Analysis of database information from patients in routine clinical settings may, in certain circumstances, provide similar results as randomized clinical trials (RCTs), according to a study published online November 20 in JAMA Internal Medicine.
The results suggest that database studies could help support supplemental drug applications for some drugs that are already approved by the US Food and Drug Administration (FDA) for other indications.
The study is one of the largest to analyze real-world data mirroring a large RCT (ONTARGET [Ongoing Telmisartan Alone and in Combination with Ramipril Global Endpoint Trial]) that had already established the clinical basis for a supplemental drug application. The drugs in question were the blood pressure medications telmisartan and ramipril.
“The fact that our case study bolstered the conclusions of a trial designed to identify a supplemental indication for a marketed medication and was done relatively efficiently using available data sets, rigorous epidemiologic methods, and modern software platforms supports the concept of conducting similar database analyses as part of routine practice for manufacturers submitting applications for supplemental indications to the FDA,” Michael Fralick, MD, from Harvard Medical School, Boston, Massachusetts, and colleagues write.
The researchers reviewed all supplemental applications to the FDA and their accompanying clinical trials between 2005 and 2014 to find a drug whose study population, inclusion and exclusion criteria, and primary clinical trial outcome could be found in Marketscan, a US healthcare database of more than 60 million commercially insured people.
That search identified the angiotensin receptor blocker telmisartan, which was approved for treating high blood pressure in 1998, and gained supplemental approval in 2009 for cardiovascular risk reduction in patients aged 55 years and older who are at high risk for major cardiovascular events but cannot take angiotensin-converting enzyme inhibitors.
Then researchers used similar inclusion and exclusion criteria as ONTARGET to identify a cohort of patients in the Marketscan database. Study patients were aged 55 years and older and were newly started on telmisartan before its supplemental approval, and had a diagnosis of coronary artery disease, peripheral arterial disease, cardiovascular disease, or diabetes. Including only new starts was one technique to decrease confounding and bias in the study.
Finally, the authors used propensity score matching to further minimize confounding and bias by adjusting for 74 patient characteristics, including demographics, comorbid conditions, concurrent medications, and healthcare use.
After propensity score matching, the analysis included 4665 patients newly started on telmisartan (mean age, 69.43 years; 51.7% men) and 4665 patients newly started on ramipril (mean age, 69.36 years; 50.2% men).
Telmisartan and ramipril had similar risk for the primary outcome, a composite of myocardial infarction, stroke, or hospitalization for congestive heart failure (hazard ratio, 0.99; 95% confidence interval, 0.85 – 1.14).
Telmisartan was also linked to much lower risk for angioedema than ramipril (hazard ratio, 0.13; 95% confidence interval, 0.03 – 0.56), suggesting the analysis could find known differences in adverse effects between the two drugs.
About half of drugs approved in the United States later gain additional approval for other indications or for modifications or expanded populations. Such applications usually rely on RCTs, the gold standard for establishing that a drug causes a particular effect. However, RCTs are costly and take a long time. For certain conditions, real-world databases, from insurance claims, registries, or electronic health records, contain much of the same information as RCTs and offer the possibility to establish clinical efficacy more quickly and at reduced cost.
For example, ONTARGET, the pivotal RCT that established that telmisartan had similar efficacy as ramipril in decreasing cardiovascular risk and led to telmisartan’s supplemental approval, took about 7 years to complete and cost tens of millions of dollars. This new database study took 12 weeks and cost a hundredth of the price.
Database studies can also include patients often excluded from RCTs, such as the elderly, pregnant women, and those with many medical conditions. Database studies are sometimes larger and may detect rare adverse events not found in RCTs.
The problem is that database studies are nonrandomized; therefore, the results may not always be valid or reproducible. However, applying new statistical techniques may improve results.
In an invited commentary, Rover M. Califf, MD, from Duke University School of Medicine, Durham, North Carolina, and Verily Life Sciences (Alphabet), South San Francisco, California, and former commissioner of food and drugs, US Food and Drug Administration, writes that the study is “valuable and technically excellent,” but represents only one case.
“Thus, it is open to the criticism that generalizing from 1 positive finding to a vast field of potential treatment comparisons with observational data is analogous to painting the target around the arrow, especially considering the high probability that the telmisartan-ramipril comparison would work,” he explains.
Also, because the study could not account for death, it used a different endpoint (a composite of heart attack, stroke, or heart failure hospitalization) than ONTARGET (a composite of all cardiovascular deaths, heart failure admission, nonfatal myocardial infarction, or nonfatal stroke).
However, he acknowledges that “excessive bureaucracy” may stand in the way of performing rigorous studies for supplemental applications. That, in turn, may encourage drug makers to push off-label use. The result is a research system that fails to answer many questions important to clinical practice.
For that reason, observational studies may have a role, he writes, while emphasizing the primacy of RCTs for establishing the comparative efficacy between drugs.
“In many circumstances, however, observational analyses will supplement RCTs for new indications and provide deeper knowledge about real-world use within labeled indications,” he concludes. “Despite the need for more examples and robust efforts to guide the use of different methods for different circumstances, observational analyses have an important place in the continuum of clinical evidence.”
The study was funded by the Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women’s Hospital, and the Laura and John Arnold Foundation. Dr Fralick was funded by the University of Toronto Clinician Scientist Training Program. One coauthor’s work is also supported by the Engelberg Foundation and the Harvard Program in Therapeutic Science. Dr Califf served as the commissioner of food and drugs, US Food and Drug Administration (FDA), from February 2016 to January 2017. Before being appointed as FDA deputy commissioner for medical products and tobacco in February 2015, he received grant funding and/or consulted for the following: the Patient-Centered Outcomes Research Institute, the National Institutes of Health, the FDA, Amylin, Eli Lilly and Company, Bristol-Myers Squibb, Janssen Research and Development, Merck, Novartis, Amgen, Bayer Healthcare, BMEB Services, Genentech, GlaxoSmithKline, Heart.org–Daiichi Sankyo, Kowa, Les Laboratoires Servier, Medscape/Heart.org, Regado, and Roche. He also holds equity in N30 Pharma and Portola. He currently receives consulting payments from Merck and is employed as a scientific adviser by Verily Life Sciences (Alphabet).
JAMA Intern Med. Published online November 20, 2017. Article full text, Commentary
For more news, join us on Facebook and Twitter
