Ralph Horwitz, an academic internist and one of my mentors from my fellowship many years ago, has been a proponent of medicine-based evidence.1 In contrast to evidence-based medicine, medicine-based evidence is designed to aggregate clinical experience and outcomes and inform clinicians about what treatment would be most helpful and least harmful for their individual patients.2 You might be thinking that I am confused—Isn't evidence-based medicine just that? Doesn't evidence-based medicine provide us with the best available evidence to treat our patients? Don't treatment guidelines aggregate the evidence so that it is clinically useful? What is the difference between evidence-based medicine and medicine-based evidence beyond the sequence of the words?
Horwitz details the failures of evidence-based medicine to inform clinical care and assist clinicians to make optimal decisions3,4 and then describes what medicine-based evidence care should (but does not yet) look like.2 Evidence-based medicine provides us with aggregate average differences between groups randomized to minimize baseline susceptibility biases. The lack of generalizability from randomized clinical trials to clinical care has been well documented and discussed, as well as the remarkable lack of evidence used in formulating guidelines (even in cardiology).5
So much of evidence-based medicine simply fails to ask the right questions—Given the medical, biological, and biographical/historical data about my patient, what would be the best treatment at this time? Horwitz references our mutual mentor, the late Alvin Feinstein, who suggested that we aggregate a library of clinical experience and then make the best match between the patient in front of us and the group of patients most similar to our patient.6 Now that we have electronic medical records, making such a library of clinical experiences is theoretically possible, but to the best of my knowledge, no such system exists. But it is not too far of a leap of the imagination to consider an optimal system in which the clinician (or someone else) inputs clinically relevant information into the electronic medical record, the system then finds all of the patients in the system who are reasonably similar to the patient, and then reports back the treatments received along with the outcomes and the relevant and informative efficacy and effectiveness trials. Then the system can recommend a set of treatment options for the clinician to review with their patient. That would be amazing.
- Horwitz RI, Hayes-Conroy A, Caricchio R, Singer BH. From evidence based medicine to medicine based evidence. Am J Med. 2017;130:1246–1250. doi:. doi:10.1016/j.amjmed.2017.06.012 [CrossRef]
- Horwitz RI, Charlson ME, Singer BH. Medicine based evidence and personalized care of patients. Eur J Clin Invest. 2018;48:e12945. doi:. doi:10.1111/eci.12945 [CrossRef]
- Horwitz RI. The dark side of evidence-based medicine. Cleve Clin J Med. 1996;63:320–323. doi:10.3949/ccjm.63.6.320 [CrossRef]
- Horwitz RI, Singer BH. Why evidence-based medicine failed in patient care and medicine-based evidence will succeed. J Clin Epidemiol. 2017;84:14–17. doi:. doi:10.1016/j.jclinepi.2017.02.003 [CrossRef]
- Tricoci P, Allen JM, Kramer JM, Califf RM, Smith SC Jr, . Scientific evidence underlying the ACC/AHA clinical practice guidelines. JAMA. 2009;301:831–841. doi:. doi:10.1001/jama.2009.205 [CrossRef]
- Feinstein AR, Rubinstein JF, Ramshaw WA. Estimating prognosis with the aid of a conversational-mode computer program. Ann Intern Med. 1972;76:911–921. doi:10.7326/0003-4819-76-6-911 [CrossRef]