Cost Analyses of the Iowa Medicaid Health Home Program

Transcription

1 Policy Report October 2014 Cost Analyses of the Iowa Medicaid Health Home Program Elizabeth T. Momany Phuong Nguyen-Hoang Peter C. Damiano Suzanne E. Bentler Dan M. Shane

2 Cost Analyses of the Iowa Medicaid Health Home Program Elizabeth T. Momany Assistant Director, Health Policy Research Program Associate Research Scientist, Public Policy Center Phuong Nguyen-Hoang Assistant Professor, Urban and Regional Planning and Public Policy Center Peter C. Damiano Director, Public Policy Center Professor, Preventive & Community Dentistry Suzanne E. Bentler Assistant Research Scientist, Public Policy Center Dan M. Shane Assistant Professor, Health Management and Policy Public Policy Center The University of Iowa

3 Executive Summary Introduction The Medicaid Health Home program began on July 1, 2012 with an initial enrollment of 308.This program is designed to enhance services to Medicaid members with chronic conditions through providers implementing Patient- Centered Medical Home best practices. Providers are paid to provide these enhanced services through per member per month payment based on the enrolled member s number of chronic conditions. Currently, there are 31 counties with MHH providers. Demographics Though enrollment did not rise rapidly due to difficulty in recruiting MHH providers, it did rise to 4,296 by December, The MHH population is primarily female, white, adult and living in an urban county. In addition, most of the study population qualified for Tier 1 or Tier 2 indicating they had 6 or fewer chronic problems. Methods The unit of analysis for this study was a per member month, with the dependent variable being per member per month (PMPM) Medicaid total costs and PMPM emergency department (ED) costs. We used Medicaid claims and enrollment files from calendar years for individuals in the study and comparison populations. This yielded 1,870,608 months of data for 66,449 members: 5,778 members in the intervention group and 60,671 members in the comparison group. Analytic method We used a fixed effects regression modeling technique that included monthly information for each member for the months they were in the study. The dependent variables for the analyses were total PMPM cost and PMPM cost for ED visits not resulting in an inpatient stay, PMPM costs for inpatient care and PMPM costs for nursing home services. The independent variables included month in the MHH program, month in the study, percent poverty, county of residence, Medicaid program of enrollment, presence of specific chronic conditions, and age. Results Regression analyses indicate that $ were saved in the first month of a person s enrollment in the program. This estimate increases by $10.70 for each additional month they are enrolled in the program. Final total savings from the program as of December 2013 are over $9 million. ii

4 Contents Executive Summary... ii Introduction....2 Eligibility for the Medicaid Health Home Program. 2 Provider Network....2 Study population....3 Results Cost Study unit of analysis...6 Study groups....7 Analytic method....7 Independent variables....7 Number of month in study and number of month in MHH...7 Dependent variables...8 ED costs...8 Inpatient costs Nursing home costs...8 Total costs....8 Changes in costs...9 Change in ED costs Change in inpatient costs Change in nursing home costs Change in total costs Limitations Appendix A: Provider Potential Income with the Quality Performance Bonus Appendix B: Case Finding Protocol for Chronic Conditions

5 Introduction The Iowa Medicaid Health Home incentivizes health care providers in Iowa to offer additional services to Medicaid members with chronic conditions through a monthly payment tied to the number and severity of the enrollee s with chronic conditions in their practice (Table 1-1). The Health Home model was authorized under a state plan amendment approved by the Centers for Medicare and Medicaid Services with enrollment beginning July 1, Health Home is a specific designation under section 2703 of the Patient Protection and Affordable Care Act and is a model of care that provides patientcentered, whole person, coordinated care for all stages of life and transitions of care specifically for individuals with chronic illnesses. For Iowa Medicaid, Health Home practices are enrolled Medicaid provider organizations capable of providing enhanced personal, the enhanced care provided, the Iowa Medicaid Enterprise (IME) offers monthly care coordination payments and the potential for annual performance based incentives designed to improve patient health outcomes and lower overall Medicaid program costs. Additional information about the Iowa Medicaid Health Home program is located at Information from the brochure entitled Provider Potential Income with the Quality Performance Bonus from the Iowa Department of Human Services is provided in Appendix A. Eligibility for the Medicaid Health Home Program To be eligible for the Health Home Medicaid enrollees must have at least two chronic conditions or one chronic condition and be at risk for developing a second condition from the following list: Hypertension Overweight (Adults with a Body Mass Index of 25 or greater/children in the 85th percentile) Heart Disease Diabetes Asthma Substance Abuse Mental Health Problems In addition, they may not be in IowaCare, PACE, Iowa Family Planning Network, QMB/SLMB, HMO or be a presumptively eligible child or adult. Table 1-1. Tier definitions Tier Sum of chronic conditions Monthly payment $ $ $ or more $76.81 Provider Network Medicaid Health Home enrolled providers include but are not limited to: physician clinics, community mental health centers, Federally Qualified Health Centers (FQHCs), and Rural Health Clinics (RHCs). Return to TOC 2

6 Figure 1-1. Map of the counties with Health Home providers as of July, (Map: Courtesy of the Iowa Department of Human Services) Study population Health Homes are expected to reduce health care costs for members through effective management of existing conditions, early detection of new conditions, and prevention efforts resulting in fewer and less costly hospitalizations, fewer nursing home admissions, and less emergency department (ED) use. This report explores the change in cost resulting from member enrollment in the Health Home. For the purpose of determining the cost effect of Medicaid Health Home (MHH) participation we included all claims for the period January 1, 2011 through December 31, This study period encompassed 18 months before the implementation of the program and 18 after implementation of the program. Due to difficulty establishing the costs associated with HMO encounters, members with enrollment in the HMO were removed from the analyses (though HMO enrollees are estimated to be less than 10% of all enrollee months in the program). This resulted in 5,808 members within the cost analyses. The MHH was slow to enroll providers and thereby members (Figure 1-2). By the end of 2013 only 5,869 members had been enrolled for at least one month during some time in that period. 3 Return to TOC

7 Figure 1-2. MHH enrollment by month and Tier level Tables 1-2 through 1-4 provide information regarding the demographics of this study population. Table 1-2. Number of months enrolled in the MHH for members in the final cost analyses Number of months enrolled N Percent 1 month % 2 months % 3 months % 4 months % 5 months % 6 months % 7 months % 8 months % 9 months % 10 months % 11 months % 12 months % 13 months % 14 months % 15 months % 16 months % 17 months % 18 months % Return to TOC 4

8 Table 1-3. Age, gender, race/ethnicity, county of residence, and tier for the MHH study population Characteristic Number Percent Gender Female 3,692 64% Male 2,116 36% Race/Ethnicity* White 3,102 53% Black or African American % Hispanic/Latino 329 6% Asian/Pacific Islander 116 2% American Indian 106 2% Multiple-other 34 <1% Undeclared 1,234 21% Age 0-17 years old % years old 4,386 76% 65+ years old % County of residence Polk 1,540 27% Woodbury 1,442 25% Black Hawk % Scott 381 7% Linn 359 6% Des Moines 207 4% Marshall 127 2% Plymouth 127 2% All others % Tier Tier 1 2,605 45% Tier 2 2,237 39% Tier % Tier % The study population is primarily female, white, adult and living in an urban county. In addition, most of the study population qualified for Tier 1 or Tier 2 indicating they had 6 or fewer chronic problems (Table 1-3). The distribution of gender by age (Table 1-4) reveals that though the study population is primarily female for those under 17 the gender distribution is more even at 55% female for those 18 years of age and over. In addition, as age increases it appears that members are more likely not to disclose their race, while children under 18 and adults over 64 are less likely to be white. The county of residence by age indicates that though all age groups are primarily in urban counties, the counties in which they reside vary by age. This most likely reflects the propensity of MHHs to take people in certain age ranges, particularly pediatric MHHs in certain counties. Finally, as might be expected, as age increases the likelihood that a member will be in a higher tier also increases. In fact, there are no children in tier 4 within the study population. 5 Return to TOC

9 Table 1-4. Gender, race/ethnicity, county of residence, and tier for the MHH study population by age Characteristic 0-17 years Number (%) years Number (%) 65+ years Number (%) Gender Female 272 (45%) 2,892 (66%) 528 (65%) Male 333 (55%) 1,494 (43%) 289 (35%) Race/Ethnicity* White 253 (42%) 2,503 (57%) 346 (42%) Black or African American 98 (16%) 716 (16%) 73 (9%) Hispanic/Latino 110 (18%) 163 (4%) 56 (7%) Asian/Pacific Islander 5 (1%) 37 (1%) 74 (9%) American Indian 21 (4%) 75 (2%) 10 (1%) Multiple-other 23 (4%) 11 (<1%) 0 (0%) Undeclared 95 (16%) 881 (20%) 258 (32%) County of residence Polk 49 (8%) 1,224 (28%) 267 (35%) Woodbury 176 (29%) 1,043 (24%) 223 (29%) Black Hawk 155 (26%) 683 (16%) 52 (7%) Scott 15 (3%) 279 (7%) 87 (7%) Linn 82 (14%) 243 (6%) 34 (5%) Des Moines 2 (<1%) 194 (5%) 11 (1%) Marshall 5 (1%) 106 (3%) 16 (2%) Plymouth 48 (8%) 64 (2%) 15 (2%) All others 72 (12%) 488 (11%) 58 (8%) Tier Tier (71%) 1,856 (42%) 318 (39%) Tier (26%) 1,759 (40%) 324 (40%) Tier 3 20 (3%) 613 (14%) 132 (16%) Tier 4 0 (0%) 158 (4%) 43 (5%) Results Cost Study unit of analysis The unit of analysis for this study was a per member month, with the dependent variable being per member per month (PMPM) Medicaid total costs, PMPM emergency department (ED) costs, and PMPM nursing home costs. We used Medicaid claims and enrollment files from calendar years for members in the MHH and the comparison population. This yielded 1,870,608 months of data for 66,449 members. Of these, 5,778 members had at least one month within the MHH. These members had 53,343 months of MHH experience, the remaining 1,817,463 months were either months when the MHH members were not enrolled in the MHH such as in the 18 months before the program started or were months of experience for the comparison group. Table 2-1 provides an estimate of the reimbursement provided to MHHs for the tier payments. This estimate is based on the August, 2014 enrollment file. Member tier level was summed across months for the study period. Though Return to TOC 6

10 these costs are broken out in the table, the cost analyses that follow include these costs in the PMPM values. Therefore, these costs should NOT be subtracted from the savings estimates provided later in the report. Table 2-1. Tier payments to the MHHs Tier Monthly payment Tier months 1 $ ,168 $283,750 2 $ ,812 $481,587 3 $ ,244 $319,755 4 $ ,357 $104,231 Reimbursement to MHHs Total 48,581 $1,189,324 Study groups The intervention group consisted of 5,778 members who had been enrolled for at least 1 month in the MHH and who had no months of enrollment in the HMO. These members were matched to 100,000 randomly selected members from the general Medicaid population on race, gender, age, and whether they had each of the seven chronic conditions needed for entry into the MHH, mental health condition, substance abuse, asthma, diabetes, hypertension, obesity and/ or coronary artery disease. This provided 60,671 members of the comparison. Matching weights were created reflecting the degree to which each comparison group member contributed to the analyses. These weights were applied when calculating the cost changes. Analytic method We used a fixed effects regression modeling technique that included monthly information for each member for the months they were in the study. The maximum number of months of data available for a member in the analyses was 36. As this model allows for data for each member in the study and comparison groups for the period before and after implementation, each member serves as his/her own control. This method of predicting cost changes is quite robust. Independent variables Number of month in study and number of month in MHH Each month was coded as to the number of months the members were in the MHH and number of the month in the study. For example, a member who was in the third month of MHH enrollment and had been in Medicaid for the full 18 months of the pre-implementation period would be considered to have a value of 3 for the month in MHH and a value of 21 for the month in the study. Members enrollment in the MHH must be verified by the MHH provider or clinic staff on a monthly basis. If the verification does not occur MHH members are given a tier value of 0 indicating Medicaid has suspended the tier payment to the MHH provider. Verification by the MHH can result in reinstatement of the member. After consultation with DHS personnel familiar with the MHH program it was determined that if a gap of 6 months or less existed between MHH suspension and reenrollment the member should be considered as enrolled in the MHH program during the entire gap period. Medicare coverage We included all months regardless of Medicare Part A status with an indicator (0=not in Medicare; 1=is in Medicare). Medicare Part B is assumed when a recipient is on Part A. 7 Return to TOC

11 Percent poverty The percent of the federal poverty level, as reported on the Medicaid enrollment file, is provided by month. Has a chronic condition In each month there are seven indicator variables (0=does not have the condition; 1=has the condition), one for each of the conditions listed as a qualifying condition for the MHH program: substance abuse, mental health problem, asthma, diabetes, coronary artery disease, hypertension, and obesity. We used case finding protocols derived from the HEDIS quality measures (Appendix B) to find the first month of the study in which there is a claim for the condition. This is the index month and the first month that the indicator is set to 1. The indicator remains 1 throughout the study period following the index month. Age Age was calculated for each month as the age on the first of the month. Program There were 10 indicator variables (0=not in the named program; 1=is in the named program) for the Medicaid program the members was enrolled in. These included: MediPASS, Fee-for-service (FFS), disability determination (SSI), foster care, waiver programs, IowaCare, Family Planning, Medicaid for Employed People with Disabilities (MEPD), dual eligible, and other. County of residence The data for each month also had 99 indicator variables (0=not in the named county; 1=in the named county) to indicate the member county of residence. Dependent variables ED costs ED visits were identified through revenue codes ( ) from institutional claims. By calculating the length of stay for the claim we were able to determine whether there was an inpatient stay associated with the visit. ED costs include all costs associated with ED visits that did NOT have an associated inpatient stay. Inpatient costs Inpatient costs include all costs associated with an inpatient stay not including those for observation in the ED. Nursing home costs Nursing home costs included all reimbursements provided to skilled and intermediate care facilities. Total costs Total costs include medical, institutional, dental, inpatient, outpatient, pharmaceutical, durable medical equipment, and any additional services provided under special programs or waivers. Essentially, if Medicaid made a payment to cover the service it is in the total cost calculation. Return to TOC 8

12 Changes in costs The regression results are provided in Tables 2-2 through 2-5. First month of MHH provides an estimate of the change in PMPM cost resulting from the first month of MHH enrollment. Monthly trend provides an estimate of the adjustment to change in PMPM costs that should be made each month. The two together are used to estimate total change in cost over time. For example, in Table 2-5 First month of MHH, indicating the change in costs during the first month is -$ and so on. We also include a lower-bound estimate to give a range of savings. This lower bound estimate is the value provided when we move one standard deviation below the original estimate. Change in ED costs Table 2-2 provides the results for the change in costs for ED visits that did not result in an inpatient stay. The savings incurred while enrolled in the MHH is $11.80 PMPM with a lower-bound estimate of $9.79. The PMPM savings do not increase over time as the monthly trend was not significant. There were 53,343 months of MHH experience, with a savings of $11.80 per month. The total ED savings were $629,447. Table 2-2. Change in ED visit PMPM costs as a result of enrollment in the MHH, average and lower-bound estimate Independent variable Change in cost Lower-bound estimate of change in cost Average monthly cost of MHH -11.8*** Monthly trend Medicare indicator -25.2*** Percent poverty Substance abuse disorder 27.8*** Asthma Diabetes Coronary artery disease 19.5*** Hypertension 9.96*** Obesity 5.51*** 7.54 Mental health condition 5.34*** 6.78 Age MediPASS Fee-for-service Disability determination -27.2** Foster care -27.4** Waiver services -32.2*** IowaCare -44.7*** Family Planning -51.4*** Medicaid for Employed People with Disabilities -38.2*** Dual eligible -33.4*** * p<0.10, ** p<0.05, *** p< Return to TOC

13 Change in inpatient costs Table 2-3 provides the results for the change in costs for inpatient stays. The savings incurred while enrolled in the MHH is $ PMPM with a lowerbound estimate of $ The PMPM savings increase over time at a rate of $4.29 per month. There were 53,343 months of MHH experience resulting in total inpatient savings of $8,423,310. Table 2-3. Change in inpatient PMPM costs as a result of enrollment in the MHH, average and lower-bound estimate Independent variable Change in cost Lower-bound estimate of change in cost Average monthly cost of MHH *** Monthly trend -4.29* Medicare indicator *** Percent poverty Substance abuse disorder 317.6*** Asthma Diabetes 98.9*** Coronary artery disease 343.9*** 380 Hypertension 122.9*** Obesity 110.9*** Mental health condition 54.1*** 74.9 Age MediPASS 122.3* Fee-for-service 186.4** Disability determination Foster care ** Waiver services *** 1156 IowaCare ** Family Planning *** Medicaid for Employed People with Disabilities Dual eligible * p<0.10, ** p<0.05, *** p<0.01 Change in nursing home costs The regression estimating changes in nursing home costs indicated there were significant savings in PMPM costs related to nursing home care of $11.70 with no significant trend (see Table 2-4). This result is somewhat unexpected as there were only 2,027 months with nursing home costs for the MHH members, reducing the power of the analyses to determine differences between the groups. Total savings in PMPM nursing home costs were $624,113. Return to TOC 10

14 Table 2-4. Change in NH PMPM costs as a result of enrollment in the MHH, average and lower-bound estimate Independent variable Change in cost Average monthly cost of MHH -11.7* Monthly trend Medicare indicator Percent poverty -0.23*** Substance abuse disorder Asthma -15.9** Diabetes 34.6*** 47.2 Coronary artery disease 22.8* 35.7 Hypertension Obesity Mental health condition Age MediPASS Fee-for-service Disability determination Foster care ** Waiver services 743.7*** IowaCare Family Planning Medicaid for Employed People with Disabilities 58.8 Lower-bound estimate of change in cost Dual eligible Change in total costs The results for change in total costs are shown in Table 2-5. The average monthly savings per member in the MHH was $ with a lower-bound estimate of $ This PMPM savings estimate rose by $10.70 PMPM with a lower-bound estimate of $7.82 PMPM. The savings by month are shown below in Table 2-6. Table 2-5. Change in Total PMPM costs as a result of enrollment in the MHH, average and lower-bound estimate 11 Return to TOC Independent variable Change in cost Average monthly cost of MHH *** Monthly trend -10.7*** Medicare indicator *** Percent poverty Substance abuse disorder 359.8*** Asthma 87.0*** Diabetes 106.3*** Coronary artery disease 420.3*** Hypertension 171.1*** Obesity 139.8*** Mental health condition 148.7*** Age Lower-bound estimate of change in cost

15 MediPASS 174.6* Fee-for-service 221.6** Disability determination Foster care Waiver services *** IowaCare *** Family Planning *** Medicaid for Employed People with Disabilities Dual eligible Table 2-6. Change in total cost for the first 18 months of the MHH program by member s month in the program Month in program Number of months PMPM savings Total monthly savings Lowerbound PMPM savings Lowerbound total monthly savings 1 5,834 $ $770,671 $ $622, ,570 $ $795,396 $ $637, ,285 $ $811,248 $ $646, ,984 $ $818,373 $ $648, ,698 $ $821,680 $ $648, ,346 $ $806,618 $ $633, ,008 $ $786,770 $ $615, ,424 $ $708,768 $ $552, ,899 $ $631,112 $ $490, ,414 $ $551,358 $ $427, ,127 $ $508,566 $ $393, ,901 $ $474,870 $ $366, ,702 $ $443,371 $ $341, ,538 $ $417,106 $ $320, ,118 $ $315,164 $ $241, $ $237,591 $ $181, $ $150,740 $ $115, $ $58,404 $ $44,573 Total 53,343 $10,107,805 $7,928,945 Less Admin Cost $268,750 $9,839,055 $7,660,195 The regression results indicate that the MHH provided over 9 million in savings to the Medicaid program during the first 18 months. Additional, analyses indicate that some of these savings were derived from lower costs for ED visits, less money spent on inpatient care and reductions in nursing home costs. The total amount spent on care for the intervention group over the 18 month period was approximately $49 M, marking a nearly 20% reduction in costs. Return to TOC 12

16 Limitations Though we attempted to match the comparison group to the intervention group we were not completely successful. We did not utilize all of the Medicaid members, but used a random sample of Medicaid members to derive a matched comparison group. Interpretation and extrapolation of these results to other members or practices must be limited. These analyses provide the results for a targeted program with both members and practices needing to meet certain criteria. Any assumption that these results would be easily replicated with members who do not have chronic conditions or practices that are not ready to intervene would be false. 13 Return to TOC

17 Appendix A: Provider Potential Income with the Quality Performance Bonus Return to TOC 14

18 Iowa Medicaid Health Home Program Provider Potential Income with the Quality Performance Bonus Medicaid Providers did you know? Beyond the usual fee-for-service reimbursement, Health Home providers are eligible for two additional types of payment: Member Enrollment Income and Quality Performance Bonus (P4P) Health Home Examples Average Monthly Member Enrollment Average Monthly Member Income Average Yearly Member Income Potential Annual P4P Bonus Clinic A 200 $4,839 $58,064 $11,613 Clinic B 500 $12,084 $145,000 $29,000 Clinic C 1500 $36,290 $435,483 $87,097 Earn an annual pay for performance (P4P) bonus that represents up to 20% of the total PMPM payments made to the Health Home. Member s Tier PMPM Rate Tier 1 (1-3 chronic conditions) $12.80 Tier 2 (4-6 chronic conditions) $25.60 Tier 3 (7-9 chronic conditions) $51.21 Tier 4 (10 or more chronic conditions) $ Return to TOC

19 Tiered Per Member per Month (PMPM) For more information contact: Pamela Lester at or Visit us online at Iowa Department of Human Services - March 2014 Return to TOC 16

20 Appendix B: Case Finding Protocol for Chronic Conditions Mental health condition Table MPT-A: Codes to Identify Mental Health Diagnosis-taken from NCQA HEDIS Mental Health Utilization measure. 08.Utilization and Relative Resource Use_partB_pp ICD-9-CM Diagnosis 290, , Substance abuse disorder Table IAD-A: Codes to Identify Chemical Dependency Diagnosis-taken from NCQA HEDIS Identification of Alcohol and Other Drug Services measure. 07.Utilization and Relative Resource Use_partA_pp ICD-9-CM Diagnosis , , 305.0, , 535.3, Asthma Table ASM-A: Codes to Identify Asthma-taken from Use of Appropriate Medications for People with Asthma measure. 03. Effectiveness of Care_Respiratory-Cardiovascular_ Description ICD-9-CM Diagnosis Asthma 493.0, 493.1, 493.8, Diabetes Table CDC-A: Prescriptions to Identify Members With Diabetes-taken from Comprehensive Diabetes Care measure. 04. Effectiveness of Care_Diabetes, Musculo-Skeletal, Behavioral_pp Medications are listed in Excel spreadsheet in Health Home folder/claims/diabetesmeds.xls Description Alpha-glucosidase inhibitors Amylin analogs Prescription Acarbose Miglitol Pramlinitide 17 Return to TOC

21 Antidiabetic combinations Glimepiride-pioglitazone Glimepiride-rosiglitazone Insulin Meglitinides Miscellaneous antidiabetic agents Sulfonylureas Thiazolidinediones Glipizide-metformin Glyburide-metformin Table CDC-B: Codes to Identify Diabetes Description Metformin-pioglitazone Metformin-rosiglitazone Metformin-sitagliptin Insulin aspart Insulin aspart-insulin aspart protamine Insulin detemir Insulin glargine Insulin glulisine Insulin inhalation Insulin isophane beef-pork Insulin isophane human Insulin isophane pork Insulin isophane-insulin regular Insulin lispro Insulin lispro-insulin lispro protamine Insulin regular beef-pork Insulin regular human Insulin regular pork Insulin zinc beef-pork Insulin zinc extended human Insulin zinc human Insulin zinc pork Nateglinide Repaglinide Exenatide Liraglutide Sitagliptin Acetohexamide Chlorpropamide Glimepiride Glipizide Glyburide Tolazamide Tolbutamide Pioglitazone Rosiglitazone ICD-9-CM Diagnosis Diabetes 250, 357.2, 362.0, , Heart disease We have opted to use ICD-9-CM to define heart disease. There are no measures that detail Return to TOC 18

22 inclusion criteria for this chronic problem. Overweight We have opted to use ICD-9-CM to define obesity and overweight. There are no measures that detail inclusion criteria for this chronic problem. Hypertension Table CBP-A: Codes to Identify Hypertension-taken from Controlling High Blood Pressure measure. 03. Effectiveness of Care_Respiratory-Cardiovascular_ Description Hypertension 401 ICD-9-CM Diagnosis Return to TOC 19