Appendix XIV. Decision Framework to Assess and Guide the Need for Searches of Existing Decision Models
Comprehensively identifying and evaluating models can be a time consuming process, particularly if it is not carefully structured in terms of process and timing. At the point that the USPSTF topic leads have determined a model is likely to be needed, a well-structured, noncomprehensive scanning process to identify existing models may prove effective as well as efficient.
Define a topic-specific scanning process to identify readily available, published decision modeling studies and to determine their apparent relevance to the USPSTF recommendation statement being developed.
Outline of Process
- The USPSTF topic leads, AHRQ Medical Officer, and EPC team (together comprising the topic workgroup) work through questions about the rationale for incorporating decision modeling, considering the extent of the prior USPSTF recommendation, public response to the prior USPSTF recommendation, prior use of modeling, current issues for clinical practice, and current state of the science.
- The USPSTF topic leads scope the decision problem to be addressed by modeling, briefly articulating the modeling objective (for the purposes of the recommendation statement), the scenarios to be modeled, required outcome measures, and modeling time horizon needed to guide the search for existing models. The main constructs of relevance should be defined, including a time frame to define how recent models should be, and any setting considerations, including policy and practice context. If there are acceptable constraints on modeling approaches (i.e., perspectives of the analysis, type of model, or other), these should be specified. The optimal timing for this activity may vary by topic, as these components of the decision problem become clear.
- A draft template to guide the articulation of the key questions, model objectives, and model scope is attached. This template can also serve to articulate the approach when commissioning a new or adapted model.
- An appropriate party (to be determined*) conducts a scanning search in MEDLINE to identify readily available existing recent models. Assessing the quality and completeness of existing models is beyond this scanning exercise.
- Based on critical PICOTS factors, are there existing relevant models?
- Do these models address the key questions/decision needs for this recommendation statement?
- Is there more than one modeling group represented by the existing relevant models?
- A summary document is prepared and distributed to the topic leads and resource allocation decisionmakers to further inform the availability of relevant models and a potential approach (i.e., use existing model or models as they are, commission USPSTF-specific modeling from existing groups, or commission de novo modeling), if modeling is determined to be a priority in addition to the systematic review.
*The appropriate party needs to be determined based on timing and extent of work required for the scanning activity. This may be the topic-specific EPC team, a separate decision modeling support team, or a commissioned modeling group. If decision modeling is already commissioned, International Society for Pharmacoeconomics and Outcomes Research best practices suggest that the modelers search for previously published modeling analyses of the same or similar problems, in order to discuss their model with respect to others. Having commissioned modelers take on this scanning activity would also support this best practice.
Internet Citation: Appendix XIV. Decision Framework to Assess and Guide the Need for Searches of Existing Decision Models. U.S. Preventive Services Task Force. July 2017.