Carolinas HealthCare System: Consumer Analytics

Carolinas HealthCare System: Consumer Analytics

9-515-060 R E V : S E P T E M B E R 8 , 2 0 1 5

Professor John A. Quelch and Research Associate Margaret L. Rodriguez prepared this case. It was reviewed and approved before publication by a company designate. Funding for the development of this case was provided by Harvard Business School and not by the company. . Professor Quelch is the Charles Edward Wilson Professor of Business Administration at the Harvard Business School and Professor in Health Policy and Management at the Harvard T.H. Chan School of Public Health. HBS cases are developed solely as the basis for class discussion. Cases are not intended to serve as endorsements, sources of primary data, or illustrations of effective or ineffective management. Copyright © 2015 President and Fellows of Harvard College. To order copies or request permission to reproduce materials, call 1-800-545-7685, write Harvard Business School Publishing, Boston, MA 02163, or go to www.hbsp.harvard.edu. This publication may not be digitized, photocopied, or otherwise reproduced, posted, or transmitted, without the permission of Harvard Business School.

J O H N A . Q U E L C H

M A R G A R E T L . R O D R I G U E Z

Carolinas HealthCare System: Consumer Analytics

In 2014, Dr. Michael Dulin, chief clinical officer for analytics and outcomes research and head of the Dickson Advanced Analytics (DA2) group at Carolinas HealthCare System (CHS), was preparing for a planning meeting with Carol Lovin, executive vice president and chief strategy officer at CHS. In the three years since DA2 was formed, the team had successfully unified all analytics talent and resources into one group that served CHS. Rapid increases in computing power and decreases in data storage costs had enabled DA2’s data architects to build predictive models incorporating complex clinical, financial, demographic, and claims data that would have been impossible to create only a few years before. Although DA2 had blazed the trail for applied analytics in healthcare, other players in the value chain were making increased investments in their own modeling capabilities.

Healthcare payers, such as Humana and UnitedHealth, were increasingly making analytics the focus of a strategic shift toward consumer-centric healthcare, going so far as to create targeted communications strategies for different patient segments and engaging behavioral health companies to provide exercise, nutrition, and other programs that would reduce the healthcare costs of their highest-risk patients. While many agreed that analytics could help the healthcare industry reduce costs and increase access to care, CHS recognized that privacy protections on patient data, as well as competitive rivalries, restricted the sharing of data among the various healthcare stakeholders.

Dulin also noted the entry of consumer tech companies into the healthcare space; in 2014, both Apple and Google announced features in their new mobile operating systems that aggregated and tracked the output from various health wearables (like heart-rate monitors or step counters), as well as electronic medical record (EMR) data. Apple’s HealthKit could even incorporate the results of lab tests into the dashboard (with the user’s permission). Although the tech giants did not yet have access to claims or clinical data, they could potentially enter the field by acquiring an EMR company. Their expertise in analytics, access to demographic and location data, as well as the broad consumer adoption of their devices, led Dulin to consider which industry players consumers would trust to integrate their healthcare data in the future and what role DA2 could play.

This document is authorized for use only by Kaili Yun (kyun9@outlook.com). Copying or posting is an infringement of copyright. Please contact customerservice@harvardbusiness.org or 800-988-0886 for additional copies.

515-060 Carolinas HealthCare System: Consumer Analytics

2

Carolinas HealthCare System Background

In 2014, the Carolinas HealthCare System, headquartered in Charlotte, North Carolina, owned and managed hospitals and acute care facilities that served over 2.2 million patientsa per year across three states (North Carolina, South Carolina, and Georgia). CHS was one of the oldest healthcare systems in the U.S.; its origins could be traced to the state’s first civilian hospital, Charlotte Home and Hospital, established in 1876. The Charlotte Home and Hospital operated until 1940 (although the name was later changed to St. Peter’s Hospital), when it was replaced by a new facility, the Charlotte Memorial Hospital, in a different location. During World War II, the hospital’s financial difficulties led Rush S. Dickson, a local businessman, to lobby the city and county governments for larger reimbursements for emergency and indigent patients, and solicited financial support from local nonprofits and corporations. In 1943, the Charlotte-Mecklenburg Hospital Authority was organized under the North Carolina Hospital Authorities Act, which provided for oversight mechanisms for Charlotte Memorial Hospital (including rules governing the construction of new facilities, funding, and management of day-to-day operations). The act also provided the legal and financial frameworks to support patients who could not afford to pay for healthcare services.

In 1990, the name of the Charlotte Memorial Hospital was changed to Carolinas Medical Center (CMC) to reflect the hospital’s increasing focus on education. That year, the facility was designated an “Academic Medical Center Teaching Hospital” by the state of North Carolina (one of only five hospitals in North Carolina to receive the designation). Five years later, the authority changed the name of the growing hospital network to CHS. In 2007, CHS opened the Levine Children’s Hospital, which housed more than 30 medical specialties. In 2010, CHS announced a 10-year, $500 million investment to advance cancer treatment strategies and research through the creation of the Levine Cancer Institute. In 2010, CMC (now a part of CHS) was designated the Charlotte Campus of the University of North Carolina (UNC) School of Medicine and hosted third- and fourth-year medical students.

By 2014, CHS had become the biggest healthcare provider in North Carolina, with more than 61,000 full-time and part-time employees and an annual budget of over $7.7 billion (see Exhibit 1 for selected financial data). Its medical education and research center included over 300 residents and fellows pursuing a variety of medical specialties and had established research relationships with Oxford (stroke), UNC (dementia), Duke University, and many other academic centers across the U.S. CHS operated 900 care locations and 7,494 licensed beds in three states, including 39 hospitals (21 of which were managed by CHS, and 18 of which were owned), as well as additional virtual care services. Roughly 75% of patients were located in North Carolina, and CHS spent $20 million each year on community outreach in the greater Charlotte area alone. CHS tracked patient satisfaction with mailed surveys or follow-ups within days of an appointment or discharge. Satisfaction was measured by a patient’s likelihood to recommend CHS. As part of its role as a public healthcare system, CHS provided healthcare services to underserved patients and communities. CHS offered financial support to patients without insurance (or who were underinsured), subsidies for Medicare and Medicaid recipients, as well as funding for its education, behavioral health, and community health clinics. CHS gave medical supplies and equipment to nonprofits valued at over $1.5 million in 2013 (see Exhibit 2 for a full list of CHS charitable expenditures for 2013). Roughly 62% of annual revenue came from Medicare and Medicaid patients.

a Patients were considered active if they had engaged with one of the CHS sites (including primary care facility, hospital, worksite clinic, a virtual visit, or a trip to a clinic inside of CVS) at least once in the prior 18 months. Roughly 13,000 patients fell off the active rolls each month, many due to relocating out of state, death, or attrition.

This document is authorized for use only by Kaili Yun (kyun9@outlook.com). Copying or posting is an infringement of copyright. Please contact customerservice@harvardbusiness.org or 800-988-0886 for additional copies.

Carolinas HealthCare System: Consumer Analytics 515-060

3

By 2014, the vision of CHS had remained unchanged for two decades: healthcare, education, and research. That year, Lovin led an initiative to renew the strategic road map to guide CHS’s future growth. She worked with her executive team colleagues to craft a strategy that provided for personalized, high-quality service across a single, unified enterprise. “Our customers are the consumers and patients first, payers second,” said Lovin. The team developed a list of strategic priorities (see Exhibit 3) and performance measures (see Exhibit 4) to guide the organization toward its goals.

Health Provider Industry Background

In 2012, healthcare expenditures in the U.S. totaled over $7,600 per capita versus an average of

$2,800 per capita among OECD countries.1 The majority of healthcare expenses were paid for by

government-sponsored coverage, such as Medicare and Medicaid,b or by health insurance companies that sold plans to employers and individuals (either to those who purchased via healthcare exchanges

or those who were eligible for Medicare Advantagec). Government reimbursements per Medicare patient to cover healthcare had declined over time (see Exhibit 5); however, the 2010 Affordable Care

Act (ACA) offered providers who were organized as accountable care organizations (ACOs)d a share of the cost savings generated in the delivery of care to Medicare patients, so long as minimum quality thresholds were met.

Key changes to the U.S. healthcare landscape over the decade prior had influenced healthcare providers like CHS and other healthcare stakeholders to revisit their care delivery models:

 Fee-for-value instead of fee-for-service After the ACA’s passage, hospital compensation was determined in part by the quality of outcomes, rather than simply on a fee-for-service basis as before. Hospitals faced penalties for high readmission rates and hospital-acquired conditions, but could also receive financial rewards for exceeding clinical quality outcome or patient satisfaction benchmarks. Many healthcare providers sought to quickly build capabilities in analytics and measurement in order to track quality improvements, and to shift the organizational focus toward the continuous improvement of patient care.

 Physician shortages In the U.S., there were roughly 2.5 physicians per 1,000 people (versus an average of 3.3 per 1,000 among comparable OECD countries).2 The ACA was expected to exacerbate the supply-demand shortfall in the future, since it increased the population covered by health insurance. In 2014 alone, nearly 32 million new people entered the healthcare system.3 The Association of American Medical Colleges estimated that, by 2025, the

U.S. would face a shortage of over 130,000 doctors.4

 Digitization of healthcare In 2011, the U.S. Centers for Medicare and Medicaid Services (CMS) established an incentive system for doctors’ offices and hospitals to switch from paper to EMRs. Hospitals that served Medicare patients could receive up to a $2 million incentive

for adopting EMRs.5 Although patients were free to view and request corrections to the data in their EMRs, the platforms made it difficult for providers to extract and model the data held

b Medicare was the federal health insurance program offered to seniors aged 65 and older. Medicaid was offered to low- income individuals who were unable to obtain healthcare via the exchanges or an employer.

c Insurers offered Medicare Advantage plans to eligible patients who chose to receive their benefits via the insurer’s network.

d Qualified ACOs agreed to be accountable for the overall care of their Medicare patients, obtain adequate participation of primary care physicians, create processes around evidence-based medicine, report on quality and costs, and coordinate care.

This document is authorized for use only by Kaili Yun (kyun9@outlook.com). Copying or posting is an infringement of copyright. Please contact customerservice@harvardbusiness.org or 800-988-0886 for additional copies.

515-060 Carolinas HealthCare System: Consumer Analytics

4

in the EMRs. In 2014, Google and Apple each announced healthcare dashboards that would aggregate data from wearable devices, including scales, running apps, and sleep trackers, into a consumer-friendly dashboard. Apple even partnered with several EMR companies to make medical records and lab results available to consumers on their iPhones via the single dashboard.

 Shift to outpatient In an effort to cut costs and increase access to care, providers encouraged patients to seek care via outpatient facilities, or even via virtual checkups, rather than at high- cost treatment locations like emergency departments (EDs). Outpatient care comprised 51% of health expenditures in the U.S., whereas the average among OECD countries was 33%.6 In 2013, CHS chief executive Michael Tarwater observed: “More than 90% of our patient encounters now take place in a setting other than the bedside of an inpatient hospital room.”7

 New entrants The shift to outpatient care led consumers to seek more convenient and inexpensive healthcare services. Retailers (including CVS, Walmart, Target, and Kroger) began opening healthcare clinics staffed by nurse practitioners in their stores as early as 2000. By 2014, there were 1,600 walk-in clinics in the U.S., and the number was expected to reach

nearly 3,000 by 2015.8 Cost of care at the clinics for three common illnesses averaged $110, versus $166 at doctors’ offices and $570 in EDs.9

HIPAA Regulations

In 1996, the U.S. House of Representatives passed the Health Insurance Portability and Accountability Act (HIPAA). HIPAA provided for both the portability of employer-provided health insurance (which enabled an individual to keep the same health insurance between jobs) and the establishment of the first set of national security and confidentiality standards for patient health

data.10 The U.S. Department of Health and Human Services (HHS) established a privacy rule to protect individually identifiable patient data, while still enabling stakeholders to access the data needed to provide care. Data protected under HIPAA included physical or mental health conditions (including those that occurred in the past), healthcare provided, payments made for healthcare received, and demographic information that could be used to identify the individual (see Exhibit 6 for examples).

Health plans (payers), healthcare providers, and healthcare “clearinghouses” (which included billing services, community health management information systems, value-added networks, and other business associates) were all subject to the privacy standards outlined in HIPAA. Healthcare organizations had to notify patients of their privacy rights (including acceptable use of personally identifiable information) and obtain signed authorization from patients for any use of individual data

beyond treatment, payment, and healthcare operations.11 Restrictions on data use could be waived if data were “de-identified,” either by the formal assessment of a statistician to prove individual anonymity was retained, or by the removal of indicators used to identify the individual and his or her relatives, employer, or household members.12 De-identified data became propriety to the company that held it.

HIPAA supported use of patient data to perform analysis necessary to make improvements to healthcare systems, including quality reviews, utilization reviews, and population reviews (often for a given condition, such as diabetes). Such information could be shared with other entities also subject to HIPAA, such as payers. Employers like CHS, who both provided healthcare and self-funded insurance to their employees, were not permitted to view their employees’ disaggregated healthcare data; in addition, employee information collected through the human resources department was kept

This document is authorized for use only by Kaili Yun (kyun9@outlook.com). Copying or posting is an infringement of copyright. Please contact customerservice@harvardbusiness.org or 800-988-0886 for additional copies.

Carolinas HealthCare System: Consumer Analytics 515-060

5

separately from employee health data. Those who contravened the HIPAA privacy rule could be subject to civil and/or criminal charges and fines of over $1 million.13

Dickson Advanced Analytics (DA2)

Following CMC’s designation as a teaching hospital in 1990, CHS established partnerships with 11 academic research centers. There was no medical school located in Charlotte at the time, so one of the independent centers was designated a branch of UNC. The Dickson Institute for Health Studies, as the center was known, provided education and training facilities to 300 residents, nurses, and graduate students. It partnered with UNC–Charlotte and UNC–Chapel Hill to provide health-data- focused projects to PhD candidates. The Dickson Institute initially focused on improving acute care quality, but the mission was later broadened to cover an array of healthcare projects under the leadership of Dr. Roger Ray, executive vice president and chief physician executive at CHS. Although the Dickson Institute conducted research, data analysis, and public reporting of key metrics for CHS, Ray noted that its activities did not influence the majority of day-to-day operations that occurred within the CHS network.

Beginning in the 2000s, CHS embarked on a visioning and process-development project to determine what data analytics capabilities would be integral to CHS’s operations in the future. It determined it would need to develop a distributed data system and create a corporate data warehouse and decided to coalesce analytics personnel who were currently working in small silos throughout the organization to achieve the vision. CHS Information Services leadership anticipated that cost of data storage would plummet, based upon their experience implementing the EMR system at CHS in 2006, so the team decided to build generous data storage to support the new analytics team. Prior to CHS’s adoption of EMR, most of the data it collected was financial data generated through transactions. The EMR rollout served as a proof-of-concept that patient data and financial data could be combined to provide decision support. With increasing computing power, CHS and other providers could collect, store, and model a variety of clinical data (including unstructured data) that it could not have assembled previously.

CHS hired consultants to advise the organization on the creation of a unified analytics group through the development of a high-level road map. Lovin was interviewed by the consultant group and asked to provide executive leadership for the initiative, and she recruited Dulin to help execute on CHS’s vision of creating a unified, data-driven system. Dulin was trained in electrical and biomedical engineering and worked as a quality control specialist for a microchip manufacturer before attending medical school. At the time, there were many groups within CHS that handled analytics, but most were tied to a particular business, function, or geography with no integration. CHS decided to differentiate on the basis of its analytics capability and made investments to raise the analytics “IQ” of the organization. CHS leadership and others believed the system could move beyond its current analytics-related key performance indicators (KPIs) to include data in key decisions that would change patient care and save money over the long term. The new analytics group could have adopted a hybrid model structure wherein descriptive analysis linked to performance metrics was performed internally and more rigorous analytics were outsourced, but CHS instead chose to develop both its foundational descriptive analytics and more advanced predictive and prescriptive models in-house.

Creation of DA2

Dickson Advanced Analytics (DA2) was launched in 2011 with an annual budget of $14 million. It was initially composed of 70 people sourced from the disparate internal analytics groups that

This document is authorized for use only by Kaili Yun (kyun9@outlook.com). Copying or posting is an infringement of copyright. Please contact customerservice@harvardbusiness.org or 800-988-0886 for additional copies.

515-060 Carolinas HealthCare System: Consumer Analytics

6

preceded DA2. Much of DA2’s capacity was devoted to providing tools to support CHS-affiliated hospitals in delivering best-in-class healthcare to patients, although, over time, D