Appendix D: Benchmarks from the NHQR
Diabetes Care Quality Improvement: A Resource Guide for State Action
More Details on Benchmarks
The NHQR provides a national set of estimates and often State estimates that can be used as benchmarks for quality improvement. A benchmark can be a baseline or point from which you start, not necessarily representing a goal or target. Or it can be the best current rate, something achievable, or a consensus of what should be achieved. It is a basis for making comparisons. Several types of benchmarks can be derived from the NHQR:
Benchmarks, Key Messages:
Several types of benchmarks can be derived from the NHQR:
- Theoretic limit benchmark: The theoretic limit refers to the maximum or minimum level that a measure can take on; for example, 100 percent for positive outcomes or 0 percent for negative, avoidable events. In an ideal world, these would be achievable, but in a world where so many factors are involved in achieving a maximum result, those benchmarks may be unrealistic. Also, some concepts might feasibly come closer to the theoretic limit than others.
- Best-in-class benchmark: The rate for the top State or top tier of States can be used for what manufacturers call a "best in class" benchmark. (The top tier can be defined as the top 5 or 10 percent of States averaged together.) Using influenza vaccination as an example, the highest rate of flu vaccination for people with diabetes across the States (64 percent) may be assumed to be a feasible goal for States to achieve. However, some may view the top State rate as an impractical target given their population and circumstances. Others may view that goal as inadequate, depending on the value of the rate and the state of medical knowledge and practice, and they may view the 100-percent goal as their target. These judgments will vary across States because States face different circumstances and environments. This Resource Guide uses the top 10 percent of States, combined in a simple average, to derive the best-in-class estimate. A simple average, rather than weighted average, was used because the denominators from the BRFSS estimates were not available in the NHQR.
Healthy People 2010 Diabetes Care Topics
- A national consensus-based goal: Some organizations propose targets that should be achieved to improve the health status of the overall population and vulnerable subgroups. For example, two decades ago, the National Center for Health Statistics of the Centers for Disease Control and Prevention developed diabetes-related goals for a healthier U.S. population. Each decade those goals are reviewed and reestablished. The current goals (go to inset of diabetes-related topics for Healthy People goals), now called Healthy People 2010 (HP2010, U.S. Department of Health and Human Services, 2000), also are included in the NHQR when relevant.
- The national average: The overall average indicates where the average member of a group stands. For example, the average of influenza vaccination rates for people with diabetes in States (37 percent according to the BRFSS data source) is the "norm" for States or is the rate for the "average" State. States with rates below the average would prefer to be at or above the average. But the average may not be an indicator of quality health care.
- The regional norm: States may prefer a regional estimate for comparison because they want to see how they perform compared to medical practice within the region. Given the wide regional variation in U.S. medical practice (Wennberg and Cooper, 1999), regional estimates may be weak goals for regions where practice should change to enhance the health care quality for people with diabetes. For this Resource Guide, the regional averages are calculated for the four Census regions: Northeast, Midwest, South, and West. The averages are simple averages because the denominators for BRFSS estimates were not available from the NHQR.
- The State rate: As noted in the Module 3: Information, the State's own rate may serve as a benchmark for various purposes-tracking changes over time, evaluating the effect of a statewide intervention to improve quality, or reporting the norm for local communities and providers to compare to their own performance. Concerns noted above about using national or regional averages as goals also apply to State rates. For provider-level estimates, the best-in-class providers may be a better indication of what is achievable and should be used as a goal rather than the State average rate. Severity adjustments are an important issue at the provider level, where populations of patients with varying severity and comorbidity levels are unlikely to be distributed evenly across providers.
The Best Benchmarks
Best-in-class estimates are the best way to view the opportunities for a State to improve. Basing a best-in-class measure on a group of best States (rather than the single top State) mitigates the effects of an extreme that other States might find unreasonable to emulate.
Table D.1 shows values for the best-in-class benchmarks (as simple averages of the top 10 percent of States) and for other benchmarks. The other benchmarks include Healthy People 2010 goals (when available for a measure), the national norm, regional average benchmarks, and State rates for the four example States used in this Resource Guide. These benchmarks are provided for all of the diabetes-related measures in the NHQR. Four of the measures-HbA1c test, eye exam, foot exam, and flu vaccinations in the past year (in addition to HbA1c test two times in the past year)—are displayed graphically in the Module 3: Information.
Benchmarks related to diabetes care for different socioeconomic groups are available from the NHDR. Those benchmarks are national averages and are not available by State. However, individual States may have data that can be analyzed by socioeconomic group (e.g., avoidable hospitalizations by racial, educational, or income group). Table D.2 shows values for the national averages for diabetes process and outcome measures by socioeconomic characteristics of the national population.
Methods of measurement and data quality can have a large impact on the value of a benchmark. For this reason, it is crucial that the methods and data used to derive various benchmarks are similar. For example, when comparing the State to the Nation, the same methods and data sources should be used to calculate the estimates. That is why this Resource Guide presents only the BRFSS estimates to compare States and the Nation. Other sources (for example MEPS) were used for national estimates of the same measures in the NHQR. However, MEPS and BRFSS use different survey methods and present different measures; the impact of the former is apparent in the HbA1c rates—90 percent (MEPS) versus 79 percent (BRFSS)—and the impact of the latter is seen in the influenza vaccination rates for people with diabetes-55 percent for those age 18 and over (MEPS) versus 37 percent for those age 18 to 64 (BRFSS).