Journal of Clinical Epidemiology 99 (2018) 164e166
COMMENTARY
Pragmatic trials revisited: applicability is about individualization Jose A. Sacristana,b,*, Tatiana Dillac,b a
Department of Preventive Medicine and Public Health, School of Medicine, Universidad Autonoma de Madrid, Madrid, Spain b Medical Department, Lilly, Madrid, Spain c Health Evaluation and Market Access Program, Universidad Carlos III, Madrid, Spain Accepted 4 February 2018; Published online 9 February 2018
Keywords: Pragmatic trials; Explanatory trials; Pragmatic studies; Applicability; External validity; Generalizability; Patient-centered medicine; Patient orieted research; Randomized database studies; N-of-1 trials
Fifty years ago, Schwartz and Lellouch [1] published in the Journal of Chronic Diseases (the predecessor to the Journal of Clinical Epidemiology) the seminal article that described the differences between ‘‘explanatory’’ and ‘‘pragmatic’’ attitudes in therapeutic trials. Explanatory trials aim to understand how and why a clinical intervention works under ideal conditions. By contrast, pragmatic trials (also referred to as ‘‘practical,’’ ‘‘naturalistic,’’ or ‘‘effectiveness’’ trials) aim to guide real-word decisions about alternative interventions. Explanatory and pragmatic trials represent the ends of a continuum, with most trials including elements from the two types. Because of the limited external validity of traditional randomized clinical trials, the need to conduct more pragmatic trials has been widely advocated [2]. The Pragmatic Explanatory Continuum Indicator Summary tool [3] was introduced to support researchers in making design choices that match the intended question, whether it is primarily explanatory or pragmatic [4]. The most important feature of a pragmatic trial is that patients, doctors, interventions, and clinical settings are selected to maximize the applicability of the results to usual practice [5]. In order to increase the clinical applicability of the research findings, pragmatic trials compare clinically relevant alternative interventions, enroll a diverse study
Conflict of interest: J.A.S. and T.D. are employees of Lilly. The views or opinions presented in this work are solely those of the authors and do not represent those of Lilly. * Corresponding author. Department of Preventive Medicine and Public Health, School of Medicine, Universidad Autonoma de Madrid, Avenida Arzobispo Morcillo s/n, 28029 Madrid, Spain. Tel.: þ34619241656; fax: þ34916635231. E-mail address:
[email protected] (J.A. Sacristan).
population of patients, recruit from a variety of practice settings, and measure a broad range of relevant health outcomes [6]. As pragmatic trials try to mimic real-world practice, ‘‘usual clinical care’’ is the ideal setting to conduct this type of research. An innovative approach of pragmatic research is the randomized database studies (presently called randomized registry trials) that were suggested in this journal by Sacristan et al. in 1998 [7] and that are considered by some experts as ‘‘the next disruptive technology in clinical research’’ [8]. By embedding clinical trials into clinical practice, randomized registry trials combine the strengths of randomization (internal validity) with the advantages of registries and electronic health records (external validity). This type of research is gaining popularity as indicated by a recent work that reviewed the characteristics of more than 70 registry-based randomized controlled trials published thus far in the medical literature [9]. The current model of designing pragmatic trials is grounded in a profound contradiction. By applying flexible selection criteria and including diverse and a heterogeneous population of patients, researchers try to generate applicable results for ‘‘average’’ patients. The contradiction is that average patients do not exist in clinical practice, and physicians do not treat average patients but individuals. The focus on ‘‘generalizability’’ may contribute to the dilution of potential differences between subgroups, hindering decision-makers’ ability to identify which option is better for a given patient [10]. In summary, the strategy aimed at increasing applicability actually reduces it. At this point, it is necessary to clarify that classically, the terms generalizability, external validity, and applicability are all used with overlapping meanings (ie, whether the results can be reasonably applied to a defined group of patients in a particular setting in routine practice) [11].
https://doi.org/10.1016/j.jclinepi.2018.02.003 0895-4356/Ó 2018 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).
J.A. Sacristan, T. Dilla / Journal of Clinical Epidemiology 99 (2018) 164e166
The root of the contradiction is that the same model that considers that a pragmatic attitude aims to inform clinical decision-making assumes that health care decision-makers speak the language of populations. In reality, while historically decisions made by policy-makers have been population based, clinical decisions are always individual based [12]. Despite this duality, the rapid development of the patient-centered medicine movement and the advances in bioinformatics and molecular biology, which progressively provide a better understanding of the interindividual variability in treatment response, are narrowing the gap in the therapeutic perspective between doctors and regulators, payers, and other health care decision-makers [12]. As modern health care systems share the common goal of improving health outcomes for individual patients, it is imperative to transform pragmatic trials into patientcentered, dynamic, relevant, and useful research. The practice of patient-centered medicine requires the development of individual patient-oriented research [13]. There is a need to revisit the true meaning of the terms ‘‘generalizability’’ and ‘‘applicability,’’ two concepts usually considered equivalent but that, in this new context, become antagonistic. As health care decisions are becoming more patient centric, the term ‘‘applicability’’ should evoke ‘‘individual patient’’ rather than ‘‘average patient.’’ To be applicable in clinical practice, pragmatic trials need to redirect their focus from generalizability to individualization. The emphasis on ‘‘averages’’ may be understandable under the population-oriented research strategy that has pervaded the scientific community in the last few decades but cannot be justified under the patient-oriented research approach that must support the practice of patient-centered medicine [10]. The challenge is to identify subgroups of patients and individuals in whom the magnitude of the effect of an intervention may differ from the effect on the average population. Advances in molecular biology (eg, genomics, proteomics, and epigenomics) and predictive analytics are accelerating the development of precision medicine. But patient-centered medicine goes beyond genes. New individual patient-oriented research strategies should 1) develop new study designs that focus on a single person 2), incorporate patients’ perspectives on their care, and 3) integrate clinical research and medical care. First, biomarker-driven trials may increase the prediction response to targeted agents that are uniquely tailored to subgroups or individuals. New trial designs take into account the variability in treatment response. As an example, basket trials assess the effect of an intervention based on its mechanism of action in different diseases. In umbrella trials, researchers test the effect of multiple drugs in a study of a single disease. These designs represent an important step toward stratified therapy, but N-of-1 trials [14] are the purest form of pragmatic patient-centered design [15]. N-of-1 trials are multiple-period, crossover experiments comparing two or more treatments within individual
165
patients. They are the optimal design to estimate individual treatment effects directly and to identify the best treatment for each individual patient in each specific setting. The Journal of Clinical Epidemiology has recently published a number of articles reviewing the main features and applications of N-of-1 trials [16]. Second, patient-oriented pragmatic trials have to incorporate the perspective of the patients. Individual patients’ preferences, objectives, and values, possibly the most robust ‘‘biomarkers,’’ are an essential component in the personalization of health care [17]. Pragmatic trials are increasingly evaluating not only traditional hard variables such as mortality and major morbidity, but also patientreported outcomes (PROs) such as quality of life, functional outcomes, symptom severity, or satisfaction [18]. In order to be applicable to individuals, patient-centered pragmatic trials should be able to identify the relevant PROs for each individual patient, as not all symptoms will likely have the same impact on all patients. Similarly, when researchers compare the effectiveness of several therapeutic options, the ‘‘minimal clinically important difference’’ is usually defined for the whole population. In order to be applicable to individuals, ‘‘minimal clinically important differences’’ should be established by asking each patient. Third, the integration of research and care (through the systematic use of N-of-1 trials) may be an excellent opportunity to develop point-of-care clinical research [19] and to fully implement the concept of learning health care systems [20] that contribute to generate new knowledge and quickly apply the best evidence to benefit current and future patients [21]. Medicine-based evidence is a prerequisite for evidence-based medicine [22], and the doctor-patient encounter is the crucial point where clinical research and medical care converge [16,23], so daily care is the ideal venue to implement individual patient-oriented research. Classically, clinical research has centered on studying groups of individuals to extrapolate the findings to the general population. It is time to walk back from the population (the average patient) to the individual patient, understanding that population-oriented research is actually exploratory and individual-oriented research is confirmatory [24]. It has been 70 years since the publication of the first clinical trial of the modern era [25]. The study aimed to evaluate the efficacy of streptomycin vs. placebo in patients with pulmonary tuberculosis. Dr. Crofton, a member of the Medical Research Council who participated in the trial, recalled [26]: ‘‘.randomized trials were not intellectually stimulating. Our greatest intellectual challenge with tuberculosis research was to identify the reasons why the treatment failed. The results of randomized trials, together with detailed investigation of drug resistance in individual patients and the appropriate organization of services, allowed our team to reach one hundred percent recovery from pulmonary tuberculosis, the most common form of the disease and one that not long ago killed half of the patients who suffered from it.’’
166
J.A. Sacristan, T. Dilla / Journal of Clinical Epidemiology 99 (2018) 164e166
We need to learn from the pragmatic attitude of this (explanatory) clinical trial, an attitude that allowed the participating physicians (in their double roles as researchers and doctors) to generate new and useful knowledge not only for future patients, but also for the individual patients who directly benefitted from the study. It is time to revisit the concept of pragmatic trials and to reorient their focus. Pragmatism is not a type of design, but a mind-set. It requires recognizing that ‘‘individual’’ and ‘‘research’’ are compatible concepts, that applicability is not only about generalizability of the results but especially about their individualization, and that the walls that divide clinical research and medical care need to be demolished. Some recent symptoms indicate that the classical separation between explanatory trials, aimed at understanding, and pragmatic trials, aimed at decision, will become as blurred as the frontier between research and care. References [1] Schwartz D, Lellouch J. Explanatory and pragmatic attitudes in therapeutic trials. J Chronic Dis 1967;20:637e48. [2] Ford I, Norrie J. Pragmatic trials. N Engl J Med 2016;375:454e63. [3] Loudon K, Treweek S, Sullivan F, Donnan P, Thorpe K, Zwarestein M. The PRECIS-2 tool: designing trials that are fit for the purpose. BMJ 2015;350:h2147. [4] Knottnerus JA. Research methods must find ways of accommodating clinical reality, not ignoring it: the need for pragmatic trials. J Clin Epidemiol 2017;88:1e3. [5] Sox HC, Lewis R. Pragmatic trials. Practical answers to ‘‘real world’’ questions. JAMA 2016;316:1205e6. [6] Tunis SR, Stryer DB, Claney CM. Practical clinical trials: Increasing the value of clinical research for decision making in clinical and health policy. JAMA 2003;290:1624e32. [7] Sacristan JA, Soto J, Galende I, Hylan TR. Randomized database studies: a new method to assess drugs’ effectiveness? J Clin Epidemiol 1998;51:713e5. [8] Lauer MS, D’Agostino RBS. The randomized registry trial e the next disruptive technology in clinical research? N Engl J Med 2013;369: 1579e81.
[9] Mathes T, Buehn S, Prengel P, Pieper D. Registry-based randomized controlled trials merged the strength of randomized controlled trials and observational studies and give rise to more pragmatic trials. J Clin Epidemiol 2018;93:120e7. [10] Sacristan JA, Dilla T. Generalizability in pragmatic trials. JAMA 2017;317:87e8. [11] Rothwell PM. External validity of randomized controlled trials: ‘‘to whom do the results of this trial apply?’’. Lancet 2005;365:82e93. [12] Breckenridge AB, Eichler H-G, Jarow JP. Precision medicine and the changing role of regulatory agencies. Nat Rev Drug Discov 2016; 15(12):805e6. [13] Sacristan JA. Patient-centered medicine and patient-oriented research: improving health outcomes for individual patients. BMC Med Inform Decis Mak 2013;13:6. [14] Guyatt G, Sacket D, Taylor DW, Chong J, Roberts R, Pugsley S. Determining optimal therapy e randomized trials in individual patients. N Engl J Med 1986;314:889e92. [15] Schork NJ. Time for one-person trials. Nature 2015;520:609e11. [16] Knottnerus JA, Tugwell P, Tricco AC. Individual patients are the primary source and the target of clinical research. J Clin Epidemiol 2016;76:1e3. [17] Ziegelstein RC. Personomics. JAMA Intern Med 2015;175:888e9. [18] Porter ME, Larsson S, Lee TH. Standardizing patient outcomes measurement. N Engl J Med 2016;374:504e6. [19] Fiore LD, Brophy M, Ferguson RE, D’Avolio L, Hermos JA, Lew RA, et al. A point-of-care clinical trial comparing insulin administered using a sliding scale versus a weight-based regimen. Clin Trials 2011;8:183e95. [20] Friedman CP, Wong AK, Blumenthal D. Achieving a nationwide learning health system. Sci Transl Med 2010;2(57):1e3. [21] Sacristan JA. Clinical research and medical care: towards effective and complete integration. BMC Med Res Methodol 2015;15:4. [22] Knottnerus JA, Dinant GJ. Medicine based evidence, a prerequisite for evidence based medicine. BMJ 1997;315:1109e10. [23] Sacristan JA, Dilla T. Non big data without small data. Learning health care systems begin and end with the individual patient. J Eval Clin Pract 2015;21:2022e3. [24] Sacristan JA. Exploratory trials, confirmatory observations: a new reasoning model in the era of patient-centered medicine. BMC Med Res Methodol 2011;11:57. [25] Medical Research Council Streptomycin in Tuberculosis Trials Committee. Streptomycin for pulmonary tuberculosis. BMJ 1948;2:769e83. [26] Crofton J. The MRC randomized trial of streptomycin and its legacy: a view from the clinical front line. J R Soc Med 2006;99(10):51e4.
