DDBA 8560 Week 4 Decision Criteria Matrix Example

Five Criteria and No Options: A Criteria Matrix for Brookfield's MRI Capacity Decision, Separating Clinical Screens From Tradeable Criteria, Attributing Each Criterion to a Source and Defending Swing Weights Against the Analytic Hierarchy Process

Student Name

Walden University

DDBA 8560: Seminar in Healthcare Managerial Decision Making

Instructor Name

Month Day, Year

Five Criteria and No Options: A Criteria Matrix for Brookfield's MRI Capacity Decision, Separating Clinical Screens From Tradeable Criteria, Attributing Each Criterion to a Source and Defending Swing Weights Against the Analytic Hierarchy Process

The Decision

The decision, framed in Week 2, is which action open to the manager of Brookfield Outpatient Imaging would most shorten the 23-day wait for routine MRI. This paper sets out how candidate actions will be judged, before any action is named, so that the criteria cannot be shaped to favor one.

Screens and Criteria

Two kinds of requirement must be kept apart. Screens are conditions every option must meet; an option that fails one is excluded however well it does elsewhere. Criteria are dimensions on which options trade against each other, where more of one can compensate for less of another. Keeney and Raiffa (1976) built their approach to multiple objectives on the idea that criteria can be traded at stated rates; a requirement that should never be traded does not belong among them.

Four screens apply. Protocol durations set by the radiology group must not change. Technologist credentialing standards must be met on every shift. MRI safety screening must be staffed by trained personnel whenever a scanner operates. Radiologist reading coverage must stay within the radiology group's contract. Treating any of these as a weighted criterion would imply that enough wait reduction could justify weakening safety screening, which no one at Brookfield would accept.

The Criteria

[after]

Each criterion comes from a source outside the writer. The wait criterion comes from the hospital's published access target. Cost comes from the budget holder's planning rules, which require incremental operating cost for any service change. Staff burden comes from the collective agreement, which limits assignments outside the standard shift. Access comes from the patient survey, in which 31 percent of patients who missed appointments cited work hours. Time to start comes from the capital planning calendar.

Table 1

Criteria for the MRI Capacity Decision

How the Weights Were Produced

Weights were set by swing weighting, following Keeney and Raiffa (1976). For each criterion, the decision group, the manager, the imaging director, a lead technologist and a scheduling supervisor, considered the swing from the worst plausible level to the best: for wait, from 31 days, where the wait would reach in a year if nothing changed, to 10 days. The group ranked the swings by how much each mattered, gave the most important swing 100 points and scored the others relative to it. Wait received 100, cost 45, staff burden and access 33 each and time to start 11. Normalizing gives the weights in the table.

Swing weighting ties each weight to the range of the criterion. That matters: if the plausible range of cost were small, cost would deserve little weight even if cost matters in general, because no option would differ much on it.

Why Not the Analytic Hierarchy Process

Saaty (1980) developed the analytic hierarchy process, in which decision makers compare criteria in pairs on a verbal scale and weights are derived mathematically from the comparisons. The method is widely used in health care and is easy to explain. But its weights are usually elicited without reference to the ranges of the criteria, asking whether wait is more important than cost in general rather than whether this swing in wait matters more than that swing in cost. Belton and Gear (1983) showed that adding an alternative to an analytic hierarchy model can reverse the ranking of existing alternatives, even when the new alternative is a copy of one already present, a result that follows from how the method normalizes scores. For a decision where options will be added and dropped as the search continues, that property is a serious weakness. Swing weighting does not produce rank reversal when options are added, because criteria scores are anchored to fixed ranges.

What the Matrix Does Not Do

The matrix does not yet contain options, and it does not settle the weights for good. If the radiology group later offered to review protocols, the screens would change, and the matrix would need revision.

Limits

Swing weights reflect the judgment of four people. A broader group, including referring physicians, might weigh access differently.

References

Belton, V., & Gear, T. (1983). On a short-coming of Saaty's method of analytic hierarchies. Omega, 11(3), 228-230. https://doi.org/10.1016/0305-0483(83)90047-6

Keeney, R. L., & Raiffa, H. (1976). Decisions with multiple objectives: Preferences and value tradeoffs. Wiley.

Saaty, T. L. (1980). The analytic hierarchy process. McGraw-Hill.