701 Pennsylvania Avenue, NW Suite 800 Washington, D.C. 20004 2654 Tel: 202 783 8700 Fax: 202 783 8750 www.advamed.org May 16, 2014 Division of Dockets Management (HFA-305) Food and Drug Administration 5630 Fishers Lane, Room 1061 Rockville, MD 20852 RE: Docket No. FDA 2013 N 0745: Action Plan for the Collection, Analysis, and Availability of Demographic Subgroup Data in Applications for Approval of Food and Drug Administration-Regulated Medical Products; Notice of Public Hearing; Request for Comments Dear Sir or Madam: On behalf of AdvaMed, the Advanced Medical Technology Association, we are pleased to submit these comments in response to the Food and Drug Administration s (FDA) request for comments on the issues and challenges associated with the collection, analysis and availability of demographic subgroup data in FDA applications. The Advanced Medical Technology Association (AdvaMed) is the world s largest trade association representing medical device and diagnostics manufacturers. AdvaMed's member companies produce the innovations that are transforming health care through earlier disease detection, less invasive procedures and more effective treatments. AdvaMed has more than 400 member companies, ranging from the largest to the smallest medical technology innovators and manufacturers. AdvaMed advocates for a legal, regulatory and economic environment that advances global health care by assuring worldwide patient access to the benefits of medical technology. The Association promotes policies that foster the highest ethical standards, rapid product approvals, appropriate reimbursement, and access to international markets. AdvaMed supports the collection, analysis and communication of demographic subgroup information by sex, age, race and ethnicity. We agree with a number of key findings in the FDA Report on the Collection, Analysis and Availability of Demographic Subgroup Data for FDA- Approved Medical Products with respect to medical devices including that: Many Premarket Approval Applications (PMAs) ( nearly one quarter ) include an additional pivotal study cohort to obtain clinical experience in a specific subpopulation. Demographic subgroups may vary by product area because a variety of factors may influence the interpretation and clinical relevance of demographic information including the intended population for use, the disease prevalence and the study sample size. Bringing innovation to patient care worldwide
Division of Dockets Management (HFA-305) May 16, 2014 Page 2 of 7 The unique nature of medical devices may mean that additional information on demographic subgroups may not be contributory to FDA decision-making (e.g., additional analyses by subpopulations may be unnecessary when the in vitro diagnostic test s analyte detection and performance are highly accurate.) 1 AdvaMed provides responses to the questions in the Federal Register below. A. Demographic Subgroup Representation in Clinical Trials 1. What approaches might be used to encourage enrollment of representative proportions of subgroup participants in clinical trials consistent with disease prevalence in the underlying population being studied? Although FDA does not indicate it will require enrollment of representative proportions of subgroup participants in clinical trials consistent with disease prevalence, AdvaMed cautions FDA in taking such an approach, particularly for medical device trials. As we indicated in our March 19, 2012 comments on the Draft Guidance for Industry and Food and Drug Administration Staff: Evaluation of Sex Differences in Medical Device Clinical Studies, FDA should not require sponsors to enroll specific percentages of subgroups consistent with disease prevalence unless there is evidence that a difference in outcomes is anticipated for the subgroup. We also commented that FDA should provide guidance as to when FDA would require collection of further data to assure there are no meaningful/significant differences in device performance when a non-significant trend develops by chance. We urged this approach in our comments on the draft guidance and we also urge them for these comments because many devices are designed for the human anatomy which shares anatomical features across all sub-groups, including male and female populations (except where there are individual congenital deficiencies), and are designed to replace or augment a function of the body and typically act locally. As an example, for many devices, the most important subgroup may be anatomical size or BMI. It is also important to note that females and racial and ethnic minorities face significant and well-documented barriers that prevent them from receiving guideline indicated devices such as stents, ICDs, and CRTDs. The medical device industry, government agencies, physician societies and advocacy groups have focused on improving access to these populations. However, gaps in device utilization still persist. The FDA should consider the current utilization of these therapies in subpopulation rather than prevalence when encouraging enrollment of subgroup participants. It is also important to recognize that most device studies are not large enough to include substantial representations of small demographic subgroups. Requiring disease prevalence representation (if validated evidence can be found to calculate disease prevalence by subgroup) will likely considerably lengthen the trial time to enrollment and substantially increase trial costs as more trial sites will need to be established in order to shorten recruitment time. It may also not be easy to define what is considered as 1 FDA Report on Collection, Analysis, and Availability of Demographic Subgroup Data for FDA- Approved Medical Products, August 2013. p. 52-53
Division of Dockets Management (HFA-305) May 16, 2014 Page 3 of 7 consistent with disease prevalence. A requirement for demographic subgroup disease prevalence representation will depress medical device innovation over time. In lieu of required demographic subgroup disease prevalence representation in device trials, AdvaMed recommends that sponsors clinical trial plans include recruitment of diverse human subjects consistent with the Guidance for Industry: Collection of Race and Ethnicity Data in Clinical Trials. In general demographic subgroup analysis should only be required as part of a clinical trial if there is prior evidence that there is an anticipated difference in patient outcomes for the demographic subgroup in question. If there is evidence of a demographic subgroup difference, sponsors could be encouraged to choose clinical trial sites that may have an overrepresentation of the demographic group in question, develop subject specific study materials, or work with relevant patient advocacy groups as some examples. If evidence of demographic subgroup differences develop during the trial, a postmarket study could be developed to assess whether the difference is significant or is by chance. In addition, to encourage enrollment of demographic subgroups, some sponsors have instituted therapy awareness groups to work within the communities of those likely to receive their product. These groups work to assure adequate understanding of the disease and device as well as access to therapy and to assure that appropriate follow-up exists across all demographic segments that receive the product. Development of trust within the research environment is essential to encourage subject participation in a clinical trial. 2. What sources could be used to define disease prevalence among subgroups? Are there priority areas for study in terms of disease/condition, or in terms of demographic subgroup? It is not clear that disease prevalence exists for all device types and their related diseases and conditions. However, peer-reviewed journals and the Centers for Disease Control (CDC) may have some information on disease prevalence for certain diseases and conditions. CMS claims data or provider data bases (private payer) may also enable racial or ethnic prevalence to be established for certain diseases or conditions. The National Institutes of Health and state public health departments may also have relevant databases. Leading clinicians in a particular field, may also be aware of disease prevalence information, particularly for rare or small demographic subgroups (e.g., pediatric subgroups). In some instances, obtaining disease prevalence information would be a clinical trial in itself. 3. What are best practices and considerations for developing inclusion and exclusion criteria for clinical trials generally and for the early stages of research? The first consideration for clinical trials generally is the trial sponsor s understanding of the disease and/or condition. One should be aware of the breakdown (whenever possible) by gender, race, age, comorbidities, BMI, or the related key medical/demographic factors for that disease or condition. Particularly for late phase trials, the population being
Division of Dockets Management (HFA-305) May 16, 2014 Page 4 of 7 recruited for the medical device trial should reasonably represent the intent-totreat/intent-to-diagnose population upon commercialization. There are times when a sponsor may decide to initially target a subpopulation likely to have lower risk and/or higher benefit from the device. For instance, one might initially exclude patients near the end stage of disease (in the protocol exclusion criteria and from labeling) and first study earlier stage disease patients to assure success in efficacy there and only later conduct a separate trial with late stage disease patients so as to extend the labeling to the later stage disease population. This may be done to de-risk the development program associated with the more challenging-to-successfully-treat subpopulations and to protect them when the benefit may not outweigh the risk of the use of the device. The sponsor must take into consideration whether labeling restrictions may result from excluding key populations. For devices, early feasibility studies often purposely target subpopulations based on disease state and condition with a preference toward more robust patients and/or those more likely to demonstrate efficacy. Again, this is done to generate clinical trial data as early as possible and to give the manufacturer reason to believe that further clinical study will lead to a product that adds value to the health care system. It also protects vulnerable, higher risk subpopulations when there are concerns or uncertainty regarding the device safety profile. 4. What approaches should FDA use to standardize the capture of race and ethnicity information, including for studies conducted outside the United States? Standardization may be an impossible task, given all the mixed racial and ethnic identities around the world. Ideally, one would know the genetics of each subject, and their resulting propensity to the outcomes being measured. However, this remains a long-term objective at present, particularly when one considers the complex interplay between an individual s genetics and the culture/environment within which the patient resides factors which may determine the outcome of an illness in a given patient. FDA could ask sponsors to take into consideration the most relevant racial or ethnic considerations for that disease state, trial and test device. FDA could also consider whether it s Guidance for Industry: Collection of Race and Ethnicity Data in Clinical Trials should be updated. Based on input from AdvaMed members, a current and growing challenge is how to handle mixed races/ethnicities in clinical trials. If FDA updates its guidance, it should provide recommendations on this growing issue. One caveat or limitation that FDA should take into account is that current practices due to privacy laws in some OUS areas (e.g., European Union) may limit the ability to collect demographic subgroup data. B. Analysis of Demographic Subgroup Data 1. What are the statistical challenges in analyzing clinical trial data to evaluate subgroup differences?
Division of Dockets Management (HFA-305) May 16, 2014 Page 5 of 7 The statistical challenges for powering a subgroup (when it is appropriate due to prior meaningful differences in clinical outcomes) are similar to the statistical challenges associated with primary populations: one must enroll sufficient subjects to obtain statistical power and to understand event rates. Assumptions regarding equivalent intervention effects and variance between the overall population and the subgroups are not always tenable, and it is important to delineate whether the subgroup analyses are being performed for specific labeling indications or for a due diligence check on unusual or unexpected results in a specific subgroup (see response to B.2. below). Additionally, the reduced subgroup sample size brings a concurrent loss of power and precision for the planned statistical inference whether it is formal hypothesis testing or confidence interval estimation. 2. Given that it is not feasible to power most studies to detect subpopulation differences, what approaches should be used to analyze subgroups to explore clinically relevant information? Appropriate statistical analysis for some demographic subgroups might require doubling, tripling or quadrupling sample sizes with concurrent increases in the time and cost of a trial a logistical and financial hurdle that could prevent many device trials from being undertaken. In cases where labeling indications are sought for a specific subgroup, then we support appropriate statistical powering of the study to establish the subgroup effect in a way that protects the overall Type I error planned for the study. In due diligence cases, where subgroups are being investigated for consistency of results with the overall population to rule out an unexpected result, the planned statistical analyses of effectiveness should include a separate analysis where the effect of the subgroups are examined by means of an interaction term and graphical displays of the results by subgroup. The goal here would be to demonstrate consistency in the effect of intervention, and to identify or rule out a possible qualitative interaction across the levels of the subgroup variable (i.e., results in opposite directions across subgroup levels). From a safety perspective, similar graphical presentations could be provided for key complications or safety parameters to establish consistency in intervention effect across subgroups. 3. How might additional clinically relevant information about subgroups be obtained in the postmarket setting? We concur that it will not be feasible to power most medical device studies to detect subpopulation differences. In general, if there is a demographic subgroup trend that is detected during a trial, it could be studied postmarket either through a postmarket study or, if there is an existing and appropriate registry, potentially through the registry. Discussions should occur between the sponsor and FDA to determine the most appropriate mechanism to study demographic subgroup trends to assess whether they are significant or by chance.
Division of Dockets Management (HFA-305) May 16, 2014 Page 6 of 7 We would also note that Medwatch forms currently contain sex and age but do not include race or ethnicity. FDA may want to consider whether revising the Medwatch forms to capture race and ethnicity might be a useful mechanism to obtain clinically relevant information about subgroups in the postmarket setting. Finally, we urge FDA not to withhold clearance or approval of products while demographic subgroup trends are being studied (assuming the clinical trial data demonstrates that the device is safe and effective for its intended use). Withholding clearance or approval, would result in the perverse situation in which devices would be withheld for some demographic subgroups (e.g., females) while studies to assess a potential demographic trend are performed (e.g., for males) to determine if it is significant or by chance for some additional period of time. C. Communication of Demographic Subgroup Information to the Public 1. What information regarding demographic subgroups is helpful to health care professionals to make informed decisions about the use of medical products? To consumers/patients? To researchers? 2. What is the best way for FDA to communicate and make accessible such information to health care professionals? To consumers/patients? To researchers? If there is a scientific basis for the belief that there will be clinically meaningful demographic subgroup difference in outcomes, the clinical trial should be designed to answer the demographic subgroup question if reasonably possible (e.g., if sufficient members of the demographic subgroup can be enrolled in the trial in a timely manner). Demographic subgroup data should only be communicated if there is a scientific basis for the information or if there is the potential for significant risk to the patient s health or well-being (e.g., a significant difference exists or a prevailing theory on safety and effectiveness in a subgroup was overturned by clinical trial findings). If a trial were not designed or powered to assess safety and effectiveness in a subgroup, no comments should be made to the public, due to the significant potential for misinterpretation. Physicians treating a patient must make judgments from imperfect information, including trial subgroup analyses. Even the data communicated to physicians should be carefully bounded to avoid misinterpretation. Information regarding demographic subgroup analyses associated with devices should be included in the product labeling for health care providers and/or patients. For clinicians, researchers and patients, the information should also be included in clinical trial results on ClinicalTrials.gov and in 510(k) Summaries and PMA Summaries of Safety and Effectiveness (SSEDs). In other testimony to the Institutes of Medicine on strategies for responsible sharing of clinical trial data, AdvaMed has recommended that FDA should make 510(k) and SSEDs more prominent on FDA s website so they can be more easily accessed by the public. In the same testimony, we also recommended that medical journals should make their articles free and accessible to the public, potentially after a
Division of Dockets Management (HFA-305) May 16, 2014 Page 7 of 7 period of exclusivity or within a defined timeframe after publication. Currently, accessing medical journal articles is prohibitively expensive for the general public. Finally, such information should also be included in patient brochures and publications targeting the population-at-large concerned with their health. For purposes of this particular docket, No. FDA 2013 N 0745, we would also like to resubmit the comments we provided to this docket on November 20, 2013 which provided our recommendations and comments on FDA s Action Plan on demographic subgroups. Many of the comments and recommendations are relevant for the questions in this docket. In conclusion, thank you for the opportunity to provide comments on the issues and challenges associated with the collection, analysis and availability of demographic subgroup data in FDA applications. Please don t hesitate to contact me if you have any questions. Sincerely, Tara Federici Vice President, Technology and Regulatory Affairs Enclosures