Recommendations vary for second-line treatment of type 2 diabetes, with little consensus even among clinical guidelines1,2 in which the primary end point is maintenance of a blood hemoglobin A1c level less than 7%. Multiple prior studies also have contradictory results in terms of risk of adverse outcomes, including cardiovascular events and kidney disease.3,4
The study by O’Brien et al5 makes an important contribution to this area, by assessing the effectiveness of second-line treatment options for type 2 diabetes in reducing the risk of cardiovascular events (stroke, congestive heart failure, ischemic heart disease, and peripheral artery disease). Analyzing a large commercial claims data set, they found that risk of cardiovascular events did not differ between either sodium-glucose cotransporter 2 inhibitors or thiazolidinediones and dipeptidyl peptidase 4 inhibitors, and that risk was increased for sulfonylureas or meglitinides and basal insulin in comparison with dipeptidyl peptidase 4 inhibitors. Their findings agree with our recent study6 examining the association between classes of second-line type 2 diabetes therapies and blood glucose control as well as risk of adverse events (myocardial infarction, kidney disorders, and eye disorders). O’Brien et al5 make a valuable contribution by examining an additional drug class, glucagon-like peptide 1 receptor agonists, and finding that, in some analyses, risk of cardiovascular events was lower for glucagon-like peptide 1 receptor agonists in comparison with dipeptidyl peptidase 4 inhibitors.
Both the study by O’Brien et al5 and our previous study leverage observational data that capture details of health care processes and patient outcomes for millions of lives, with significant longitudinal coverage. Such data are increasingly accessible via commercial claims databases available for purchase and data provider networks such as the Observational Health Data Sciences and Informatics initiative,7 which also provides a common data model and an associated suite of open source analytic tools. Observational studies carried out at the scale enabled by such resources make it possible to investigate the efficacy of an ever-growing pool of treatments and their associated risks of adverse outcomes. The increasing ease of conducting such studies with these large data sources makes it essential both to conduct rigorous analyses and to make the details of these analyses transparent.
The study by O’Brien et al5 and the experience in the Observational Health Data Sciences and Informatics network has surfaced the following essential properties of rigorous and meaningful observational studies: (1) clearly stating the type of question and study; (2) exploratory analyses of the data source to ensure that the data are appropriate to answer the question of interest in terms of existence of necessary variables (eg, hemoglobin A1c blood test results) and years covered, especially when studying drugs that enter the market in specific years; (3) reporting precise and reproducible cohort definitions; (4) conducting sensitivity analyses and reporting their results in the main conclusions; (5) empirically calibrating significance thresholds8 and confidence intervals9; and (6) emphasizing study replication across sites and data sources.10
Study replication is especially important because if multiple researchers ask the same clinical question over time, across data sources and settings, and with differing study designs, the conclusions drawn from them are collectively stronger. The study by O’Brien et al5 targets an area of significant clinical uncertainty with the potential to inform the treatment of millions of individuals with type 2 diabetes. The study states the question type, clearly reports cohort definitions, and conducts sensitivity analyses (eg, note that after performing sensitivity analyses, they found that the seemingly lower risk of composite cardiovascular events with glucagon-like peptide 1 receptor agonists was no longer significant).
To advance a learning health system as a community, clinical researchers need to take advantage of observational data sources. To generate findings that can be trusted, clinical researchers need to adopt good practices in the analyses of observational data. JAMA Network Open, with its emphasis on ensuring rigorous, transparent, and reproducible studies, is leading the way in making the design and conduct of such research a norm.
Published: December 21, 2018. doi:10.1001/jamanetworkopen.2018.6119
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2018 Callahan A et al. JAMA Network Open.
Corresponding Author: Nigam H. Shah, MBBS, PhD, Center for Biomedical Informatics Research, Stanford University School of Medicine, 1265 Welch Rd, Room X-235, Stanford, CA 94025 (firstname.lastname@example.org).
Conflict of Interest Disclosures: Dr Shah reported receiving grants from the National Institutes of Health during the writing of this article. No other disclosures were reported.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Callahan A, Shah NH. A Second Opinion From Observational Data on Second-line Diabetes Drugs. JAMA Netw Open. 2018;1(8):e186119. doi:10.1001/jamanetworkopen.2018.6119
Customize your JAMA Network experience by selecting one or more topics from the list below.
Create a personal account or sign in to: