Dynamic HTA for digital health solutions: opportunities and challenges for patient-centered evaluation.

The special features in Germany’s DiGA regulations (the breadth of positive-care effects, the unique and flexible nature of initial evidence generation, and the need for continuous evaluation) highlighted above will require new evaluation methods. One crucial factor for a successful widespread application of dynamic HTA will be the use of RWD and RWE. FDA provides clear and helpful definitions of RWD and RWE, which we rely upon here: “Real-world data are the data relating to patient health status and/or the delivery of health care routinely collected from a variety of sources.” ( 27 ) These may include data from electronic health records (EHRs), claims and billing activities from the healthcare delivery system, product and/or disease registries, patient-generated data (including from in-home use and digital devices such as wearable sensors), as well as data from other sources that are relevant to the health status, such as data or metadata from mobile devices ( 27 ). Relatedly, but distinctly, “real-world evidence is the clinical evidence regarding the usage and potential benefits or risks of a medical product derived from analysis of RWD.” Generally speaking, “RWE can be generated by different study designs or analyses, including but not limited to, randomized trials, including large simple trials, pragmatic trials, and observational studies (prospective and/or retrospective)” ( 27 )– that is, RWE explicitly covers many of the study design options that are outlined in the regulations that govern acceptable forms of clinical evidence for DiGA in the German market. Indeed, RWD and RWE can support all three dynamic HTA features described above. First, SaMDs can be designed to record data to evaluate positive-care effects for users over time. For example, metadata from apps themselves can track the duration and frequency of use and therefore measure adherence to a prescribed treatment regimen that involves a SaMD product. Improvements in health status can also be documented via sensors and connected devices, such as smart glucometers for patients with type I diabetes. Other positive care effects can be identified from existing data sources. For example, health insurance claims data can be used to document reductions in illness duration and increases in patient survival. Finally, features such as in-app questionnaires lend themselves easily to the digital environment, providing a standardized framework for collecting data within the context of product use. In-app questionnaires can be used to document several relevant categories of positive-care effects ranging from established scales for measuring patient quality of life to nearly all measures of “structure and process” (see Table 2 ), such as facilitation of access to care, health literacy, patient sovereignty/autonomy in health management, and reducing therapy-related expenses and burdens for patients and their relatives. Of course, the scope for gathering RWD goes well beyond health insurance claims and in-app questionnaires, and in the future, the increasing digitization of health systems combined with the establishment and growth of patient registries will create new opportunities for the generation of RWD. Looking forward, RWD from a number of sources will be an important input to assessing the benefits and costs/savings associated with new healthcare technologies. RWD and RWE employed as suggested above also have the potential to support initial evidence generation in the DiGA process. As noted, in order to gain reimbursement in Germany via the DiGA Fast-Track, some forms of initial evidence must be provided to the BfArM. RWE, for example, from claims data, can deliver this initial evidence at costs far below those associated with traditional RCTs. Most importantly, to make ongoing evaluations feasible and the corresponding HTAs truly dynamic, both RWD and RWE will be necessary because full RCTs are simply not feasible for every product update – nor would their requirement be desirable, as it would slow the pace of innovation and/or roll-out of improvements and additional beneficial features to patients. Indeed, many non-digital medical devices undergo incremental innovation, whereby new product versions can be released without a full clinical evaluation involving RCTs. RWE has been widely used in many countries. Pongiglione et al. ( 28 ) provide a review of sources of RWD and to what extent they are known and used in medical, epidemiological, and economic research in 13 European countries. One example shows that cancer survival rates are higher for participants of an RCT versus an RWE cohort ( 29 ). The National Institute for Health and Care Excellence regularly accepts RWE in the evaluation of cancer drugs ( 30 ) as does FDA, which has published formal regulatory guidance on the use of RWD and RWE for studies of biologic drugs ( 31 ) and medical devices ( 32 ) as well as specific recommendations on the use of HER data in clinical investigations ( 33 ). Even though expert interviews from Germany reveal a generally positive attitude toward RWD and RWE ( 34 ), none of the DiGA approved to date have used this form of data and evidence generation, suggesting a clear opportunity for the introduction of new tools and data sources going forward. There are already important steps toward that goal with new tools currently developed for example in the framework of data fusion ( 35 ) while a new research data center for healthcare data currently established in Germany can serve as a hub for RWE studies. 1 Of course, the use of RWE has many limits. Causal inference is more difficult than in RCTs and managing challenges such as patient selection, data representativeness, and data privacy/security will be key issues for the practical success of implementing RWE in the German context. Further, researchers have highlighted key areas that should be prioritized for the use of RWE in digital medical product evaluation and in the promotion of international harmonization of best evidentiary practices. These include the establishment of best practices around topics including missing data, study endpoints, comparator group(s), multimodal interventions, study question(s), equity, generalizability, confounders, and fit-for-purpose approaches ( 36 ). However, the promise of more and richer data, better tools, and more patient-centered data collection and product launches also provide an overwhelming promise for learning to implement RWE approaches thoughtfully. Institutionalizing continuous evaluation with RWE for SaMDs, however, could be the first step toward a broader movement of assessing benefits in healthcare systems both on the disease and system levels ( 37 ). More broadly, pioneers of RWE for dynamic HTA in the SaMD setting may also garner insights from studying other approaches that take advantage of dynamic, ongoing data generation – and in particular, the use of RWD for health economic decision-making internationally. For example, “coverage with evidence development” (CED) has many parallels with the DiGA Fast-Track. Experts on CED have noted that when used “selectively” and for “innovative” interventions, this approach can “provide patient access…while data to minimize uncertainty are collected” ( 38 ). Other work on CED has described the types of settings in which such an approach is likely to be most appropriate, arguing for its use when “there are reasonable grounds for believing that a technology will offer significant benefits” and remaining uncertainty “around the clinical or cost effectiveness…can be overcome through evidence that can be generated in an appropriate time frame and is the main source of equivocality in a coverage decision” ( 39 ), a set of circumstances that also directly applies to the second core characteristic of HTAs in the context of Germany’s Fast-Track pathway for DiGA.

Digital health applications and the accompanying demand for dynamic HTA present both a significant challenge and great opportunity for contemporary healthcare delivery. In Germany, new policies now allow for the use of both a broader set of research designs and more flexible approaches for their demonstration. However, early experience with evidence generation has shown that manufacturers are still hesitant to focus on nontraditional endpoints and nontraditional evidence-generation strategies. In particular, recent changes to German policy have facilitated reimbursement of SaMD products that do not necessarily fit with either current HTA approaches nor are they well-matched to the unique characteristics and additional needs of digital health applications (in particular, ongoing development and, as a corollary, a need for ongoing evaluation). As such, a new, dynamic HTA will be important both to facilitate continuous improvement and ongoing reimbursement of innovative healthcare solutions as well as the basis for their fair, evidence-based, and efficient reimbursement after launch. Additionally, the approaches presented here may have implications for the development of HTA for non-digital products such as orphan drugs, where approval decisions may be made based on limited evidence and subsequently supported by RWD and RWE from routine medical practice (e.g., claims data) or registries. The next generation of HTAs for SaMD in general – and for DiGAs in Germany in particular – will need to take advantage of new sources of RWD and methodological innovations, including improvements and best practices in the use of RWD, while managing the challenges unique to using RWE. If successful, such approaches will facilitate patient access to demonstrably beneficial SaMD products and ensure that their prices are value-based and, by doing so, improve the healthcare delivery system for all parties.