February 13th, 2014 2:33 pm ET - Guest Blogger
Alison Stewart Guest Blogger, and Muin J. Khoury, Director, Office of Public Health Genomics, Centers for Disease Control and Prevention
A general practitioner recently writing in the BMJ, said that evidence-based medicine is polluted with “fraud, sham diagnosis, short term data, poor regulation, surrogate ends, questionnaires that can’t be validated, and statistically significant but clinically irrelevant outcomes”, all leading to “overdiagnosis and misery”. In more temperate tones, Goldberger and Buxton recently suggested in a JAMA Viewpoint article that personalized medicine and guideline-based medicine “present conflicting priorities”, with evidence-based guidelines derived from clinical trial data failing to recognize the heterogeneity of the patient population to which they will be applied. The existence of these guidelines then acts, they say, as a barrier to the development of personalized approaches that would be more appropriate for different population subgroups – including those likely to gain no benefit from the intervention.
These are forceful arguments, but are they aimed at the right target? Is it really “evidence” that’s to blame for overtreatment and spiralling pharmaceutical usage, and are evidence-based guidelines standing in the way of the development of personalized medicine? This is especially pertinent in an era of breathtaking developments in genomics that promise a new era of precision or personalized medicine.
We have previously argued that an evidence-based approach is essential if we are to reap the promised benefits of genomic medicine – including the goal of more precisely targeted or “personalized” approaches to prevention and treatment. The methods of the population sciences are needed to enable us to integrate the complex set of biologic, social, economic, cultural and physical factors that interact to determine each individual’s health, disease susceptibility, and response to treatment or preventive interventions. The analytic frameworks of epidemiology and health services research provide methodologies for evaluating the benefits, harms, and costs (including opportunity costs) of implementing diagnostic and therapeutic interventions based on genomic factors, and of comparing genomic approaches to existing clinical practice.
We must also keep in mind that, however “personalized” at the level of individual patient and medical practitioner, integration of genomics into clinical care takes place within the wider context of healthcare organizations, families, communities and state and federal policies. The public health imperative is to ensure that validated applications can reach all segments of the population, to protect patients and the wider community from the premature implementation of tests or interventions that are minimally effective, ineffective, or even harmful. It’s not “evidence” that leads to overtreatment and harm, but poor evidence and overinterpretation of evidence.
No-one would deny, though, that there are serious challenges to face in assembling and appraising the evidence base to support personalized medicine in the era of genomics. As disease entities become more finely subdivided on the basis of histologic and molecular features, and genomic factors join “traditional” characteristics such as age and sex in defining population risk subgroups, it becomes more challenging to design clinical and population studies with enough power to yield statistically significant results. New ways of adapting or applying evidentiary standards will be needed, and consideration of the relative weight given to different types of evidence derived from comparative RCTs, observational studies, natural experiments, adaptive trials, pragmatic trials, evidence synthesis, and modeling.
In the meantime, should we abandon the concept of the “evidence-based guideline”? We think this would be throwing the baby out with the bath water. Guidelines will never be perfect, and should always be supplemented by patient empowerment and clinician knowledge of contextual factors including personal characteristics, social circumstances, values and preferences. If viewed in this way, guidelines developed according to sound principles of evidence-gathering and appraisal, explicit statement of the processes by which they were formulated (including steps taken to avoid bias and conflicts of interest), and which clearly set out the clinical scenario and patient population to which they apply, can serve as an important reference point for clinical decision-making. Guideline developers can also play a valuable role in flagging areas in which evidence is lacking or limited, and in identifying issues that need further research. Even in the rapidly developing field of genomic medicine, evidence on the balance of benefits and harms will always be required to make informed health related decisions by healthcare providers, patients and policy makers.
CDC - Blogs - Genomics and Health Impact Blog – Guidelines We Can Trust are Crucial for the Successful Implementation of Genomic Medicine
February 13th, 2014 2:33 pm ET - Guest Blogger
Alison Stewart, Guest Blogger and Muin J. Khoury, Director, Office of Public Health Genomics, Centers for Disease Control and Prevention
In a previous blog post, Michael Douglas and David Dotson from our office asked the question “So what are health care providers to do today when considering ordering a genomic test to diagnose, prevent or ameliorate a medical condition?” If we set aside the “genomic” bit of this for a moment, and think about how a health care provider might come to a decision about using any kind of medical test, the answer might well be that he or she would look to see whether any professional practice guidelines had been published that were relevant to the clinical situation in which use of the test was being considered.
Most clinical guidelines are published by professional groups and societies. While many are of high quality, an important contribution can also be made by independent groups evaluating a wide range of genomic applications using standardized criteria and methodology. The CDC sponsored EGAPP working group has developed recommendation statements for clinical uses of genomic tests, based on rigorous evidence reviews supplemented by a defined process in which the EGAPP Working Group (an independent, multidisciplinary panel) assesses the “magnitude of the net benefit and certainty of evidence, and consideration of other clinical and contextual issues”. The first EGAPP methodology was published in 2009 and was further refined to keep up with both a rapidly growing list of candidate applications, and a shifting technological landscape including next-generation and whole-genome sequencing.
The quality and trustworthiness of clinical guidelines is a topic that has attracted quite a bit of attention in the last few years; for example, the Institute of Medicine’s report “Clinical Guidelines We Can Trust” sets out seven standards such guidelines should meet. These include criteria for transparency; disclosure and managing of any conflicts of interest; stating the evidence base underlying any recommendations; methods for rating the strength of that evidence; a sufficiently rigorous process for external review, and provision for updating.
More recently, Gopalakrishna and colleagues have systematically reviewed the approaches used by groups developing guidelines for medical tests, considering both their “methodological characteristics”, defined as how the evidence base used for the guidelines was assembled, appraised and applied, and their “process characteristics”, which relate to the procedure followed in development of the guideline itself. Six methodological and five process categories are defined, and these are further divided into a total of 35 sub-categories.
So, how well has EGAPP done? Although the analysis by Gopalakrishna et al. was not intended to be a “beauty parade”, it does provide a useful check-list for guideline developers to assess how complete and robust their system is, and to identify any gaps or inadequacies that could be addressed. Among the 23 subcategories of methodological characteristics and 12 for process characteristics, EGAPP recommendations fulfilled 25 overall, including many areas relating to the soundness of its methodology for structuring the evidence search, gathering and appraising evidence, and formulating guidelines. The 10 subcategories not fulfilled by EGAPP included two relating to explicit criteria for bringing together and appraising different bodies of evidence, such as different systematic reviews or cost-effectiveness studies.
Interestingly, though, the other eight mainly concerned ways of making guidelines and recommendations more user-friendly for their intended recipients, such as piloting the guideline among stakeholders. Of all the groups evaluated by Gopalakrishna et al, the EGAPP working group is the only group that has focused exclusively on genomic medicine applications.
New diagnostic and prognostic tests are constantly being developed. Arguably, one of the biggest growth areas is in genomic tests, which now cut across virtually every clinical discipline including oncology, cardiology, neuroscience and infectious diseases, to name just a few. Although representing only a tiny fraction of genomics publications, most of which report primary research, the numbers of published guidelines for genomic applications are growing too. Our office’s ongoing horizon scanning effort identified 38 such articles published in a single year from May 2012 to May 2013. The articles reflect guidelines and recommendations from several organizations and expert groups in the US and around the world and cover topics across the lifespan.
As genomics becomes part of mainstream healthcare and public health, many if not most professional societies and guideline developers will begin to integrate genomic topics in their recommendation making. Because of the rapidly developing nature of the field, it will become increasingly crucial for minimum standards and review methods be developed and applied by various groups. Guidelines we can trust will become a cornerstone of the successful implementation of genomic medicine in practice.