The CEA will require the effectiveness to be measured separately for each NMEP activity. Analysis of process outcomes (e.g., www.Medicare.gov Web site hits) is straightforward. Data are gathered and collected in a simple database where they may be analyzed for a cost-consequence study.
Analysis of cost data depends on data availability. Budgeting or accounting records may be organized in a simple database. For other forms of costing, one could use a “bottom-up” method, in which individual pieces of an activity are separately valued and summed to estimate total costs. A model of all major components of an activity must be developed and costs attached to each component. The first challenge is to identify a full list of components that will generate an accurate depiction of costs. For example, suppose that total costs of one NMEP activity equal the sum of labor costs, facilities and capital costs, and materials costs. Within these are multiple classes of workers (hourly, salary, and temporary), a single source of capital, several buildings, and three different types of materials used in the activity. These major headings are further broken down to the extent that components can be identified and separately costed. If worker compensation is not available for a finer level of detail (than hourly, salary, and temporary) or if there is little variation within subclasses, then no further gradations are needed. At each level, then, costs are gathered as availableeither directly from the source (e.g., CMS wage and salary records)or from some comparable external source (e.g., Bureau of Labor Statistics average wage data by occupation and industry).
Once effectiveness (outcome) and cost data are available, the cost-effectiveness calculations are quite simple. For a specific NMEP activity, the cost-effectiveness is the costs of that activity divided by the “effect” of that policy. For example, a 10-unit increase in the beneficiary knowledge index may be obtained for an investment of $1 million in a hypothetical activity. To compare and rank activities, estimates are sometimes transformed into single unit changes; in this case, a one-unit increase for a cost of $100,000. It is important to note, however, that transforming the estimates into such changes is not always meaningful or correct. Not all policies can be reduced in size, nor does the effectiveness necessarily change equally. To address these concerns, the incremental ICER is preferred. Moving from policy A to B, the ICER is defined as the change in the costs [(costs of activity B minus costs of activity A)/(effects of activity B minus effects of activity A)].