Glycemic Control and Type 2 Diabetes Mellitus: The Optimal Hemoglobin A_1c Targets. A Guidance Statement from the American College of Physicians Free

Anthony J Bleyer

Section on Nephrology, Wake Forest University School of Medicine, Winston-Salem, NC 27157

September 25, 2007

Hemoglobin A1c in Dialysis Patients

The publication, "Clinical Guidelines: Glycemic Control and Type 2 Diabetes Mellitus: The Optimal Hemoglobin A1c Targets. A Guidance Statement from the American College of Physicians"(1) discussed the importance of achieving hemoglobin A1c (HbA1c) targets in patients with diabetes. Although they mention the lack of existing evidence supporting targets in those with co-morbid conditions, they failed to mention inherent inaccuracies of the HbA1c assay in those with severe kidney failure on hemodialysis. This group is at extremely high risk for cardiovascular complications(2). Approximately 200,000 Americans on dialysis have diabetes mellitus(3). Glycohemoglobin levels are lower than expected in this population, compared with ambient blood glucose measurements and relative to the glycated albumin assay. This effect appears related, in part, to uremia- associated reductions in erythrocyte lifespan (2). Indeed, an association between glycohemoglobin levels and survival in United States hemodialysis patients only becomes evident after controlling for hemoglobin levels(4).

Managing diabetes in the dialysis population is exceedingly difficult. These patients are at high risk for diabetic complications, in the face of increased risk for hypoglycemia(5). They also cannot utilize all pharmacologic therapies (e.g., metformin). Physicians need to be cognizant of the inaccuracy of glycosylated hemoglobin in this population. Glycosylated hemoglobin values below 7% in the hemodialysis population do not necessarily indicate optimal control, and may provide clinicians and patients a false sense of security.

Reference List

(1) Qaseem A, Vijan S, Snow V, Cross T, Weiss KB, Owens DK. Glycemic control and Type 2 diabetes mellitus: The Optimal Hemoglobin A1c targets. A guidance statement from the American College of Physicians. Ann Int Med. 2007;147:417-22.

(2) Inaba M, Okuno S, Kumeda Y, Yamda S, Imanishi Y, Tabata T et al. Glycated albumin is a better glycemic indicator than glycated hemoglobin values in hemodialysis patients with diabetes: effect of anemia and erythropoietin injection. J Am Soc Nephrol. 2007;18:896-903.

(3) National Institutes of Health: National Institute of Diabetes, Digestive, and Kidney Disease. USRDS 2007 Annual Data Report: Atlas of Chronic Kidney Disease and End Stage Renal Disease in the U.S. Bethesda,MD: 2007.

(4) Kalantar-Zadeh K, Arnovitz J, Kopple JD, McAllister CJ, Regidor DL, Whellan D et al. Hemoglobin A1C and survival in maintenance hemodialysis patients. Diabetes Care. 2007;30:1049-55.

(5) Fischer KF, Lees JA, Newman JH. Hypoglycemia in hospitalized patients. N Engl J Med. 1986;315:1245-50.

Conflict of Interest:

The authors have conducted a study sponsored by Asahi Kasei Pharma Corporation evaluating an assay of glycated albumin.

Valerie A. Palda

Guidelines Advisory Committee and University of Toronto

September 29, 2007

Using the AGREE to compare multiple guidelines

We read with interest the ACP guidelines on glycemic control in Type 2 diabetes mellitus (1). The use of the AGREE instrument to compare the methodologic rigour of various guidelines is a meaningful step forward, to our knowledge used regularly only by a few groups such as the Guidelines Advisory Committee in Ontario (http://www.gacguidelines.ca) and more widely in Europe by the Belgian Centre for Evidence-Based Medicine (http://www.cebam.be) and others (http://www.agreetrust.org/links.htm). Using well-developed guidelines as a basis for forming recommendations is an important strategy to improve quality of care and may also lead to efficiencies in the development process.

Nevertheless, we have two concerns with the methodology used to assess the guidelines. First, using a total AGREE score as the authors do in Table 2 (1), averaging the scores of all six domains, assumes each domain is of similar importance. However the developers of the AGREE Instrument specifically state that the domain scores should not be aggregated into a composite score because they are independent (2). Each domain score is calculated by standardising the summed up scores of individual items. Moreover, the relative importance of the domains is not static and depends on the values of the raters. In other words, total scores provide no data on performance within each domain and currently, we do not know in which of these domains bias is more likely to influence recommendations or practice (3).

Our second concern regarding the methodology of the ACP guideline is the use of only two raters. It is considered the standard for systematic reviews to have two abstractors for data, but the inter-rater agreement is optimized when at least four trained raters are used to apply the AGREE score. While current efforts with the AGREE Next Steps Project (Phase 1 and Phase 2) (4) are aimed to address this human resource demand, the current standard to achieve acceptable reliability is four (3).

The methodology of guideline development is evolving rapidly, with groups such as the Guidelines International Network (http://www.g-i-n.net) and the ADAPTE project (http://www.adapte.org) providing collaborative guidance to interested developers (5). We hope that the American College of Physicians will consider participating in these activities, providing wisdom acquired over many years of evidence-based guideline development.

References:

1. Qaseem A, Vijan S, Snow V, Cross T, Weiss KB, Owens DK. Glycemic control and Type 2 diabetes mellitus: The Optimal Hemoglobin A1c targets. A guidance statement from the American College of Physicians. Ann Int Med. 2007;147:417-22.

2.The AGREE Collaboration. The Appraisal of Guidelines for Research & Evaluation (AGREE) Instrument, 2001. London: The AGREE Research Trust. (http://www.agreetrust.org)

3. The AGREE Collaboration. Development and validation of an international appraisal instrument for assessing the quality of clinical practice guidelines: the AGREE project. Quality and Safety in Health Care 2003; 12(1): 18-23.

4. see http://www.agreetrust.org/projects.htm Accessed September 29 2007

5. Fervers B, Burgers JS, Haugh M, Latreille J, Mlika-Cabanne N, Paquet L, Coulombe M, Poirier M, Burnand B, for the ADAPTE working group. Adaptation of clinical guidelines: a review of methods and experiences. Int J Qual Health Care 2006; 18: 167-176.

Conflict of Interest:

None declared

Thomas E. Finucane

Johns Hopkins Bayview Medical Center

October 2, 2007

The hegemony of "tight control"

The ACP guidelines on glycemic control and type 2 diabetes present the familiar chestnut that, "Various trials have validated the need for tight glycemic control." The authors provide three references for this remark. Two of these are from the Diabetes Control and Complications Trial, which studied type 1 diabetes, and thus are not directly relevant. The third is UKPDS 33, a trial of tight control which found no significant difference in two aggregate endpoints; all-cause mortality and any diabetes-related death. The third aggregate endpoint showed a difference in a composite of diabetes-related events and found significant reduction.. The benefit was limited to microvascular events and was due largely to a reduction in the need for retinal photocoagulation.

The burdens of "tight glycemic control" can be considerable, and the "need" for tight control hasn't actually been validated. This is particularly true for the frail and elderly.

Conflict of Interest:

None declared

Beatrice Fervers

SOR and Centre Leon Berard, Lyon, France

October 12, 2007

Guideline adaptation: an appealing alternative to de novo guideline development

The Adapte Collaboration (www.adapte.org), an international group aiming to enhance the use of evidence through more efficient development and implementation of guidelines by guideline adaptation read with great interest the process used by the ACP to develop their statement for glycemic control (1). The development and maintenance of high quality guidelines require substantial time, expertise and resources and the ACP has used an attractive alternative to de novo guideline development. Yet, guideline adaptation also presents challenges. While guideline developers often consider existing guidelines as a source of evidence, few are using a formal process for reviewing and selecting high quality guidelines among the heterogeneous, ever-expanding volume of sometimes differing or even contradictory recommendations.

The ACP performed a comprehensive search to identify existing guidelines and used the AGREE instrument to assess their quality. This instrument has become a widely accepted standard for assessing guideline methodological quality, but does not assess guideline content. We think it is important to explicitly assess how up-to-date source guidelines are, whether recommendations are consistent with the underlying evidence, clinically useful and applicable within the targeted context of use (2, 3). Evidence supporting recommendations may be outdated within three years (4). Not all guidelines assessed by the ACP provide an explicit link between the recommendations and the underlying evidence and seven of the 9 guidelines are older than three years. Furthermore, organisational and cultural differences between countries can lead to legitimate variations in recommendations, even if the evidence based is the same (5). Thus, guidelines produced in one setting may not be appropriate for another, without modification.

The Adapte Collaboration has developed a systematic approach for the adaptation of guidelines and produced a manual and a resource toolkit. The process is designed to help customize existing guidelines to a different setting while preserving evidence-based principles as well as to encourage confidence in, acceptance, and use of adapted guidelines by targeted users.

1. Qaseem A, Vijan S, Snow V, Cross T, Weiss KB, Owens DK. Glycemic control and type 2 Diabetes Mellitus: The optimal Hemoglobin A1C targets. A guidance statement for the American College of Physicians. Ann Intern Med 2007; 147:417-422

2. Graham ID, Harrison MB, Brouwers M, Davies BL, Dunn S. Facilitating the use of evidence in practice: evaluating and adapting clinical practice guidelines for local use by health care organizations. J Obstet Gynecol Neonatal Nurs 2002;31:599-611.

3. Fervers B, Burgers JS, Haugh M, Latreille J, Mlika-Cabanne N, Paquet L, et al. Adaptation of clinical guidelines: literature review and proposition for a framework and procedure. Int J Qual Health Care 2006;18:167-76.

4. Shekelle P, Eccles MP, Grimshaw JM, Woolf SH. When should guidelines be updated? BMJ 2001; 323:155-7.

5. Eisinger F, Geller G, Burke W, Holtzman NA. Cultural basis for differences between US and French clinical recommendations for women at increased risk of breast and ovarian cancer. Lancet 1999;353:919-20.

Conflict of Interest:

None declared

Amir Qaseem

American College of Physicians

December 21, 2007

In Response

We would like to thank the authors of the letters for their comments on ACP's guidance statement on "Glycemic Control and Type 2 Diabetes: The Optimal Hemoglobin A1c Targets". We concur that the AGREE instrument is not comprehensive enough to capture issues such as those raised by the Adapte Collaboration. In the paper, we acknowledged the limitations of the AGREE instrument such as the inability to capture guideline development process that lies outside the guideline document. Due to these limitations, ACP developed additional criteria that are listed in table 1 of the guidance statement document that addresses concerns such as translation of evidence into recommendations. However, we agree with the authors that our criteria did not account for organizational and cultural difference between different countries and hope for a new and improved evaluation instrument that can account for shortcomings such as the ones raised in the letter.

The concern related to the use of aggregate scores vs. individual domain scores is valid and that is why we included a table in the paper showing scores for each individual question to give the reader an idea how guidelines faired against each other. In addition, we individually evaluated each guideline and summarized their strengths and weaknesses that were not necessarily based on the AGREE instrument.

We also considered the issue of using two raters. In addition to the two raters, other members of the committee also reviewed selected parts of the guidelines. We agree that more reviewers are helpful, however, and are investigating ways to increase the number of reviewers for the assessments with the AGREE instrument.

Conflict of Interest:

None declared

Robert E. Hoyt

Naval Hospital Pensacola

February 12, 2008

In Response

TO THE EDITOR: After admitting a 66 year old compliant type 2 diabetic with severe hypoglycemia due to oral hypoglycemic agents, I read the ACP guidance statement on glycemic control with great interest (1). Clinicians recognize the importance of decreasing practice variability by following evidence based guidelines. On the other hand, clinical guidelines may be a harder sell if the supporting evidence is lacking or there is risk related to the treatment. Unfortunately, much of what is written about the association of hyperglycemia and outcomes is based on observational studies and not on clinical trials. Importantly, the widely reported clinical trial, the United Kingdom Prospective Diabetes Study (UKPDS) failed to show a reduction in macrovascular disease with tighter glucose control. Although the UKPDS showed improvement in microvascular disease with tighter control, there was not a difference in visual acuity or blindness in spite of the reduced need for photocoagulation and there was no reduction in renal failure, in spite of reduced microalbumin levels. Not surprisingly, patients in the intensive treatment group had more episodes of hypoglycemia (2).

As pointed out in the editorial by Pogach in the same issue of the Annals, the 2007 diabetic guidelines by the American Diabetes Association (ADA) state "there are no clinical trial data available for the effects of glycemic control in patients with advanced complications, the elderly (>65 years of age) or young children"(3)(4). How many busy clinicians will make the effort to read this 37 page guideline to find this caveat and how many local guidelines warn clinicians about the need to be less stringent with the elderly or with patients with co-morbidities? How many physicians spend the majority of their visit with diabetic patients trying to improve glycemic control with little time left for intensified blood pressure and hyperlipidemia control and smoking cessation?

Two important clinical trials were reported this month. The Action to Control Cardiovascular Risk in Diabetes (ACCORD) stopped its intensive glycemic control arm (HbA1c <6.0%) due to an unexplained 20% relative risk increase in all cause mortality (5). The follow-up STENO-2 trial studied type 2 diabetic patients with microalbuminuria, treating multiple risk factors in accordance with ADA guidelines, compared to conventional therapy. They were able to show that at 13.3 years of follow-up there was a 20% absolute risk reduction in all cause mortality (6).

Until we better understand the true risk of intensive glycemic control we should exercise caution and focus on multiple risk factor reduction and individualization of glycemic control, as recommended by the ACP guidelines.

Robert E. Hoyt MD, FACP Pensacola, Fl 32508

Potential Financial Conflicts of Interest: None disclosed

References

1. Qaseem A, Vihan S, Snow V, Cross JT, Weiss KB, Owens DK. Glycemic Control and Type 2 Diabetes Mellitus: The Optimal Hemoglobin A1C Targets. A Guidance Statement from the American College of Physicians. Ann Intern Med. 2007;147:417-422 [PMID: 17876024]

2. UK Prospective Diabetes Study (UKPDS) Group. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). Lancet. 1998;352:837-53 [PMID: 9742976]

3. Pogash LM. "Doctor, How CERTain Are We That This Diabetes Medication Is Best For Me?" Ann Intern Med. 2007;147:428-9 [PMID: 17652707]

4. American Diabetes Association. Standards of medical care in diabetes"”2007. Diabetes Care. 2007;30 Suppl 1:S4-S41. [PMID: 17192377]

5. Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial. NHLBI. www.nhlbi.nih.gov . February 6 2008 (Accessed February 7 2008)

6. Gaede, P, Lund-Andersen, H, Parving HH, Pedersen O. Effect of a Multifactorial Intervention on Mortality in Type 2 Diabetes. N Engl J Med 2008;358:580-91