How do Medicaid, Medicare, and Commercial Insurance Vary in Community-Level Performance? Using Claims Data from the Wisconsin Health Information Organization (WHIO) to Assess Variation in Population Health Processes Donna Friedsam, Daphne Kuo, and Kristen Voskuil UW Population Health Institute January 13, 2012
Background WHIO received Medicare data from the Dartmouth Atlas/Brookings Institute collaboration, with support from the Markel Foundation, to support its understanding of variation among various markets in payer performance. UW Population Health Institute conducted exploratory analyses to determine the utility of the Medicare data provided in aggregate at the county level. Dartmouth (Medicare) Ingenix (Commercial and Medicaid)
Analyses Conducted 1. By County, along process measures for diabetes quality: HbA1c, LDL annual, annual retinopathy exam. 2. For three service areas participating in pilot projects for the Partnership for Health Care Payment Reform (PHPR): Three process measures and the composite measure, by payer and for all payers. 3. Factors Associated with county variation in diabetic care measures: health care, socioeconomic status, and social integration
1. Diabetes Process Measures by County Percent of persons, by county, in each payer group receiving the recommended tests HbA1c and LDL tests and the difference in rates between the payers Commercial, Medicaid, and Medicare. Analysis of Variation from County Mean, by Payer Payer (COM, MCR, MCD) variation, by county, from the county mean for HbA1c and LdL testing Counties performing outside the bounds of one standard deviation above (better) or below (worse) than the overall mean of all counties. Beyond the zone of standard deviation indicate performance either significantly above or significantly below the mean for all counties on that performance measure.
Summary of Findings For nearly all counties, Medicaid shows significantly lower performance than Commercial and Medicare in these counties. Medicaid statewide performs approximately 35% lower than Commercial and Medicare on LDL testing. Medicaid statewide performs approximately 30% lower than Commercial and Medicare on HbA1c testing. The performance of Commercial and Medicare are statistically similar in most counties, except In Jefferson, Juneau, Kenosha, Racine and Sauk Counties, Medicare performance for LDL testing significantly outpaces Commercial payer performance. In Kenosha, Racine, and Sauk Counties, Medicare performance of HbA1c testing significantly outpaces Commercial payer performance.
County Performance Relative to Overall Mean for All Counties, Medicaid Counties Performing Above Overall County Mean for Medicaid HbA1c (County Mean = 59%) LDL (County Mean = 47%) Door Door Langlade Langlade Milwaukee Richland Oneida Shawano Polk Portage Price Richland Waushara Counties Performing Below Overall County Mean for Medicaid HbA1c (County Mean = 59%) LDL (County Mean = 47%) Green Green Iowa Green Lake Kenosha Iowa LaCrosse Jackson Monroe LaCrosse Sauk Monroe Sheboygan Sauk Trempeleau Trempeleau Vernon
Composite, All-Payer Variation by County on Diabetes Testing Performance Diabetes tests HbA1c, LDL, and retinopathy All-payer basis (Commercial, Medicare, and Medicaid) Weighted to adjust for varying composition of payer groups in each county
2. Payment Reform Service Areas Three service areas participating in the pilot projects for the Partnership for Health Care Payment Reform (PHPR). Defined service areas in two levels: 1. Counties named by the providers participating in the PHPR, and 2. As derived from the zip codes associated with the Hospital Service Area within the Dartmouth Atlas.
Payer Performance in Each Service Area Relative to Statewide Rates Commercial: Both NEWHVN and Monroe exceed the all payer individual test rates and composite rate. Milwaukee trails the statewide rate for each individual test rate and for the composite rate. Medicaid: Monroe substantial trails the statewide rate for HbA1c and LDL testing and for the composite rate All Payer: Only minor variations appear in the comparison of service area all-payer rates to the statewide rates, particularly once the service area rates are weighted to correct for differing payer composition among the population.
Summary of Quality Measure Compliance for Selected Diabetes Measures by County and Product Category, Year Ending 9/30/2009, Patients within last 12 months
Table 2c: Commercial HbA1c LDL Eye exam All Tests Composite PRODUCT COUNTY DEN (N) NUM Rate NUM Rate NUM Rate Rate COM Oconto 199 181 91% 168 84% 70 35% 70% COM Kewaunee 123 119 97% 106 86% 46 37% 73% COM NEWHVN 1,454 1,322 91% 1,208 0.83 632 43% 72% COM Calumet 448 405 90% 379 85% 209 47% 74% COM Outagamie 823 753 91% 713 87% 387 47% 75% COM Shawano 217 201 93% 192 88% 88 41% 74% COM Waupaca 382 358 94% 351 92% 199 52% 79% COM Waushara 140 134 96% 118 84% 57 41% 74% COM Winnebago 638 593 93% 572 90% 293 46% 76% COM NEWHVN-out 2,648 2,444 92% 2,325 0.88 1,233 47% 76% COM NEWHVN-TOTAL 4,102 3,766 92% 3,533 0.86 1,865 45% 74% COM Milwaukee 6,801 5,851 86% 5,387 79% 2,532 37% 67% COM Waukesha 2,839 2,475 87% 2,287 81% 1,161 41% 70% COM IPN 9,640 8,326 86% 7,674 0.80 3,693 38% 68% COM Kenosha 866 643 74% 618 71% 236 27% 58% COM Ozaukee 719 622 87% 577 80% 280 39% 69% COM Racine 1,246 912 73% 860 69% 363 29% 57% COM IPN-out 2,831 2,177 77% 2,055 0.73 879 31% 60% COM IPN-TOTAL 12,471 10,503 84% 9,729 0.78 4,572 37% 66% COM Green/Monroe 298 275 92% 258 87% 148 50% 76% COM Statewide 37,401 32,663 87% 30,093 80% 14,701 39% 69%
Table 2d: Medicaid HbA1c LDL Eye exam All Tests Composite PRODUCT COUNTY DEN (N) NUM Rate NUM Rate NUM Rate Rate MCD Oconto 184 114 62% 97 53% 76 41% 52% MCD Kewaunee 80 45 56% 34 43% 33 41% 47% MCD NEWHVN 1,538 932 61% 797 52% 714 0.46 53% MCD Calumet 321 175 55% 156 49% 155 48% 50% MCD Outagamie 514 324 63% 272 53% 221 43% 53% MCD Shawano 215 142 66% 126 59% 95 44% 56% MCD Waupaca 270 147 54% 135 50% 132 49% 51% MCD Waushara 174 121 70% 91 52% 92 53% 58% MCD Winnebago 631 353 56% 301 48% 295 47% 50% MCD NEWHVN-out 2,125 1,262 59% 1,081 51% 990 0.47 52% MCD NEWHVN-TOTAL 3,663 2,194 60% 1,878 51% 1,704 0.47 53% MCD Milwaukee 10,891 7,332 67% 5,783 53% 4,364 40% 53% MCD Waukesha 756 423 56% 354 47% 338 45% 49% MCD IPN 11,647 7,755 67% 6,137 53% 4,702 40% 53% MCD Kenosha 884 421 48% 389 44% 345 39% 44% MCD Ozaukee 162 100 62% 80 49% 59 36% 49% MCD Racine 1,179 654 55% 554 47% 418 35% 46% MCD IPN-out 2,225 1,175 53% 1,023 46% 822 0.37 45% MCD IPN-TOTAL 13,872 8,930 64% 7,160 52% 5,524 0.40 52% MCD Green/Monroe 154 71 46% 61 40% 86 56% 47% MCD Statewide` 32,086 19,497 61% 15,731 49% 14,632 46% 52%
Table 2e: Medicare HbA1c LDL Eye exam All Tests Composite Product County DEN (N) NUM Rate NUM Rate NUM Rate Rate MCR Brown 5,410 4,783 88.41 4,388 81.11 4,031 74.52 81% MCR Oconto 1,350 1,144 84.76 1,051 77.82 915 67.78 77% MCR Kewaunee 605 515 85.16 501 82.79 374 61.76 77% MCR NEWHVN 7,365 6,442 87% 5,940 81% 5,320 72% 80% MCR Calumet 530 458 86.37 450 84.90 382 71.99 81% MCR Outagamie 3,530 3,209 90.92 3,076 87.14 2,691 76.22 85% MCR Shawano 1,120 990 88.37 942 84.12 806 71.93 81% MCR Waupaca 1,770 1,551 87.64 1,580 89.24 1,368 77.27 85% MCR Waushara 1,225 1,030 84.09 923 75.34 788 64.36 75% MCR Winnebago 3,270 2,867 87.68 2,725 83.32 2,261 69.13 80% MCR NEWHVN-out 11,445 10,105 88% 9,695 85% 8,294 72% 82% MCR NEWHVN-TOTAL 18,810 16,548 88% 15,635 83% 13,614 72% 81% MCR Milwaukee 26,910 23,064 85.71 21,273 79.05 17,755 65.98 77% MCR Waukesha 9,270 8,150 87.92 7,784 83.97 6,875 74.16 82% MCR IPN 36,180 31,214 86% 29,057 80% 24,630 68% 78% MCR Kenosha 4,660 4,077 87.50 3,962 85.03 2,918 62.61 78% MCR Ozaukee 2,185 1,910 87.42 1,781 81.49 1,572 71.92 80% MCR Racine 6,310 5,468 86.65 5,107 80.93 4,088 64.79 77% MCR IPN-out 13,155 11,455 87% 10,850 82% 8,577 65% 78% MCR IPN-TOTAL 49,335 42,669 86% 39,907 81% 33,207 67% 78% MCR Green 1,240 1,131 91.21 1,091 87.95 806 64.97 81% MCR Statewide 146,815 129,231 88% 120,159 82% 104,398 71% 80%
County Den (N) Num Rate Weighted NUM Rate Weighted NUM Rate Weighted Rate Weighted Table 2f: All Payers Composite HbA1c LdL Eye exam All Tests Composite Brown 7,816 6,578 84% 86% 5,988 77% 78% 5,152 66% 68% 76% 76% Oconto 1,733 1,439 83% 83% 1,316 76% 76% 1,061 61% 61% 73% 72% Kewaunee 808 679 84% 84% 641 79% 79% 453 56% 57% 73% 72% NEWHVN 10,357 8,696 84% 85% 7,945 77% 78% 6,666 64% 66% 75% 75% Calumet 1,299 1,038 80% 84% 985 76% 81% 746 57% 67% 71% 75% Outagamie 4,867 4,286 88% 88% 4,061 83% 83% 3,299 68% 69% 80% 78% Shawano 1,552 1,333 86% 87% 1,260 81% 82% 989 64% 65% 77% 76% Waupaca 2,422 2,056 85% 85% 2,066 85% 85% 1,699 70% 71% 80% 79% Waushara 1,539 1,285 84% 85% 1,132 74% 74% 937 61% 60% 73% 72% Winnebago 4,539 3,813 84% 85% 3,598 79% 80% 2,849 63% 64% 75% 75% NEWHVN-out 16,218 13,811 85% 86% 13,101 81% 81% 10,517 65% 67% 77% 76% NEWHVN-TOTAL 26,575 22,508 85% 86% 21,046 79% 80% 17,183 65% 67% 76% 76% Milwaukee 44,602 36,247 81% 84% 32,443 73% 76% 24,651 55% 60% 70% 72% Waukesha 12,865 11,048 86% 84% 10,425 81% 79% 8,374 65% 67% 77% 75% IPN 57,467 47,295 82% 84% 42,868 75% 77% 33,025 57% 62% 71% 73% Kenosha 6,410 5,141 80% 81% 4,969 78% 78% 3,499 55% 56% 71% 70% Ozaukee 3,066 2,632 86% 84% 2,438 80% 78% 1,911 62% 64% 76% 74% Racine 8,735 7,034 81% 81% 6,521 75% 75% 4,869 56% 58% 70% 69% IPN-out 18,211 14,807 81% 81% 13,928 76% 77% 10,278 56% 58% 71% 70% IPN-TOTAL 75,678 62,102 82% 84% 56,796 75% 77% 43,303 57% 61% 71% 72% WI, Green 1,692 1,477 87% 86% 1,410 83% 82% 1,040 61% 62% 77% 75% Statewide 216,302 181,391 84% 85% 165,983 77% 78% 133,731 62% 65% 74% 74%
3. Factors Associated with County Variation in Diabetic Care Measures Health Care, Socioeconomic Status, and Social Integration Daphne Kuo, PhD
Research Questions and Design Characteristics associated with compliance with diabetic care measures Economic resources Health care Healthy places Social integration Analytical issues: structural equation models All characteristics correlated with one another A global view of diabetic cares: all three elements together Four different health care payers (commercial plans, Medicaid, Medicare, or No Insurance)
Figure 1. County characteristics and variation in diabetes care measures % Black % Hispanic % Male % Age 45 & + % single female hh % public assistance % family < poverty % rural % high school grad % diabetics % obesity % smoking % heavy drinking % English % PCP Segregation Recreation % teen birth Healthy food % violent crime HbA1c Blood Lipid Eye- Exam Commercial rate Medicaid rate Medicare rate Commercial rate Medicaid rate Medicare rate Commercial rate Medicaid rate Medicare rate
Results % Black % Hispanic % Male % Age 45 & + % single female hh % public assistance % family income % rural % high school grad % diabetics % obesity % smoking % heavy drinking % English % PCP degree segregation Recreation % teen birth % Healthy food % Crime HbA1c Blood Lipid Eye- Exam Commercial rate Medicaid rate Medicare rate Commercial rate Medicaid rate Medicare rate Commercial rate Medicaid rate Medicare rate
Conclusions Proportion of minority is not the issue, but how integrated into the community. Effective explanations for county variations in diabetic care measures Primary care Social integration SES Additional data needs Analysis of role of personal responsibility requires individual level data Characteristics of providers and medical facilities
Further Information Donna Friedsam, MPH Researcher and Health Policy Programs Director UW Population Health Institute 608.263.4881 dafriedsam@wisc.edu http://uwphi.pophealth.wisc.edu http://www.evidencebasedhealthpolicy.org