Deconstructing the RADV: The Past, Present, and Future of RADV

Deconstructing the RADV: The Past, Present, and Future of RADV May 24, 2016 Presented by: Kim Browning, CHRS, PMP, CHC Executive Vice President

Agenda 2 PY 2007 What We ve Learned PY 2011 Expectations & Beyond Plan Selection Insights HCC Selection Highlights Internal Controls Building a RADV Survival Toolkit Future of RADV References Q&A

3 PY 2007 RADV What We ve Learned Increased sophistication o Sampling o Extrapolation methodology o More Coding Clinics Published after 2007 Greater knowledge o CMS o Health Plan Annual rhythm starting with 2011 dates of service PY 2016 Final Call Letter 1 o Authority to use HHS to conduct RADV OIG also looks at risk adjustments

Still waiting for average error rate o Annual rhythm to continue o Leverage annual window to submit redacts outside reporting year 2 o OIG Audit focus includes risk adjustment 3 PY 2011 RADV Expectations & Beyond 4 Can breathe again once you ve been selected o One plan selected in 2007, 2011, and 2012 Develop CDAT competency Hardships o Really worth the time?

PY 2011 RADV Expectations & Beyond, cont. 5

Higher than average risk scores o Bleeding edge of Prospective Does a lot of Prospective Plan Selection Insights Disproportionate volumes of average HCC Recent merger or acquisition Large membership Issues with timely RAPS and Encounter Data submission Little or no deletes Fast growing 6

HCC Selection Highlights 7 Health Plan 1 Health Plan 2 Targeted Diagnosis No dx detected HCC for provider s with high RAF scores More claims driven prospective and supplemental Long inpatient stay Cross year inpatient claims o Admitted in 2014 o Discharged in 2015

Common concerns o Single source claims o Problematic HCCs HCC Selection Highlights, cont. HCCs with < 1% average distribution 8

HCC Selection Highlights, cont. 9 HCC HCC 1 HCC 6 HCC 8 HCC 17 HCC 27 HCC 34 HCC 46 HCC 70 HCC 71 HCC 73 HCC 74 HCC 76 HCC 80 HCCs with < 1% Avg Distribution Aids Opportunistic Infections Metastatic Cancer and Acute Leukemia Diabetes with Acute Complications End Stage Liver Chronic Pancreatitis Severe Hematological Disorders Quadriplegia Paraplegia Amyotrophic Lateral Sclerosis and Other Motor Neuron Disease Cerebral Palsy Muscular Dystrophy Coma, Brain Compression/Anoxic Damage HCC HCC 82 HCCs with < 1% Avg Distribution Respirator Dependence/ Tracheostomy Status HCC 104 Monoplegia and Other Paralytic Syndromes HCC 110 Cystic Fibrosis HCC 115 Pneumococcal Pneumonia, Emphysema, Lung Abscess HCC 157 Pressure Ulcer of Skin with Necrosis Through to Muscle, Tendon or Bone HCC 162 Severe Skin Burn or Condition HCC 166 Severe Head injury HCC 173 Traumatic Amputations and Complications HCC 186 Major Organ Transplant or Replacement Status

Best defense is strong offense Continue to audit vendors Still need to audit prospective and supplemental Risks lurk in claim-driven risk adjustments o Start/increase auditing effort Collaborate with internal audit Enhance claim filters Look for the infrequent HCCs Single source claims If only one diagnosis submitted in review year 10 Internal Controls: Pre-RADV YIKES!

11 Internal Controls: Pre-RADV, cont. Adopt consistent discipline of self audits o Once and done is not enough Know your outliers o Regionally o Nationally

Medicare Advantage outliers 12 Internal Controls: Pre-RADV, cont. Risk Score Group Strata RAF Average Percent of Population Low < 1.0 60% Medium > 1.0-2.5 35% High 2.6-3.1 < 3% Very High > 3.2 < 3%

Duals outliers 13 Internal Controls: Pre-RADV, cont. Risk Score Group Strata RAF Average Percent of Population Low < 1.6 15% Medium 1.6-2.3 < 20% High 2.3-3.2 > 20% Very High > 3.2 > 25%

14 Internal Controls: Pre-RADV, cont. Effective validation techniques o Members with seven or more HCCs o Members with + 1 in RAF score from prior year o Top 1/3 paid stratum o High distribution HCCs o Presumed red flags Active versus history of Vascular disease Diabetes with complications Major depression Consider a combination 2016 2014 Cognisight, LLC

15 Internal Controls: Pre-RADV, cont. Know extrapolation methodology and how it applies to your risk adjustments You got this, right?

RADV EXTRAPOLATION EXAMPLE # RISK SCORE STRATA MA PAYMENT Hypothetical MA Pymt. Variance Weighted Payment error Deviation Deviation Squared Variance (div by 67-1) 1 5.073 TOP 3RD $ 3,934.20 $ 3,000.00 $ 934.20 $ 13,943.28 $ 822.02 $ 675,710.99 2 4.975 TOP 3RD 3,957.67 3,957.67 - - (112.18) 12,585.16 3 4.087 TOP 3RD 3,171.02 3,200.00 (28.98) (432.54) (141.16) 19,927.16 4 3.737 TOP 3RD 2,900.11 2,900.11 - - (112.18) 12,585.16 5 3.547 TOP 3RD 2,753.05 2,753.05 - - (112.18) 12,585.16 63 1.277 TOP 3RD 996.02 996.02 - - (112.18) 12,585.16 64 1.251 TOP 3RD 975.90 975.90 - - (112.18) 12,585.16 65 1.246 TOP 3RD 972.03 972.03 - - (112.18) 12,585.16 66 1.242 TOP 3RD 968.93 968.93 - - (112.18) 12,585.16 67 1.240 TOP 3RD 967.39 967.39 - $ - $ (112.18) $ 12,585.16 $ 109,128.51 $ 101,612.21 $ 7,516.30 $ 112,183.58 $ 0.00 $ 12,710,440.10 192,582.426 Error % 6.89% Enrollee Weight 14.925 Weighted Enrollee Payment Error $ 112,183.58 68 1.224 MIDDLE 3RD $ 955.00 $ 940.00 $ 15.00 $ 223.88 $ (21.11) $ 445.68 69 1.219 MIDDLE 3RD 951.13-951.13 14,195.97 915.02 837,259.42 70 1.207 MIDDLE 3RD 941.84 941.84 - - (36.11) 1,304.02 71 1.204 MIDDLE 3RD 939.52 939.52 - - (36.11) 1,304.02 72 1.168 MIDDLE 3RD 911.65 911.65 - - (36.11) 1,304.02 130 0.685 MIDDLE 3RD 542.04 542.04 - - (36.11) 1,304.02 131 0.685 MIDDLE 3RD 537.81 537.81 - - (36.11) 1,304.02 132 0.683 MIDDLE 3RD 536.25 536.25 - - (36.11) 1,304.02 133 0.679 MIDDLE 3RD 533.16 533.16 - - (36.11) 1,304.02 134 0.679 MIDDLE 3RD 533.16 533.16 - $ - $ (36.11) $ 1,304.02 $ 48,392.30 $ 45,972.85 $ 2,419.45 $ 36,111.19 $ (0.00) $ 2,112,386.58 32,005.857 Error % 5.00% Enrollee Weight 14.925 Weighted Enrollee Payment Error $ 36,111.19 135 0.649 BOTTOM 3RD $ 509.94 $ 515.00 $ (5.06) $ (75.52) $ (16.05) $ 257.74 136 0.649 BOTTOM 3RD 516.19 516.19 - - (10.99) 120.87 137 0.645 BOTTOM 3RD 506.84 506.84 - - (10.99) 120.87 138 0.635 BOTTOM 3RD 499.11 499.11 - - (10.99) 120.87 139 0.617 BOTTOM 3RD 485.17 485.17 - - (10.99) 120.87 197 0.177 BOTTOM 3RD 147.91 145.00 2.91 43.43 (8.08) 65.35 198 0.177 BOTTOM 3RD 144.60 144.60 - - (10.99) 120.87 199 0.177 BOTTOM 3RD 144.60 144.60 - - (10.99) 120.87 200 0.177 BOTTOM 3RD 144.60 144.60 - - (10.99) 120.87 201 0.177 BOTTOM 3RD 144.60 144.60 - $ - $ (10.99) $ 120.87 $ 20,348.32 $ 19,611.71 $ 736.61 $ 10,994.18 $ 0.00 $ 264,963.28 4,014.595 Error % 3.62% Enrollee Weight 14.925 Weighted Enrollee Payment Error $ 10,994.18 2 3 4 4 4 Strata Total $ 177,869.13 $ 167,196.77 $ 10,672.36 $ 159,288.96 Standard Error (SE) (Sq of PE) 58,412.18 Total CMS Pymt $ 3,900,000 Sample Population (Extrapolated) Confidence Interval CI (2.575*SE) $ 150,411.37 Average Error % 6% 4% 5 POINT ESTIMATE (PE) $ 159,288.96 PE + Confidence Interval $ 309,700.33 8 PE - Confidence Interval 8,877.58 9 16 Nh=3000 1 Roadmap Summary 1 RADV eligibles 2 Total CMS payment (based on monthly premiums) 3 Divide sample into three stratum 4 Weight the stratum 5 Point estimate 6 Standard Error 7 Confidence Interval 8 Upper and lower bounds 9 Lower + FFS adjuster 6 7

Building A RADV Survival Toolkit 17 Gulp! Dear CEO, You ve been selected for a contract level RADV

Attend CMS training o Project manager o Coders Include risk adjustment coders Contract externally if not available in-house Ideally with audit competency Building A RADV Survival Toolkit, cont. Form cross functional team o Not all coders can be auditors 18 o Finance (actuary, revenue manager, etc.) o Provider relations o Rented network liaisons o Internal audit o Compliance o Consider data governance Future filter modifications and pend actions

Medical records o Chart acquisition is key Building A RADV Survival Toolkit, cont. Can burn a lot of runway Hire competency if internal competency or capacity is lacking Precision over volume 19 Attestations o Get out right away even if not sure about using medical record Don t wait until the chart gets ranked We re playing 52 card pickup

CMS hardships Building A RADV Survival Toolkit, cont. o Why you can t get medical records Need sub-team to document hardship cases Mixed reaction from plans regarding usefulness 20

Ranking Building A RADV Survival Toolkit, cont. o Up to five medical records per HCC o New competency o Develop strategy and include an actuary Rank charts A through E o Support targeted diagnoses and new HCCs CMS stops when they get what they need Extrapolation at audited HCC level Example: 5 members with HCC 19 21 2 strong with no new HCCs + 3 strong with new HCCs No evidence we re being given credit for new HCCs

CDAT o New four-letter word o Enter and save while work is in progress Allows re-sequencing Building A RADV Survival Toolkit, cont. o Once submitted, it s final o At least three people should be copied and available to complete CDAT info and uploads 22

Other o External RADV support Building A RADV Survival Toolkit, cont. ~ > $300k including chart acquisition o Unclear fate for Up to five medical records 3:1 ratio for one plan o 500 HCCs > 1500 medical records o Timeline feels long but really isn t Don t waste time o Blind coding isn t recommended for RADV o Blue Card via Verisk isn t applicable for vendors 23

Future of RADV 24 RAC R A D V

25 Future of RADV: A Little Bit on RAC Program Recovery Audit Contractor (RAC) Program o Selection based on items with a propensity of error o All provider types o Procedure code vs. diagnoses code o Level of care and medical necessity o Looks at both over and under payments o Ongoing audit not an event And don t forget about

Improve coding intensity calculation 26 Future of RADV: GAO Recommendations o Only include three most recent years of risk adjustment data o Standardize change in disease risk scores to account for expected increase in all Medicare Advantage contracts o Develop method of accounting to separate provider submissions from supplemental record review o Include beneficiaries renewed from different contracts under same Medicare Advantage organization doing pair-year period

Modify contract selection for contract-level RADVs to focus on those with high rate of improper payments o Select contracts with 27 Highest coding intensity score High rates of unsupported diagnosis in prior contract level RADV audits For contracts no longer in operation, select contract under same Medicare Advantage organization and service area Combination of above for plans with high enrollment o Exclude contracts with Low coding intensity score Future of RADV: GAO Recommendations, cont.

Improve RADV timeliness o Closely align timeframe to national RADV o Reduce timeframe between plan selection and audit sample notifications o Improve record transfer reliability and performance o Require defined number of days for audit Improve timeliness of appeals process Future of RADV: GAO Recommendations, cont. Develop plan and timeline for incorporating RAC into Medicare Advantage 28

References 1. Final Call Letter PY 2016 (slide 3) https://www.cms.gov/medicare/health- Plans/MedicareAdvtgSpecRateStats/Downloads/ Announcement2016.pdf 29 2. Window to submit redacts outside reporting year (slide 4) http://www.csscoperations.com/internet/cssc3.nsf/files/overpayment%20 Memo02182015.pdf/$FIle/Overpayment%20Memo02182015.pdf 3. OIG Audit focus includes risk adjustment (slide 4) http://oig.hhs.gov/reports-and-publications/archives/workplan/2016/oigwork-plan-2016.pdf

Q&A For more information, please contact: Kim Browning kbrowning@cognisight.com 585.662.4215 www.linkedin.com/in/kimbrowning1 30