In January of this year, the Federal Joint Committee (G-BA) commissioned the Institute for Quality and Efficiency in Health Care (IQWiG) to prepare a scientific report on the methodology for generating and evaluating care-related data in the context of ongoing data collection (AbD). A preliminary rapid report (1) has been available since the end of July. Interested parties were able to submit written comments until August 19. Just under a month later, on September 16, the IQWiG conducted a scientific discussion.

>> The IQWiG’s preliminary Rapid Report is entitled „Scientific elaboration of selected aspects for the generation of care-related data and the
evaluation for the purpose of benefit assessment of medicinal products according to Section 35a SGB V”.
In its conclusion, the institute states that the benefit assessment of pharmaceuticals requires data for comparison with the therapeutic standard. The necessary data collection should be conducted as comparative studies without randomization. IQWiG recognizes that, provided certain quality requirements are met, studies based on registry data can close this knowledge gap.
According to the IQWiG's definition, target trial emulation is a recommended approach when planning comparative studies without randomization based on data from real-world care settings to minimize systematic (avoidable) bias. The prerequisite for optimal emulation of a hypothetical RCT using observational data is that the required data are contained in the registry dataset with the necessary completeness and depth. High data quality can only be achieved across the board if the data generation and utilization of registry data are practical and can be achieved with as little effort as possible. The Institute considers the establishment and maintenance of a permanently available (functional) data infrastructure to be beneficial. This could be supported by pharmaceutical companies that wish to use the registries to conduct registry-based studies.
Furthermore, according to the preliminary rapid report, in comparative studies without randomization for the purpose of comparing treatment effects, a systematic identification of relevant confounders and their consideration in the analysis are necessary for adequate confounder control. The Institute considers confounder identification according to the approach of Pufulete 2022 to be feasible and, in the opinion of the reporters, represents a fundamentally sensible approach. However, before clinical experts assess the relevance of the confounders, an intensive consolidation of the identified confounders is recommended.
Likewise, the concepts for a pre-clinical study should assess whether sufficient patients can be included within an acceptable period of time to generate meaningful results for a benefit assessment. However, the information available for this assessment is generally uncertain. If sufficient information on the intervention and the comparator is available, an indicative sample size estimate will be performed for this purpose. If the necessary information for the assumptions underlying a pre-clinical sample size estimate is lacking, an indicative consideration of sample size scenarios should be conducted to demonstrate potentially detectable effects. Both approaches should follow the same principle (be based on identical parameters) and differ only in the parameter to be estimated using the other parameters. Furthermore, the IQWiG recommends:
• For long-term data collection (and patient support) in routine care, incentives are needed that compensate for the effort involved in data generation and motivate both centers and patients to collect as complete data as possible.
• To ensure that the effort and benefit of PRO surveys are balanced, digital surveys (e.g., via an app or patient portal) are recommended. This allows for low-threshold PRO surveys, independent of physician visits. Digital tools are already being successfully used for data collection in registries and in routine care, enabling PRO data to be made readily available for research purposes.
• A challenge in the evaluation of care-related data without randomization is determining the start date of observation (index date [to]). If a new therapy is initiated in both treatment groups at the start of observation (to) (switch indication), the ACNU design can be used in a study. To ideally corresponds to the day of the treatment decision. In a treatment situation in which patients in the control group continue treatment with an established therapeutic standard, the PNU design represents a suitable alternative.
• Regardless of the choice of design and evaluation strategies, the primary outcome should always be the ITT analysis (according to the treatment policy strategy). This requires that all patients (regardless of treatment changes during the observation period) be evaluated according to their original group allocation. The confounding influence of treatment changes during the course of the AbD can be addressed with sensitivity analyses.
by: MVF editor-in-chief Peter Stegmaier