A patient-centric perspective on population health The psychosocial - PDF document

5/31/2017 An Optum Company Alejandro Reti, MD Chief Medical Officer; Optum Analytics Social determinants of health and population health analytics Alejandro Reti, MD Chief Medical Officer, Optum Analytics A patient-centric perspective on population health • The psychosocial foundations for good (or poor) health • Evolution and maturation of analytics in service of population health management • Patient reported and collected data in service of patient engagement and patient directed care An Optum Company 2 1

5/31/2017 Images courtesy of Harry Burns, MD Scotland NHS Scotland saw a substantial relative decrease in life expectancy beginning in the 1950s Life expectancy: Scotland & other Western European Countries, 1851-2005 Source: Human Mortality Database 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 3 7 1 5 9 3 7 1 5 9 3 7 1 5 9 3 7 1 5 9 3 7 1 5 9 3 7 1 5 9 3 7 1 5 9 3 7 1 5 5 5 6 6 6 7 7 8 8 8 9 9 0 0 0 1 1 2 2 2 3 3 4 4 4 5 5 6 6 6 7 7 8 8 8 9 9 0 0 8 8 8 8 8 8 8 8 8 8 8 8 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 1 5 9 3 7 1 5 9 3 7 1 5 9 3 7 1 5 9 3 7 1 5 9 3 7 1 5 9 3 7 1 5 9 3 7 1 5 9 3 5 5 5 6 6 7 7 7 8 8 9 9 9 0 0 1 1 1 2 2 3 3 3 4 4 5 5 5 6 6 7 7 7 8 8 9 9 9 0 8 8 8 8 8 8 8 8 8 8 8 8 8 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 An Optum Company Images courtesy of Harry Burns, MD Scotland NHS 4 2

5/31/2017 The coherence theory of human health For health to exist, life must be: This leads to motivation & engagement • Comprehensible • Manageable • Meaningful When any of these do not exist, it leads to chronic stress Aaron Antonovsky An Optum Company 5 Evidence from all corners – environmental and psychological stress translates into physical stress 00-046 The Founders’ Network Evening Cortisol Levels Increase with Months of Orphanage Rearing * -0.2 -0.4 -0.6 -0.8 *linear trendline -1 -1.2 0 10 20 30 40 50 Months of Orphanage Rearing Stress and grade of employment: men 20 18 Salivary cortisol levels 16 14 12 nmol/l Higher Grade 10 Lower Grade 8 6 4 2 0 08-8.30 10-10.30 12-12.30 14-14.30 16-16.30 18-18.30 20-20.30 22-22.30 Time of day An Optum Company Images courtesy of Harry Burns, MD Scotland NHS 6 3

5/31/2017 Engaging patient reported data helps us move upstream A window into the health, wellness, and preparedness of the whole person to not only deal with, but thrive within, their life circumstance will help us better address that person’s health needs An Optum Company 7 What are ACOs focusing on today? HCC Code Accuracy Complex Gaps In Reduction of Care Care Preventable Managemen Managemen Hospitalizations t t Readmission Prevention Transitions Scorecards in Care & Managemen Dashboards t Identification Patients of High-Cost Missing Patients Basic Care An Optum Company 8 4

5/31/2017 Risk characterization and stratification are common building blocks for prospective patient care What can I do to lower Who is highest risk? that risk? For how long have they had this risk profile? Why are they this way – what is driving that risk? An Optum Company 9 A brief retrospective of risk stratification techniques reveals a focus on increasing precision, timeliness, and relevance Rising risk – single scale While this methodology is simple to implement and High (3-5%) understand, it has disadvantages: Rising (10-15%) • It is backward looking and not particularly predictive Well • “Risk” or outcome not clearly defined – thus people tend to mean different things when they talk about it • Metrics designed to capture patients likely to need, and A typical methodology – 12 point scale respond to, intervention can end up filtering out patients • Age who still have very high risk of bad outcomes • Utilization (IP, ED, ambulatory) • Current active meds • Chronic conditions • Some include concepts like disease control, care gap counts and other measures for “actionability” or relevance to improve the relevance of the derived patient list An Optum Company 10 5

5/31/2017 A real world patient population selected using these criteria will have meaningful utilization and chronic disease Average age of 71 3 ED visits and 2 hospitalizations in the last year Saw an outpatient provider 10 times in the last year Taking 14 different medications Has 2 chronic conditions Costs about 1.5-2x the average beneficiary An Optum Company 11 Predictive models introduce a prospective view on outcomes of inherent interest Prediction has introduced a number of benefits to the risk stratification process: Likelihood Future of • Far superior precision in identifying target groups (capture cost hospital- >50% of events in top 10-15% of patients) ization • Model event-based risk profiles with higher fidelity High cost with contribution from hospitalizations Initial hospitalization Readmission Patients by risk of CHF-related hospitalization 100 80 Period of heightened 60 vulnerability & 40 Patient count management needs 20 0 0-79% 80-89% 90-95% >95% Time An Optum Company 12 6

5/31/2017 Clinical data greatly improves predictive model precision What drives a claims model for heart failure What drives a clinical model for heart failure admission? admission? • Historical heart failure admissions • Historical heart failure admissions • Historical non-heart failure admissions • Historical non-heart failure admissions • General utilization profile • General utilization profile • Presence of heart failure diagnosis • Time of first heart failure diagnosis • Presence of comorbidities (atrial fibrillation, mitral • Most recent ejection fraction valve disease, conduction disorders) • Recent BNP value • Relative change in BNP over time • Patterns of treatment (i.e. beta blockers, diuretics and nitrates) • Markers of kidney and pulmonary function An Optum Company 13 Active Patients Pts. w/ CHF-Related Total CHF-Related with CHF IP Admissions IP Admissions An Optum Company 14 7

5/31/2017 A predictive methodology will yield a population that is reasonably similar by age, utilization history, and comorbidity profile A typical methodology – 12 point scale Methodology driven by likelihood of admission • Age • Age • Utilization (IP, ED, ambulatory) • Utilization (IP, ED, ambulatory) • Current active meds • Current active meds • Chronic conditions • Chronic conditions • Laboratory values 71 74 Average age 3 3 ED visits 2 2 Hospitalizations 10 11 Outpatient visits 2.0 1.8 Charlson comorbidity index An Optum Company 15 …The previously described historically-based methodology captures only 40% of the patients at highest risk for admission Highest admission risk patients captured by the non-predictive methodology Captured Not captured An Optum Company 16 8

5/31/2017 Five key social determinants of health – not well documented in clinical data • Poverty Economic stability • Employment • Food security • Housing stability Education • High school graduation • Enrollment in higher education • Language and literacy • Early childhood education and development Social and community context • Social cohesion • Civic participation • Discrimination • Incarceration • Access to health care Health and health care • Access to primary care • Health literacy • Access to healthy food Neighborhood and built environment • Quality of housing • Crime and violence • Environmental conditions An Optum Company 17 IOM recommendations for collection of social data An Optum Company 18 9