Leveraging implementation science to improve implementation outcomes in precision medicine

Abstract

Background and purpose: Introduction of omics technologies in clinical practice means increased use of validated biomarkers, through precision medicine (PM). Although implementation science (IS) affords an array of theoretical approaches that can potentially explain PM intervention uptake, their relevance and applicability in PM implementation has not been empirically tested. This article identifies and examines existing implementation frameworks for their applicability in PM, demonstrating how different IS theories can be used to generate testable implementation hypotheses in PM.

Methods: A three-step methodology was employed to search and select implementation models: a scoping search in Google Scholar produced 15 commonly used models in healthcare; a systematic search in PUBMED and Web of Science using the names of each model as keywords in search strings produced 290 publications for screening and abstraction; finally, a citation frequency search in the 3 databases produced most cited models that were included in the narrative synthesis.

Results: Main concepts and constructs associated with each of the 15 models were identified. Four most cited frameworks in healthcare were: REAIM, CFIR, PRISM and PARiHS. Corresponding constructs were mapped and examined for potential congruence to PM. A generalized PM implementation conceptual framework was developed showing how omics biomarker uptake relates to their evidence base, patient and provider engagement and Big data capabilities of involved organizations.

Conclusion: We demonstrated how implementation complexities in PM can be addressed by explicit use of implementation theories. The work here may provide a reference for further research of empirically testing and refining the identified implementation constructs.

Keywords: Implementation science; biomarkers; genomic medicine; omics technologies; precision medicine.

Figure 1

Developmental phases of an omics-based biomarker (OBM). Biomarker discovery pipeline is here juxtaposed along traditional pharmacological drug discovery process. The biomarker discovery process results in products that may be applicable at single or multiple drug development stages. The traditional drug development process divides translational studies into five phases, T0 through T4. OBM discovery and development mirrors this translational process. For instance, use of predictive pharmacogenetic OBMs can improve drug development by increasing the size of the treatment effect by stratifying patients based on disease type at the beginning of a clinical trial.

Figure 2

PRISMA Flow Diagram of literature search and selection process for the admitted papers.

Figure 3

A Pareto chart showing implementation models according to their citation frequencies. The bars, arranged in descending order to depict significance, show individual models and their citations; the line graph shows the cumulative total in percentage. (Calculations based on frequency of citations of publications citing each of the implementation models in 3 databases: PubMed, Web of Science and Google Scholar in the stated period). RE-AIM = Reach, Effectives, Adoption, Implementation and Maintenance; CFIR = Consolidated Framework for Implementation Research; PARIHS = Promoting Action in Research Implementation in Health Services; PRISM = Practical, Robust Implementation and Sustainability Model; CM-EBM = Conceptual Model of Evidence-based Practice Implementation in Public Service Sectors; KTE = Knowledge Translation and Exchange; NPT = Normalization Process Theory; ARC = Availability, Responsiveness & Continuity; AIFs = Availability Implementation Frameworks; SK = Sticky Knowledge; CM = Conceptual Model of Implementation Research.

Figure 4

A precision medicine implementation conceptual framework illustrating six identified precision medicine factors and their constituent constructs.

Figure 5

A typical bioinformatics workflow for the analysis of genomic data based on DNA/RNA sequencing from next-generation sequencing (NGS) platforms. It graphically describes the shifting complexity of sequencing from sample preparation, data processing, downstream analysis, data management and finally data dissemination. The arrow indicates decreasing complexity. Tools are essential to the understanding and interpretation of the multiple and complex data-sets generated and analytical outputs. The workflow provides the means of integrating diverse outputs to generate novel medically actionable insights.