Image: loveguli / Getty Images The role of data analytics in present and future claims cost containment Dr. Andreas Bayerstadler Senior Consultant Business Analytics Dubai, 12 th September 2018
Wearables and Digital Health Services will change the medical world 500 million wearable devices will be sold by 2020» monitoring daily activities and providing analyses and instructions» customisation of products and services 40% estimated cut of healthcare costs by 2022 through digital health» customer centricity» interconnection with health ecosystem accessible by physicians to have a complete picture of the patient» prevention of diseases, real-time health records etc. Image: loveguli / Getty Images 2
Bringing together data and technology with highly skilled people and advanced methods generates value Data & Technology Skills & Methods Value Image: loveguli / Getty Images 3
How Data we and at technology Munich Re is are the applying crucial basis this formula future success Data Lake Big Data & Analytics Platform Data Hunting Data & Technology Data Democracy - Simple Access Skills & Methods Data Governance Value Image: loveguli / Getty Images 4
Additionally we need an outstanding skillset throughout the company Recruiting Talent sourcing Training of existing staff Skills & Methods Statistical modelling & ML capabilities Artificial intelligence & DL capabilities Further data-related capabilities Image: loveguli / Getty Images 5
On this basis, Munich Re will create new strategic options and gain substantial business advantage Value Innovation Enablement IoT (e.g. Telematics, Wearables) Cyber Digitally Augmented UW & Claims Data Analytics Solutions & Digital Cooperation Models Analytics platforms Strategic Analytics consulting AI / Analytics use cases & pilots Image: loveguli / Getty Images Analytics @ Munich Re 6
Health Analytics Suite Comprehensive Data Analytics Platform for primary health insurers APPROACH BENEFITS Coverage of the whole primary insurance value chain Implemented analytics solutions can be applied without additional IT investments and very low efforts Advanced Analytics is used to gain new insights in the portfolio, individual members and individual providers www.munichre.com/analytics-suite Results are prepared in very easy-to-use visualisations through dynamic reports Image: Ipopba / Getty Images 7
Analytics solutions support operations within all areas of the health insurance value chain Pricing & product development Sales and marketing Medical underwriting Claims & network management Re-pricing of existing health plans or existing individual risks Pricing of new health plans or new individual risks New product development based on big data Identification of new potential clients for health products Identification of cross- and up-selling potential for existing business Prevention of lapses & win-back opportunities Digital analytics Market specific adjustment of global medical underwriting solutions Proof of existing loadings for regulator Optimization of UW processes Measuring quality of providers Steering DMPs Identifying fraud and abuse Optimization of Claims processes Creating rules for automated claims evaluation Goal: Manage top- and bottom-line development in health business Premium growth: Optimized existing health plans Identified new sales potentials Medical cost control: Optimized control of utilization Decelerated increase of average claims costs 8
Fraud and Abuse Identification of systematic fraud and abuse RESULTS BENEFITS Identification of complex patterns (over time) and consolidation to meaningful provider scores High increase of efficiency in fraud and abuse detection Optimization of provider investigations Substantial strengthening of the position in provider network negotiations Image: Ipopba / Getty Images 9
Member 1 Member 2 Member 1 Member 7 Member 1 Member 2 Member 6 Member 7 Member 3 Member 2 Member 5 Member 1 Member 3 Member 2 Member 5 Member 8 Fraud and Abuse Major challenge is the identification of systematic behaviour Classical claims adjudication Analytical fraud and abuse detection Provider A Provider A Provider B Provider B Invoice-by-invoice perspective Time Time series perspective Time No irregularities Same-time treatment and kickback -referral Analytical fraud and abuse detection facilitates a holistic view on systematic behavior of different players in health insurance 10
Real time Retrospective Fraud and Abuse To identify systematic fraud and abuse, several components need to play together Deterministic Probabilistic Reporting factory (more than 150 reports in 4 dimensions) Technical rules Medical rules STOP Experience based rules Regulatory rules Live scores (for each incoming claim in 3 categories) 11
Fraud and Abuse The reporting factory is measuring historic provider behaviour Deterministic reports Probabilistic reports Simple reporting methods focusing on detecting fraud and abuse Reporting categories: Service providers Insured members Social networks General F&A statistics Examples: Fraud and abuse frequencies Frequent referral constellations Large number of treatments etc. More complex statistical techniques focusing on unusual behaviour Four reporting categories, as with standard reports Examples: Risk-adjusted claimed amounts Length of stay per service provider using regression Claims distribution using clustering and factor analysis etc. 12
Examples Fraud and Abuse The first component of real-time claims scoring are various rule sets Technical rules Examples: Bundling / unbundling Mutual exclusions Waiting periods / treatment intervals Once-in-a-lifetime procedures... Medical rules Experience based rules Examples: Missing follow-up treatment Overnight stays for outpatient treatment Consulting time limits Parallel inpatient and outpatient treatment... Regulatory rules Rule Type Diagnosis/ Treatment Appendicitis & appendectomy Appendicitis & tissue sample Diagnosis/ Drug Respiratory infection & Augmentin tb. Respiratory infection & Norvasc tb. Treatment/ Gender Ultrasound prostate & male Ultrasound prostate & female Examples: Lab / radiology included in DRG (cannot be charged separately) Limited length of stay according to DRG Private vs. public health cover. 13
Fraud and Abuse The second component are scores derived from analytical models Step 0: Retrospective reports Historic F&A cases Historic claims data Provider reports Member reports Network reports Further F&A statistics Rep. Rep. Rep. Rep. Rep. Rep. Rep. Rep. Rep. Rep. Rep. Rep. Rep. Rep. Rep. Rep. Step 1: Aggregation Provider scores Member scores Network scores Claim statistics Step 2: Scoring Invoice N Invoice 2 Invoice 1 Provider information Member information Invoice information Policy information Analytical model Probabilistic scores for each invoice Depending on scoring result: Manual F&A evaluation or usual claims processing 14
Fraud and Abuse All these components lead to a more targeted investigation process The identification of fraud and abuse can be compared with searching a needle in the haystack. Analytical methods will not be able to find the needle Image: used under license from Shutterstock.com Image: used under license from Shutterstock.com however they can make the haystack considerably smaller! 15
Medical Quality Individual provider and hospital assessment RESULTS BENEFITS Medical Quality assessment combines medical expertise and data analytics The provider quality score includes complex adjustments of the health status and demographics of the patient as well as surgeryspecific risks Active steering to better performing providers for elective surgeries New quality insights which substantially strengthen the position in provider negotiations Image: Ipopba / Getty Images 16
Foreseeable surgeries Medical Quality Areas of treatment where quality measurement is particularly useful Orthopedics Knee surgery Hip surgery Shoulder surgery Spinal disc surgery (General) surgery Colon and rectum surgery in case of colorectal cancer Appendectomy Removal of gallbladder in case of gallstones Removal of the thyroid Cardiac & vascular surgery Angioplasty (primarily stent angioplasty)
Foreseeable surgeries Medical Quality The basic idea is the development of so called quality measures Orthopedics Quality profiles Knee surgery Quality measure 1: Re-admission Quality measure 2: Bacterial inflammation Quality measure n: Follow-up costs Hip surgery Quality measure 1: Repeated surgery Quality measure 2: LOS Quality measure n: Use of painkillers over long period of time Spinal disc surgery Quality measure 1: Athetosis Quality measure 2: Paralysis / Nerve damage 18
Medical Quality Quality Measures (QM) are combined to a provider performance score Goal: Risk-adjusted performance measures for providers based on claims data Claims data Examples: Surgery for colon carcinoma Total knee/hip arthroplasty Examples: Mortality Number of complications Resubmission rates Claimed amount Relevant procedures Key quality indicators Procedure specific ranked lists of providers Hospital D Hospital C Hospital A Hospital F Hospital B Hospital E List of providers 19
Conclusions and Outlook Where will claims management go in the future? 1 New medical devices will increase the availability of (external) data 2 Data Analytics will reveal new saving potentials in claims and network management 3 Investing in Analytics and digital cooperation models is a must for (health) insurers 4 The combination of medical, insurance and analytical expertise is key 20
Innovation Enablement Data Analytics Solutions & Digital Cooperation Models Medical Do you have Quality any questions? Individual provider and hospital assessment RESULTS BENEFITS Medical Quality assessment combines medical expertise and data analytics Data & Technology Contact details: Dr. Andreas Bayerstadler abayerstadler@munichre.com www.munichre.com Image: loveguli / Getty Images Skills & Methods The provider quality score includes complex adjustments of the health status and demographics of the patient as well as surgeryspecific risks Value Active steering to better performing providers for elective surgeries New quality insights which substantially https://edoc.ub.uni- strengthen the position in provider negotiations muenchen.de/21687/1/b ayerstadler_andreas.pdf 21