The IHI Triple Aim team has put together a prudent set of suggested measures that also help operationally define the IHI Triple Aim.
The primary issue with measurement is how to use it at various levels and for various purposes. By now the IHI team has heard a set of frequently asked questions or concerns regarding measures. Below is a list of questions or concerns and proposed answers.
Frequently Asked Questions and Concerns
The measures you propose are too high level. They are not actionable.
High level measures, such as years of potential healthy life lost, are useful for operationally defining health and for establishing long range, 5-10 years, plans. Publically available data such as Hospital Compare or CAPHS from CMS or some the NCQA measures would be useful at a project level. Hospitals, doctors and other providers can and should develop the capability to use their existing data for improvement.
The data are not current and therefore not useful.
It is true that some publically available data are a year or more behind. This limits their use in providing real time or even quarterly feedback to guide improvement efforts. However, they can be useful in several ways. One is to compare trends in the data against other comparison groups such as national or state results. Without some aggressive initiatives to change the relative results high level measures will be stable with respect to these comparison groups. They then are useful for building will and setting priorities. For more timely feedback these publically available data can be supplemented locally by utilization data from health plans for the under 65 year old population. Quality Improvement Organizations contracted in each state by CMS have access to current Medicare data. The trends in some of these “drill down” data sources can be confirmed by the “official” data when they are available. For example, suppose a region provided equitable access to primary care and developed an integrated approach to chronic disease prevention and treatment. The region might expect to see an improvement in the self reported health status for people making less than 50,000 per year when the data becomes available.
How long will it take to move these high level measures?
Of course for some initiatives there will be a delay between cause and effect. However, important measures such as self reported health status, readmission rates, admissions for exacerbations of chronic illness, trends in per capita cost can all be impacted in a three year period. It is useful to develop goals for 1, 3 and 5 year periods. Perhaps the most important factor in determining the speed at which results are produced is the capability and capacity for improvement in the region.
The data are not aggregated at a useful level.
This is often the case. Perhaps county data is needed but the data are aggregated only at the state level. Sometimes county data are available but target populations are mostly resident in a few zip codes. Often the reason the data are not available is that the researchers believe that the sample sizes are too small at these small geographic regions. This is probably true for one time snapshots. But a region with a sustained focus on the Triple Aim will accumulate years of data in these focused areas that can be plotted over time and analyzed using control chart methods. The Commonwealth Fund Team (now residing at IHI) that produces the state scorecards is completing a project to disaggregate the data into smaller geographic groupings.
A general principle to be used when establishing a regional measurement strategy is that the publically available data will be somewhat useful but will not be exactly on target. Ingenuity will be needed to define surrogate measures that are more targeted but less rigorous. These measures can provide useful feedback for learning. For example, it may be very difficult to get overall health spending in a region for the commercially insured population. However, the hospitals and some of the businesses could contribute their figures on yearly health care premium increases as an indicator of per capita cost trend. Emergency departments must report data on volume and condition to local health departments forsurveillance purposes. These data could be used a source of data on trends in population health or health care experience.
