to review trials investigating the effects of SMBG in noninsulin-treated patients with T2DM. We did not focus only on the effects on glycemic control, but also on the effects on health-related quality of life and treatment satisfaction as well as on the methodological quality of the trials.

Methods

Inclusion criteria of studies

Randomized controlled trials (RCTs) of at least 12 weeks’ duration were included for this review. These studies should have included noninsulin-treated patients with T2DM, the intervention group should use SMBG, and the control group should use any other form of measurement of glycemic control or should continue to receive usual care. Trials should have included measurements of glycosylated hemoglobin (HbA1c).

Search strategy

Medline was searched until June 29, 2009 using PubMed with the following combination of free text words and MeSH-terms:

(((self-measur* OR self-monitor*) OR (“Blood Glucose Self-Monitoring”[Mesh])) AND ((“Diabetes Mellitus”[Mesh]) OR (diabete* OR diabeti* OR diabeto*)))) AND (((randomized controlled trial[Publication Type] OR (randomized[Title/Abstract] AND controlled[Title/Abstract] AND trial[Title/Abstract]))) OR systematic[sb].

Articles were selected for full text reading based on their title and abstract by two of the authors (NK and JH). The references of the systematic reviews were checked for additional studies as well.

Data selection

The selected RCTs were carefully read and data on the following topics were gathered systematically by two of the authors (NK and KJJH): patient characteristics, treatment modality, duration of intervention and control, type of intervention and control, and the number of patients in each treatment arm. Data on changes in HbA1c and quality of life and treatment satisfaction were extracted both within and between treatment groups. Methodological quality of the trials was assessed using the same method as Welschen et al used in their Cochrane review.4 Eleven items pertaining to internal validity were applied:

Was the method of randomization adequate?

Was the treatment allocation concealed?

Were the groups similar at baseline regarding the most important prognostic indicators?

Was the patient blinded to the intervention?

Was the care provider blinded to the intervention?

Was the outcome assessor blinded to the intervention?

Were co-interventions avoided or similar?

Was the compliance acceptable in all groups?

Was the withdrawal/drop-out rate described and acceptable?

Was the timing of the outcome assessment in all groups similar?

Did the analysis include an intention-to-treat analysis?

Each item had a rating scale of “yes”, “no” or “don’t know”. Studies fulfilling 6 or more of the 11 quality criteria were considered to be of “high quality”. All studies scoring less than 6 of the criteria were rated as “low quality”. Studies of low methodological quality were not excluded.

Results

Studies

The initial search identified 427 publications, of which 31 papers were selected for full text reading based on title and abstract. These 31 studies included 17 potential systematic reviews and 14 potential RCTs. From the RCT selection, 1 trial was excluded because of the inclusion of patients using insulin.6 In 2 studies, the intervention as well as the control group used SMBG.7,8 For 1 study 2 references9,10 were found and for another study 3 references were found.11–13 One additional RCT14 was selected from the reference list of a systematic review.15 Consequently, 9 original RCTs were selected and will be discussed.9–14,16–21 Both authors (NK and JH) obtained the same results for the literature search. Of the 17 potential systematic reviews, 14 references qualified.4,15,22–33 Four references, however, were all based on the Cochrane review of Welschen et al.4,24,27,28 Two references were based on the review of Coster et al.31,32

Patient characteristics and treatment modality (see Table 1)

The included trials were almost all performed in Western Europe or the United States. One study included patients from several Eastern Europe countries and from Malaysia and Iran.14 Except for the study of Farmer et al11–13 all studies included patients in poor glycemic control indicated by HbA1c values ranging from 8.1% to 12.0% (mean values). The study by O’Kane et al included only patients with new onset diabetes.21 The duration of diabetes in the other studies ranged from 2.8 to 12.7 years. The majority of the included patients was treated with oral blood glucose-lowering agents. Furthermore, most patients were obese, and did not use or infrequently used SMBG prior to the study.

Study characteristics (see Table 2)

All studies had a duration of 24 weeks or longer, with the longest study duration being 12 months. Studies differed substantially regarding requested number of blood glucose values, ranging from 6 to 42 times a week, with most studies collecting some postprandial values. Patients in most studies received target glucose values to strive for. These targets were different between studies, and more recent studies used lower target values. Glucose values were mostly part of collected information which was used to give advice on lifestyle responses. In 4 studies strict algorithms were used to adjust blood glucose-lowering therapy.14,16,17,21 From the published data from the remaining 5 studies it is not clear whether and if so on what basis therapy was adjusted.9–13,18–20 A dietician participated in most studies or specific advices about diet were given (in 1 study education about counting carbohydrates was given). Patients in the less intensive SMBG group of the study by Farmer et al did not receive such advice.11–13 In some studies much time and effort of a multidisciplinary team was invested to support the patient with SMBG, while in another study SMBG was given to the intervention group without support or education. In 2 studies, the control group performed urine testing of blood glucose.16,18 In other studies usual care was continued or HbA1c was measured periodically in the control group. In the studies by Allen, Davidson, Fontbonne, Guerci, and in the less intensive group in the study by Farmer, it appears that except performing SMBG or not, no differences exist between the intervention and control group.11–13,16–19 In other studies, the SMBG group received also additional information or got personal advice about diet and exercise.

Methodological quality of studies (see Table 3)

Four studies were of low quality, 5 studies were of high quality. The studies published by Farmer, Davidson, and O’Kane are the studies with the best methodology. However, all these studies had problems with the compliance of patients performing SMBG.11–13,17,21 In the study by Davidson et al compliance was less than 50%.17 The study described by Farmer et al reached compliance rates of 67% and 52% in the less and more intensive intervention groups, respectively. In the study by O’Kane et al 34% of the patients performed less than 80% of the requested measurements.

Effect on glycemic control (see Table 4)

Three studies found a significant beneficial effect on HbA1c of SMBG compared to the control group.9,10,19 In the study by Guerci et al HbA1c decreased from 9.0% to 8.1% after 6 months. This decrease in HbA1c was ~0.3% more compared to the decrease in found in the control group (P = 0.009). In the study by Schwedes et al HbA1c decreased from 8.5% to 7.5%, which was ~0.5% (95% CI −0.77, −0.11) more compared to the control group. In the study by Barnett et al HbA1c decreased from 8.12% to 6.95%, which was 0.24% (95% CI −0.45, −0.03) more compared to the control group.14 Muchmore et al found a nonsignificant benefit of SMBG of ~0.7%.20 In the other studies nonsignificant differences between SMBG and control groups were between −0.2% and 0.1%.

Effect on quality of life and treatment satisfaction (see Table 5)

In 4 studies, quality of life and treatment satisfaction were assessed using questionnaires. In the study of Muchmore et al no differences in quality of life were found between the intervention (SMBG) and control group (usual care).20 Although no significant difference was found for the total score on the patient well-being questionnaire in the study by Schwedes et al, 2 sub items (“depression” and “lack of well-being”) improved with SMBG compared to control.9,10 In contrast with these findings, 2 more recent trials reported that SMBG was associated with lower quality of life.11–13,21 In the study by Farmer et al quality of life, as measured with the EQ-5D questionnaire, was lower in the more intensive intervention group compared to the control group.11–13 In the study by O’Kane et al the depression scale of the well-being questionnaire was significantly worse in the SMBG group compared to the control group.21 No differences were found concerning treatment satisfaction.

Discussion and conclusion

Nine RCTs investigating the efficacy of SMBG in noninsulin-treated patients with T2DM have been published so far.9–14,16–21 One third of the studies reported a positive effect of SMBG on HbA1c ranging from a 0.2% to 0.5% decrease. None of the studies published found a negative effect of SMBG on HbA1c. The evidence is more conflicting about the effects on quality of life. SMBG, as discussed in the introduction, is a complex intervention, although one can debate whether or not SMBG should be classified as an intervention in patients who “only” self-monitor without any form of self-regulation.

The differences in patient and study characteristics are distinct and complicate direct comparison between the studies. For example, mean HbA1c ranges from 7.5% to 12%. Furthermore, timing and frequency of SMBG vary widely, some studies advise patients which values to strive for, some studies give advice how to achieve these goals, and in some studies patients receive additional help from a dietician or a diabetes specialist nurse. These large differences in study design illustrate the many different ways that SMBG is integrated in the management of noninsulin-treated patients with T2DM and make it almost impossible to draw a single overall conclusion about its effectiveness.

Regardless of the heterogeneity of the studies, it is remarkable that there is an inverse relationship between study quality on the one hand, and efficacy on glycemic control and quality of life parameters on the other. Only 1 study of high methodological quality found a significant improvement of HbA1c of 0.24% (95% CI 0.03, 0.45) with SMBG compared to control.14 The other 4 well designed studies found no effects of SMBG on glycemic control, but did report a worsening of quality of life parameters compared to control.11–13,16,17,21 When comparing the study by Barnett et al with the other well designed studies in order to find a possible explanation for the differences in effects on HbA1c, no major differences were found for the interventions used. What was somewhat different in the study by Barnett et al is the shorter diabetes duration and the lower body mass index (BMI) of patients included (Table 1).

Table 1

Randomized controlled trails of self-monitoring of blood glucose (SMBG) in patients with type 2 diabetes not using insulin: patient characteristics

Reference; country	Schwedes 2002; Germany and Austria9,10	Farmer 2007; UK11–13	Allen 1990; USA16	Davidson 2005; USA17	Fontbonne 1989; France18	Guerci 2003; France19	Muchmore 1994; USA20	O’Kane 2008; UK21	Barnett 2008; Seven countries14
Previous use of SMBG	No use of SMBG in the last 6 months	No use of SMBG twice a week or more over the previous 3 months	No	?	?	No	No use of SMBG in the last 3 months	No	No current management with SMBG
Male (%)	52	57	100	26	62	55	39	60	52
Age (years)a	~60 (45–70)	65.7 (≥25)	58	~50	~55	~62 (40–75)	~59 (40–75)	~59 (<70)	~56 (40–80)
HbA1c(%)a	~8.4 (7.5–10)	7.5 (≥6.2)	~12.0	~8.5	~8.3	9.0 (7.5–11)	~10.4 (9.5–13.5)	8.7	8.1 (7–10)
Diabetes duration (years)a	~5.3 (≥0.25)	3	~7.9	~5.6	~12.7 (≥3)	~8.0 (≥1)	~5.5 (≥1)	0b	2.8
Body mass index (kg/m2)a	~31.5 (>25)	~31.3	? (weight: ~93 kg)	32.5	~26.7	~30.1	~34.2 (27.5–44)	~33	~30.4
Treatment	Diet (?%) or diet and oral blood glucose-lowering treatment (?%)	Diet (27%) or diet and oral blood glucose-lowering treatment (73%)	Diet (15%) or diet and oral blood glucose-lowering treatment (85%)	Diet (2%) or diet and oral blood glucose-lowering treatment (98%; 14% triple therapy)	Diet (?%) or diet and oral blood glucose-lowering treatment (?%)	Oral blood glucose-lowering treatment	Diet (26%) or diet and oral blood glucose-lowering treatment (74%)	Newly diagnosed patients	Diet (29%) for ≥3 months or diet and oral blood glucose-lowering treatment (71%)

Presented data are means or median with between brackets the inclusion range.

Newly diagnosed patients with type 2 diabetes were included.

Two recent systematic reviews concluded that the effect of SMBG on HbA1c was on average 0.21% and 0.22%.15,22 One may question, however, whether it is justified to perform a meta-analysis including all the trials, keeping in mind that the benefits of SMBG were found primarily in low quality trials. A Cochrane review did not perform a meta-analysis because of the differences in study quality and because of the differences in baseline data of the patients and type of interventions between the studies.4,34

A collaboration of authors of previous trials will investigate whether a certain subgroup of patients can be found who might benefit more from SMBG than the effects found in the systematic reviews.5 Perhaps, a short diabetes duration and a low BMI could be factors as hypothesized earlier. On top of that, a large RCT will be performed in the Netherlands including 600 patients with noninsulin-treated T2DM, in which the patients will be randomized into 3 groups: SMBG, self-measurement of urine glucose, and control.35 Perhaps, these 2 studies will give some new perspectives about the usefulness of SMBG in certain subgroups of patients. Until then, we recommend that SMBG should not be used generally in a population of noninsulin-treated diabetic patients. Effectiveness on glycemic control has not been established, and quality of life could diminish.

Table 2

Randomized controlled trails of self-monitoring of blood glucose (SMBG) in patients with type 2 diabetes not using insulin: study characteristics

Study	Duration	Treatment arm		Number of patients
		Intervention	Control	Intervention/Control
Schwedes9,10	24 weeks	SMBG 6 times a day (pre- and 1 hour postprandial values), 2 days a week(1 week day and Sunday)Documentation of eating habitsDocumentation of state of well-beingPatients were explained that combined information allowed them to make diet and lifestyle changes in order to improve glycemic control counseling to increase self-perception, to promote self-reflection and to enhance self-regulationTherapy adjustment?	Nonstandardized counseling with a focus on diet and lifestyleTherapy adjustment?	113/110
Farmer11–13	12 months	1. Less intensiveSMBG 3 times a day (1 fasting and 2 pre- or postprandial values), 2 days a weekStrive for preprandial glucose concentrations of 4–6 mmol/l and postprandial 6–8 mmol/LA diary was used to record goals, activities, and blood glucose resultsNo information about how to interpret glucose values was givenTherapy adjustment?2. More intensiveIn addition to ‘1. less intensive’:Training and support in timing, interpretation and using results also to enhance motivation and maintain adherence to diet, physical activity and drug regimensEncouragement to experiment with SMBG to explore the effects of specific activityTherapy adjustment?	HbA1c every 3 monthsA diary was used to record self-care goals and strategies for achieving themUsual (standardized) careInformation about eating and physical activityTherapy adjustment?	150/151/152
Allen16	6 months	SMBG before meals every other day, at least 36 times a month Strive for glucose concentrations <7.7 mmol/L fasting and <8.8 mmol/L before lunch and dinnerAlgorithm to adjust blood glucose-lowering treatmentInstructed by dietician about diet, fiber intake, ideal bodyweight, activity level	Urine testing before meals every other day, at least 36 times a month Strive for negative urine checksAlgorithm to adjust blood glucose-lowering treatment. Instructed by dietician about diet, fiber intake, ideal body weight, activity level	27/27
Davidson17	6 months	SMBG before and between 1–2 hours after meals during 6 days a week; 2 times breakfast, 2 times supper and 2 times dinnerRecord meals 5 visits to dietician; education about meal size and components on rise of postprandial glucose valuesStrive for glucose concentrations <7.15 mmol/LAlgorithm to adjust blood glucose-lowering treatment	Visits to dietician; education about meal size and components on rise of postprandial glucose valuesAlgorithm to adjust blood glucose-lowering treatment	43/45
Fontbonne18	6 months	SMBG twice every other day; fasting and 2 hours after dinner and on Sundays after lunchConsulting physician was allowed to adjust therapy	1. Urine testing twice every other day; fasting and 2 hours after dinner and on Sundays after lunchConsulting physician was allowed to adjust therapy2. Every 2 months HbA1c testing with physician’s comment about metabolic control to patientConsulting physician was allowed to adjust therapy	68/72/68
Guerci19	24 weeks	SMBG at least 6 times a week on 3 different days including a weekend day. Dietary advice about ideal bodyweight. Walking was encouraged as a form of exerciseTherapy adjustment?	HbA1c every 12 weeksDietary advice about ideal body weightWalking was encouraged as a form of exerciseTherapy adjustment?	345/344
Muchmore20	44 weeks	SMBG: first 4 weeks: 6 times daily (pre- and 2 hour postprandial values)Next 16 weeks: 2 times daily (pre- and 2 hour postprandial values). Last 24 weeks: individual choiceStrive for a postprandial increase of 2.2–3.9 mmol/L with, if necessary, adjustment of carbohydrate quantity and quality or change in exercise timing or intensityEducation about carbohydrate countingReceived behavioral weight program (90 min a week during the first 8 weeks). In addition, there were 3 and 4 sessions with a dietician and diabetes nurse, respectivelyGoal was to lose 0.5–1.0 kg a weekTherapy adjustment?	Received general information about nutritionReceived behavioral weight program (90 min a week during the first 8 weeks). In addition there were 3 and 4 sessions with a dietician and diabetes nurse, respectivelyGoal was to lose 0.5–1.0 kg a weekTherapy adjustment?	12/11
O’Kane21	12 months	SMBG 8 times a week; 4 fasting and 4 postprandial valuesReceived advice on interpretation and appropriate (lifestyle) responses to high or low readingsEducational program3-month visits with doctor, diabetes nurse, and dieticianAlgorithm to adjust blood glucose-lowering treatment	Educational programEvery 3 months visits with doctor, diabetes nurse, and dieticianAlgorithm to adjust blood glucose-lowering treatment	96/88
Barnett14	27 weeks	SMBG 5 times a day (before each meal, 2 hours after the main meal and before bedtime), 2 days a week (1 working and 1 nonworking day). Once a month postprandial measurements after each mealDiet and lifestyle adviceDiary was used to record symptoms of hypoglycemia and actions takenAlgorithm to adjust blood glucose-lowering treatment	Diet and lifestyle adviceDiary was used to record symptoms of hypoglycemia and actions takenAlgorithm to adjust blood glucose-lowering treatment	311/299

Table 3

Randomized controlled trails of self-monitoring of blood glucose (SMBG) in patients with type 2 diabetes not using insulin: methodological quality

Criteria	Schwedes9,10	Farmer11–13	Allen16	Davidson17	Fontbone18	Guerci19	Muchmore20	O’Kane21	Barnett14
Randomization adequate	+	+	+	+	+	+	+	+	+
Treatment allocation concealed		+	−					−	+
Groups similar at baseline	+	+	+	+	+	+	+	+	+
Patient blinded	NA	NA	NA	NA	NA	NA	NA	NA	NA
Care provider blinded	NA	NA	NA	NA	NA	NA	NA	NA	NA
Outcome assessor blinded		+/?a						+/?a
Co-interventions avoided	−	+	+	+		+		+	+
Compliance acceptable	+	−	+	−	+		+	−
Withdrawal/drop-out rate acceptable	+	+	+	+	+	−	+	+	+
Timing of outcome assessment similar	+	+	+	+	+	+	+	+	+
Intention to treat analyses	−	+	−	+		+		+
Total quality score	5	(7)8	6	7	5	5	5	(6)7	6
Overall quality conclusion	Low	High	High	High	Low	Low	Low	High	High

Outcome assessor was only blinded for HbA1c.

Abbreviation: NA, not applicable.

Table 4

Randomized controlled trails of SMBG in patients with type 2 diabetes not using insulin – study results: HbA1c

Study	Treatment arm
	Intervention	Control	Intervention vs Control
Schwedes9,11	8.47 → 7.47	8.35 → 7.81	−0.46 (95% CI −0.77, −0.11)
Farmer11–13	1. 7.41 → 7.282. 7.53 → 7.36	7.49 → 7.49	1. − 0.14 (95% CI −0.35, 0.07)2. −0.17 (95% CI −0.37, 0.03)
Allen16	12.4 → 10.4	11.7 → 9.7	~0.0 (P > 0.95)
Davidson17	8.5 → 7.7	8.4 → 7.8	−0.2 (95% CI −1.1, 0.6)
Fontbonne18	8.2 → ~7.8	1. 8.6 → ~8.52. 8.2 → ~7.7	1. ~ −0.2 (NS)2. ~0.1 (NS)
Guerci19	9.0 → 8.1	8.9 → 8.4	~ −0.3 (P = 0.009)
Muchmore20	10.29 → 8.75	10.45 → 9.6	~ −0.7 (NS)
O’Kane21	8.8 → 6.9	8.6 → 6.9	−0.07 (95% CI −0.38, 0.25)
Barnett14	8.12 → 6.95	8.12 → 7.20	−0.24 (95% CI −0.45, −0.03)

Table 5

Randomized controlled trails of self-monitoring of blood glucose (SMBG) in patients with type 2 diabetes not using insulin – study results: Quality of life and treatment satisfaction

Study1st author	Questionaire	Treatment arm
		Intervention	Control	Intervention vs Control
Muchmore20	Diabetes Quality-of-life inventory			NS
Schwedes9,10	1. Patient well-being questionnaire
	General well-being (total score)			NS
	Subitem ‘Depression’			Reduction in intervention compared to control P = 0.032
	Subitem ‘Lack of well-being’			Reduction in intervention compared to control P = 0.02
	2. Diabetes Treatment Satisfaction Questionnaire (DTSQ)	27.6 → 31.1	27.0 → 30.6	~ −0.1(P = 0.9)
Farmer11–13	1. DTSQ	1. 29.4 → 29.72. 29.7 → 30.1	29.3 → 30.0	NS
	2. W-BQ12	1. 24.3 → 24.52. 25.2 → 24.9	25.1 → 25.9	NS
	3. EQ-5D	1. 0.781 → 0.7552. 0.807 → 0.733	0.799 → 0.798	−0.029 (95% CI −0.084, 0.025)−0.072 (95% CI −0.127, −0.017)
O’Kane21	1. Well-being questionnaire depression (1 of the subitems, other NS)			6.05 (p = 0.011)
	2. DTSQ			NS
	3. Diabetes attitude scale (3 of 7 subscales were included)			NS