Clinical trials: conventional or pragmatic?

The aim of conventional large, randomized, controlled, clinical trials (RCTs) is to answer relevant clinical questions. What may be good is that RCTs are simple, with clear design, predefined sample size, including old and comorbid patients, with expected duration, without the need for complex equipment, and with essential dataset. Predefined sub-analyses on subgroups or phenotypes can be incorporated in the design. The bad consists in rigid and costly designs, mostly funded by companies and managed by Clinical and Research Organizations with slowly and cumbersome realization. Ancillary investigations to answer pathophysiological questions are limited by costs, time and the scarce interest of the sponsors and often also of the researchers. Representativeness of the results are often debatable. Comorbidities remain an open issue, typically clicked in the database as ‘physician-reported’, without further information. With time, the bad of RCTs has prevailed in the considerations of the scientific community and the word ‘pragmatic’ started to be used without a clear meaning. An abrupt methodological shift has brought about by the COVID-19 outbreak. In response to the emergency, thousands of studies started. The majority were methodologically poor, uncoordinated, with irrelevant results. Others were excellent, most with adaptive designs, flexible and adjustable to the circumstances, testing several drugs simultaneously in parallel arms.1 Nevertheless, after 2 years, more than 5.5 million people died worldwide and we are facing the fifth wave of the COVID-19 new variant without the discovery of any new options except vaccines but… clinical trials were forced to become ‘pragmatic’. ‘Pragmatism’ arose from concerns that several trials failed to impact on daily practice,2 thereby RCTs should be mostly reoriented to real-world data, focused on outcomes and symptom management and, possibly, digitalized. Two decades ago the idea of a digitalized health system as ‘universal’ source of big data, linked to artificial intelligence, learning from everyday clinical experience was put forward and this was the really important novelty.3 What was lacking was an effective digitization of the health system. In 2012 the US National Institutes of Health (NIH) created the Health Care Systems Research Collaboratory4, 5 and launched 12 practical (pragmatic) projects with primary community goals such as: opioid abuse, suicide prevention, heart failure, end-stage renal disease, colo-rectal cancer, and others. The platform for the NIH Collaboratory initiative was the electronic health record (EHR) system, storing all health data of each citizen. Indeed, the EHR system, which represents the ultimate reference for any clinical trial, is a huge, essential, and difficult programme. In most countries, health digital systems and related knowledge are suboptimal or randomly distributed, hampering the achievements of a trustable network. In April 2019, just before the COVID outbreak, the US NIH and the National Science Foundation held a workshop among experts in clinical trials, digital technology, and analysis to implement the use of digital technologies. The conclusions of the workshop are reported in a position paper6 and in Table 1. In 2021, in the middle of the COVID-19 catastrophe, the US National Academy of Medicine (NAM) 50 released a document on pragmatic clinical trials (PCTs),7 that reads ‘the generation of real-world evidence without the limitations associated with observational studies, or the time, expense, and lack of generalizability that are barriers to conducting conventional randomized clinical trials.’ PCTs should be strongly integrated with standard care, and ‘involve existing clinicians instead of dedicated research staff, and use routinely collected clinical data.’ The NAM 50 has further specified the characteristics of pragmatic trials. First, PCTs should aim to answer a ‘practical care’ need. Second, PCTs should target conditions affecting several patients, have the potential to improve quality of care, and demonstrate benefit over a short period (1 to 3 years). Third, the research question should be a service or a treatment added to usual care, possibly delivered as part of it. Thus, PCTs are attractive for the health systems, offering actionable information with low barrier to participate in. Conceptually, pragmatism should lead to a large generalizability of the results (compatible with the current health strategy of precision medicine?), a larger participation of physicians, including those not attracted by explanatory research (so increasing patient enrollment), and should make the trial shorter, less expensive, and easier to manage. However, in reality, what makes a trial pragmatic? In 2016, Ford and Norrie provided the answer within ‘The Changing Face of Clinical Trials’ series.8 It reads Schwartz and Lellouch2 proposed a distinction between explanatory trials, which confirm a physiological or clinical hypothesis, and pragmatic trials, which inform a clinical or policy decision by providing evidence for adoption of the intervention into real-world clinical practice.8 An indicator has also been proposed to assess the degree of a trial's pragmatism by tools like the PRECIS-2 (Pragmatic Explanatory Continuum Indicator Summary-2), based on nine non-categorical domains, reported in Figure 1.9 Actually, the proposed criteria are of help to scientifically profile any trial. Besides these criteria in the current cardiology literature, pragmatic means a digital trial. So, what makes a clinical trial “pragmatic and digital” (PDCTs)? The classical answer is “one that uses digital technology to improve recruitment, data collection, and analysis.”6 At present, most countries cannot perform PDCTs, as they do not have an operative EHR system. Therefore, they are excluded. Even in countries with implemented EHR experience, operational barrier may arise mainly because of the heterogeneity of software commercially available which may mine homogeneity. Schematically, in remote or virtual trials (labelled as digital – but also as pragmatic), personal contacts among patients, investigators and the coordinating structure are minimized or absent, replaced by web communications. In general, patients receive an invitation describing the trial by mail, the details for enrolment and the informed consent, written or presented by a video. No visits or direct contact with doctors or monitors are planned. Once the enrolment is confirmed, the enrollee will receive information on the drug to be tested and on the appropriate comparator, together with timing of re-contacts and follow-up if something more than EHR is needed. The visits, if there will be one, will be conducted remotely. The collection of physical data will be by appropriate telemetric methods, which is expected to rapidly increase.10 Of course, in remote trials the EHR system is vital for capturing clinical events (endpoints) and understand the disease evolution. Information taken remotely reduces time, effort, and participants' travelling expenses. However, this may lead to an unintended selection of enrollees limited to those able to cooperate with the trial management, while a number of patients may be excluded for linguistic, technical or cognitive reasons, the so-called ‘digital divide’. Thus, representativeness of the population enrolled and generalizability of the results still need to be confirmed trial by trial. For instance, comorbidities are frequent and may be of paramount importance in heart failure. Thus, the diagnosis should be reasonably verified instead of being reported by a computer click, as it often occurs. Progressive use of digitalization in clinical research is logic and necessary but, in view of heterogeneity of both culture and lifestyle across the world, the organization of international trials needs to be accurately prepared and still monitored because the risk of moving from ‘simple’ to ‘facile’ is high, as suggested by the ambiguous results of a few large trials conducted by expert investigators.11-13 Most pragmatic trials aim to answer categorical questions: yes or no, A or B. Thus, the temptation to formulate ‘categorically’ most clinical questions and make ‘pragmatic’ most trials, expecting ‘definitive’ answers with less efforts and costs, may be high. Nevertheless, considering the proposed characteristics of PCTs (Table 2),14 one can perceive that this model is not immediately applicable to all RCTs. For instance, the clinical research of new drugs first requires explanatory trials to test efficacy and safety, to identify patients who could best benefit from the new medicaments and try to understand why. In summary, from the current ‘pragmatic’ experience with PDCTs (online supplementary Appendix S1), the good relates to short performance times, organizational agility, low costs and transferability to clinical practice. The risk is an over-simplification of the trial questions. The bad would be uncertain results difficult to correctly translate into clinical practice. The European Society of Cardiology is trying to implement a platform with continuous registration of harmonized patient data during hospital stay or in the ambulatory setting. The project will encompass different cardiovascular diseases across several European nations with the possibility of linking the data with the national claim records. Such approach, if properly conducted, will allow implementation of more PDCTs in the general population while substantially reducing cost and time.15 The digitalization of the health system is expected to expand in the future and clinical trials will benefit from machine learning and artificial intelligence. What we are living today will be reported in the history of medicine as the first worldwide approach to control an unexpected pandemic. Literally, a war to survive which, if one likes seeing the glass half full, has had a positive effect on the conduction of RCTs which, by necessity, have become pragmatic and adaptive. We thank Elisa Ghizzardi for her professional assistance on this manuscript. Conflict of interest: none declared. Appendix S1. Some examples of remote pragmatic trials on heart failure, most still in progress. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.