How Population Based Casemix can Improve the Delivery of Primary Care Karen Kinder, PhD, MBA Johns Hopkins Bloomberg School of Public Health Presented at the 5th Nordic Casemix Conference Oslo, Norway June 1, 2012 Goals of Presentation • To introduce concepts surrounding population-based risk adjustment • To review examples of applications worldwide • To briefly describe the Johns Hopkins “ACG” case-mix adjustment / predictive risk modeling methodology Copyright 2007 Johns Hopkins University 2 More than just EMRs Copyright 2007 Johns Hopkins University IMPROVED POPULATION HEALTH STATUS FEEDBACK LOOP 3 Understanding populationbased risk adjustment Copyright 2007 Johns Hopkins University 4 Working Definitions • Case mix (clinical) / risk adjustment (economic) is the process by which the health status of a population is taken into consideration when setting budgets or capitation rates, evaluating provider performance, or assessing outcomes of care. • Predictive risk modeling is the prospective (or concurrent) application of risk adjustment measures and statistical forecasting to identify individuals with high medical need who would likely benefit from care management interventions. Copyright 2007 Johns Hopkins University 5 Population Case-Mix vs. Episodic Case -Mix DRGs categorize individuals, • considering their experience with the particular provider • An illness or event, • over a specified period of time ACGs categorize individuals, • considering their entire experience with the health care system, • their whole morbidity, • over a period of time Copyright 2007 Johns Hopkins University 6 Overview of Johns Hopkins ACG System • TOTAL POPULATION – Not just those who have been in hospital and includes non-users. • TOTAL EXPERIENCE - Applied using all diagnoses describing the person. They do not focus on individual visits. Ideally they are derived from primary and specialty ambulatory contacts as well as inpatient. • TOTAL PERSON -Comprehensive measure of a population’s risk and morbidity burden. They do not just categorize organ system-based diseases. Copyright 2007 Johns Hopkins University 7 Not all persons have the same need for health care Percent of Population Percent of Health Care Dollars Consumed 1% 30% 10% 70% 50% 97% Copyright 2007 Johns Hopkins University 8 Clinical Observations Underpin the ACG System • Morbidity is NOT randomly distributed across individuals. – 1) Morbidity “clusters”. – 2) Diagnoses co-occur. • The “illness burden” of providers’ practices is NOT randomly distributed. – 1) Some providers care for “sicker” patients. – 2) Sick patients choose certain providers referentially. Copyright 2007 Johns Hopkins University 9 Co-morbidity is the norm among older adults Diabetes 9% Heart Disease 11% Arthritis 12% Hypertension 22% 21% 22% 17% 20% Condition + 1 Source: Partnership for Solutions 21% 27% 25% 24% 19% 23% 22% 21% 24% 0% Single Condition 21% 23% 40% 20% 60% Condition + 2 16% 80% Condition + 3 100% Condition + 4+ Copyright 2007 Johns Hopkins University 10 Co-morbidity Total morbidity is not the same as the sum of different diseases, because diseases cluster and are inter-related in various ways. A more accurate way of characterizing morbidity is to characterize the pattern of diseases in people and populations. Copyright 2007 Johns Hopkins University Starfield 03/06 CM 3371 11 These patterns are linked to the prevalence of chronic co-morbidities (Data for Americans 65+) # Chronic Co-morbidities % Pop. Relative Cost (Per Pt.) Est. % of Total Medicare Costs Avg. # Unique MDs/Yr. Avg. # Filled Rx / Yr. 5+ 20% 3.2 66% 13.8 49 3-4 27% .9 23% 7.3 26 0-2 53% .1 11% 3.0 11 Data Source: G. Anderson et. al., Johns Hopkins Univ. 2003. (Derived from Medicare claims and beneficiary survey.) Copyright 2007 Johns Hopkins University 12 What Can Be Achieved with Case Mix Adjustment • Equity and fairness • To identify those patients most in need of health care resources • To facilitate providers who specialize in treating patients with higher than average illness burden. • Create incentives to encourage providers to match services to needs (appropriateness) • Ensure appropriate comparisons for research and performance assessment Copyright 2007 Johns Hopkins University 13 Why interest in risk adjustment is increasing globally • Population health care needs are rising, resource availability is not; focusing on “higher risk” patients makes sense. • Data systems and data collection are improving. • Management systems are integrating primary, secondary, and community care. • There is an increased interest in the equitable delivery of health care. Copyright 2007 Johns Hopkins University 14 Some Real-World Applications Copyright 2007 Johns Hopkins University 15 Domains one system. many tools. many solutions. many benefits.Copyright 2007 Johns Hopkins University 16 Possible Applications • Population based need-assessment across patient populations (e.g.,regions, vulnerable patient groups) • Assessing performance of providers (e.g. hospital clinics, doctors, regions). • Resource allocation / budgeting across clinics, regions or other care units. • “Predictive Risk” measurement to assist in chronic care management. • Quality improvement comparisons. Copyright 2007 Johns Hopkins University 17 Population Profiling Copyright 2007 Johns Hopkins University 18 Benefits of Health Status Monitoring • Understanding population risk and overall morbidity patterns • Detection of life style issues that may lead to health problems • Ability to identify changes in population health • Development of education or outreach programs Copyright 2007 Johns Hopkins University 19 Types of Morbidity Varies by Region Copyright 2007 Johns Hopkins University 20 Morbidity Distribution by ADG ADGs Total Group 1 Group 2 1: Time Limited Minor 14.7% 14.8% 14.4% 3: Time Limited Major 5.5% 4.0% 12.3% 9: Likely to Recur Progressive 2.0% 0.8% 7.7% 10: Chronic Medical: Stable 12.9% 7.4% 37.1% 11: Chronic Medical: Unstable 8.6% 4.0% 28.8% 25: Psychosocial: Recurrent or Persistent Unstable 5.8% 2.5% 20.1% 26: Signs/Symptoms: Minor 16.9% 15.3% 24.4% 32: Malignancy 1.0% 0.3% 4.0% Copyright 2007 Johns Hopkins University 21 Stratifying Risk within Diseases by RUB distribution Patient Count RUB 1 RUB 2 RUB 3 RUB 4 RUB 5 Ischemic Heart Disease (excl AMI) 32 0 0 31.2 40.6 28.1 Cardiac Arrhythmia 33 0 0 18.1 42.4 39.3 Disorders of Lipoid Metabolism 173 0 9.2 61.8 20.2 8.6 Hypertension w/o Major Complications 166 0 7.8 55.4 24.6 12.0 Hypertension w/Major Complications 33 0 0 60.6 18.1 21.2 Type 2 Diabetes, w/o Complication 51 0 13.7 52.9 21.5 11.7 Chest Pain 88 5.6 6.8 51.1 25 11.3 Obesity 43 0 9.3 44.1 41.8 4.6 Anxiety, Neuroses 160 0 11.8 62.5 21.2 4.3 Tobacco Use 57 0 3.5 43.8 40.3 12.2 Depression 40 0 10 57.5 22.5 10 EDC Description Copyright 2007 Johns Hopkins University 22 Example from Sweden Relisted 1-Oct-2005-1-Oct-2006 and differences in comorbidity 66.22 0 11.33 1 Not relisted 17.35 2 4.90 3 4 0.19 0 50.00 1 Relisted 10.31 26.32 2 3 4 13.16 0.22 0 20 40 % 60 80 RUB:resource utilization band Source: Zielinski A, et.al. Impact of comorbidity on the individual's choice of primary 2007 Johns Hopkins University . health care provider. Scand J Prim Health Care. 2011Copyright Jun;29(2):104-9 23 Capitation, Budgeting & Other Financial Issues Copyright 2007 Johns Hopkins University 24 Determining the Healthcare Budget Involves a Variety of Factors - Available Budget - Political Forces - Actuarial Forecasts Size of the Healthcare Pie Copyright 2007 Johns Hopkins University 25 Risk Adjustment Can Be Used To Slice The Pie Risk Adjustment Copyright 2007 Johns Hopkins University 26 Reasons why Risk Adjusted Payment & Budgeting May Be Necessary • To protect doctors, practices, or organizations that care for costlier than average patient populations. • To help ensure that those that finance care pay their fair share (neither too high or low). • To deter providers from selectively attracting healthier patients . • To facilitate organizations or providers wishing to specialize in treating people with higher than average illness burden. Copyright 2007 Johns Hopkins University 27 Plans defer in the morbidity burden of their patients 1,3 Families/Children Disabled Risk Ratio 1,2 1,1 1 Average Risk 0,9 0,8 0,7 Plan A Plan B Plan C Plan D All Using ACGs, risk ratios were determined for each contracting managed care organization / health plan. Expected values were determined separately for the two enrollee groups with this State Medicaid program. Copyright 2007 Johns Hopkins University 28 Alternative Ways to Apply Case-Mix to Payment • Applied concurrently, budgets are adjusted retrospectively based on the experienced morbidity profile of the population. • Applied prospectively, capitation amounts are adjusted based on the anticipated need for health care resources. • The portion of the payment which is case-mix adjusted is arbitrary. Copyright 2007 Johns Hopkins University 29 Risk Adjusted Performance Profiling Copyright 2007 Johns Hopkins University 30 Interpreting Profiling Results… 140 100 80 60 40 20 Potential Access Issues / Witholding Services Performance Feedback / Contracting / Incentives Over Utilization / Potential Fraud/Abuse 0 <. 70 0, 70 0, 75 0, 80 0, 85 0, 90 0, 95 1, 00 1, 05 1, 10 1, 15 1, 20 1, 25 1, 30 >1 .3 Number of Physicians 120 Efficiency Index Copyright 2007 Johns Hopkins University 31 Clinic 1 vs. 2, Expected Costs ACG Clinic 1’s Panel N (Expected) Clinic 2’s Panel N (Expected) Expected (mean costs) 0100 5 ($500) 8 ($800) $100 0200 2 ($500) 6 ($1,500) $250 0300 3 ($900) 3 ($900) $300 0400 10 ($5,000) 3 ($1,500) $500 Total 20 ($6,900) 20 ($4,700) $315 Copyright 2007 Johns Hopkins University 32 Using ACGs to Risk-Adjust Performance “Profiles” of Provider Groups Group Actual Cost Unadjusted Relative Cost ACG “Illness Burden” ACG Adjusted Efficiency Ratio #1 $157 1.22 1.02 1.20 #2 153 1.19 1.21 0.99 #3 144 1.12 0.92 1.22 #4 98 0.76 0.69 1.11 All* $129 1.00 1.00 1.00 The relative cost is the ratio of the actual cost to the plan average cost. “Illness Burden” describes the overall morbidity level of the population served by this provider.. “Efficiency” is the ratio of “actual-toexpected.” Copyright 2007 Johns Hopkins University 33 Risk Adjusted Practice Efficiency of Doctor Group #3 By Service Category Type of Service Relative Cost ACG Illness Burden Efficiency Inpatient 0.91 0.90 1.01 Primary Care 1.20 1.15 1.04 Surgery 2.23 0.91 2.45 Medical Specialties 1.61 0.92 1.75 Lab & x-ray 1.77 0.85 2.08 Pharmacy .86 0.85 1.01 1.12 0.92 1.22 Total Copyright 2007 Johns Hopkins University 34 Clinic Profiling Pharmacy cost x patient: observed ( ) and expected ( Efficiency Index: 0,79 ) Efficiency Index: 1,27 21% undercost 943.000 € 27% overcost 737.000 € 400 350 300 250 200 150 100 50 0 001 002 003 004 005 006 007 008 009 Mean Cost (€) 182,58 291,57 274,75 212,19 337,71 289,03 328,99 287,14 196,36 Average 270,49 Mean cost (€) expected 231,02 271,59 293,94 243,63 296,59 295,57 258,10 280,21 241,01 270,49 Overcost or undercost, undercost related to standard Efficiency Index Impact (€) 0,79 943.068 1,07 510.658 0,97 0,87 1,14 0,98 1,27 280.254 481.278 715.386 121.540 736.869 1,02 144.487 Copyright 2007 Johns Hopkins University 0,81 281.209 35 Risk-Adjusted O/E (Efficiency) Profiling Ratios for GPs Across UK Primary Care Trust (PCT) (2005) 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 GP1 GP2 GP3 GP4 GP5 GP6 GP7 GP8 No of referrals No of unique prescriptions / month GP9 GP10 GP11 GP12 No of unique radiology tests Copyright 2007 Johns Hopkins University 36 How Profiling Results are Typically Applied • Developing financial incentives – Distributing bonuses – Defining tiered networks – Differentiating fee schedules • Profiling/Assessment tool – To stimulate voluntary changes in behavior by sharing valid data presented in a useful format. – To identify potential fraud & abuse. – Can be developed on a number of entities, to include: physicians, employers, networks, or health systems Copyright 2007 Johns Hopkins University 37 Care Management & Predictive Modeling Copyright 2007 Johns Hopkins University 38 Uses • Predicting the future medical and pharmacy expense of individuals • Identifying individuals who would benefit from disease, case management and care management interventions • Identifying persons for inclusion in consumer-oriented education/outreach programs. • Identifying persons for inclusion in multi-disease intensive case-management programs. • Effectively target single-focus disease management programs. • Providing information useful to individual clinicians for patient-specific quality improvement measures. Copyright 2007 Johns Hopkins University 39 Using PM Risk Scores to Target Disease Management Program Participants % Enrollees in Rx-MG Risk Category Condition Below 90% Diabetes Congestive Heart Failure Resource Use of Cohort Relative to Total Population 90-95% Above 95% Below 90% 90-95% Above 95% 44.97 42.1 11.9 1.34 4.90 7.44 19.75 53.5 26.75 1.14 6.02 7.93 Tier 1 Tier 2 Copyright 2007 Johns Hopkins University Tier 3 40 Clinical profile of a population by “risk levels” (Based on JH-ACG risk scores) ACG Risk Groups: Percentile Rank (Year 1) Characteristic 0-10 11-25 26-50 51-75 76-90 91-95 95-99 Age, mean 9.1 22.5 28.1 37.1 42.9 47.3 49.4 % female 25.5 29.4 56.6 59.8 63.3 64.0 64.0 # chronic conditions .1 .1 .2 .6 1.1 1.9 3.1 Total # of medical conditions 1.9 1.8 2.8 3.9 5.6 7.6 10.5 # of drug therapeutic classes used .2 .4 1.3 2.5 4.8 7.4 10.8 Data Source: n=904,007, Privately Insured below the aged of 65 Copyright 2007 Johns Hopkins University 41 The ACG Software provides patient risk information in support of nurse case managers • Numerous co-morbidities • Seeing 13 doctors • At risk for future hospitalization • ER Visit with no admission • Poly-pharmacy use • Tobacco Use Copyright 2007 Johns Hopkins University 42 Identify, Stratify, Intervene with Persons with Diabetes Our goal: High Complexity Level 1 Moderate Complexity Level 2 1. Identify all persons with diabetes, and 2. Stratify them into three levels of complexity, and 3. Intervene appropriately. Each level of complexity has an appropriate level of care management intervention Low Complexity Level 3 Copyright 2007 Johns Hopkins University 43 Stratification: The Simplified Version High Complexity n = 65 Moderate Complexity n = 352 Low Complexity n = 3,332 Diabetes Diagnosis HbA1C > 9 ACG-pm > .6 Diabetes Diagnosis HbA1C >7 and < 9 ACG-pm > .2 and < .6 Diabetes Diagnosis HbA1C <7 ACG-pm <=0.2 Copyright 2007 Johns Hopkins University 44 Our services for Level 1, 2, and 3 Level 1 High risk with multiple chronic illness Intensive Case Management: Guided Care •RN or Social Work Case Manager •Individualized Assessment •Care Plan •Self-Management Plan Level 2 Moderate risk patients with single chronic illness or risk factors Level 3 Low risk Disease Management: Health Coaching and Lifestyle Management •Remote monitoring with TeleWatch •Programs to modify diet, increase exercise, smoking cessation, weight loss Health Education and Promotion •Healthwise information—online and in print, handbooks and mailing •Direct messaging via mail and web •Healthy lifestyle program promotions Copyright 2007 Johns Hopkins University 45 Predicting Hospital Events Acute care hospitalization excluding childbirth and injury 2 prediction periods: 6 and 12 months Intensive care hospitalization (ICU/CCU) Extensive length of stay (12+ days cumulative) Injury-related acute care hospitalization Copyright 2007 Johns Hopkins University 46 46 Four Coordination Markers A majority source of care (and percent of outpatient visits provided) A count of the number of unique providers A count of the number of specialty types (not the same as number of specialists seen) A marker for the ABSENCE of a generalist 6/6/2012 Copyright 2010, Johns Hopkins University Copyright 2007 Johns Hopkins University 47 47 Two Outputs • Probability of Unexpected Pharmacy Cost is a numerical probability score predicting individuals with moderate or high morbidity who have unusually large pharmacy expenditures • High Risk for Unexpected Pharmacy Cost is a binary flag that indicates individuals with a Probability of Unexpected Pharmacy Cost greater than 0.4. Copyright 2007 Johns Hopkins University 48 Why Does This Matter? High Pharmacy Utilization could suggest: • Inappropriate Care: Physicians who prescribed expensive medications • Uncoordinated Care: Potentially replicated prescriptions • Drug Abuse: Patients shopping different providers for prescriptions • Data Problems: Need to be ruled out 6/6/2012 Copyright 2010, Johns Hopkins University Copyright 2007 Johns Hopkins University 49 49 Benefits of PM for Care Management • Provides robust clinical information – Diagnosis-based condition markers – Pharmacy-based morbidity markers • Administratively efficient – Rapid assessment – Reductions in case finding and case preparation – Allows for better allocation of scarce case management resources • Identifies up to 25% unique individuals for case management compared to traditional methods – Using pharmacy, this holds true with as little as 1 month of data Copyright 2007 Johns Hopkins University 50 Overview of the ACG System Copyright 2007 Johns Hopkins University 51 Basis of the Johns Hopkins ACG System • ACGs are generally applied using all diagnoses describing the person. They do not focus on individual visits. Ideally they are derived from primary and specialty ambulatory contacts as well as inpatient • Comprehensive measure of a population’s risk and morbidity burden. They do not just categorize organ system-based diseases. Roots were primary care / population based. • There is a comprehensive ACG “suite” of risk/case-mix measures. Copyright 2007 Johns Hopkins University 52 ACG Actuarial Cells Reflect the Constellation Of Health Problems Experienced by a Patient Time Period (e.g., 1 year) Treated Morbidities Visit 1 Diagnostic Codes Morbidity Groups Code A ADG10 Code B Visit 2 Code C ADG21 Visit 3 Code D ADG03 Clinician Judgment Clinical Grouping ACG Category Data Analysis Copyright 2007 Johns Hopkins University 53 What is an ADG? • Definition: An ADG is a morbidity cluster that indicates severity and persistence of a patient’s condition treated over time. • Diagnoses within the same ADG are similar in terms of clinical criteria and expected need for health care resources. • ADGs are not mutually exclusive. Copyright 2007 Johns Hopkins University 54 Criteria Used to Assign Diseases/Conditions Into ADGs: • Duration 9Acute, chronic or recurrent • Severity 9Minor/stable versus major/unstable • Diagnostic certainty 9Symptoms versus disease • Etiology 9Infectious, injury or other • Specialty care involvement Copyright 2007 Johns Hopkins University 55 Assignment of ICD* Codes to ADGs: Diabetes Mellitus ICD-9 Code Description ADG 250.0 Diabetes Mellitus Uncomplicated 10: Chronic Medical Stable 250.03 Diabetes Mellitus without complications 11: Chronic Medical Unstable 250.1 Diabetes with Ketoacidosis 09: Likely to Recur, Progressive 362.0 Diabetes Retinopathy 18: Chronic Specialty, UnstableEye * ICD – WHO’s international classification of disease. Can also be use with Read codes and ICPC codes Copyright 2007 Johns Hopkins University 56 ADGs and Health Care Needs Evidence from BC Relationship between No. of ADGs and Hospitalization & Death, BC Adults 1996/97 60% Percent 50% 40% Hospitalization 30% Death 20% 10% 0% 1 2-3 4-5 6-9 10 + Number of Ambulatory Diagnosis Groups (ADGs) Copyright 2007 Johns Hopkins University 57 ACG Actuarial Cells Reflect the Constellation Of Health Problems Experienced by a Patient Time Period (e.g., 1 year) Treated Morbidities Visit 1 Diagnostic Codes Morbidity Groups Code A ADG10 Code B Visit 2 Code C ADG21 Visit 3 Code D ADG03 Clinician Judgment Clinical Grouping ACG Category Data Analysis Copyright 2007 Johns Hopkins University 58 Examples of ACG Categories ACG 0200 0600 1722 2800 4430 5322 Description Acute Minor, Age 2-5 Likely to Recur, without Allergies Pregnancy: 2-3 ADGs, no major ADGs, not delivered Acute Major and Likely to Recur 4-5 other ADG combinations, Age > 44, 2+ major ADGs Infants: 0-5 ADGs, 1+ major ADGs, low birth weight 59 Copyright 2007 Johns Hopkins University Expanded Diagnosis Clusters (EDCs) • EDCs categorize different diseases and conditions • Based only on ICD codes (no procedure codes) • EDCs complement ACGs by serving as a surrogate for chronic condition episodes. • Provides a greater clinical context to the case-mix • Useful for examining the epidemiology of a population or comparing two populations • Can help identify patients for inclusion in DMPs Copyright 2007 Johns Hopkins University 60 27 Major EDCs • • • • • • • • • • Administrative Allergy Cardiovascular Dental Ear, Nose, Throat Endocrine Eye Female Reproductive Gastrointestinal/Hepatic General Signs and Symptoms • General Surgery • Genetic • Genito-urinary • • • • • • • • • • • • • • Hematologic Infections Malignancies Musculoskeletal Neonatal Neurologic Nutrition Psychosocial Reconstructive Renal Respiratory Rheumatologic Skin Toxic Effects and Adverse Events Copyright 2007 Johns Hopkins University 61 Example: The Cardiovascular EDCs CAR01 Cardio Signs & Symptoms CAR03 Ischemic Heart Disease CAR04 Congenital Heart Disease CAR05 Congestive Heart Failure CAR06 Cardiac Valve Disorders CAR07 Cardio-myopathy CAR08 Heart Murmur CAR09 Cardiac Arrhythmia CAR10 Generalized Atherosclerosis CAR11 Disorder of Lipoid Metabolism CAR12 Acute Myocardial Infarction CAR13 Cardiac Arrest/Shock CAR14 Hypertension w/o Major Comp CAR15 Hypertension w/ Major Comp Copyright 2007 Johns Hopkins University 62 Assignment of ICD Codes to EDCs: Diabetes Mellitus ICD-9 Code Description ADG EDC 250.0 Diabetes Mellitus Uncomplicated 10: Chronic Medical Stable END06: Type 2 diabetes, w/o complication 250.03 Diabetes Mellitus without 11: Chronic Medical complications, Unstable uncontrolled END08: Type 1 diabetes, w/o complication 250.1 09: Likely to Recur, Diabetes with Ketoacidosis Progressive END07: Type 2 diabetes, with complication 648.0 Gestational Diabetes 11: Chronic Medical Unstable FRE04: Pregnancy and Delivery with Complications 362.0 Diabetes Retinopathy 18: Chronic Specialty, Unstable-Eye EYE13: Diabetic Retinopathy Copyright 2007 Johns Hopkins University 63 How are EDCs Used? • EDCs are used primarily for looking at disease prevalence and Standardized Morbidity Ratios (SMRs) -- which tell us, is the prevalence of the sub-group of analyses different than the overall population from which the sub-group was drawn. • EDCs are also used to demonstrate variability of cost within disease category Copyright 2007 Johns Hopkins University 64 ACG Predictive Models Copyright 2007 Johns Hopkins University 65 The ACG Predictive Models • Identifies persons likely to have serious medical needs and could benefit from case management programs • Predicts future resource use of patient groups within a population Copyright 2007 Johns Hopkins University 66 Value of Predictive Modeling Population of Persons Enrolled Across Two Year Period Prior High Cost Year-1 Predicted High Risk Year-2 (Prior Use) Not High Risk (Using Year-1 Data) Actual High Cost Year-2 High Risk, Current Costs Low, Future Costs High Copyright 2007 Johns Hopkins University 67 Risk Factors in the Johns Hopkins Predictive Model Age Overall Disease Burden Gender (ICD-10 → ACG) Risk Score Medications (ATC Codes → Rx-MG) Selected Medical Conditions (ICD-10 → Expanded Dx Clusters) Special Population Markers Selected Resource Use Measures ($) (ICD-10 → HOSDOM, Frailty) Copyright 2007 Johns Hopkins University 68 The Johns Hopkins ACG System: An Expanding Suite of Measures and Tools • Dx-PM – a “predictive model” that uses diagnoses to calculate a score representing future risk. Based on ACGs, EDCs and special high risk markers. • Rx-PM - a predictive model that calculates a score representing future risk and expected resources use based only on pharmacy use history. • DxRx-PM - The Rx-PM and Dx-PM measures can be combined if both sources are available to calculate a predictive score. Copyright 2007 Johns Hopkins University 69 ACG Pharmacy Model Copyright 2007 Johns Hopkins University 70 Combining both diagnoses and prescription data provides expanded information Total # of patients identified (ICD or pharm.) Percent of patients identified by diagnosis Percent of patients uniquely identified pharmacy Hypertension 59,937 70% 30% Disorders of Lipoid metabolism 37,736 61% 39% Congestive Heart Failure 11,223 61% 39% 1646 81% 19% Depression and Anxiety 20,863 23% 77% Diabetes 27,656 55% 45% Condition Chronic Renal Failure Source: US HMO claims dataset of elderly n=90,000 in 2001; Copyright 2007 Johns Hopkins University 71 Why Look at Pharmacy Data? • Pharmacy data capture a unique constellation of clinical information – Non-Concordance • Expediency -- Pharmacy-based claims are usually processed within 24 hours while office or hospital claims can takes several months for adjudication – Avoids or shortens the “claims-lag” common to much diagnosis-based risk adjustment • Provides an alternative data source when claims are NOT available Copyright 2007 Johns Hopkins University 72 Clinical Criteria for Rx-MG Assignment 1) Morbidity-type - symptom v disease 2) Duration of morbidity - chronic v time-limited 3) Stability of morbidity - stable v unstable 4) Route of administration - oral, inhaled, topical, intramuscular, intravenous 5) Therapeutic goal - curative, palliative, preventive Copyright 2007 Johns Hopkins University 73 The Major Rx-MG Categories • • • • • • • • Allergy/Immunology Cardiovascular Ears, Nose, Throat Endocrine Eye Female Reproductive Gastrointestinal/Hepatic General Signs & Symptoms • Genito-urinary • Hematologic • • • • • • • • Infections Malignancies Musculoskeletal Neurologic Psychosocial Respiratory Skin Toxic Effects/ Adverse Reactions • Others / non-specific medications Copyright 2007 Johns Hopkins University 74 Rx-MG Example: Corticosteroids Active Ingredient Route of Administration Rx-MG Description methylprednisolone-neomycin topical SKNx020 Skin / Acute and Recurrent Prednisolone compounding ALLx030 Allergy/Immunology / Immune Disorders Prednisolone injectable MUSx020 Musculoskeletal / Inflammatory Conditions Prednisolone oral ALLx030 Allergy/Immunology / Chronic Inflammatory Prednisolone ophthalmic EYEx020 Eye / Acute Minor: Palliative Prednisolone-sodium sulfacetamide ophthalmic EYEx010 Eye / Acute Minor: Curative Beclomethasone Compounding RESx030 Respiratory / Cystic Fibrosis Dexamethasone Nasal ALLx010 Allergy/Immunology / Acute Minor Betamethasone Injectable ENDx020 Endocrine / Chronic medical Dexamethasone Compounding ALLx030 Allergy/Immunology / Chronic Inflammatory ciprofloxacin-dexamethasone otic Otic EARx010 Ears, Nose, Throat / Acute Minor Dexamethasone Intravenous MUSx020 Musculoskeletal / Inflammatory Conditions beclomethasone inhalation RESx040 Respiratory / Airway Hyperactivity betamethasone-calcipotriene topical topical SKNx030 Skin / Chronic Medical Copyright 2007 Johns Hopkins University 75 How are Rx-MGs Used? • Used for patient profiles • Understanding disease prevalence and Standardized Morbidity Ratios (SMRs) -- (which tell us, is the prevalence of the sub-group of analyses different than the overall population from which the sub-group was drawn). • Identifying patients w/ morbidities NOT identified by diagnoses Copyright 2007 Johns Hopkins University 76 In Summary The future: Risk adjustment / predictive modeling are part of the solution for many health system challenges Costs are rising, quality levels are not. Areas where casemix should play a role: • • • • • Cost saving / efficiency focused interventions Care coordination and improvement for chronically ill Performance monitoring Access for special populations (disparities) Financial incentives for quality and efficiency Copyright 2007 Johns Hopkins University 78 Key non statistical considerations for model selection • Transparency – How easily can the model be understood and explained to all participants? • Clinical Texture – Does the system make sense to clinicians/physicians? • Flexibility – Does the system support a range of applications? • Support – What types of services are offered w/ the product? The reputation of the vendor? • Customizability Copyright 2007 Johns Hopkins University 79 Overview of Johns Hopkins ACG System • TOTAL POPULATION – Not just those who have been in hospital and includes non-users. • TOTAL EXPERIENCE - Applied using all diagnoses describing the person. They do not focus on individual visits. Ideally they are derived from primary and specialty ambulatory contacts as well as inpatient. • TOTAL PERSON -Comprehensive measure of a population’s risk and morbidity burden. They do not just categorize organ system-based diseases. Copyright 2007 Johns Hopkins University 80 The Johns Hopkins ACG System • Comprehensive measure of risk and morbidity burden. They do not just categorize organ system-based diseases. • Roots were primary care / population based. • ACGs are applied using all diagnoses (and/or pharmacy information) describing the person. They do not focus on individual visits. Ideally they are derived from primary and specialty ambulatory contacts as well as inpatient • Based at internationally respected academic institution provides for stability, transparency, as well as ongoing support and development. We have been doing this for 30 years. • Johns Hopkins has been developing collaborative IT / consulting and academic support infrastructure around the globe. Copyright 2007 Johns Hopkins University 81 Over a Dozen nations • Several Provinces in Canada • Numerous County Councils in Sweden • Several Regions of Spain • Multiple Primary Care Trusts in the UK • Sickness Funds in Germany • The largest Health Plan in Israel • Two Medical Schemes in South Africa • Active piloting in Denmark, Italy, Hong Kong, Turkey and Chile • Research in Japan, Lithuania, Thailand, and Taiwan • Interest expressed in numerous other countries Copyright 2007 Johns Hopkins University 82 ACGs around the Globe : Concurrent R-squared Key independent variables in model Age, gender , ADGs ACGs alone Country Dependent Variables Age, gender United States Total Costs (including pharmacy) 0.13 0.55 0.37 Canada - Manitoba Ambulatory costs 0.08 0.50 0.43 Taiwan Physician Costs 0.12 0.52 0.47 Dr. Visits 0.06 0.58 0.53 Primary care costs 0.11 NA 0.38 GP visits 0.13 0.59 0.53 Sweden Spain United Kingdom GP visits 0.54 Copyright 2007 Johns Hopkins University 83 ACG System is Customizable Recent Developments enable customizability for • Local Diagnostic coding systems • Local Pharmaceutical coding systems • Incorporation of local resource measures (costing measures) • Local Practice Behaviour Patterns • Incorporation of available data on socio-economic measures, individual’s functionality, living arrangement, and other non-morbidity based markers • Language Copyright 2007 Johns Hopkins University 84 Concluding Comment Case Mix is critical to ensuring the equitable delivery of health care, promoting the continuity of care and enabling the targeting of limited resources. Copyright 2007 Johns Hopkins University 85 Opportunities for learning more about the Johns Hopkins ACG System • Web Site: – www.acg.jhsph.edu • To learn more, contact: – Dr. Karen Kinder, Executive Director • [email protected] Copyright 2007 Johns Hopkins University 86
© Copyright 2024