An initial goal of this project was to obtain a set of comparable HEDIS and CAHPS data for all study states, to track the change in measures over time, and to compare measures across states. The data were to be aggregated to the state level for as many years as possible.
We obtained data for all study states, but the variability in definitions of the measures has severely limited the number of measures that can be compared across time and across states. The effort to identify comparable measures resulted in six comparable HEDIS measures and eight comparable CAHPS measures. They are:
- Percent of pregnant women with timely prenatal care (a visit in the first trimester)
- Percent of children age 3–6 with a well-child visit in the past year
- Percent of children age 2 with up-to-date immunizations (Combination 214)
- Percent of adult women with a breast cancer screen in the past two years (ages vary by state)
- Percent of women age 21–64 with a cervical cancer screen in past three years
- Percent of adult diabetics with an HbA1c test in the past year
- Overall satisfaction with health plan (adults)
- Overall satisfaction with health care (adults)
- Overall satisfaction with personal doctor (adults)
- Overall satisfaction with specialist (adults)
- Overall satisfaction with health plan (children)
- Overall satisfaction with health care (children)
- Overall satisfaction with personal doctor (children)
- Overall satisfaction with specialist (children)
Notably, all of the common HEDIS measures are measures of preventive care services. There are no commonly reported measures of health outcomes (for example, the outcomes from screening tests) across the time period. In addition, there are not enough study states reporting CAHPS measures in the first half of the decade for a cross-state analysis.
Even for these commonly reported measures, not all 20 states collected data in a common way across the time period. Appendix C contains a set of figures showing state-by-state data for all of these measures for the earliest year in the study period for which the state reported data and for the most recent year.
All states reported some HEDIS data for the six common measures, although no study states reported on all of them. Between 17 to 19 states reported on each of the six HEDIS measures for Medicaid (depending on the measure). The most commonly reported measure was well-child visits (19 states for Medicaid). Separate reporting for CHIP was less common; only six states reported separately on well-child visits, and only three reported separately on up-to-date childhood immunizations (Combination 2).
Eighteen states reported any CAHPS data for 2006–2010. In addition, the way that CAHPS data are reported varies by state. For example, on a satisfaction scale of 1 to 10, some report the percent scoring 7–10, some report the percent scoring 8–10, and some report the percent scoring 9–10, all very different measures. So even in the 20 states with the longest-established and largest risk-based managed programs, a very limited set of quality measures are available with common definitions for cross-state analysis. In addition to the limited number of quality and satisfaction measures for comparison, the different measurement approaches mentioned above and variations in definitions over time limit the comparability of data.15
With these limitations in mind, we created box plots to illustrate the level and range of quality/satisfaction measures for the most commonly reported HEDIS and CAHPS measures as shown in Figures 9–12. These figures demonstrate considerable variability in both HEDIS and CAHPS quality/satisfaction measures. The first box plot (Figure 9) shows variation in three pregnancy/pediatric HEDIS measures: timeliness of prenatal care, well-child visits, and immunizations (Combination 2). The mid-point of each box plot (represented by a triangle) is the median across all the states with data. The box plots also indicate the extreme values and the 25th (Q1) and 75th (Q3) percentiles. For well-child visits and immunizations, HEDIS Medicaid and CHIP measures are both included, since there are not enough CHIP measures to analyze separately. For all three measures, we present data for the first half of the study period (2001–2005) and the second half of the study period (2006–2010).
The range of values for these quality measures varies considerably across states and over time. For example, the percent of pregnant women with timely prenatal care varies from 56 to 82 percent during the first half of the study period (median = 74 percent), rising to from 68 to 91 percent (median = 84 percent). The improvement in this measure is impressive, and the median for the second time period is very similar to the rate for all women in the United States in 2007 at 82 percent (National Center for Health Statistics [NCHS], 2011), and higher than for African American women in that year (75 percent). In addition, the median rate is slightly higher than the Medicaid average for plans that are either NCQA accredited or that voluntarily report data, at 81.5 percent in 2007. However, it is substantially below the rate for commercial enrollees (91.5 percent) as reported by NCQA in the same year (NCQA, 2011).
The figure shows that there has been substantial improvement in both of the child HEDIS measures (well-child visits and immunizations) among the states reporting over the decade. In addition, the study states reporting data compare favorably to both Medicaid and commercial enrollees reporting data to NCQA. For example, the median percent with a well-child visit for study states was 73 percent, and 65 percent (Medicaid) and 68 percent (commercial) for plans reporting to NCQA for 2007. The NCQA rates both rose to about 70 percent by 2010. All the caveats suggested earlier in this report should be kept in mind in making these comparisons, including the variations in how quality measures are developed and reported, and how intensively the quality of data is reviewed (for example, by EQROs).
Turning to adult HEDIS measures, Figure 10 shows the range and trends in rates of screening for diabetes among diabetics, and cervical and breast cancer among adult women. The adult rates of use of these preventive care services are slightly lower than for the obstetric/pediatric preventive care services, and the range is generally greater across states. However, as with children, there is an improved trend over time for diabetic screening and cervical cancer screening. The NCQA (2011) found rates of 65 percent (Medicaid) and 82 percent (commercial) for cervical cancer screening nationally in 2007, which compares to a median of 67 percent among study states. Breast cancer screening rates are lower for study states and for NCQA, and have not shown substantial improvement over the decade in either case. For example, for NCQA reporting plans the rate was only 50 percent for Medicaid women in 2007 (69 percent for commercial women). This compares to a median of 54 percent among the 18 study states reporting on all their plans, essentially the same throughout the study period.
Figures 11 and 12 show satisfaction with various aspects of care for children and adults, as reported in CAHPS surveys. The range across the limited number of states that report CAHPS data is much less than for the HEDIS measures. For example, for the child CAHPS, the percent of parents who report they are highly satisfied with their child’s health plan varies from 77 percent to 90 percent with a median of 82 percent. The median values for the other child CAHPS measures are very similar (83 percent for health care, 86 percent for personal doctor, and 82 percent for specialist satisfaction). The ranges in these values across states are no more than 7 percentage points. Unfortunately there were not enough study states reporting CAHPS data in the first half of the time period (only three states reported), so it is not possible to discern a trend in satisfaction over time.
The levels of satisfaction for adults are consistently lower for adults than for children. While this pattern could be due to some measurement differences of various types (such as which states are included in the figure), it appears from this limited information that adults are less satisfied with their own care under risk-based Medicaid managed care than parents are with their children’s care. This could be due to many factors, including differences in the types of providers used and the fact that adults are in worse health, on average, than children.
The median percentage of adults who are satisfied with their health plan (reporting a satisfaction level of 8–10 on a scale of 0–10), across the nine reporting states, is similar at 75 percent to the percent for Medicaid adults reporting to NCQA at 72 percent. These rates of satisfaction for Medicaid enrollees are higher than satisfaction of commercial enrollees reporting to NCQA at 64 percent (NCQA, 2011).