Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Effect of the Frequency of Self-Monitoring Blood Glucose in Patients with Type 2 Diabetes Treated with Oral Antidiabetic Drugs—A Multi-Centre, Randomized Controlled Trial

  • Werner A. Scherbaum,

    Affiliations Department of Endocrinology, Diabetes and Rheumatology, WHO Collaborating Centre for Diabetes, European Training Centre in Endocrinology and Metabolism, University Hospital Düsseldorf, Düsseldorf, Germany, German Diabetes Clinic, German Diabetes Centre at the Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany

  • Christian Ohmann,

    Affiliation Coordination Centre for Clinical Trials, University Hospital Düsseldorf, Düsseldorf, Germany

  • Heinz-Harald Abholz,

    Affiliation Department of General Medicine, University Hospital Düsseldorf, Düsseldorf, Germany

  • Nico Dragano,

    Affiliation Department of Medical Sociology, University Hospital Düsseldorf, Düsseldorf, Germany

  • Mark Lankisch

    scherbaum@uni-duesseldorf.de

    Affiliation German Diabetes Clinic, German Diabetes Centre at the Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany

Effect of the Frequency of Self-Monitoring Blood Glucose in Patients with Type 2 Diabetes Treated with Oral Antidiabetic Drugs—A Multi-Centre, Randomized Controlled Trial

  • Werner A. Scherbaum, 
  • Christian Ohmann, 
  • Heinz-Harald Abholz, 
  • Nico Dragano, 
  • Mark Lankisch
PLOS
x

Abstract

Objective

Recommendations on the frequency of self-monitoring of blood glucose (SMBG) vary widely among physicians treating patients with type 2 diabetes (T2D). Aim of this study was to investigate two testing regimen of SMBG in patients with stable metabolic control.

Research Design and Methods

Patients with T2D treated with oral antidiabetic drugs were randomized to two groups: either one SMBG (low) or four SMBG (high) per week. Subjects were followed up after 3, 6 and 12 months. Primary outcome parameter was the change in HbA1c between baseline and 6 months. Primary outcome criterion was tested by a one-sided t- test for non- inferiority. Secondary outcome parameters were safety, compliance and HbA1c at 3 and 12 months.

Results

There were no differences in the 202 subjects for demographic and sociodemographic parameters and drug treatment. HbA1c (%) at baseline was similar in both groups (7.2±1.4 vs. 7.2±1.0). Non- inferiority was demonstrated for the low group (p = 0.0022) with a difference from baseline to 6 months of 0.24 in the low and of 0.16 in the high group. Compliance with the testing regimen was 82–90% in both groups. There were no statistical significant differences for compliance, HbA1c at 3 and 12 months and serious adverse events (SAE).

Conclusion

One SMBG per week is as sufficient and safe as four SMBG per week to maintain HbA1c in non-insulin treated T2D close to metabolic target. The results of this study are in contrast to current international consensus guidelines.

Trial Registration

Controlled-Trials.com ISRCTN79164268

Introduction

Self-monitoring blood glucose (SMBG) improves glycaemic control in patients with type 1 diabetes [1] and possibly also in insulin-treated type 2 diabetes (T2D) [2], especially when treated with multiple insulin injections per day [3]. However, the value and frequency of SMBG in non-insulin-treated patients with T2D is a matter of controversy.

A consensus opinion among a group of experts from the UK suggested that patients with T2D using oral antidiabetic drugs (OAD) should monitor their blood glucose at least once daily, varying the time of testing between fasting, preprandial and postprandial levels during the day [4]. A global consensus conference on SMBG recommended eleven measurements a week in these patients [5] and another recent consensus conference noted that patents with T2D on OAD may use SMBG but specific recommendations with respect to frequency were not made [6].

A cross-sectional and longitudinal study of patients with T2D in Australia showed that HbA1c was not significantly different between SMBG users and nonusers, either overall or within diabetes treatment groups such as diet, OAD or insulin, with or without OAD [7]. Although such observational data can be useful in determining the effect of an intervention [8], conclusive evidence of this assumption is not available from randomized controlled trials.

A recent study reported on the effect of a more and less intensive diabetes education combined with recommendations on the frequency of SMBG in patients with T2D [9]. They found that a more intensive education did not result in an improved HbA1c (%) compared to standard information and care.

Our study was designed to strictly test the effect of the frequency of SMBG without changing the other diabetes management.

Our hypothesis was that one SMBG measurement a week in a stable phase of metabolic control close to glycemic target is no less effective than more measurements a week with regard to metabolic control, hypoglycaemia and/or hyperglycemias, or adverse events (AE).

Methods

The protocol for this trial and supporting CONSORT checklist are available as supporting information; see Checklist S1, Protocol S1 and Protocol S2.

Study design

A prospective multi-centre randomized controlled trial was performed evaluating non-inferiority of low vs. high frequencies of SMBG with respect to metabolic control T2D. The study was approved by the ethical committees of all six institutions participating, and all patients provided their written informed consent.

Study population

Patients with T2D treated with one or more OAD (not combined with insulin therapy and stable oral medication for the last three months) aged between 35 and 80 years were enrolled. Patients with following conditions were excluded: type 1 diabetes, advanced renal insufficiency (creatinine level at ≥2.5 mg/dl), at least two episodes of hypoglycaemia requiring external support within the previous three months, one or more severe metabolic events (hypoglycemic shock, hyperosmolar coma, inpatient stay due to severe hyperglycaemic events) within the past three months, pregnancy, severe impairment of vision or communication problems due to language.

All patients had received a structured education on diabetes mellitus and instructions on SMBG in practice before the start of the study, and were not specifically re-educated.

Originally, it was planned to establish a second trial arm with insulin treated patients, comparing one measurement a week (low group) with 11 measurements a week (high group). Due to poor recruitment this arm was terminated early by the trial steering committee. The reasons for poor recruitment were mainly changes in diabetes treatment in Germany due to a move from pre-mixed insulin to a different type of insulin treatment.

Study intervention

Patients were randomly assigned to two strategies of SMBG. Low group: SMBG with one measurement a week and additional measurement in the event of suspected hypoglycaemia or severe hyperglycaemia or high group: four measurements a week on Tuesdays, Thursdays and one day of the weekend before dinner and one additional measurement before lunch, and also additional measurement in the event of suspected hypoglycaemia or severe hyperglycaemia. Hypoglycaemia was defined as an SMBG<3.2 mMol/L (<60 mg/dl). Severe hypoglycaemia was defined as any hypoglycaemia with the need for assistance by another person.

The randomization list was generated and centrally applied with concealment by the Coordination Centre for Clinical Trials with the use of randomly selected block sizes of 4, 6 and 8 patients stratified according to centre.

In our case-management approach, all patients were asked to report back to their physician in the event of inappropriate diabetes control according to the targets set between the patient and his or her doctor.

Evaluation and follow-up

Potential study patients were screened according to inclusion and exclusion criteria. After randomization, patients were instructed on the strategy of SMBG to be applied for a period of 6 months (formal instruction, written instruction, patient diary, delivery of equipment for measuring blood glucose). Patients were evaluated at the initial visit and after 3 and 6 months by clinical investigation, assessment of (serious) adverse events, quality of life (at baseline and at 6 months only), compliance with intervention, change in diabetes treatment, socioeconomic effects and measurement of HbA1c. After 12 months only HbA1c was assessed. In addition, blood glucose measurements were documented by the patient in a diary, and the data from the first week each month were used to assess intervention compliance from the patient's viewpoint. Compliance with treatment was defined as follows: maximal deviation of one measurement from the allocated frequency (e.g. 3 to 5 in the high group, 0 to 2 in the low group).

Trial outcomes

The primary outcome criterion was the change in HbA1c levels between baseline and six months after enrolment. Secondary outcomes were number and type of hypoglycaemic and hyperglycaemic events, quality of life, compliance and satisfaction with interventions, socioeconomic effects and HbA1c after 3 and 12 months. Data on quality of life, satisfaction with intervention and on the socioeconomic background will not be presented in this paper.

Statistical analysis

The study was designed to test non- inferiority of the low group compared to the high group. Sample-size calculation was based on data from the German Diabetes Centre (590 patients, mean value (standard deviation) of HbA1c: 9.95±1.83), a one-sided t-test for non-inferiority assuming an acceptable deviation of 0.5% for the difference in HbA1c levels between baseline and 6 months, statistical power of 80% and an alpha error of 5% leading to an estimated sample size of 167 patients per group. Taking drop-outs into consideration (e.g. informed consent withdrawn, lost to follow-up), sample size was calculated at 400, stratified into two strata: patients with oral medication (n = 200) and patients with insulin treatment (n = 200). All analyses were performed according to intention to treat and included all randomized patients excluding the drop-outs. The primary outcome criterion was tested by a one-sided t-test for non-inferiority. The alternative hypothesis was stated as follows: the difference in the low group is equal to or better than in the high group, taking a deviation of 0.5% into consideration [10]. Comparability between study groups and secondary endpoints were investigated by chi-square test for qualitative and t-test, Mann- Whitney- test for quantitative data.

Results

A total of 202 patients from six centres underwent randomization in the oral group between Dec 2003 and Oct 2005; 100 were randomly assigned to the low and 102 to the high group. There were 12 drop-outs in the low and 11 drop-outs in the high group (Fig. 1). The baseline characteristics in the two groups were similar (Table 1, 2). Mean diabetes duration was 7.8±6.4 years in the low and 8.2±6.5 in the high group.

thumbnail
Figure 1. Flow of participants in the trial (according to CONSORT statement).

* with respect to primary outcome criterion HbA1c.

https://doi.org/10.1371/journal.pone.0003087.g001

Compliance and adverse effects

Compliance with the assigned frequency of SMBG was measured from the doctor's and the patient's viewpoint (Table 3). Compliance in the low group assessed by the doctor was 61% at 3 months and 73% at 6 months. In the high group, compliance to the SMBG regimen assigned was 77% at 3 months and 83% at 6 months. Analysis of each patient's diary revealed compliance in 82–83% (0 to 3 months) and 85–88% (4 to 6 months) in the low group and 87–90% and 84–88% in the high group. A statistical significant difference was observed according to doctor's assessment at 3 months (p<0.03).

Clinicians were asked to record all AE's and SAE's according to Good Clinical Practice (GCP). The number and type of SAE and AE was similar between the groups except for hypoglecaemia, which was increased in the high group (Table 4). Twenty patients from the low group and fifteen in the high group showed SAE. No hypoglycaemic shock or hyperosmolar coma occurred. Inpatient hospital treatment was necessary in 19 patients in the low group and 14 in the high group. There was one death in the low and one death in the high group. Both deaths were unrelated to the study intervention (pancreatic cancer, heart failure).

Outcome

HbA1c at baseline was similar in both groups (7.2±1.4 vs. 7.2±1.0) (Table 4). The primary outcome criterion (difference of HbA1c value at baseline and after 6 months) was assessed in 178 patients (low group: n = 87, high group: n = 91); non-inferiority was demonstrated for the low group (p = 0.0022) with a mean difference of 0.24 in the low and of 0.16 in the high group. In addition, no statistically significant differences in HbA1c values at 3 and 12 months were observed (Table 4). There were no statistically significant differences between the groups with respect to health care utilization (physician visits, inpatient treatment, or days off work) and changes in diabetes treatment (Table 4).

Discussion

To our knowledge, this is the first randomized, controlled prospective multi-centre trial on the frequency of SMBG in patients with T2D treated with OAD with a strict focus on the specific impact of the frequency of SMBG. Patients in this cohort were in a stable phase of T2D close to glycemic target and they had been on the same anti-diabetic treatment for at least three months. In a large observational cohort study (ROSSO), we recently demonstrated that patients with T2D who apply SMBG had better outcomes in terms of morbidity and mortality than patients that did not [11]. However, the database from this cross-sectional study did not allow to provide any answer to the optimal frequency of SMBG in T2D.

The need for SMBG in patients with T2D has been discussed controversially. A recently published review of randomized controlled trials came to the conclusion that SMBG is likely to be helpful in non-insulin-treated patients with T2D, but the methodological quality of the studies analyzed was considered to be low [12]. Welschen and colleagues defined nine quality criteria that should ideally be met in this type of study. In their review on the literature concerning SMBG, they found 1572 citations, but only six studies were considered as appropriate for further analysis [13][18]. However, the design of these six studies only met 5, 5, 5, 5, 6, or 7 out of nine methodological quality criteria. Drawbacks were, among others, failure to conceal treatment allocation (all 6), blinding of the outcome assessor to the intervention [13][16], [18], unaccounted or unacceptable withdrawal or drop-out rate [18], or failure to include an intention-to-treat analysis [13][16]. The conclusions drawn from these studies are limited. All of the above quality criteria except blinding of the outcome assessor were met in our study.

We considered one SMBG measurement as clinically important in order to allow the patient to detect unexpected metabolic deterioration between doctor's visits. Four SMBG measurements a week in our high group is by far less than the eleven measurements a week that were suggested by a global consensus panel [5]. The decision to suggest a lower number of SMBG in patients with T2D treated with OAD and an HbA1c close to target was based on our own experience in outpatient care of patients with T2D. There is also a substantial disparity between the actual and recommended frequency of SMBG testing. A previous analysis of data from German health care providers demonstrated that patients with T2D on insulin therapy or on OAD have a test strip use of 1.8 and 0.8 strips a week, respectively [19]. It is also common experience that patients with HbA1c close to target may not show much change, and differences between groups would most probably be minimal [20].

Our primary endpoint of change in HbA1c demonstrated non-inferiority in the low compared to the high group. One concern we had with respect to the performance of the study was that patients assigned to the low group would measure more frequently and patients in the high group would measure less frequently. However, most patients in the two groups kept a high level of compliance to their assigned testing regimen as assessed by patient diary.

As mentioned above, our study was designed to strictly evaluate the effect of the frequency of SMBG on metabolic control without mixing this tool with different education procedures or intensities in diabetes management. In the study by Farmer et al. [9] it appeared that only about half of the patients in the more intensive group followed the recommendations regarding SMBG and the other performed less than 2 SMBG per week. It may be critisized that the failure to show an effect on HbA1c in this study is due to a lack of adherence to the study regimen. In our trial, however, the adherence in both groups, i.e. low and high frequency of SMBG, was as high as 82–90%.

No differences were observed with respect to AE's and SAE's, except for hypoglycaemia, which occurred more often in the high group probably due to the higher frequency of measurements.

Although our data show no advantage of more than one SMBG measurement a week in non-insulin treated T2D patients it is important to emphasize that these results only refer to patients with stable metabolic control and without any necessity of changing antidiabetic medication. The conclusions drawn from our study do not refer to patients with newly diagnosed T2D or patients with intercurrent diseases disturbing metabolic control. Although prospective controlled trials are still missing to this point a change in the treatment regimen and frequency of SMBG measurements may be necessary in such patients [5].

Conclusion

This study shows that in patients with T2D treated with OAD one SMBG measurement a week is not associated with any deterioration in metabolic control (HbA1c) or therapeutic safety as compared to four measurements a week. Under study conditions compliance for a low frequency of SMBG was surprisingly high. Implementation of the study results in clinical routine would result in considerable cost-savings and it would be very important for the sake of type 2 diabetic patients not to have to measure SMBG more than once weekly.

Supporting Information

Protocol S2.

Trial Protocol, in English.

https://doi.org/10.1371/journal.pone.0003087.s002

(0.02 MB PDF)

Acknowledgments

We thank the following physicians for including patients from their practice into the study Dr. Axel Dost (Borken), Dr. Thorsten Dost (Borken), Dr. Gero Wallenfang (Bottrop), Dr. Baerbel Lobscheid (Meerbusch), Dr. Rainer Lux (Juelich).

We thank Selime Emiklibas and Petra Heidkamp for their help in seeing the participants, Jannett Schindler for her help with study coordination and Pablo Verde for his biostatistical support.

We also thank ROCHE Diagnostics (Mannheim, Germany) for providing blood glucose meters and meter strips for patients in the study.

Author Contributions

Conceived and designed the experiments: WAS CO HHA ND ML. Performed the experiments: WAS ML. Analyzed the data: CO. Contributed reagents/materials/analysis tools: WAS CO ML. Wrote the paper: WAS CO HHA ND ML.

References

  1. 1. The Diabetes Control and Complications Trial Research Group (1993) The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med 329: 977–986.The Diabetes Control and Complications Trial Research Group1993The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus.N Engl J Med329977986
  2. 2. Karter AJ, Ackerson LM, Darbinian JA, D'Agostino RB Jr, Ferrara A, et al. (2001) Self-monitoring of blood glucose levels and glycemic control: the Northern California Kaiser Permanente Diabetes registry. Am J Med 111: 1–9.AJ KarterLM AckersonJA DarbinianRB D'Agostino JrA. Ferrara2001Self-monitoring of blood glucose levels and glycemic control: the Northern California Kaiser Permanente Diabetes registry.Am J Med11119
  3. 3. Ohkubo Y, Kishikawa H, Araki E, Miyata T, Isami S, et al. (1995) Intensive insulin therapy prevents the progression of diabetic microvascular complications in Japanese patients with non-insulin-dependent diabetes mellitus: a randomized prospective 6-year study. Diabetes Res Clin Pract 28: 103–117.Y. OhkuboH. KishikawaE. ArakiT. MiyataS. Isami1995Intensive insulin therapy prevents the progression of diabetic microvascular complications in Japanese patients with non-insulin-dependent diabetes mellitus: a randomized prospective 6-year study.Diabetes Res Clin Pract28103117
  4. 4. Owens DR, Barnett AH, Pickup JC, Kerr D, Bushby P, et al. (2004) Blood glucose monitoring in type 1 and type 2 diabetes: reaching a multidisciplinary consensus. Diabetes and Primary Care 6: 8–16.DR OwensAH BarnettJC PickupD. KerrP. Bushby2004Blood glucose monitoring in type 1 and type 2 diabetes: reaching a multidisciplinary consensus.Diabetes and Primary Care6816
  5. 5. Bergenstal RM, Gavin JR III (2005) The role of self-monitoring of blood glucose in the care of people with diabetes: report of a global consensus conference. Am J Med 118: 1S–6S.RM BergenstalJR Gavin III2005The role of self-monitoring of blood glucose in the care of people with diabetes: report of a global consensus conference.Am J Med1181S6S
  6. 6. Nathan DM, Buse JB, Davidson MB, Heine RJ, Holman RR, et al. (2006) Management of hyperglycemia in type 2 diabetes: A consensus algorithm for the initiation and adjustment of therapy: a consensus statement from the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care 29: 1963–1972.DM NathanJB BuseMB DavidsonRJ HeineRR Holman2006Management of hyperglycemia in type 2 diabetes: A consensus algorithm for the initiation and adjustment of therapy: a consensus statement from the American Diabetes Association and the European Association for the Study of Diabetes.Diabetes Care2919631972
  7. 7. Davis WA, Bruce DG, Davis TM (2006) Is self-monitoring of blood glucose appropriate for all type 2 diabetic patients? The Fremantle Diabetes Study. Diabetes Care 29: 1764–1770.WA DavisDG BruceTM Davis2006Is self-monitoring of blood glucose appropriate for all type 2 diabetic patients? The Fremantle Diabetes Study.Diabetes Care2917641770
  8. 8. Benson K, Hartz AJ (2000) A comparison of observational studies and randomized, controlled trials. N Engl J Med 342: 1878–1886.K. BensonAJ Hartz2000A comparison of observational studies and randomized, controlled trials.N Engl J Med34218781886
  9. 9. Farmer A, Wade A, Goyder E, Yudkin P, French D, et al. (2007) Impact of self monitoring of blood glucose in the management of patients with non-insulin treated diabetes: open parallel group randomised trial. BMJ 335: 132–139.A. FarmerA. WadeE. GoyderP. YudkinD. French2007Impact of self monitoring of blood glucose in the management of patients with non-insulin treated diabetes: open parallel group randomised trial.BMJ335132139
  10. 10. Armitage P, Berry G, Matthews JNS (2002) Statistical methods in medical research. Oxford: Blackwell Science. P. ArmitageG. BerryJNS Matthews2002Statistical methods in medical researchOxfordBlackwell Science
  11. 11. Martin S, Schneider B, Heinemann L, Lodwig V, Kurth HJ, et al. (2006) Self-monitoring of blood glucose in type 2 diabetes and long-term outcome: an epidemiological cohort study. Diabetologia 49: 271–278.S. MartinB. SchneiderL. HeinemannV. LodwigHJ Kurth2006Self-monitoring of blood glucose in type 2 diabetes and long-term outcome: an epidemiological cohort study.Diabetologia49271278
  12. 12. Welschen LM, Bloemendal E, Nijpels G, Dekker JM, Heine RJ, et al. (2005) Self-monitoring of blood glucose in patients with type 2 diabetes who are not using insulin: a systematic review. Diabetes Care 28: 1510–1517.LM WelschenE. BloemendalG. NijpelsJM DekkerRJ Heine2005Self-monitoring of blood glucose in patients with type 2 diabetes who are not using insulin: a systematic review.Diabetes Care2815101517
  13. 13. Fontbonne A, Billault B, Acosta M, Percheron C, Varenne P, et al. (1989) Is glucose self-monitoring beneficial in non-insulin-treated diabetic patients? Results of a randomized comparative trial. Diabete Metab 15: 255–260.A. FontbonneB. BillaultM. AcostaC. PercheronP. Varenne1989Is glucose self-monitoring beneficial in non-insulin-treated diabetic patients? Results of a randomized comparative trial.Diabete Metab15255260
  14. 14. Muchmore DB, Springer J, Miller M (1994) Self-monitoring of blood glucose in overweight type 2 diabetic patients. Acta Diabetol 31: 215–219.DB MuchmoreJ. SpringerM. Miller1994Self-monitoring of blood glucose in overweight type 2 diabetic patients.Acta Diabetol31215219
  15. 15. Allen BT, DeLong ER, Feussner JR (1990) Impact of glucose self-monitoring on non-insulin-treated patients with type II diabetes mellitus. Randomized controlled trial comparing blood and urine testing. Diabetes Care 13: 1044–1050.BT AllenER DeLongJR Feussner1990Impact of glucose self-monitoring on non-insulin-treated patients with type II diabetes mellitus. Randomized controlled trial comparing blood and urine testing.Diabetes Care1310441050
  16. 16. Schwedes U, Siebolds M, Mertes G (2002) Meal-related structured self-monitoring of blood glucose: effect on diabetes control in non-insulin-treated type 2 diabetic patients. Diabetes Care 25: 1928–1932.U. SchwedesM. SieboldsG. Mertes2002Meal-related structured self-monitoring of blood glucose: effect on diabetes control in non-insulin-treated type 2 diabetic patients.Diabetes Care2519281932
  17. 17. Davidson MB, Castellanos M, Kain D, Duran P (2005) The effect of self monitoring of blood glucose concentrations on glycated hemoglobin levels in diabetic patients not taking insulin: a blinded, randomized trial. Am J Med 118: 422–425.MB DavidsonM. CastellanosD. KainP. Duran2005The effect of self monitoring of blood glucose concentrations on glycated hemoglobin levels in diabetic patients not taking insulin: a blinded, randomized trial.Am J Med118422425
  18. 18. Guerci B, Drouin P, Grange V, Bougneres P, Fontaine P, et al. (2003) Self-monitoring of blood glucose significantly improves metabolic control in patients with type 2 diabetes mellitus: the Auto-Surveillance Intervention Active (ASIA) study. Diabetes Metab 29: 587–594.B. GuerciP. DrouinV. GrangeP. BougneresP. Fontaine2003Self-monitoring of blood glucose significantly improves metabolic control in patients with type 2 diabetes mellitus: the Auto-Surveillance Intervention Active (ASIA) study.Diabetes Metab29587594
  19. 19. Koster I, von FL, Ihle P, Schubert I, Hauner H (2006) The cost burden of diabetes mellitus: the evidence from Germany–the CoDiM study. Diabetologia 49: 1498–1504.I. KosterFL vonP. IhleI. SchubertH. Hauner2006The cost burden of diabetes mellitus: the evidence from Germany–the CoDiM study.Diabetologia4914981504
  20. 20. Soumerai SB, Mah C, Zhang F, Adams A, Barton M, et al. (2004) Effects of health maintenance organization coverage of self-monitoring devices on diabetes self-care and glycemic control. Arch Intern Med 164: 645–652.SB SoumeraiC. MahF. ZhangA. AdamsM. Barton2004Effects of health maintenance organization coverage of self-monitoring devices on diabetes self-care and glycemic control.Arch Intern Med164645652