What is the best evidence of evidence-based medicine?

Evidence-based medicine is clinical medicine that follows scientific evidence. It advocates combining the personal clinical practice and experience of clinicians with objective scientific evidence to provide the most correct diagnosis, the safest and most effective treatment and the most accurate prognosis estimation for each specific patient.

Evidence-based medicine is different from traditional medicine. Traditional medicine is based on empirical medicine, that is, treating patients according to non-experimental clinical experience, clinical data and basic knowledge of diseases. Evidence-based medicine does not want to replace clinical skills, clinical experience, clinical data and medical expertise, but only emphasizes that any medical decision should be based on the best scientific evidence.

Evidence quality classification of evidence-based medicine has the following classification methods:

1. The classification method of American Preventive Services Working Group can be used to evaluate the quality of treatment or screening evidence:

* Level I evidence: evidence obtained from at least one well-designed randomized controlled clinical trial;

Grade II-1 evidence: evidence obtained from well-designed non-randomized controlled trials;

* Level II-2 evidence: evidence from well-designed cohort studies or case-control studies (preferably multicenter studies);

* Level II-3 evidence: evidence from multiple time series studies with or without intervention. The results of non-controlled experiments with extremely obvious differences can sometimes be used as evidence of this level;

* Third-level evidence: authoritative opinions from clinical experience, descriptive research or expert committee reports.

The NHS uses another letter-based evidence grading system. The above American grading system is only applicable to interventional therapy. Many studies are needed to provide evidence for evaluating the accuracy of diagnosis, the natural history of disease and prognosis. Therefore, the Oxford Center for Evidence-based Medicine has put forward another set of evidence evaluation system, which can be used for research evaluation in the fields of prevention, diagnosis, prognosis, treatment and harm research:

* Grade A evidence: consistent randomized controlled clinical studies, cohort studies, all or inconclusive studies and clinical decision-making rules that have been verified in different groups;

* Grade B evidence: consistent retrospective cohort study, prospective cohort study, ecological study, outcome study, case-control study, or conclusions extrapolated from Grade A evidence;

* Class C evidence: the conclusion drawn from the case sequence research or inference of Class B evidence;

* D-level evidence: expert opinions without critical evaluation, or evidence based on basic medical research.

Generally speaking, the quality of evidence guiding clinical decision-making is determined by the quality of clinical data and the clinical "positioning" of these data. Although there are differences between the above-mentioned evidence grading systems, their purpose is the same: to let users of clinical research information know which studies are more likely to be the most effective.

In addition, in clinical guidelines and other writings, there is a set of recommendation and evaluation system to guide the communication between doctors and patients by measuring the risks and benefits of medical behavior and what level of evidence the operation is based on. The following are the evaluation criteria recommended by the American Preventive Services Working Group:

* A-level recommendation: Good scientific evidence shows that the benefits brought by this medical behavior actually outweigh its potential risks. Clinicians should discuss medical behavior with applicable patients;

* Class B recommendation: At least there is acceptable evidence that the benefits brought by this medical behavior exceed its potential risks. Clinicians should discuss medical behavior with applicable patients;

* Class C recommendation: There is at least fair scientific evidence that the medical behavior can provide benefits, but the benefits and risks are very close, so it is impossible to make general recommendation. Clinicians do not need to provide this kind of medical behavior unless there are some individual considerations;

* D-level recommendation: at least there is fair scientific evidence that the potential risks of this medical behavior exceed the potential benefits; Clinicians should not routinely perform this kind of medical behavior on asymptomatic patients;

* The first type of recommendations: medical behaviors lack scientific evidence, or the quality of evidence is low, or conflict with each other, for example, risks and benefits cannot be measured and evaluated. Clinicians should help patients understand the uncertainty of this medical behavior.