Most new drug approvals are based on data from large randomized clinical trials (RCTs). However, there are sometimes contradictory conclusions from seemingly similar trials and generalizability of conclusions from these trials in limited (see Chapter ___) [1]. Furthermore, even when a RCT shows a convincing benefit of an intervention the benefit is often not equally distributed amongst recipients [[2], [3], [4]]. These limitations impose gaps between evidence from RCTs, real-world evidence (RWE) derived from real-world data (RWD). This gap is particularly critical for haematological cancers where interventions are complex, costly and with substantial potential of adverse events. RWE derived from analysing RWD is receiving increasing attention, We discuss and the potential of real-world evidence (RWE) in drug approvals and clinical decision-making in haematology.

RCTs are considered the highest level of evidence for safety and efficacy of an intervention. Randomization of sufficient numbers of subjects maximizes the likelihood differences in outcome results from an intervention rather than selection biases and known and unknown (latent) confounders and co-variates [5]. However, RCTs require subject selection and study-eligibility criteria which ki\\limit generalizability of the conclusions [6]. Subjects receive the intervention in highly controlled settings unlike in clinical practice. Compliance in RCTs is much better than in clinical settings [7] RCTs require considerable time from inception to final reporting subject participation is typically brief such that almost all RCTs transform into observational databases. These and other considerations limit generalizability of conclusions of RCTs to clinical practice settings [[8], [9], [10], [11]].

There are possible alternatives to RCTs to determine safety and efficacy in the real world. Pragmatic clinical trials (PCTs) have more liberal inclusion criteria resembling those used in clinical practice [12,13]. However, methodological, ethical and legal standards and costs are as high as conventional RCTs [14]. Therefore, less expensive and alternative study-designs are needed for generating RWE [15,16]. In a registry or observational database (ODB, a form of registry) the intervention and outcome occur before the analyses. Data from ODBs address issues such as safety and efficacy of an intervention in the real world, of special interest in rare haematological cancers. However, selection biases and confounding are major concerns because of unrecognized baseline differences.

Sources for generating RWE are electronic health records, claims and billing activities, disease or drugs registries and patient-generated data. However, there are several constraints on informing clinical practice using RWD. 1st, subject-level data are needed. 2nd, a population-based data archive is needed. 3rd, the population sample should be large, especially when dealing with new interventions and uncommon phenotypes/genotypes and/or diseases. A possible solution to these requirements is using of Electronic Healthcare Utilization (EHU) data created to pay providers of health care services [17]. These EHU data have several advantages: (1) The electronic format database can be obtained without great cost, over long intervals and quickly; (2) A unique anonymized identifier assigned to each person could be linked to datasets to track healthcare given over time; (3) Informed consent is not usually required for collecting and storing EHU data [18]: and (4) The data reflect clinical practice especially in the context of a national health care system [7].

The real barrier to using EHU data is that data are collected for health care management and not for research. Therefore, important biological, clinical and therapy information may not be captured or, if captured, may not be in a useable, in a compatible form or both. Consequently, data sharing processes are needed to capture additional information and outcomes [19]. Examples are cancer registries, health data from referring centres, smart home apps and wearable digital medical devices.

There are limitations when analysing EHU as RWD. 1st, there is a need to collect population-based data to avoid or limit selection biases and confounders. 2nd, generalizability of RWD is not always possible. 3rd, data sharing requires universal or at least inter-operable technical standards [20]. 4th, data protection is critical [21].

It is unlikely a RWD repository could instantaneously increase our knowledge of the real-world. The challenge is to interrogate these data and generate useful and credible evidence. The latter refers not only to capture “big data” (i.e. large volumes of structured and unstructured data from several sources) but the ability to design appropriate studies and use correct analyses and scientific methods [20,22,23]. Explanatory (hypothesis testing) and exploratory studies can be done with RWD. Explanatory studies typically evaluate a pre-specified effects focusing on effect size sharing with RCTs hypothesis testing. Exploratory investigations are a first step in learning about possible effects of an intervention such as who in a population benefits.