NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Goodman SN, Gerson J. Mechanistic Evidence in Evidence-Based Medicine: A Conceptual Framework [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2013 Jun.

Cover of Mechanistic Evidence in Evidence-Based Medicine: A Conceptual Framework

Mechanistic Evidence in Evidence-Based Medicine: A Conceptual Framework [Internet].

Show details


We used multiple resources and sought different perspectives to develop a framework for the identification of research gaps. We carried out six steps. We first attempted to identify, enumerate and describe frameworks that have been used (Steps 1 to 3). We then developed, tested and refined a framework (Steps 4 to 6). The six steps are:

Step 1.

Focused literature review

Step 2.

Development of draft framework

Step 3.

Workshop with technical experts

Step 4.

Refinement of framework

Step 5.

Development of two case studies

Step 6.

Pilot test of framework on case studies

Step 1. Focused Literature Review

We conducted comprehensive literature reviews in two broad areas: evaluation of surrogate endpoints and the value and use of animal models in translational research. Both searches encompassed the publication dates between 2000 and 2009, with additional references found before and after those dates through reference and citation searches. Criteria for inclusion in this report was the degree to which the paper provided either a high-level perspective or a case example directly relevant to the development of a framework. These two searches were used to compile an annotated bibliography (Appendices A and B). From each included article we also mapped its focus into the draft framework we were using at that time.

Step 2. Development of Draft Framework

On the basis of preliminary reading in each of the domains described previously, the following draft framework was proposed. This was both discussed at the invited workshop, and used for the mapping exercise in the annotated bibliography.

Step 3. Workshop With Technical Experts

We identified technical experts representing a variety of disciplines relevant to the development of this framework. These included translational medicine, biomarker development, philosophy, evidence-based medicine, toxicology and animal research. The proceedings of the workshop are presented in Appendix C.

The discussion in the workshop covered a very broad territory, reflecting the range of expertise among the participants. Each participant presented an example from their own domain of work of the use of mechanistic information in the development, evaluation, or prediction of the efficacy of a therapy. These perspectives then informed conversation focused specifically on the components of the framework, as well as providing potential examples for its application. The discussions in this workshop were quite rich, going far beyond the issue of the framework itself. However, both the examples and ensuing discussion of the framework highlighted the following potential weaknesses in the draft framework as proposed:

  1. Virtually all pathways underlying biologic mechanisms are incompletely known.
  2. Most therapies are developed based on partial knowledge of mechanisms, and rarely affect more than one step in a pathway.
  3. The quality and relevance of experiments in animals to human disease and therapeutics are widely recognized as deeply problematic.
  4. Experimentation on animals serves many purposes other than documentation of whole organism responses. The use of transgenic organisms, and other genetically manipulated animals makes them occasionally excellent experimental models for specific therapeutic effects.

As a result of these observations plus other discussion in the workshop along with further literature searching the originally proposed framework was modified as described in the next section.

Step 4. Refinement of the Framework

To be maximally useful, an evidential framework for mechanistic knowledge must be applicable to all forms of interventions in humans to prevent or treat disease. That requires a high degree of generality for the overall structure, with elements that are customizable for particularly contexts, e.g., drugs, devices or behavioral interventions. Second, it to should have the irreducible minimum of elements, capturing only those that are absolutely essential for the task. Finally, the potential application and context in which this framework is expected to be used must be clear. The main domains of application are in those settings where empirical information directly linking intervention to human outcome is absent or weak, such as typically occurs in early phase clinical testing or technological development, but also arises in many other contexts, listed in Table 1.

Table 1. Contexts and mechanisms for Step 4.

Table 1

Contexts and mechanisms for Step 4.

These considerations lead to the development of an alternative framework which included many of the ideas embedded in the draft framework, but organize them in a way that reflected the developmental processes of therapeutics as well as the minimally sufficient set of conditions and categories for evidential measurement.

Step 5. Pilot Test of the Framework on Two Case Studies

As part of a companion project, two in depth case studies were developed to see how the conceptual framework being developed would apply to actual examples. The two case studies were of Gleevec for the treatment of chronic myelogenous leukemia and estrogen use in menopausal women for the prevention of heart disease. These case studies were summarized and mapped into the conceptual framework.


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...