Intended for healthcare professionals

Editorials Christmas 2017

Big Science for patient centred care

BMJ 2017; 359 doi: https://doi.org/10.1136/bmj.j5600 (Published 08 December 2017) Cite this as: BMJ 2017;359:j5600
  1. Victor M Montori
  1. Mayo Clinic, Rochester, MN, USA
  1. montori.victor{at}mayo.edu

Only massive and fearless collaboration can produce evidence that works for patient care

Advancing the problematic situation of patients requires clinicians and patients to draw from their experience, expertise, and evidence. Working together, they should be able to identify a way forward that makes intellectual, emotional, and practical sense.1 How research is conducted, however, offers challenges to achieving patient centred care. The solution may lie in Big Science.

Little Science

For many situations, funders prefer to obtain answers efficiently, and researchers win the competition for resources when they propose the smallest and briefest study. To succeed, these Little Science trials must enrol high risk patients, use comparators that favour the experimental intervention, and use responsive endpoints—surrogate or composite endpoints. Hiding unflattering results completes the illusion. Clinicians and patients must deal with imprecise, inconsistent, and incomplete results about the effect of impertinent comparisons on endpoints of unclear relevance to the experience and destiny of patients. Little Science may work to support regulatory action and marketing of new drugs and devices, but falls short of what patients and clinicians need for care.

Big Data

Database observational research, or Big Data, can explore the effect of disease and care on many patients (with enough participants with rare conditions or from important subgroups), across many outcomes (with enough occurrence of rare outcomes), at a low cost per question. Big Data are more representative of the “real world” than Little Science trials that recruit a few selected patients from referral centres. Clinical trialists can use Big Data to design more efficient and useful trials, and to formulate and validate prediction models to support decision making and individualise care.2 Sophisticated analyses cannot overcome inaccurate data—common in these large clinical and administrative repositories—or permit causal inferences. Too often, Big Data are not-so-great data.

Big Science

Patient centred care demands trustworthy evidence that applies directly and confidently to the problems of patients. Studies designed to practically meet the needs of decision makers3 will need to be much bigger and longer than Little Science trials. Systematic reviews can help, but only when the body of small trials is trustworthy. Rather, randomised mega-trials, prospective meta-analyses, and other designs—some yet to be invented—respond better to the demands of care. When they include thousands of patients and follow them completely for long periods, they can compare options as used in the “real world” and estimate their effect on outcomes that really matter to patients. Their design and conduct can be planned to protect against bias, rendering their results more trustworthy than Big Data studies. Their feasibility requires massive collaboration between scientists, academic institutions, clinics and health systems, patients, and communities. This is Big Science.

To work, Big Science requires generous collaboration. We have no evidence that relying on competition to cull ideas worth funding produces the best science. A zero-sum game favours competitors who offer timid improvements; bold innovations or reproduction of previous studies need not apply.4 Competition produces rivals where we need partners; secrecy, redundancy, and waste where we need transparency and efficiency. Young talent is convinced—through repeated failure that underinvestment makes more likely—that their ideas are not worthy of realisation. Competition hinders our progress to Big Science. Funding, scientific, and healthcare agencies must bring together a few talented people in a room and thousands online to openly share ideas, methods, and protocols. Promotion and recognition programmes must be engineered to foster fearless collaboration, beyond collaboration’s inherent intellectual and emotional rewards. The ownership of ideas, resources, data, results, and credit in this culture must remain subsidiary to the celebration of problems solved.

The evaluation of new methods and the testing of alternative interventions will require a close and trusting partnership between clinical research and practice. Mega-trials must take place across diverse health systems, on which scientists must impose the smallest possible footprint. Learning healthcare systems must engage with this minimally disruptive research and participate in Big Science, but they must also apply its results. For instance, healthcare systems can prevent suffering and waste by forgoing the use of inefficient or unsafe interventions and instead implement better options. Big Science is most likely to produce the kind of nuanced comparative data necessary to support shared decision making conversations.5 Patient centred care demands practical evidence. To produce it, we must shift from a culture of competition to one of collaboration: from Little Science and Big Data to Big Science.

Footnotes

  • Adapted from a keynote address at the first Patient-Centered Outcomes Research Institute (PCORI) Annual Meeting in Washington, DC, in 2015, and from a report of that address. Montori VM. Big Science; research collaboration for evidence-based care. Circ Cardiovasc Qual Outcomes 2016 Nov;9(6):688-92.

  • Competing interests: The author has read and understood BMJ’s policy on declaration of interests and has no relevant interests to declare.

  • Provenance and peer review: Commissioned, not externally peer reviewed.

References

View Abstract