Chapter 4 Research Methodology & Design
4.2 Consensus Development Process – Expert Measures of Integration
As noted in Section 2.2.3 there are few validated measures of health system integration. The use of
expert opinion is an established methodology in health research (Kuzel, 2010) , particularly when
developing indicators where there is no data or consensus available, or where there is limited
evidence (Campbell & Cantrill, 2001). Consensus development processes such as the nominal group
technique have a long-standing history of use in the healthcare setting, however the methodology is
dependent upon the credibility of the participant experts and the methodological rigour of the process
(Campbell & Cantrill, 2001). For these reasons a methodology that engaged a broad selection of
experts in consensus development around appropriate measures of system integration was used.
4.2.1 Purpose
A modified, iterative consensus development process, called a nominal group technique, was used to
gather an expert advisory panel’s input into how system integration might be measured through
electronic information exchange. The nominal group technique or process (NGT) is preferred over
other consensus techniques as it allows for face-to-face discussion, and facilitates more equitable
contributions from groups of participants due to its structured format and anonymous ranking process
(Campbell & Cantrill, 2001). The process has been widely used in exploratory health service research
and for preliminary investigations prior to more formal research methods such as surveys. In one
example, valid, important and useful financial indicators in the health care sector led Pink and
colleagues (2007) to use an expert panel, literature review, and survey methodology to select
indicators and methods of calculation. This combined methodology produced indicators which were
4.2.2 Procedures
Experts were identified by “social acclamation”, a reasonable strategy according to Shanteau and
colleagures (2002). Their experience and training gave them the ability to understand complex
systems due to their relational and causal knowledge (Abernethy, Horne, Lillis, Malina, & Selto,
2005); it is this comprehensive understanding of systems which was required in this conceptually
challenging performance measurement task. The participants were purposively sampled to include
recognized experts from a variety of sectors and disciplines, such as primary care and public health,
with a sound knowledge of the regional health system, and an understanding of electronic information
systems (Van de Ven & Delbecq, 1972). Participation was also determined by availability and
willingness to participate. Recommended group size for nominal group process is no larger than eight
to twelve individuals (Fink, Kosecoff, Chassin, & Brook, 1984; Van de Ven & Delbecq, 1972).
Figure 8 presents the four phases of the nominal group technique through which the panel of
Figure 8. Nominal Group Technique Phases for Expert Panel
Ideally, consensus processes should be informed by a concise summary of existing empirically-
derived data (Fink et al., 1984). Introductory materials identifying indicator selection criteria, as well
as interoperability frameworks, were distributed ahead of the meeting and introduced participants to
both the domain of interest through a literature review, and the procedures for the nominal group
technique (see Appendix A). Participants were asked to read the introductory material ahead of time
and consider the question: “What metrics satisfy the indicator selection criteria (scientifically sound, Phase 1 - Introduction to
Literature & Nominal Group Technique - Literature review &
frameworks sent in advance of the meeting
- Participants instructed to note their views on prompt question & come prepared to discuss
Phase 2 - Brainstorming & Group Discourse
- Discussion of both the NGP and literature
- Individual reflections - Round Robin of metric suggestions until exhustion - Group discussion of each metric for clarity
Phase 3 - Ranking
- Experts anonymously rank all
metrics on a 9-point Likert scale using selection criteria - Results collated in real time and fed back to panel for brief discussion
Phase 4 - Refining - Summary report of
proceedings & rankings emailed to all participants
Conference call to further define the indicators, their relevance and feasibility Refinement of indicators based on feasbility & availability of data
relevant, feasible and communicable), and might contribute to our goal of measuring system
integration using between-provider electronic health information exchange?”
While there are no accepted benchmarks to identify when consensus is reached, consensus
must be defined in advance and the more demanding the criteria the better (Fink, Kosecoff, et al.,
1984). Consensus was achieved in this process, when the mean score of all participants’ rankings for
an indicator was greater than or equal to 7, and 70 per cent or more of the participants ranks the
indicator at 7 or higher. Despite its acceptance as an established practice in performance
measurement, the risk remains that expert panels in a nominal group process may reach consensus but
agree on indicators that do not satisfy the measurement objective. In order to mitigate this risk the
literature review provided an empirical foundation for methodological decisions.
However, we know that healthcare system measurement constructs must be clearly defined
and consistently understood in order to populate performance measures with reliable data (The
National Quality Forum, 2008). This is particularly important in nascent domains such as that
addressed by this study, where it is necessary to decompose concepts into smaller foundational
elements. In order to share this knowledge with a wider community for future validation and use, an
explicit conceptualization of regional electronic health information exchange and system integration
was proposed through the development of an ontology to formalize the Conceptual Model and
Measurement Framework (see Figure 5). Gruber (as cited inUschold & Gruninger, 2004) defines an
ontology as “a formal, explicit specification of a shared conceptualization”. Gandon's (2010)
definition is a little more accessible; it is as a “hierarchical organization of the relevant concepts and
relevant relationships between these concepts, as well as rules and axioms that constrain these