Analysis of the ADA’s Standard of Diabetes Care Recommendation for Continuous Glucose Monitoring System

Abstract

Transparency is a key factor to understand how recommendations were reached and to decide whether to adopt them or not. Currently, the American Diabetes Association (ADA) recommends the usage of real-time continuous glucose monitors in certain subgroups of people living with type 1 diabetes mellitus. In this commentary we sought to critically appraise this recommendation, mainly regarding the outcomes assessed, the evidence used for each outcome, and how the balance of benefits and harms was made. We found that the decision-making process followed to reach the ADA recommendation was not clear, and that the final recommendation does not seem to be supported in the assessed evidence.

Introduction

The American Diabetes Association (ADA) “Standards of Medical Care in Diabetes” ¹ is intended to provide guidelines of diabetes care to clinicians, patients, researchers, payers, and other interested individuals. Its recommendations are considered as the ADA’s current position and are taken into account by many practitioners around the world.

The Standards of Medical Care in Diabetes 2020, ² recommendation number 7.9, state that “When used properly, real-time continuous glucose monitors in conjunction with insulin therapy are a useful tool to lower A1C levels and/or reduce hypoglycemia in adults with type 1 diabetes who are not meeting glycemic targets, have hypoglycemia unawareness, and/or have episodes of hypoglycemia.”

Although different criteria could be used to reach a recommendation, in this document we focus on the choice of outcomes and the benefits/harms balance, which are essential components of an evidence-based decision-making, as stated by different organizations that work on decision-making in health such as GRADE (Grading of Recommendations Assessment, Development and Evaluation),^3,4 and by tools used to critically assess clinical practice guidelines such as AGREE-REX (Appraisal of Guidelines REsearch and Evaluation – Recommendations EXcellence).^5,6

Discussion on Evidence for ADA Recommendation 7.9

Although the ADA Standards of Medical Care in Diabetes are transparent in stating what studies they are reviewing to provide treatment recommendations, it is not specified how these studies were selected (ie, how was the literature review process carried out). Moreover, the reasoning used to reach the ADA recommendations is not completely clear, and the discussion of other important criteria (such as risk of bias of the studies, precision of the results, and publication bias) is absent. Altogether, these failures in clarity prevent the readers from seeing the whole picture and complicate the critical appraisal of the recommendation.

Regarding evidence of benefit–harm balance, to justify its 7.9 recommendation, the ADA states that real-time continuous monitoring of glucose (CMG) has a beneficial effect on two outcomes: hemoglobin A1c (HbA1c) and hypoglycemia. The guideline fails to present information regarding how the selection of these outcomes was performed, or whether evidence for other potential important outcomes for patients was searched or not. This is pertinent, since other patient important outcomes such as quality of life or ease of use of CMG are important and need to be taken into account when assessing the evidence regarding CMG. ⁷

Regarding HbA1c, the ADA mentions that

“In two larger studies in adults with type 1 diabetes that assessed the benefit of real-time CGM in patients on MDI (multi daily injections), there were significant reductions in A1C: –0.6% in one and –0.43% in the other. No reduction in A1C was seen in a small study performed in underserved, less well-educated adults with type 1 diabetes. In the adult subset of the JDRF CGM study, there was a significant reduction in A1C of –0.53% in patients who were primarily treated with insulin pump therapy.”

This statement references only four randomized controlled trials (RCTs) (one that found no benefits, and the other three reported HbA1c mean differences of –0.6%, –0.43%, and –0.53% after 24-26 weeks).^8-11 However, this has two important limitations:

• It fails to reference the other RCTs, and also systematic reviews (SRs) published. The most recent SR before the ADA guideline was released was published in 2017 ¹² ; it included 11 RCTs, and found a reduction in HbA1c that is not clinically important (mean difference: –0.26%, 95% confidence interval [CI]: –0.46% to –0.05%), and which was much lower than the RCTs cited in the guideline. This is consistent with an SR published in 2020, which included 15 RCTs and found a reduction of –0.16% (95% CI: –0.25 to –0.06). ¹³

• There is no discussion regarding the clinical relevance of the mean differences found in the cited RCTs, which is pertinent taking into account that none of their CIs exceed a 0.5% reduction of HbA1c, which is considered a clinically important threshold. ¹⁴

Regarding hypoglycemia, the ADA mentions that “in studies in adults where reduction in episodes of hypoglycemia was the primary end point, significant reductions were seen in individuals with type 1 diabetes on MDI or CSI. In one study in patients who were at higher risk for episodes of hypoglycemia, there was a reduction in rates of all levels of hypoglycemia.”

To support this statement, the ADA only cites three RCTs.^15-17 One of the trials found a 4.7% reduction of the percentage of time in hypoglycemia (95% CI: –5.9 to –3.4) and a reduction of 1.1 hours per day (95% CI: –1.4 to 0.8) ¹⁵ in hypoglycemia, the second one found a reduction of 1.24 hours per day (95% CI: –1.48 to –1), ¹⁶ and the third study found differences in the proportion of hypoglycemic blood glucose measurements in those blinded to measurements and those who used real-time CMG (10.1% vs 7.5%, P value = .04). ¹⁷ Once again, this has two important limitations:

• It fails to reference the most recent SR published before the ADA guideline was released, which included four RCTs in the analysis of time in hypoglycemia (<60 mg/dL) and three RCTs that assessed incidence of hypoglycemia (<70 mg/dL). This SR did not find an effect of CMG in reducing hypoglycemia. ¹²

• It fails to discuss the clinical relevance of their findings, in other words, what reduction in time in hypoglycemia or incidence of hypoglycemia would be considered as clinically relevant.

Given the inconsistencies found in the supposed benefits, the available evidence should be carefully assessed before rushing into a conclusion. In addition, ADA do not mention what potential harms of the use of real-time continuous glucose monitors were considered such as potential skin reactions (pain, erythema, itching, and allergic reactions among others) that can increase diabetes-related emotional distress and disease burden.^18,19 Thus, the tradeoffs between the benefits and the potential harms that are supposed to be weighted to reach the recommendations are not clear.

Discussion on Other Important Factors for Decision-Making, for ADA Recommendation 7.9

Regarding other criteria that should be taken into account to reach to the recommendation, it is not clear if the patients’ and the clinicians’ acceptability were taken into account, nor how that was done. In the case of patients’ acceptability, it would be important to mention if patients’ important outcomes were identified, if information related to these outcomes was searched, and how it was assessed. Given that greater clinical benefits have been associated with adherence to device usage,^1,20 factors related to adherence such as barriers perceived by patients (intrusiveness of the device, quality of life) or emotional distress caused by the device and self-efficacy should be taken into account as patient important outcomes. ²⁰

Moreover, ADA starts the recommendation with the phrase “When used properly,” which seems to leave the decision to the clinicians. However, no additional information is given regarding how clinicians should decide regarding the use of CMG. ADA should directly state whether this recommendation should be applied to every subgroup of patient they mention (“adults with type 1 diabetes who are not meeting glycemic targets, have hypoglycemia unawareness, and/or have episodes of hypoglycemia”) and, if not, it should provide guidance on how to choose in which patient it would be adequate to use these devices.

Finally, we would like to stress the importance of using a rigorous and transparent methodology for the development of clinical practice guidelines and issuing of recommendations. The methodology used to reach recommendations varies widely between guideline development groups ²¹ and it has been reported that several actors assessing the same clinical question can arrive to different recommendations, which forces the reader to assess which recommendation better adapts to his/her context. ²² Moreover, conflicts of interest and political pressure could influence on the decision-making process and using a structured and transparent framework to make decisions in health could help deal with potential problems.

For the above-mentioned reasons, we recommend using a transparent and systematic methodology such as GRADE to develop recommendations, and following basic steps: ²³ (1) define questions to be addressed, (2) find or prepare systematic reviews to answer the question, (3) assess the relative importance of outcomes, (4) summarize the evidence and assess its certainty, and (5) decide on the direction and strength of the recommendation. In the case of clinical practice guidelines’ users, we recommend to use a tool such as AGREE or AGREE-REX^5,6 to understand how guidelines recommendations were reached and to what extent can they be used in their contexts.

Conclusion

To sum up, for the assessed recommendation, the ADA uses two outcomes (HbA1c and hypoglycemia), seems to cite studies at convenience, and does not aim to use all the available evidence, nor discusses the clinical importance of the benefits and harms. These shortcomings prevent making an adequate benefit/harm balance and therefore to state an evidence-based recommendation. Thus, a more transparent and comprehensive reporting of the methodology would be more appropriate to provide reliable recommendations.

Currently, RCTs that have assessed CMG usually do not evaluate hard outcomes such as diabetes complications or death, and there is scarcity of cost/benefit evaluations on this subject. In this context, guideline developers must give clinicians enough information to understand how a decision was maken, in terms of both the evidence taken into account and the reasoning behind the evidence-to-decision process.