promising practice pathways€¦ · includes “entry options” for pathology practices without...

cap.org© 2012 College of American Pathologists.

Promising PrACtiCe PAthwAys™

cap.orgPromising Practice Pathways™

© 2012 College of American Pathologists.

Coordinated Population Care services

Promising Practice Pathways™

College of American Pathologists180

The new performance-based de-mands of health care delivery reform will ultimately shift compensation to “alternative,” value-based purchas-ing arrangements such as shared savings, bundled, or global payment. These changes are driving health care systems and physician groups, such as Accountable Care Organiza-tions (ACOs), toward a new paradigm called population health manage-ment, a model that holds the ACO accountable for both the health and costs of an assigned patient popula-tion.

This new care delivery model de-mands that ACOs perform compre-hensive performance measurements, including process of care, outcomes, and patient experience metrics. To im-prove quality and reduce costs, ACOs need to both systematize care with pathways and embed “individualized medicine” into routine care. They also need to segment their populations and predict risks to direct interven-tions, improve health, and avoid pre-ventable illnesses and complications.

Population health management (PHM) is challenging because it spans three dimensions of care. As shown in Figure 1, PHM requires 1) managing populations that range from the com-munity down to a physician’s prac-tice, 2) managing care along the time continuum from prevention through

end of life, and 3) managing across all conditions and procedures.

Pathology and laboratory medicine are competitively advantaged in hav-ing expertise and services that add value across all three of these dimen-sions, arguably uniquely so in medi-cine.

Furthermore, information technology systems and robust, thorough clinical data are critical components in this new model. But beyond IT solutions, ACOs are looking for guidelines, order sets, and other electronic medical record (EMR) applications for which pathology knowledge is a necessity. These ACO PHM needs provide an opportunity for pathology and labora-tory medicine to develop services and EMR/EHR applications that go beyond supplying laboratory results.

The key to meeting these needs intel-ligently involves a dual approach that is unique to pathology. Pathologists can marry “top down” population lab informatics applications that sup-port systematizing care with “bottom up,” patient-specific, molecular test-ing tools that can individualize those guidelines in a way that accounts for the art and variability of medicine. This is our second, critical competitive advantage in delivering PHM value to ACOs. See Figure 2.

eXeCUtiVe sUmmAry

College of American Pathologists


181

ACOs and other providers engaged in PHM need not only our data, but our expertise and services to generate the value they need to be success-ful under these new payment models. But the move to ACOs has significant momentum and the models and roles supporting their needs are becom-ing established rapidly. Without taking action to develop these services and applications that extend beyond test results, ACOs will sideline us as provid-ers of data.

ACOs are now estimated to number 3001, and are projected by some ana-lysts to double to 600 in 2013.2 That implies that we will all be exposed in some manner to ACOs of some type in the near term. “You’re either at the table or on the table” is a favorite phrase among the health care market stakeholders immersed in reform these days.

We have tremendous value to bring to the ACO table in terms of PHM ser-vices, but we need to take action to do so. We do not need to have infor-matics capabilities to start this journey, for we will have resources and guides along the way to build them, and there are many services we can offer that do not require informatics.

The “Coordinated Population Care Ser-vices Practice Pathway” we introduce

includes “entry options” for pathology practices without informatics expertise, which can begin by adding value in their hospitals, for example, through population health applications and services we’ll describe. Many of these are collaborative programs in which the pathologist’s expertise adds a vital component that can greatly improve the initiative’s impact.

For example, one of the leading op-portunities for cost avoidance for ACOs is infectious diseases, particularly given the ubiquity of resistant patho-gens, both inpatient and in community settings. Pathologists can play a sig-nificant population health, value-gen-erating role in teaming with infectious and pharmacy colleagues to develop high-powered surveillance and stew-ardship programs, including the target-ed use of molecular testing to guide antibiotic use.

There are a number of other opportu-nities for pathology groups who have not yet acquired informatics capabili-ties that rely solely on our expertise. For example, optimal management of even moderate glycemic control targets in surgical patients is challeng-ing, as hypoglycemic events drive avoidable costs and adverse events. One problem that contributes to this situation is that some of the hand-held glucometers widely used in hospitals

1Susan Dentzer, Editor of Health Affairs, remarks at the Third National Accountable Care Organization Summit in June 20122Jim Bloedau, “ACO Market Will More Than Double in 2013,” Information Advantage Group, August 9, 2012



have clinically significant accuracy problems. Pathologists could play a role in ensuring the quality and accu-racy of these and other point-of-care testing devices throughout an ACO.

This glucometer example may seem like just a routine service, but we need to understand and quantify the tre-mendous clinical quality and econom-ic value this generates and develop this one example into a population-wide program that we lead. This point-of-care testing program might involve strategic recommendations regarding when and where to introduce such testing (including home care), selec-tion of instruments, certification, as well as data collection of some sort. In this way, we have evolved a simple quality responsibility into a valuable popula-tion health management program that can be measured for results and com-pensated based on the value it gener-ates.

This chapter is designed as an intro-ductory guide to this population health management opportunity, offering context on population health man-agement, explaining the population

care needs of ACOs, exploring the business dynamics and opportunity size, and describing the pathway at a high level.

The guide is structured as follows:

rationale

1) Market Need: ACOs Need Pathologists to Manage Population Health

2) Market Opportunity: Advancing Population Health Management

Description

3) About the "Coordinated Population Care Services Practice Pathway"

4) Strategic Planning

implementation

5) Launch Phase 2014-2016 Pathway Planning Guide

6) Evolution Phase 2017-2019 Planning Guide

7) Expansion Phase 2020-2022 Planning Guide



183

Figure 1: Population health management spans three Dimensions of Care

Populations:Communityto Practice

Conditions andProcedures:Diabetes toCABG

Time:Preventionto End ofLife

PopulationHealth

Management

Source: Eleanor Herriman, MD, MBA, G2 Intelligence



Figure 2: Coordinated Population Care Practice Pathway’s Dual Approach: Population informatics and individualized testing

Population-Based:Lab Informatics• Patterns for identifying risk• Algorithms inform guidelines• Statistics for performance management

Individualized:Molecular or Biomarker Testing• Results and algorithms identify risk• Results individualized guidelines/interventions

Guidelinessystematize

care

Molecularindividualizes

care




185

Population health management: A Paradigm shift in Care DeliveryA major component of the current U.S. health care market reinvention is care delivery reform, and one of the pri-mary new paradigms in care delivery is population health management (PHM). Providers being compensated under “alternative payment” programs such as ACOs need PHM by definition; they are responsible for both the health and costs of an assigned patient population.

“To cope with these new de-mands, health care systems and physician groups are moving toward an approach known as ‘population health management.’ The goal of population health management (PHM) is to keep a patient population as healthy as possible, minimizing the need for expensive interventions such as emergency department visits, hospitalizations, imaging tests, and procedures.”3

To manage population care and suc-cessfully operate this business model, an ACO must be able to perform comprehensive performance mea-

surements, including process of care, outcomes, and patient experience metrics. In order to qualify for Medi-care shared savings payments, ACOs are required to meet specified thresh-olds on a set of 33 such quality perfor-mance measures. They also, of course, need to measure and project their costs.

When we look at the span of condi-tions, providers, and settings that need to be measured in order to be man-aged by ACOs, or indeed by any pro-vider organization that is being com-pensated based on an “alternative” or value-based purchasing arrangement (e.g., bundled or global payment), it is quite evident that potent informa-tion technology systems and robust, abundant clinical data are critical to success.

information technology: essential for Population health managementA variety of electronic health record (EHR) tools and informatics database applications have become important for ACOs and Medical Homes and, so far, these applications have helped enable PHM:

mArket neeD – ACos neeD PAthoLogists to mAnAge PoPULAtion heALth

3“Population Health Management: A Roadmap for Provider-Based Automation in a New Era of Healthcare,” Institute for Health Technology Transformation, 2012



• Describing his experience with a partnership ACO with Norton Healthcare, launched in 2010, Dr. Tom James, Humana’s corporate medical director, said, “ACOs are really changing our work from (health) utilization management to more data and information sharing. It is more extensive than we have ever seen before.”4

• The Institute for Health Technology Transformation puts it very suc-cinctly: “Automation makes popu-lation health management fea-sible, scalable and sustainable.”5

The majority of ACOs using IT tools in a PHM setting recognize that pathology and laboratory testing results are an important component of these data-bases and tools. Yet without involving pathologists in the analysis and utiliza-tion of this information, these organiza-tions are working below full potential.

What ACOs really need are standard-ized and coordinated applications for implementing the insights generated by the laboratory test informatics. Par-ticularly they need to imbed “individu-alized medicine” into routine care. This translates into guidelines, order sets, and other EMR applications for which pathology knowledge is a necessity.

systematization with intelligenceSuch systematization, however, may not be immediately embraced by physicians, especially if they perceive that standardized guidelines and co-ordinated IT applications equate to “managed care” medicine. Yet the reality is that physicians cannot care for individuals without the collective intelligence of the entire health care system, from appointment manage-ment to pharmacy to laboratory and beyond. Further, ACOs cannot man-age quality and costs without control-ling variations in provider care.

Pathologists have a unique opportunity to contribute intelligence that trans-forms “managed care” medicine into astute care. The key to this transforma-tion is a dual approach that is unique to pathology. Pathologists marry “top down” population lab informatics ap-plications that systematize care with “bottom up,” patient-specific mo-lecular testing tools that individualize guidelines in a way that allows for the art and variability of medicine.

Our data, tests, and knowledge en-able both population-level analytics and individualized patient-level appli-cations. Using this dual approach (see Figure 2), we can employ informatics and molecular testing to utilize both population-based and individualized care management.

4Steve Ivey, “Humana, Norton use information to keep employees healthier under ACO,” Business First, December 2, 20115“Population Health Management: A Roadmap for Provider-Based Automation in a New Era of Healthcare,” Institute for Health Technology Transformation, 2012



187

For example, on the population level, informatics analysis of serial antibio-grams taken from locations throughout the community, as well as other near-by locations, can be used to formulate guidelines for antibiotics based on predicted pathogens and resistance patterns. At the same time, we can offer individualization of those guide-lines for particularly high-risk patients using point-of-care molecular testing to identify the pathogens and imme-diately direct antibiotic treatment. A patient with COPD and heart failure, for instance, presenting with a recur-rence of a pneumonia would be a good candidate for such testing.

Another illustration of our dual popula-tion and individualization approach might be in the area of venous throm-boembolism (VTE) prevention. Based on statistical data mining of an ACO database containing both lab and clinical data, we might identify par-ticular patterns of lab test values that would inform the complex risk stratifica-tion involved in current VTE prevention guidelines. Turning to the individualized level, we might provide a personal-ized coagulation profiling service for challenging patients. This service might include pharmacogenetic testing to ensure proper dosing of warfarin, as well as for the hereditary thrombophil-ias, for example.

In this way, our expertise and testing information are applied both to ad-vancing guidelines at the population level and to individualizing care at the patient management level – a dual

approach that represents a strong competitive advantage.

Any provider organization that is man-aging patient populations under alter-native, non- fee-for-service payment programs needs a population care focused pathology group to optimize their performance. We are needed to develop applications and services di-rected at population health care – via informatics, molecular technologies, and programs that impact care at the community, hospital, or practice popu-lation level.

risk management and Prevention: Central to PhmIn addition to applications to support clinical decision making, information sharing, and reporting, PHM requires a different way of looking at patients. For instance, one of the core functions that determines how effectively an or-ganization can manage a population is population segmentation, such as risk stratification.

For example, population segmenta-tion can be performed according to a patient or cohort’s clinical risks, testing results, demographics, lifestyle factors, and/or claims history. Since this seg-mentation is typically used for deci-sions regarding resource allocation, care management services, and pro-vider interventions, the “accuracy” of the segmentation in terms of predict-ing risks and optimally matching needs with resources impacts the ACO’s quality and cost results.

Health risk stratification is the most criti-



cal type of segmentation, as it is mostly closely linked to value – i.e., costs and outcomes. It is also used in population health management to direct care management – which patients are at high risk and need ongoing support, which have serious chronic conditions warranting an active intervention plan to prevent complications, and which are fairly healthy and just need pre-vention and education.

The tricky part of risk stratification is that a patient’s risk status is quite dynamic, so ACOs need recent data to continu-ously refresh risk stratifications. In fact, the true power comes with being able to use risk assessment to predict future health care spending.

Studies show that more than 70 per-cent of patients in the high cost cat-egory in a given year were in a lower cost category the prior year.6 In other words, who will be the highest cost pa-tient mostly changes year to year.

While payers have developed predic-tive algorithms for risk assessment, they do not completely meet the needs of ACOs. “Health insurers use predictive modeling algorithms that can help forecast which patients are likely to have significant health costs. Some health plans are giving provider or-

ganizations these kinds of tools, which can be valuable in identifying patients who may be hospitalized or suffer complications in coming months. But, since these programs are not designed for providers and have limited utility in clinical settings, organizations must define and develop more appropriate tools.”7

For example, Robert Fortini, vice presi-dent and chief clinical officer Bon Sec-ours Medical Group, Richmond, Va., described his organization’s IT needs for population health management: “In the future, we’d like to be able to predict which patients are most likely to get sick and incur major treatment costs.”8

The critical capability of population segmentation and risk prediction needed for successful ACO financial management requires pathology and lab testing more than most any other part of medicine. Why? Because we deliver the most potent of predic-tive tools in health care – molecular diagnostics and lab biomarkers. Fur-thermore, because lab testing is per-formed frequently, it provides a steady supply of temporal data with which to refresh risk assessments.

6Ian Duncan, "Healthcare Risk Adjustment Predictive Modeling," Winstead, CT: ACTEX Publications, 20117“Population Health Management: A Roadmap for Provider-Based Automation in a New Era of Healthcare,” Institute for Health Technology Transformation, 20128“Population Health Management: A Roadmap for Provider-Based Automation in a New Era of Healthcare,” Institute for Health Technology Transformation, 2012



189

Pathology and Laboratory medicine: At the Core of PhmIn fact, pathology and laboratory medicine are at the core of popula-tion health management, and there-fore inexorably linked to its success.

Pathologists are critical to success be-cause pathology and laboratory test data are critical to PHM:

• These data go beyond being just a component in the ACO’s EHRs and integrated databases. Lab and pathology results are one of the leading sources of information powering the IT systems so vital to population care: lab and pathol-ogy data are estimated to com-prise 70 percent of EMR content.9

• Population health management requires tools that span a wide clinical breadth – few health care applications can rival the “impact range” of pathology and lab testing. They are performed in the diagnosis and/or management of almost all major clinical condi-tions.

• Population health management requires tools that cover both the population and patient care continua – again, pathology and lab medicine are among the only clinical implements with significant roles across these care continuum ranges:

– Prevention and wellness – e.g., cervical cancer screening and

HPV, hepatitis screening, lipid profiles

– Diagnosis – e.g., histopathology, coagulation testing, microbiol-ogy

– Therapeutic guidance – e.g., pharmacogenetics, molecular testing for HCV, oncology mark-ers

– Chronic care management – e.g., HA1c, warfarin INR, lipid profile

– Acute care – e.g., microbe identification with molecular tests, perioperative glucose control

• Care coordination is integral to population health management. Because pathology and lab test-ing span clinical conditions and care settings/continua (as de-scribed above), the information can be used as a coordination glue to track and manage pa-tients, services, and providers. However, this requires recognition of this resource and development of the asset into a coordination system that is managed by an expert in pathology/lab medicine. For example, because lab tests are ordered during the majority of provider visits, lab test records can be used as a proxy for clinician interactions.

• We deliver the most potent of

9L Pantanowitz et al., “Medical laboratory informatics,” Clin Lab Med, 2007;27:823-43



predictive, risk assessment tools in health care: individualized medi-cine diagnostics and biomarkers. Furthermore, lab testing is per-formed frequently, so it provides a steady supply of temporal data with which to refresh risk assess-ments.

not by Data Alone: expertise neededACOs and other providers engaged in PHM need not only our data, but our expertise and services to generate the value they need to be success-ful under these new payment models. Moreover, we bring the ability to infuse individualized medicine into guideline-driven care, and to merge the two – population informatics with person-alized medicine testing, as shown in Figure 2.

In addition to pathology and labora-tory test data, pathologists are needed to:

• InterpretthisdataforACOsGiven the complexities gener-ated by the increasing numbers of molecular and “omics” diag-nostic tests, the complexities of pathophysiology, and the com-plexities of the aging population with multiple chronic conditions and polypharmacy, providers and payers need pathology experts to interpret and disentangle the interrelationship of test results. This is essential when conducting data mining and developing predictive algorithms. For example, without knowing a patient’s albumin lev-

els, one cannot properly interpret calcium levels.

• Selectthecorrectdiagnosticdatato analyzePathologists are needed to iden-tify the new evidence regarding tests with predictive power that can serve as clinical biomarkers.

• AssurethatdataisusedcorrectlyPathologists’ expertise in labora-tory analytics can identify issues that can influence the degree of validity of informatics analyses. For example, a little-appreciated problem with potentially signifi-cant consequences on a popula-tion management level is that of mixing test results from different labs, as they often use different reference ranges for the same tests. Thus, problems can occur when trying to interpret test results for one patient from a variety of labs.

• Assurethatdiagnosticdatais leveraged optimallyPathologists can have a tremen-dous impact on the effectiveness of PHM because there are many opportunities to utilize pathology and laboratory testing in the re-duction of downstream clinical costs and improvement of out-comes, but they necessitate pa-thology/laboratory expertise. For example, by applying informatics-determined, cohort-specific test reference ranges, we can often detect early signs of pathol-



191

ogy in what would otherwise be considered “normal” lab results. Laboratories have used data repositories to set analyte-specific reference ranges based on gen-der, age groups, and pediatric Tanner development stages.

Population health management: A Clear opportunity for Pathology and Laboratory medicineTaken together, the superior value of pathology and lab medicine for popu-lation health value seems clear. Yet this value cannot be fully realized unless pathologists are engaged in applying the information to population health management.

In the “Coordinated Population Care Services Practice Pathway” our

pathology group plays a key leader-ship role by applying laboratory infor-matics, molecular testing, and pathol-ogist expertise to produce applications and services for optimally managing care that are used by the ACO provid-ers, hospitals, and community.

But we need to move quickly to secure our role and prove our value in these ACOs. They have significant adoption momentum behind them, with an es-timated 450 ACOs in existence today, so the models and roles are becoming established rapidly. Without taking action to develop these services and applications that extend beyond test results, and to present a business case, ACOs will sideline us as providers of data.



oUr mArket oPPortUnity – ADVAnCing PoPULAtion heALth mAnAgement

The emergence of value-based pay-ment models and coordinated popu-lation care have created a tremen-dous opportunity for our profession. The information contained in the myr-iad of clinical pathology and labora-tory tests performed on patients every year is essential for efficient and effec-tive population care and it is waiting to be fully unlocked. Pathologists are one of the primary key holders.

This information, in conjunction with molecular testing applications and ser-vices we provide, can advance popu-lation care through, for example:

• Predictive informatics, biomarkers – lower costs and better outcomes through avoided complications and admissions.

• guidelines that systematize “indi-vidualized medicine” – protocols that implement value-based de-terminations of when to use phar-macogenetic testing for selecting therapies and/or dosing.

• Lab informatics for care coordina-tion – e.g., IT systems to track and analyze lab results for medical home management of provider referrals, and facilitation of patient transitions between care settings.

• Lab informatics for managing ser-vice utilization – e.g., blood prod-ucts, provider test ordering varia-tions that impact downstream costs.

• Hospital and community patho-gen programs.

There are abundant opportunities for us to add value in advancing PHM given the wide array of provider orga-nizations managing populations, and their diverse needs in carrying out that management. The challenge may be in selecting the best area to begin.

Like all businesses, the answer rests with the customer’s needs, so that trans-lates into understanding your local ACO, integrated delivery network, or medical home.

target Client: Focus on the ACoWe can start with a high-level analysis of the needs of a typical ACO in terms of population health management. Managing the health of the population for an ACO translates into achieving the value-based outcomes defined by the payer who developed the ACO program.

As background, an ACO has essential-ly three levels of populations on which it is measured and hence needs to



193

manage. To give a sense of population scope, the ACO program offered by the Center for Medicare and Medicaid Services (CMS) requires a population of at least 5,000 beneficiaries.

The largest population level is the com-munity, which likely includes healthy

people and might be demarcated by a geographical region, depending on the rules of the ACO. The mid-level is the provider setting, such as a practice group or a hospital population. And finally, the smallest population unit is a clinician’s panel of patients.

Figure 3: medicare’s Quality measures for tiered Population Levels

Source: Adapted from graphic in presentation: “Quality Metrics as a Foundation for Value-Based Care” Shari M. Ling, MD, Deputy Chief Medical Officer, Center for Medicare and Medicaid Services, Office of Clinical Standards and Quality, 9th Annual World Health Care Congress, Washington, DC, April 16-18, 2012

• CMS measures have denominators based on physician’s patient population or “panel”• Individual physician’s panel of patients• Applies to all physicians/clinicians

• CMS measures have denominators based on practice setting• For example, hospital, practice group

• CMS measures have community-based denominators; e.g., county• Applicable to all providers

PatientPanel

Population

Practice Group orFacility Setting

Population

Community Population



To reiterate, ACOs need to manage care along all three population levels, both to meet quality performance tar-gets and to lower costs.

On the quality side, most ACO con-tracts dictate specific quality mea-sures for which performance is tied to ACO compensation. These measures typically span all three ACO popula-tions. For example, a quality measure regarding what percentage of the community population is receiving a vaccination or preventive screening, a measure at the hospital population level regarding 30-day readmission rates, and a measure at the physician practice level regarding diabetic pa-tients’ HA1c levels.

ACO’s management of cost savings also involves all three population levels, as well as a focus on improving quality, because decreasing poor outcomes means cost savings. The cost equa-tion for ACOs is essentially a capitated benchmark based on historical claims for the population. In other words, if the ACO can decrease the average spend per person based on the prior few years for their patient population then they receive a share of the net savings.

To manage costs, an ACO needs to use a number of tactics – e.g., co-ordinating care across the commu-nity population, using pathways to minimize provider variations, reducing adverse events in facility populations, focusing on predicting risk amongst chronic conditions within physician pa-

tient panels in order to intervene early, investing in prevention and wellness at the community level, etc.

Revisiting our Figures 1 and 2, let’s start to look at our competitive advantages involving 1) a dual approach of top down population informatics and bot-tom up, individualized molecular test-ing, and 2) value to all three dimen-sions of care can bring us front and center to the ACO PHM table.

Given that ACOs are essentially our clients, in that we are delivering value to them and looking for value-based compensation from them, we need to begin by understanding how they are being measured and where they are looking for cost savings. Let’s start with quality.

ACo Analysis: measuring QualityCMS’s ACO program, the “Medicare Shared Savings Program,” measures 33 nationally recognized quality measures in four domains. In 2013 the measures will be implemented as pay for report-ing, meaning they must be reported in order to earn bonuses, and in 2014, they will go into effect in a pay-for-performance mode, meaning ACOS will need to meet certain performance levels on these measures in order to get payments.

Further, an ACO must meet a quality performance standard for at least 70 percent of the measures within each of the four domains, or else Medicare places the organization on a correc-tive action plan. If the ACO continues to underperform, Medicare terminates



195

the agreement. The quality measures are also used for calculating points that impact their shared savings pay-ments.

The quality measure domains are as follows (bolding indicates measures of particular interest for our pathway):

• Domain 1: Patient/caregiver experience – 7 measures; e.g.:

“Health promotion and educa-tion”; “how well your providers communicate”

• Domain 2: Care coordination/ patient safety – 6 measures; e.g.:

All condition 30-day readmissions

Ambulatory sensitive conditions admissions – congestive heart failure (CHF)

Ambulatory sensitive conditions admissions – chronic obstructive pulmonary disease (COPD)

• Domain 3: Preventive health – 8 measures; e.g.:

Colorectal cancer screening; mammography screening

• Domain 4: At-risk population:

Diabetes (6 measures); e.g.:

Percent of Diabetics with HA1c below 8 percent

Hypertension (1 measure)

ischemic Vascular Disease (2 measures)

Heart Failure (1 measure)

Coronary Artery Disease (2 measures)

Looking at how we might support ACOs’ meeting these quality mea-sures, major opportunities might in-clude managing diabetes and chronic cardiovascular diseases, and predict-ing and preventing readmissions.

The need to predict and prevent re-admissions is a high-profile opportunity applicable to all hospitals, regardless of ACO status, because of Medicare’s “Readmissions Reduction Program,”10 which starts in 2013 with acute myocar-dial infarction, heart failure, and pneu-monia. It targets 30-day readmissions.

One study11 of readmissions among 11,855,702 Medicare beneficiaries found that across all medical and surgical cases, the first most frequent cause of a readmission was heart failure, and the second pneumonia. Postoperative infections and septi-cemia were also among the top four most frequent reasons in the majority of the surgical procedures analyzed.

A potentially promising opportunity for pathology to impact preventable

10http://cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/ Readmissions-Reduction-Program.html/ 11SF Jencks et al. “Rehospitalizations among Patients in the Medicare Fee-for-Service Program,” N Engl J Med, 2009;360:1418-28



30-day readmissions is to tackle the resistant pathogen problem. With the emerging maturity of rapid molecular testing for identifying microbes, and powerful informatics for integrating surveillance data across laboratories, microbial surveillance programs are advancing to the point where they might be able to help meaningfully reduce resistant pathogen levels and hospital-acquired infections.

ACo Analysis: reducing CostsTo understand how an ACO might ap-proach reducing costs, we can refer-ence an influential and vetted source, the Prometheus Payment model,12 which was recently adopted by CMS as the engine behind their bundling programs. The Prometheus model bundles all care services for specific conditions and procedures, called “ev-idence-informed case rates (ECRs),” into base payments with severity ad-justments (what should be done) and “potentially avoidable complications (PACs)” (what should not happen).

The developers write, “We separated the costs attributable to patient-relat-ed factors from those attributable to providers’ actions. These latter costs are critically important in terms of accountability. In Prometheus, these potentially avoidable costs are called PACs and are recognized as the result of “care defects” – problems neces-sitating technical care that are under

the professionals’ control and that, with the best professional standards, could have been avoided. PACs might include the cost of hospitalization of a patient with uncontrolled diabetes or the readmission for a wound infection of a patient who had recently been discharged after cardiac bypass sur-gery.”13

The Prometheus model has been de-veloped for 21 ECRs, together com-prising about 30-35 percent of total spending. These 21 do not include any oncology related conditions, other than colonoscopy and colon resection procedures. HCI3 is currently develop-ing ECRs for the majority of major con-ditions and procedures, and for the Medicare population.

Because PACs represent the costs that are in providers’ control and have been deemed “potentially avoid-able” by groups of expert physicians who advised on the development of the program, these dollars and areas should be good targets for ACO cost reduction, and therefore for our PHM services.

HCI3 has made available the results of an analysis of a privately insured popu-lation of 4.7 million adult beneficiaries for the starting set of 21 conditions and procedure ECRs and their associated PACs. Using these data, we have ana-lyzed the share of PAC dollar opportu-

12Developed by the Healthcare Incentives Improvement Institute (HCI3) – see http://www.hci3.org/what_is_prometheus 13F de Brantes et al., “Building a Bridge from Fragmentation to Accountability – The Prometheus Payment Model,” N Engl J Med, 361;11



197

nity by ECR condition/procedure and service setting and type in Figure 4.

The following summarizes the share of PAC dollars for this commercial popu-lation – in other words, what percent of cost reduction opportunity does each area have:

By conditions and procedures: percent of PAC dollars

Chronic Medical = 61 percent of PAC dollarsOutpatient procedural = 21 percentInpatient procedural = 12 percentAcute medical = 6 percent

By setting/charge area: percent of PAC dollars

Inpatient = 57 percent of PAC dollarsOutpatient = 27 percent Professional = 15 percent

The top 10 conditions or procedures (in order of highest share): percent of PAC dollars

GERD (gastroespohageal reflux) = 17 percent of PAC dollarsDiabetes = 15 percentCOPD = 9 percentHypertension = 6 percentPCI (Percutaneous Coronary Intervention) = 6 percentAsthma = 5 percentColonoscopy = 5 percentCHF = 5 percent

Prometheus also provides a detailed analysis of the types of PACs that drive these costs. Some of the most costly PAC areas as a percent of all PAC dollars include:

Emergency room visits = 15 percent of PAC dollarsGastritis, Ulcer, GI Hemorrhage = 7 percentPneumonia = 7 percentSepsis, meningitis, hepatitis = 5 percentFluid and electrolyte disorders = 5 percentDVT and pulmonary embolism = 4 percentStroke = 4 percentDiabetic emergencies – hypo- and hyper-glycemia = 4 percentUrinary tract infections = 3 percent

What can we take from these data to inform a starting roadmap for our PHM applications and services? A few observations:

• Inpatient and chronic conditions areas dominate the opportunities for savings.

– This is driven in part by admis-sions for chronic conditions that are deemed avoidable.

• Diabetes also offers an oversized share of avoidable costs.

• Infections and coagulation dis-orders comprise a large share of PAC dollars.



Figure 4: Potentially Avoidable Costs (PACs) from Prometheus Payment model – example for Commercially insured Population of 4.7 million Adults

Ref – Analysis of data from Healthcare Incentives Improvement Institute website, http://www.hci3.org/what_is_prometheus/framework/evidence_informed_case_rates

inPAtient oUtPAtient ProFessionAL totAL PAC shareFor a privately insured population of 4,723,747 adults; Total claims cost for one year = $93,959,198,000

Chronic medical 61 percentDiabetes $73,146,717 $87,177,569 $61,949,863 $222,274,149 15 percentCAD $31,968,633 $16,277,081 $14,382,656 $62,628,370 4 percentCHF $34,567,824 $19,283,541 $15,871,551 $69,722,916 5 percentGERD $138,623,866 $70,787,277 $47,226,541 $256,637,684 17 percentAsthma $22,238,841 $35,084,926 $21,845,902 $79,169,669 5 percentCOPD $54,307,843 $43,771,418 $29,970,775 $128,050,036 9 percentHypertension $12,680,732 $43,391,292 $33,642,908 $89,714,932 6 percent

Acute medical 6 percentAMI $26,517,544 $0 $0 $26,517,544 2 percentPneumonia $39,949,528 $0 $0 $39,949,528 3 percentStroke $19,308,490 $0 $0 $19,308,490 1 percent

inpatient Procedural 12 percentCABG $53,230,340 $0 $0 $53,230,340 4 percentBariatric surgery $26,735,367 $0 $0 $26,735,367 2 percentHip replacement $14,669,590 $0 $0 $14,669,590 1 percentKnee replacement $33,552,754 $0 $0 $33,552,754 2 percentColon resection $33,552,754 $0 $0 $33,552,754 3 percent

outpatient Procedural 21 percentPCI $75,885,702 $7,935,749 $0 $83,821,451 6 percentLow-risk pregnancy and delivery $52,767,034 $1,767,581 $0 $54,534,615 4 percent

Knee arthroscopy $5,994,561 $5,078,196 $0 $11,072,757 1 percentGall bladder $44,106,893 $8,844,822 $0 $52,951,715 4 percentHysterectomy $21,367,938 $4,495,240 $0 $25,863,178 2 percentColonoscopy $16,315,169 $61,955,282 $0 $78,270,451 5 percent

totALs $848,778,733 $405,849,974 $224,890,196 $1,479,518,903PAC share 57 percent 27 percent 15 percent



199

the ACo movement: growth and DistributionAs of August 2012, there are 154 ACOs in the Medicare Shared Savings pro-gram. Almost half of the most recent group of 89 ACOs announced this summer is physician-led ACOs with fewer than 10,000 beneficiaries, in-dicating that this model is being ad-opted beyond the major, integrated health systems.14 Physician ACOs may be advantageous pathway custom-ers for us, because they might have greater laboratory testing informatics and coordination needs than hospital system ACOs.

In addition to the 154 Medicare ACOs, there are an estimated 150 ACOs in the private sector, bringing the current total estimate to about 300 ACOs.15

In terms of geographic distribution, the 154 Medicare ACOs are spread across 40 states and the District of Columbia, and cover 2.4 million beneficiaries. Figure 5 shows the regional service areas for about 200 ACOs that Leavitt Partners was tracking as of May 2012.16 The East and Midwest have a greater concentration of ACOs compared to the South, Great Plains, and Mountain

West regions. As far as we have been able to identify, there are currently only two states, Delaware and South Dakota, without ACOs.

How fast might the ACO movement grow? That depends, of course, on how successful the model is in gen-erating cost savings, which won’t be known for likely two years. Jonathan Blum, principal deputy administrator and director of the Center for Medi-care, reported that an estimated $940 million over four years might be saved by the Medicare Shared Savings Pro-gram.17

Indicators of near-term interest include Medicare’s comment that it has re-ceived about 400 notices of intent from provider groups regarding the ACO program at the next start date of January 1.18 Furthermore, a recent, national survey of 1,070 executives in provider organizations (66 percent hos-pitals or health systems and 14 percent physician orgs) found that 53 percent indicated their organization would join an ACO within the next five years.19 These and other market signs point to a brisk growth in the number of ACOs over the next few years, barring any

14John Commins, “89 ACOs Open for Business,” HealthLeaders Media, July 10, 201215Susan Dentzer, editor of Health Affairs, remarks at the Third National Accountable Care Organization Summit in June 201216David Muhlestein et al., “Growth and Dispersion of Accountable Care Organizations: June 2012 Update,” Leavitt Partners, June 2012 17John Commins, “89 ACOs Open for Business,” HealthLeaders Media, July 10, 201218“Blum: Diversity, Number Of ACOs Shows Program Is Spreading Fast,” Inside Health Policy, July 11, 201219“Healthleaders Media Industry Survey 2012,” www.healthleadersmedia.com/intelligence, HealthLeaders Media, a division of HCPro, Inc., 2012



major changes politically or reports of the model failing.

In fact, some analysts are projecting about 600 ACOs, a doubling in num-ber, for 2013.20 It is important to note

that even if the Affordable Care Act were to be repealed, the private mar-ket momentum behind ACO-like mod-els is sufficiently strong that analysts predict it would have little impact on the commercial insurance side.

20Jim Bloedau, “ACO Market Will More Than Double in 2013 by Jim Bloedau,” Information Advantage Group, August 9, 2012

Figure 5: numbers of Private and medicare ACos by state, may 2012

Source: Adapted from Leavitt Partners, “Growth and Dispersion of Accountable Care Organizations, June 2012 Update.” Note: Leavitt Partners’ tracking of ACOs, current as of May 2012; their total at that time was 221 ACOs in 45 states. This map shows the 198 ACOs for which they were able to define geographic limits.

ACOs20+10-197-104-62-310



201

Fitting into the ACo market There are a multitude of business fla-vors among ACOs, as many are imple-mented in the private payer market. Even within the Medicare Shared Sav-ings program, there are ACOs that are run and owned by physician organiza-tions, others by hospitals, and others by payers partnered with either physi-cian groups or hospitals. It is beyond the scope of this chapter to review the contractual options available to

pathology groups across various ACO models. In most circumstances, a pa-thology group that was already con-tractually tied to either an ACO hospi-tal or ACO physician organization for diagnostic testing services will be the best candidate for this pathway.

Figure 6 illustrates how the Medicare Shared Savings “one-sided” program works in terms of payment flows, in a simplified depiction.

Figure 6: medicare ACo shared-savings Payment Mechanism–SimplifiedIllustration

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%Savings

15%

ActualSpend

85%

Benchmark spending for ACO basedon prior 3 years CMS claims data

Share of Savings for Medicare’sTrack 1 Upside Only Program

Medicare

ACO

Perc

ent o

f Dol

lars

ACOSource: Eleanor Herriman, MD, MBA, G2 Intelligence



No matter how it is contractually con-figured, the business arrangement we seek with ACOs is compensation based on value payments – e.g., ACO savings bonuses. We will demonstrate the value of our Coordinated Popula-tion Care Pathway to the ACO in terms of a business case that documents opportunities for cost savings and for achieving quality measures. Based on this business case, our goal will be to negotiate a share of the ACO’s net savings.

According to Milliman, the savings pool is generally allocated between three provider pools, primary care, specialty, and hospital (when a hospi-tal is part of the ACO). As an example methodology, the report provides the table in Figure 7 as a starting point that shows how an ACO might divide the providers’ pools first against service line dollars, according to that pool’s level of responsibility for that service line. Thus, in their example, the hospital has no responsibility for PCP service costs.

The exact share can be determined using one of several other methods they detail. For example, for hospitals, they describe incorporating important cost-saving measures into the deter-mination of the hospital’s allocation, including:

• Length of stay

• Readmission rates

• Infection rates

• Preventable events

• Admissions through the emergency room

• Intensive care unit days

• ER visits (top five diagnoses)

• Utilization of high-cost or frequently overutilized interventions – e.g., MRI, CT, PET, and endoscopy

In a scenario in which the hospital’s allocation of the savings pool is being determined based on the cost-saving measures listed above, we may have another negotiating approach: some of our pathway interventions can improve measures such as readmis-sions rates, infection rates, length of stay, and intensive care unit days. If our pathology group is already aligned or part of the hospital, we can build a case that using our new services and applications, we can position the hos-pital to capture a greater share of the ACO bonus pool, and that we there-fore deserve a share for our group.



203

Figure 7: example method for Allocating ACo savings by Provider Pool and service Line, milliman21

inpatient Costs outpatient Costs

PCP Costs specialist Costs

rx Costs Ancillary Costs

Hospital 30-60 percent 30-60 percent 0 percent 0-10 percent 0 percent 0 percent

Primary care provider (PCP)

15-30 percent 15-30 percent 100 percent 20-50 percent 50-75 percent 25-75 percent

Specialty care provider

10-25 percent 10-25 percent 0 percent 30-60 percent 25-50 percent 25-60 percent

TOTAL 100 percent 100 percent 100 percent 100 percent 100 percent 100 percent


sizing the Potential market opportunityTo model the five-year pool of po-tential ACO revenues from our “Co-ordinated Population Care Services Practice Pathway,” we projected the growth in ACOs, the average number of beneficiaries per ACO, average expenditures per beneficiary, and the average cost savings achieved on average by ACOs.

In order to model what share of the ACO pool our pathology pathway group might receive, we considered that we would likely be negotiating from the hospital pool, given that most of our groups will be affiliated closely with an ACO hospital if they are pur-suing this pathway. Most commonly, according to Milliman22, the hospital receives about 50 percent of the total ACO bonus pool. Although there are numerous variables concerning the health system, politics, hospital, and pathology group size, etc., we as-sumed for modeling purposes a range

of 2 percent to 8 percent of the hos-pital’s share of savings could be paid to our group if we make our business case.

The rationale is that 2 percent to 8 per-cent of the hospital pool translates to 1 percent to 4 percent of the total sav-ings pool. Given that laboratory and pathology charges comprise about 3 percent to 4 percent of all spending, and that some providers allocate sav-ings based on the percent of charges, we reasoned that the negotiation could be supported on this basis.

We also assume that these shares can increase over time as we prove our value to the system and increase the number of services and applications provided.

Using these projections and assump-tions we can project revenues for 2013 to 2017 for a potential, total market "Coordinated Population Care Services Practice Pathway" ACO market, as

21GN Berry and R Kipp, “ACO gain/loss sharing,” Milliman Healthcare Reform Briefing Paper, 201222GN Berry and R Kipp, “ACO gain/loss sharing,” Milliman Healthcare Reform Briefing Paper, 2012



seen in Figure 8. The caveats are that models such as these are to be taken as directional only, given the impre-cisions inherent in dealing with such complex and highly uncertain systems.

Our model estimates that in 2014, for example:

• The average payment per pathol-ogy group from a CMS ACO cov-ering 15,000 lives is ~$472,000.

• The average payment per pathol-ogy group from a private ACO covering 30,000 lives is ~$416,000.

– This is due to the fact that Medi-care patients have double the per cap expenditures vs. private populations.

• The total, potential ACO market opportunity in 2014 is estimated at $332 million.

In 2017 the model estimates the poten-tial total market for pathology groups to be $739 million.

Again, these payments are in addi-tion to diagnostic testing service pay-ments and Medicare Part A revenues. It is also true, however, that many are predicting that our fee-for-service test-ing revenues will be declining over this time period, especially hospital lab test volume.

The model directionally supports an attractive opportunity from a financial perspective.

More importantly, however, if we do not pursue demonstrating our value with ACOs, the risk of our becoming a commodity vendor is real and possibly permanent.



205

mArket estimAtes 2013 2014 2015 2016 2017medicare ACos

Estimated number of CMS ACOs 154 350 450 500 550

Estimated number of covered lives per ACO

15,000 15,000 15,000 15,000 15,000

Estimated per capita medical expenditures for Medicare beneficiaries in 2013

$11,000 $11,440 $12,012 $12,613 $13,243

Conservative scenarioAssumed percent cost savings retained by ACO (assumes Medi-care gets 50 percent under Track 1 program)

5 percent 5 percent 5 percent 5 percent 5 percent

Assumed shared of net cost savings earned by pathology group


Total potential Medicare ACO mar-ket for pathology pathway – conser-vative scenario

$12,705,000 $60,060,000 $121,621,500 $141,891,750 $163,884,971

Estimated potential Medicare market per ACO – conservative

$82,500 $171,600 $270,270 $283,784 $297,973

Aggressive scenarioAssumed percent cost savings retained by ACO (assumes Medicare gets 50 percent under Track 1 program)


Assumed share of net cost savings earned by pathology group


Total potential Medicare ACO market for pathology pathway – aggressive scenario

$91,476,000 $270,270,000 $437,837,400 $510,810,300 $589,985,897

Estimated potential Medicare market per ACO – agressive

$594,000 $772,200 $972,972 $1,021,621 $1,021,702

Total potential Medicare ACO market for pathology pathway – average

$52,090,500 $165,165,00 $279,729450 $326,351,025 $376,935,434

Potential Medicare ACO market for pathology pathway per ACO – average

$338,250 $471,900 $621,621 $652,702 $685,337


Figure 8: market estimates for medicare ACos



2013 2014 2015 2016 2017Private market ACos

Estimated number of private market ACOs

300 400 500 550 600

Estimated number of covered lives per ACO

30,000 30,000 30,000 30,000 30,000

Estimated per capita medical expenditures for commercial beneficiaries

$4,800 $5,040 $5,292 $5,557 $5,834

Conservative scenarioAssumed percent cost savings retained by ACO (assumes payer gets 50 percent)




Total potential private ACO market for pathology pathway – conservative scenario

$21,600,000 $60,480,000 $119,070,000 $137,525,850 $157,529,610

Estimated potential commercial market per ACO – conservative

$72,000 $151,200 $238,140 $250,047 $262,549

Aggressive scenarioAssumed percent of cost savings retained by ACO (assumes payer gets 50 percent)




Total potential private ACO market for pathology pathway – aggressive scenario

$155,520,000 $272,160,000 $428,652,000 $495,093,060 $567,106,596

Estimated potential commercial market per ACO – aggressive

$518,400 $680,400 $857,304 $900,169 $945,178

Total potential private ACO market for pathology pathway – average

$88,560,000 $166,320,000 $273,861,000 $316,309,455 $362,318,103

Potential private ACO market for pathology pathway per ACO – average

$295,200 $415,800 $547,722 $575,108 $603,864

total, estimated, potential ACo market for pathology pathway

$140,650,500 $331,485,000 $553,590,450 $642,660,480 $739,253,537

estimated, potential pathology pathway payment per ACo – average

$316,725 $443,850 $584,672 $613,905 $644,600


Figure 9: market estimates for Private market ACos



207

ABoUt the CoorDinAteD PoPULAtion CAre serViCes PrACtiCe PAthwAy

The "Coordinated Population Care Services Practice Pathway" offers one model for becoming an ACO popu-lation care management center. As we’ve been describing, there are a number of dimensions along which we can generate value, including:

• The population level – either com-munity, hospital, or provider’s patient population.

• The application areas – includ-ing population informatics for risk prediction, individualized medi-cine guidance, utilization man-agement, quality programs, and coordination programs.

• The provider group users – primary care providers, hospitalists and ER clinicians, surgeons/proceduralists, and ACO executive leadership who manages the population.

The primary inputs for our practice are our pathology expertise, lab informat-ics, and molecular testing. The “out-puts” – i.e., our pathway’s offerings – take the form of guidelines, algorithms, EMR applications, and risk prediction reports, for example (See Figure 8)

In pursuing this pathway, we can choose to target only a particular pop-ulation level or provider user group, especially to start out and prove our ability to generate value. For that reason, we will describe the pathway applications, services, and programs in a matrix fashion across “provider user groups” and “application areas,” as shown in Figure 9, “Pathway Offerings Matrix.”



Figure 10: Coordinated Population Care Pathway: key inputs and outputs

LabInformatics

PathologyExpertise

Guidelines,Algorithms,

EMR Tools, RiskPredictions

MolecularTesting




209

Figure 11: Coordinated Population Care Practice Pathway offerings matrix

inFormAtiCs-BAseD risk PreDiCtion

inDiViDUALizeD meDiCine gUiDAnCe

UtiLizAtion mAnAgement

QUALity AnD CoorDinAtion ProgrAms

Primary Care Providers

• Reports on patients at high risk

• Tests to order based on elevated risk

• Tools – MDx for rapid, targeted antibiotic selection

• Tools – PGx for statins, anticoags, SSRIs, – Rx selection and adherence

• PCP test order-ing variations for tests with high downstream costs

• EMR apps – changing Rx, ordering Rx levels, ordering LFTs, protocols for referrals to specialists

• High-risk patient reports – triage care manage-ment services

hospitalists and er • Reports on in-patients at high risk – e.g., daily; protocols with re-flex testing based on risk

• Early detection based on labs - alerts

• Suggested meds and tests based on risk in certain settings

• Tools – MDx for rapid, targeted antibiotic selec-tion

• Programs for preventing ER admissions using MDx – e.g., en-terovirus

• Specialist test ordering varia-tions for tests with high downstream costs

• Glycemia control and glucometer accuracy prob-lems

specialists – medical specialties, surgeons, Proceduralists

• Reports and protocols for patients at high risk – anticoagu-lation, fluids and electrolytes, …

• Tools – MDx for rapid, targeted antibiotic selec-tion

• PGx tools – e.g., Clopidogrel

• Oncology – see Oncology path-way

• Blood products/ transfusions

• Specialist test ordering varia-tions for tests with high downstream costs

• Lab testing proto-cols for referrals between provid-ers

Population-wide management/ACo Leadership

• Subpopulation reference ranges for more precise disease detec-tion

• Prediction of next year’s high-cost patients

• Identification of high adverse event areas

• Development of guidelines for PGx and MDx testing by risk status and cost-effectiveness modeling

• Blood products utilization program

• Lab-Pharmacy – decision for antibiotics based on cost-effectiveness + antibiograms

• Pathogen surveil-lance

• POC testing vali-dation and data networking

• Home testing – glucometers, etc.

• Lab phlebotomy network

• Digital pathology




Pgx = pharmacogenetic testing; mDx = molecular (e.g. DnA) testing

This Matrix is not an exhaustive ac-counting of the potential solutions we can offer for ACO population health management, but provides a range of examples that illustrate the model and its value.

Belowwedescribespecificapplicationexamples and, for some, their poten-tial value generation in terms of ei-ther quality or cost reduction for each practice category.

offerings for Primary Care Providers (PCPs):informatics-Based risk Prediction for PCPs – Reports on patients at high risk and recommended tests to order.

Example: EMR report on high-risk dia-betics. Based on trends in glucose, HA1c, and levels of biomarkers such as CRP, lipid profiles, etc. We produce an EMR report that identifies particular diabetes patients in a PCP’s practice as very high risk for thromboembolic events. If we have a program linking us to pharmacy data, we can check on statin or other anticoagulation prescriptions, and apply protocols that consider adherence and resistance questions. Test recommendations could include both other biomarkers, such as hs-CRP, and perhaps phar-macogenetic tests for aspirin or Plavix resistance, for example.

• Value generation potential: when the Prometheus model’s PAC costs for diabetic stroke, AMI, DVT,

and pulmonary embolisms are scaled up from the commercial population to the nation, they represent $1.2 billion in potentially avoidable costs. Each avoided myocardial infarction admission is estimated to be worth ~$13,600. If, using our lab data/biomarker al-gorithm informatics we can identi-fy high-risk diabetics, and thereby direct more intensive interventions and therapeutics, we might help prevent some of these adverse events.

Example: Reports alerting PCP of pa-tients with even low levels of hypona-tremia, as these have been associated with risks of falls, hip fractures, and osteoporosis. This is especially risky in patients with congestive heart failure, cirrhosis, and renal failure, as it por-tends a worse prognosis. Also, thiazide diuretics and SSRIs can worsen the condition.

• Value generation potential: by intervening, hospital admissions, adverse events like falls or frac-tures while in the hospital, and poor outcomes might be averted.

individualized medicine guidance for PCPs 1 – MDx tools for rapid, targeted antibiotic selection. EMR guidelines or decision support for using point-of-care or near care molecular testing to identify resistant pathogen strains and select targeted antibiotics. Guidelines will be based on informatics indicat-ing patient risk and antibiograms de-termining local resistance patterns.



211

The resistance problem is impacting the empirical treatment of all types of infections, and significant proportions of patients are not effectively treated based on empirically directed antibiot-ics. Molecular testing for rapid identifi-cation of the pathogen and thus more specific antibiotic selection has been shown to significantly improve treat-ment – i.e., more rapid recovery for more patients, and thus cost savings and better outcomes.

Example: an EMR application or guide-line for using rapid molecular infectious testing to guide antibiotic selection for pneumonia. This might be cost effec-tive, for example, in populations at high risk for adverse events with pneu-monia, such as patients with multiple chronic conditions. In addition, in the ER, this might allow for avoided ad-missions if molecular testing confirms a viral versus bacterial etiology (an estimated 20 percent to 30 percent of community-acquired pneumonias).

• Value-generation potential: mul-tiple studies23 show that delays in effective treatment of resistant pathogens in pneumonia increase hospital lengths of stay for several days, typically five to seven, rep-resenting a difference of at least $20,000 per patient in costs.

individualized medicine guidance for PCPs 2 – PGx tools for anticoags, statins, SSRIs, adverse reactions, etc. – pharmacogenetics (PGx) is beginning to gain both clinical and payer trac-tion, and our ACO population practice center should take the lead in answer-ing all the major questions regarding how these tests are implemented for population care in an ACO setting. An important data point here is that the two leading pharmacy benefit man-agers (PBMs), Medco and CVS Care-mark, are currently operating major PGx testing programs for millions of their client payers’ beneficiaries. The PBMs have publicly stated that these programs are economically attractive for their clients. Thus, PGx testing should have a value-generating role to play in an ACO population. For a list of drugs with PGx tests included on their FDA drug labels see http://www.fda.gov/drugs/scienceresearch/researchareas/pharmacogenetics/ucm083378.htm

Example: Program (EMR applications, testing and customized patient report) to increase patient adherence to statin therapy using data on their KIF6 carrier status as a motivator.

• Medco reports 50 percent to 90 percent of patients stop taking their prescribed therapies at end

23For example, SE Cosgrove, “The Relationship between Antimicrobial Resistance and Patient Outcomes: Mortality, Length of Hospital Stay, and Health Care Costs,” Clinical Infectious Diseases, 2006; 42:S82–9



of first year –this includes depres-sion, diabetes, cholesterol, and hypertension. For statins, at one year the difference between high and low adherence translates into almost a 10 percent difference in patient survival rates.24

• Medco trial with Celera just pub-lished – “The AKROBATS trial dem-onstrated that 6-month statin therapy adherence and persis-tence are significantly improved by providing the patient with KIF6 genetic information. This study provides the first prospective evidence that pharmacogenetic testing has utility in modifying pa-tient adherence.”25

Utilization management for PCPs – Identification of lab tests that have high PCP ordering variations and also high downstream cost consequences, using data-mining of LIS systems. Pro-vide feedback to PCPs of their test ordering relative to peers, along with explanation of optimal use of test, and test menu decision support for EMR/CPOE as available.

Example – A program to reduce varia-tions in ordering of CA 125 testing. This testing has a high false positive rate and generates high downstream costs

due to imaging testing.26

Quality and Coordination Programs for PCPs – There are a number of applica-tions that we can deliver for supporting chronic care management based on an analysis of an individual’s lab test-ing results, especially if we link our data with the pharmacy database. Need-less to say, we also need access to the ACO’s clinical database/EMRs and EHRs.

Example – guidelines implemented into the EMR for when to either refer a diabetic to an endocrinologist or start them on insulin based on lab values.

• Value-generation potential – studies show improvements in HA1c levels and complication rates when diabetics are started on insulin early, yet many PCPs are reluctant to do so.

Example – reminders to PCPs or directly to patients about testing for drug levels like warfarin, digoxin, etc.

• Value-generation potential – studies show keeping patients within the target INR zone for war-farin significantly decreases the rate of hemorrhages and thrombo-embolic adverse events. This may reduce ER visits and admissions.

24Rasmussen, J. N. et al., JAMA 2007;297:177-18625SL Charland et al., “Patient Knowledge of Pharmacogenetic Information Improves Adherence to Statin Therapy: Results of hte Additional KIF6 Risk Offers Better Adherence to Statins (AKROBATS) Trial,” JACC, March 27, 2012, Volume 59, Issue 1326G2 Intelligence interview with Philip Chen, M.D., Ph.D., director of informatics for the Department of Pathology at the University of Miami, June 15, 2011, in “Creating a Value-Driven Laboratory: Opportunities in the New Marketplace,” G2 Intelligence©, 2011 Kennedy Information, LLC



213

offerings for hospitalists and erinformatics-based risk Prediction for hospitalists and er – Our practice pathway mines our lab testing data and clinical data to identify inpa-tients who are at particularly high risk for a complication or adverse event (beyond what would be evident to a clinician), and to alert clinicians to tests that might be indicated based on lab results that point to early signs of a particular problem. For hospital-ists, the reports on patients at high risk for an adverse event, complication, or mortality are needed on a frequent reporting period, for example daily. For patients in critical units, it may be necessary to access point-of-care test results, as well, if a particularly chal-lenging clinical situation requires an intensive solution.

Example – EHR applications for fluid and electrolyte problems. Our applica-tion could first alert the clinician to a low sodium upon admission; next, if the sodium declined during hospitalization, further alerts with suggestions for test-ing such as serum and urine osmolality could be sent; finally, based on results of tests, decision support algorithms for etiology of the disorder could be of-fered (e.g., is this hypovolemic, hyper-volemic, normovolemic, etc.)

Fluid and electrolyte disturbances, hyponatremia being the most com-mon, are a major hospital problem with significant quality and cost conse-quences. One study found that 11 per-cent of surgical ICU patients had hypo-natermia at admission to the ICU and 31 percent acquired hyponatermia while in the ICU.27 The inpatient costs of patients with hyponatermia are more than twice those with normal sodium. When hyponatremia is corrected, the heightened mortality associated with it decreases to levels associated with normonaterima.28

There are a number of studies that report that clinician management of fluid and electrolyte disorders is quite problematic, and thus is playing a causal role. Diagnosing the patho-physiology underlying the electrolyte disorder is complex and important to the treatment. Furthermore, laboratory testing plays a significant role in unrav-eling the etiology of the disorder. Im-portantly, one investigation found that “the adequacy of laboratory assess-ments during hospitalization directly relates to patient outcomes in hypona-tremia.”29

• Value-generation potential – Based on the Prometheus data,

27Y Sakr, “Fluctuations in serum sodium level are associated with an increased risk of death in surgical ICU patients,” Critical Care 2012, 16(Suppl 1):P14528SS Waikar et al., “Mortality after hospitalization with mild, moderate, and severe hyponatremia,” Am J Med, 2009 Sep;122(9):857-6529M Whyte et al. “Lack of laboratory assessment of severe hyponatremia is associated with detrimental clinical outcomes in hospitalized patients,” Int J Clin Pract, 2009;63(10):1451-1455



fluid and electrolyte disorders rep-resent 1.5 percent of all PAC costs. When scaled to the national level and expanded beyond the com-mercial population based on our modeling, this represents potential avoidable costs of $1.14 billion nationally.

Our analysis using Prometheus data showed that pathway ser-vice offerings that supported the clinician in diagnostic testing and management of hyponatremia could reduce PAC costs for this disorder by 42 percent.

individualized medicine guidance for hospitalists and er – Molecular testing to guide targeted antibiotic therapy for hospitalized patients presents a ma-jor opportunity to improve outcomes and generate cost savings. According to the Prometheus Payment Model, the leading infections (pneumonia, sepsis, meningitis, and urinary tract in-fections) comprise at least 10 percent of all avoidable costs, and most of these are inpatient costs.

One recent study estimated that the cost of an antimicrobial-resistant infec-tion is $18,588 to $29,069 per patient, with an excess duration of hospital stay of 6.4 to 12.7 days and attributable mortality of 6.5 percent.30

Also, a growing evidence base sup-ports the ability of antibiotic therapy

based on molecular testing to improve outcomes and shorten lengths of stay – in other words, generate cost savings. In the next few years these tests will increasingly be available for routine in-patient use for guiding antibiotic ther-apy, and we can bring the needed expertise of assessing which tests to use and interpreting results. Likely we will collaborate with pharmacy and infec-tion control in these programs.

Example – an EMR application with guidelines specifying who is at suffi-ciently high risk to warrant molecular infectious testing for antibiotic selec-tion for UTIs and pneumonia.

Our analysis using Prometheus data showed that molecular testing services to guide urinary tract infections in high-risk patients could reduce PAC costs for UTIs by 17 percent.

Quality and Coordination Programs for hospitalists and er – One area for our pathology practice pathway to ad-dress in hospital and emergency room settings is glucose control – including diabetics with hypoglycemic episodes due to illness, operative procedures, or iatrogenic insulin therapy, and other patients undergoing procedures and having their glucose managed. Dia-betic episodes of hypo and hypergly-cemia are identified by Prometheus to constitute 3 percent of all PAC costs, a substantial opportunity for savings.

30Roberts RR, Hota B, Ahmad I, et al., "Hospital and societal costs of antimicrobial-resistant infections in a Chicago teaching hospital: implications for antibiotic stewardship," Clin Infect Dis, 2009;49(8):1175-1184



215

Further, hypoglycemia problems in hospitalized patients during attempts to manage their glucose are common, dangerous, and costly.

The National Quality Forum prioritized inpatient glycemic control as one of the 34 Safe Practices it recommends to optimize hospital care of patients, admonishing hospitals to “Take actions to improve glycemic control by imple-menting evidence-based intervention practices that prevent hypoglycemia and optimize the care of patients with hyperglycemia and diabetes.”

Example – quality control and valida-tion of bedside glucometers. Many studies have produced evidence that some of the hand-held glucometers used commonly in hospitals are not re-liable in detecting clinically important hypoglycemia.31 When clinicians are trying to calibrate critically ill patients’ glucose levels using insulin, very precise glucose measurements are needed. Despite the recent questioning of the value of intensive insulin control, mod-erate glucose targets are still pursued in inpatient diabetics and surgical patients.

The problem of hypoglycemic epi-sodes persists with less stringent pro-tocols, although likely at lower rates.

For example, a Kaiser study found that even with a “Glycemic Control Team” program to manage moderate glu-cose targets in surgical patients, hy-poglycemic episodes (glucose levels below 70 mg/dL) occurred in 4.6 per-cent of patient days.32

Our practice can contribute to im-proving glycemic control programs by exposing the issue of point-of-care glu-cose meter inaccuracies and propos-ing solutions to address the problem. These might include cost-effectiveness analyses of implementing alternative glucose analytic instruments, quality assurance and training programs, and protocol methodologies to mitigate inaccuracies.

Our analysis using Prometheus data showed that such a quality control program of bedside glu-cometers in which an accurate instrument solution is implemented for inpatient glycemic control, could reduce PAC costs for hypo and hyperglycemic episodes by 41 percent.

offerings for specialists – medical specialties, surgeons, Proceduralists individualized medicine guidance for specialists – There are a number of areas where MDx and PGx will begin

31C Voulgari et al., “Accuracy and Precision of Glucose Monitoring Are Relevant to Treatment Decision-Making and Clinical Outcome in Hospitalized Patients with Diabetes,” Diabetes Technology & Therapeutics, July 2011, 13(7): 723-73032Linnea Baudhuin, PhD, “The Use of Genetics in Guiding Therapy,” Mayo Medical Laboratories Hot Topics, 2012



playing major roles in improving qual-ity and generating cost savings across specialty areas. We’ve discussed in-fectious disease and how MDx can begin to move therapy from empiric to targeted. Oncology is another signifi-cant opportunity, but that is covered in another Practice Pathway, “High-Performance Pathology for High-Value Oncology.” PGx is a bit behind in the adoption curve, but as described above, the fact that the national PBMs are rolling out PGx testing for many tests and therapeutics indicates there is an economic rationale there. PGx tests are being developed rapidly and can have impact in selecting therapies that are most effective, driving dos-ing decisions, avoiding adverse events and bad outcomes in general, and saving pharmaceutical costs.

Example – Program for clopidogrel PGx testing – CYP2C19 for high-risk patients. Evidence is accumulating rapidly that patients with loss of function CYP2C19 alleles do not get benefit from clopido-grel and thus have much higher rates (53 percent in one study) of mortal-ity from cardiovascular events and a three-fold increase in stent thrombosis. A recent presentation by The Mayo Clinic commented about clopidogrel PGx testing, “Some would argue that there is sufficient evidence to provide this testing for certain groups of pa-tients, such as those undergoing coro-nary stenting.”33

Our practice application could be a program that enables clopidogrel PGx testing for stenting patients – guide-lines, a testing solution, interpretation of results with recommendations on changing therapies, and a means of tracking results/registry. This would all be developed in conjunction with in-terventional cardiologists/radiologists.

Pharmacogenomic guidelines, ref-erence databases, and electronic platforms to integrate personal phar-macogenomic data into the clini-cal workflow would all be helpful in overcoming these barriers. Innovative strategies are needed to promote the widespread adoption of pharmacoge-nomic clinical applications by cardio-vascular providers.

A National Heart, Lung, and Blood Insti-tute Working Group recently described how IT systems could integrate PGx into workflow. “One such strategy would be the use of information technology to integrate pharmacogenomics into the clinical workflow, which would fol-low a “learning by doing” model. For example, a provider using a computer-ized provider order entry system to pre-scribe clopidogrel for a patient would be asked whether the patient is known to have reduced-function CYP2C19 alleles and, if not, would be prompted to consider ordering a genotyping test. Extending this model, if the computer-ized provider order entry were linked

33K Musunuru et al., “Cardiovascular Pharmacogenomics: Current Status and Future Directions – Report of a National Heart, Lung, and Blood Institute Working Group,” Journal of the American Heart Association, 2012



217

to the patient's medical record and found that the patient was known to have reduced-function CYP2C19 alleles, the system would prompt the provider to consider using an alter-native antiplatelet medication. Thus, preemptive genotyping – placing data in computerized provider order entry-enabled electronic medical records – is another strategy that needs further evaluation.”34

Utilization management for specialists – Blood product and transfusion man-agement – There is a large body of evidence indicating that blood utiliza-tion is suboptimal and that managing utilization would result in substantial savings – in the hundreds of millions of dollars, according to Barry Portugal, president of Health Care Development Services, Inc. (HCDS).

One study found that over 70 percent of transfusion orders were deemed inappropriate among medical staff at an academic hospital.35 Studies also show that utilization management programs can lower usage rates by 15 percent to 25 percent.

Portugal stated36 that most hospitals have no effective blood management program and that laboratories are in an excellent position to apply test-ing algorithms to reduce unnecessary transfusions and save hospitals signifi-cant money as well as reduce adverse events.

Example – our practice pathway blood transfusion utilization management program would have multiple com-ponents, perhaps including an assess-ment of suppliers, protocols for when to transfuse, coagulation consultative services to help evaluate complex operative cases with microvascular bleeding, etc.

Quality and Coordination Programs for specialists – Lab testing protocols for referrals between providers – our practice can contribute to ACO care coordination by developing auto-mated protocol recommendations, based on lab and clinical data, for referring between specialists and PCPs. These protocols would be developed in conjunction with the specialists, and might also incorporate pharmacy data, which has the advantage of tak-

34K Musunuru et al., “Cardiovascular Pharmacogenomics: Current Status and Future Directions—Report of a National Heart, Lung, and Blood Institute Working Group,” Journal of the American Heart Association, 201235Rothschild et al., Transfusion 2007; 47 cited in “How Should the Blood Center-Hospital Relationship Evolve?," presentation by Timothy Hannon, MD, MBA, Medical Director, Blood Management Program, St. Vincent Hospital Indianapolis, CEO, Strategic Healthcare Group LLC, http://www.aabb.org/programs/nbf/Documents/lf11hannon.pdf36G2 Intelligence interview with Barry Portugal, president of Health Care Development Services, Inc., June 2, 2011, in “Creating a Value-Driven Laboratory: Opportunities in the New Marketplace,” G2 Intelligence, 2011 Kennedy Information, LLC



ing into account patient adherence to therapies. Philip Chen, MD, PhD, direc-tor of informatics for the Department of Pathology at the University of Miami, is developing protocols such as these and implementing them as EMR appli-cations.37

The protocols could be developed as a hybrid between “standardization” to mitigate care variations and improve quality – this is critical for ACOs to achieve their quality measures – and individualized care, based on risk as-sessment from lab and clinical data.

These protocols might also include indications for procedures, although obtaining physician buy-in for this will be more difficult. However, if the pro-tocols are created and positioned as evidence-based, individualized medi-cine algorithms, rather than ACO rules, they may be adopted. These could be very powerful cost-savings programs, as unnecessary procedures are a huge driver of spending.

Example – our practice pathway, in collaboration with cardiologists and PCPs, could develop an evidence-based, individualized algorithm and protocol for referrals to cardiologists based on risk for cardiovascular/hromboembolic events. The algorithm might take into account not only ba-sic cardiovascular lab tests, and more esoteric biomarkers if the patient has other risk factors, but also effectiveness

of therapeutics and adherence data if known. A further protocol might make recommendations for stenting versus medical therapy based on patient risk factors, likely outcomes, etc.

offerings for Population-wide management/ACo LeadershipInformatics-based risk prediction for Population/Leadership – depending on the scope and depth of laboratory and clinical data we have available to us for informatics and data-mining, there may be many opportunities to produce very valuable applications for ACO population management.

Subpopulation reference ranges for more precise disease detection – With sufficient population data, we can use statistical techniques to derive more “individualized” reference ranges for lab tests. Applying these more person-alized ranges equates to more clini-cally powerful testing, including earlier disease detection and more potent risk assessment.

Prediction of next year’s high-cost pa-tients. A minority of patients comprise the majority of health care spending. A typical split is 20 percent of patients representing 80 percent of costs.

In Chen’s research he observed that 3 percent of the population each year went from the low spending groups to the highest spending group. In other words, the 5 percent highest spend-

37G2 Intelligence interview with Philip Chen, MD, PhD, on May 25, 2012



219

38G2 Intelligence interview with Philip Chen, MD, PhD, director of informatics for the Department of Pathology at the University of Miami, June 15, 2011, in “Creating a Value-Driven Laboratory: Opportunities in the New Marketplace,” G2 Intelligence, 2011 Kennedy Information, LLC

ing patients are not all chronically high spenders. This presents an opportunity for a laboratory to develop analytics to help identify these 3 percent, high-risk, “time bombs.” He commented that patients typically in this 3 percent were diabetics who are not well managed and are first time myocardial infarction patients.38

With an understanding of how to read changes in lab results in the context of various combinations of disorders, combined with statistical, or perhaps artificial intelligence programs (avail-able off-the-shelf), we may be able to find algorithms that identify patients at high risk of becoming the 5 percent highest cost patients in the coming year. This would enable interventions that might mitigate that risk.

Identification of high adverse event ar-eas – Applying our expertise in statisti-cal quality control to unusual laborato-ry test results may allow us to develop applications that alert ACO leadership to potential problems in particular pro-vider areas.

Example: An early warning surveillance application to identify care units ex-periencing/about to experience high rates of adverse events. Using labora-tory QC process control charting tech-niques and critical value test reporting frequency “spikes,” a researcher at Ohio State University Medical Center

developed an application that pre-dicted units with 2.5 times higher rates of adverse events in the two weeks fol-lowing the critical value spikes.

A patient adverse event was defined as cardiac injury, cerebrovascular adverse events, major bleeding epi-sode, prolongation of hospital stay, or death. With the lead time provided by the early warning, the care unit can be assessed for operational, qual-ity, and resource problems. Also, ad-ditional care management services and interventions can be directed at particularly high-risk patients. Although this application likely needs validation, it appears quite promising and would have significant value from both a cost and quality perspective.

individualized medicine guidance for Population/Leadership – Development of guidelines for PGx and MDx testing by risk status and cost-effectiveness modeling.

• A component of our practice pathway’s leadership for the ACO is developing a type of molecular testing “formulary,” with associ-ated guidelines or decision trees. Again, the guidelines/protocols for when to test and which tests to use should be individualized based on a patient’s risk status, determined in part by lab results. A major driver will, of course, be



the level of clinical evidence sup-porting a particular indication. Also, our analytic assessment of the various assay platforms will be important in selecting and validat-ing testing systems.

• A new role in this regard for many of us involves an economic assess-ment of these tests, not from the perspective of the test cost versus its revenues, but from a popula-tion health care perspective. This essentially entails conducting value-based economic modeling whereby the test’s impact on sub-sequent spending and outcomes is compared to not performing the test in various population cohorts with different risk and clinical pro-files.

• By working with economic spe-cialists or the ACO, we can help develop these economic analy-ses, and use them to inform deci-sions regarding when to introduce new MDx and PGx tests into ACO guidelines.

Quality and Coordination Programs for Population/Leadership – Pathogen surveillance and/or antibiotic steward-ship programs – the opportunity for a tremendous value impact is simply too large to not take a leadership role in advancing whatever infection con-trol/antibiotic stewardship might be in place at the ACO’s facilities at the start. There are a number of ways we might take these programs to a new level of effectiveness.

As we’ve emphasized, resistant organ-isms are a major problem, both in com-munity and hospital settings. Although MRSA rates seem to be declining na-tionally, gram-negative organisms are showing disturbingly accelerating rates of resistance. With the advent of mo-lecular testing for pathogen identifica-tion we have a truly effective answer in hand, and we need to utilize our expertise to bring this into routine care in the near future.

A recent study used rigorous analytic techniques and a robust database to estimate the differences in costs between resistant and susceptible pathogen infections in community and health care-acquired infections. They found, “Consistent with previ-ous findings, our results from all mod-els indicate that resistant infections lead to generally higher charges and length of stay for both healthcare- and community-acquired infections.” The differences in hospital charges were significantly higher, with an average of $15,626 higher for resistant health care-associated infections and $25,573 for community-associated infections. The infections studies included pneumonia, blood stream, and urinary tract.39

Antimicrobial stewardship programs reportedly generate from $200,000 to $900,000 in annual savings from drug costs alone.40

Example – components of a steward-ship or surveillance program we could lead include:



221

39MJ Neidell et al., “Costs of Healthcare- and Community-Associated Infections With Antimicrobial-Resistant versus Susceptible Organisms,” Clinical Infectious Diseases Advance Access, published June 14, 201240Shira Doron, MD, and Lisa E. Davidson, MD, “Antimicrobial Stewardship,” Mayo Clin Proc, 2011;86(11):1113-112341Frederick L. Kiechle, MD, PhD, FCAP, medical director of Clinical Pathology at Memorial Healthcare System, Hollywood, Fla., “How to Build and Fund a Financially Viable Molecular Lab,” College of American Pathologists webinar, 2010

• MDx testing protocols integrated into antibiotic pre-prescription guidelines.

• Using MDx testing to direct antibi-otic de-escalation.

• Conducting, analyzing, and dis-seminating frequent antibiograms throughout the inpatient and out-patient ACO network, and work-ing with the pharmacy to continu-ously update and correlate drug usage data.

• Methicillin Resistant Staph Aureus (MRSA) and C difficile screening programs.

– Frederick L. Kiechle, MD, PhD, FCAP, medical director of clinical pathology at Memorial Healthcare System, Hollywood, Fla., estimates that his institu-tion’s MRSA has generated $8.8 million in cost savings.41



strAtegiC PLAnning

The "Coordinated Population Care Services Practice Pathway" is suited primarily to pathology practices with hospital, health system, and ACO rela-tionships, though there are many dif-ferent options for serving other clients, including payers or specialized ACO management organizations such as Collaborative Health Systems, a subsid-iary of the Medicare Advantage pro-vider Universal American. Collabora-tive Health Systems jointly owns several Medicare ACOs, most physician-led, and provides IT analytics, coordinated care, care management, and other services.

This section details the strategic deci-sion making that practices interested in pursuing this Pathway may follow as they plan to implement population care practice pathway services. Plan-ning steps include choosing a strategic direction that will be a good fit for your practice, identifying a compensation model, and setting a course for long term investment in this Pathway.

Finding the right FitThe first step in pursuing this pathway is developing a strategic plan to guide our implementation. In this context, strategy means assessing internal and external factors, and choosing an area of competitive advantage for initial focus. There are a plethora of potential applications, services, and programs we might deliver for population man-

agement, and unless we have sub-stantial human and capital resources we cannot implement them all initially.

Furthermore, as we have emphasized previously, some of these services can be delivered by other types of provid-ers. We bring special advantages, but in order to realize these advantages we need to choose a population manage-ment specialty area so that we can establish our expertise and “own it.”

To determine an initial focus, we need to consider the relative strengths and weaknesses of our group in two key areas:

• Molecular testing

• Informatics

We will eventually need to bring in expertise and resources to ensure both of these capabilities are strong, given they are key engines for this opportu-nity.

We also may choose a particular population level or provider user group to define the service target space in which we will prove our ability to gen-erate value. We have distilled some of these choices into four possible axes to help frame the selection space for a population management specialty area: Provider Setting; Application; Disease Area; and Technology. (See Figure 12)



223

Figure12:PossibleAxesforDefiningaPopulationCare Practice Pathway “specialty Area”

Provider Setting• PCP Focus• Hospital Focus• Specialty Focus• Community Focus

Application• Risk Prediction• Utilization Management• Chronic Care Management• Safety/Adverse Events

Disease Area• Diabetes• Cardiovascular• Infectious• Oncology• Etc.

Technology• Informatics• Molecular Testing

Source: Eleanor Herriman, MD MBA, G2 Intelligence



Applying these axes allows for mixing and matching of skills, technology, medical specialties, and provider set-tings to identify population care prac-tice pathway services that are a good fit for an individual practice. Examples include:

• Molecular diagnostics for infec-tious disorders.

• Clinical laboratory informatics for risk prediction.

• Adverse event reduction in hospi-tals, ERs.

• Decision support services for chronic care management – e.g., diabetes.

• Targeted pharmacogenetic test-ing for value generation.

Ideally, we find an area to start that fits our practice’s capabilities/plans for acquiring the expertise and has high perceived financial value for our tar-get client.

If our practice is neutral about what to stake out as our own “specialty,” we might target the ACO leadership’s high-est priorities and needs. For example, are they prioritizing inpatient adverse events, PCP chronic care manage-ment, or infections most highly? Per-haps provider coordination is a con-cern, or clinician guideline adoption.

Unless we have prior experience with the types of population management services described for this pathway, we need to begin the implementation of

population care practice pathway ser-vices by providing services in our cho-sen “specialty” focus area for at least one to two years. These years of expe-rience are needed in order to build the financial and clinical evidence base that enables us to properly “price” our “at risk” offerings, design disease-spe-cific models, and develop more com-prehensive solutions.

identifying a Compensation modelWe want to have a plan for a pro-posed compensation model before we initiate business discussions with our potential client, as this is a key element of the negotiations so we should have a position.

Our alternatives for models will, of course, be defined by our practice setting and what types of clients we’re targeting. Here we’ve described five basic, potential compensation mod-els, as shown in Figure 13, that range from a salary with variable compensa-tion model to one with a share of the ACO’s global profits (i.e., the “gain-sharing” or shared savings payment they receive). Figure 11 contains two tables, the first showing what compen-sation models might best fit by type of practice setting (independent, hospital employed, AMC, or specialty/refer-ence lab), and the second showing compensation models according to target client type (ACO, Health System managing a population, health plan, or firm).

The colored green boxes indicate compensation models (the rows) that represent likely alternatives for a prac-



225

tice either in a particular setting (top table) or targeting a particular type of client (bottom table).

For example, for a practice employed by a hospital or other provider insti-tution, such as the columns labeled “Hospital Practice” and “Academic Medical Center/Health System Prac-tice,” the compensation model will likely involve having a portion of staff time and thus salary linked to Pathway services work. Ideally there would be a variable component of this salary con-tribution that is incentive based, and the performance-based models are selected to indicate that these might be used to structure that variable in-centive pay.

An independent practice has a range of possibilities. If the practice is not contracted with an organization that is becoming an ACO, then it might consider launching the Pathway with a health system or large hospital that is likely to form one and is interested in population care management. In this case, a good way to start is with a service contract to cover the Pathway services, which might be negotiated under Medicare Part A or clinical pa-thology professional component pay-ments.

This might also be structured as a short-term service contract in which the practice goes partially at risk with part of its compensation based on the suc-cessful generation of cost savings and quality improvement from specific pro-grams. This has the great advantage

of providing experience and evidence for the practice, which will be invalu-able in future negotiations.

If the independent practice is target-ing a hospital, provider organization, or payer that is forming an ACO there are still more options. ACO models vary widely and are still in flux, so compensa-tion models will depend greatly on the individual flavor in your local market.

Based on case reports from the field, it seems that at least two alternatives are gaining some acceptance; these include the “Performance – share of specific programs” and “Performance – share of ACO” models. Each model involves the practice negotiating a share of value-based compensation, either cost savings generated from specific programs that the practice leads or participates in, or a share of the ACO’s total value-based compen-sation pool.

In the case of the share of specific pro-grams model, the practice might go at financial risk for the results, meaning payment is contingent on successful generation of cost savings for the spe-cific initiatives the practice is propos-ing. In one such case, a practice con-sidered a 10 percent share of the costs savings for its contract, which was a long-term contract.

For the share of the ACO’s total pool model, the practice will likely need to be involved in a wide variety of pro-grams in order to contract for a share of the entire ACO compensation pie. In one recent case a pathology prac-



tice was awarded a 5 percent share of any ACO shared savings generated.

Pathologists working at a specialty lab or independent group might also con-sider approaching a firm that provides specialized services to ACOs, such as Collaborative Health Systems. This area is likely the most open for innovation and entrepreneurship, because there will be interesting commercial ecosys-tems of service and tool companies arising to power the new value-based care models.

We can “bottle” our expertise in a number of ways and work with these firms as channels to distribute novel products, tools, and services of our

own. Here compensation might go beyond service contracts and perfor-mance pay and enter the realm of equity and royalties.

Because the value-based market is an emerging market that is in different stages in local areas, we recommend reaching out to as many candidate clients as possible. This is especially important as your local market begins to make the leap into the value-based market. Once one institution or payer in a local market launches an ACO, the experience is that a domino effect quickly follows. Starting discussions with potential clients as soon as you hear rumors is wise.



227

Figure 13: Potential Compensation models for Population Care Practice Pathway by Practice setting and Client type

Practice settingCompensation model independent

Practicehospital Practice Academic

medical Center/health system

Practice

specialty or reference

Lab

Salary + Variable

Service Contract

Service + Performance

Performance – share of specific programs Performance – share of ACO

Client type

Compensation model ACo or hospital within ACo

health system managing Population

health Plan

Firm offering Coordinated

Care/Pop management/

it servicesSalary + Variable Service Contract Service + Performance Performance – share of value from specific programs Performance – share of ACO "profit"/gainsharing Note: The check marks indicate compensation models (the rows) that represent likely alternatives for a practice either in a particular setting (top table) or targeting a particular type of client (bottom table).Source: Eleanor Herriman, MD MBA, G2 Intelligence



Considering the Potential risksAs with all new ventures and emerg-ing markets, risks abound. Rather than ignoring them, we are best advised to educate ourselves and, when appro-priate, create plans to mitigate them. Having said that, these challenges shouldn’t be reasons for inaction. There are likely far more risks associ-ated with the status quo.

Here we discuss a few of the potential risks. They include the regular resetting of the cost-savings bar, others usurping our value propositions, and market tim-ing challenges.

the rising Cost Bar For those new to value-based pay-ment programs, one frequent concern is that the cost bar is continually being reset. In other words, the bar is always rising. Even as value-generating pro-grams are introduced and successfully generate cost savings, payers will use that new, lower cost basis as the base-line for the next round of bundling or global payment contracts.

As you pursue this pathway and face this concern yourself, from colleagues and from providers, remember that all health care providers are in this same boat, so the playing field is level. Also, there are few alternatives to cost avoid-ance initiatives for generating the vast majority of value over the next decade. Most importantly, pathology is in a rela-tively advantaged position for meeting this challenge for three reasons:

• The sheer number of opportunities we have across conditions,

procedures, settings, and provid-ers for generating clinical cost savings is enormous. Implementing these services can keep us busy for several years.

• As the science and technology of diagnostic testing, and ge-nomics in particular, continues to advance, we will be able to bring new knowledge and increased precision to clinical decision mak-ing. This will translate into new sources of cost savings.

• Our informatics advantage is inherently one that offers an expe-riential learning curve. This means increased knowledge over time that, again, translates into new opportunities for cost savings.

Lab Data Analytics – Competing to Add ValueNo one disputes the tremendous power, indeed the essential role, of diagnostic testing data in population health management. That is why most ACO databases used to make clini-cal and resource decisions will contain diagnostic test data.

The potential concern or challenge is that other providers and informat-ics experts mat be able to evaluate and analyze these lab data as well as pathologists for purposes of population health management needs.

First, it is unlikely to be this simple. In other words, as we’re describing in this chapter, there are a multitude of ways in which lab data is needed for popu-



229

lation health management. It is most likely that the optimal implementation of many of these programs will be a collaborative team that brings differ-ent, complementary skills, and certain-ly, in most cases, includes pathologists.

There may be a few areas, such as specific cost-effectiveness analyses, where assuming pathologists do not have cost-effectiveness analytic skills, and assuming the lab data as it relates to that particular clinical context is not complex, pathologists may not be needed.

However, we will need to demonstrate the differential expertise we bring, or else we risk being marginalized as not necessary for many of these analyses.

We bring specialized value to analyz-ing lab data in a wide variety of ways, as has been discussed previously in this chapter. For example, understanding how reference ranges from different labs must be accounted for in analyz-ing data that aggregates from many locations; identifying patterns and meaning from certain test values and dynamics that others miss because of our deeper understanding; and applying knowledge of how to differ-ently interpret multiple tests results from certain patients due to interactions or particular metabolic states.

Besides the traditional, pathologist-spe-cific expertise we bring, we might also consider learning about cost-effective-ness and other economic tools used in population health management. In this way, we can further mitigate concerns

regarding our competitive value prop-osition in lab data analysis.

specialty Population management Companies for ACosWhenever major new market areas emerge, such as coordinated care and ACOs, capitalism generates new business ecosystems to feed them. So it has begun for care delivery reform, with Collaborative Health Systems be-ing just one example.

The potential risk is that some of these new ventures might encroach upon our Pathway service programs, and out-compete us due to their size, re-sources, or ability to specialize and reinvest. A further concern is that gen-eralist pathologists offering a wider range of Pathway services to ACOs may be more vulnerable to such a firm that specialized in a particular man-agement service area.

No doubt there will be competition, and we will likely not be able to pro-vide all of the services described in this Pathway with a competitively bul-letproof advantage. It is unlikely that any such company could find a way to package a pathologist’s expertise for population health management services in a way that could be widely distributed, except for computer deci-sion support tools.

The utility of such tools without effec-tive pathologist services is unclear. Many clinicians highly value knowing, trusting, and consulting with patholo-gists when it comes to clinical decision making.



The questions are –

• Is there a sufficiently sized market for us to participate, own a seg-ment, and thrive?

• What can we do strategically to mitigate this risk?

The answer to the first question, we believe, is absolutely yes. Our estimates project a substantial market, with ample room for losing share to competitors.

In terms of mitigating this risk, one of the most effective tactics is getting to market early. If we establish relation-ships and effective, value-generating programs with an ACO, they are far less likely to sever those ties for a new service management company.

Secondly, we should be working to-ward expanding our capabilities as we implement the pathways. Continu-ing education, on-the-job training, whatever we want to call it needs to be the default, not an option. Staying ahead of the wave of market partici-pants who will be scrambling for value positioning will require honing our own areas of expertise.

Below, in “Long Term Strategic Position-ing” we also discuss other approaches to solidifying our competitive value and market position.

Finally, a rather contrarian way to miti-gate this risk is a “coopetition” strategy, whereby we might partner with the firm in order to expand the range of programs and value we can provide the ACO. This might be of interest if

we are already working with an ACO and a competitive company that offers EMR tools that would be valu-able for our programs approaches the ACO. By partnering we might both be more successful.

the market entry ChallengeTiming a market is notoriously difficult – as made clear on Wall Street every day. Likewise, knowing when to cut back on our fee-for-service work and begin investing time on value-based services such as the Promising Practice Pathways is challenging.

This decision can be examined from a number of perspectives – risks ver-sus benefits, financial and strategic, and short versus long-term. It may, of course, be a matter of what is practi-cal. If reform comes knocking on our local hospital door in the form of an ACO, it makes the decision easier.

Perhaps all the more there is to say about this challenge is the optimal an-swer likely depends on which of those perspectives you chose to prioritize.

Longer term strategic PositioningBefore we move on to pathway Imple-mentation, let’s look out a bit farther down the road for options for our stra-tegic positioning.

In general, our best opportunities in a value-based market for controlling our economics and margins are to accept financial risk in contracting with ACOs, providers or payers, and to develop complete solutions, as this optimizes value delivery.



231

To advance our value position and economic control over the longer term, as payment and delivery reform spreads even further, we will want to progress our offerings toward either or both of these two paths (see Figure 14):

Developing Full Potential SolutionsDeveloping complete solutions in-volves aggregating individual services and tools to target a particular condi-tion, procedure, or clinical problem area. Complete solutions have the dual advantage of yielding synergies between the individual interventions and aligning with popular episode-of-care, or bundled, payment programs. By approaching providers with a mul-ticomponent program that addresses numerous sources of value gaps, we can demonstrate that our contribu-tions warrant equitable shares of the bundle.

By applying knowledge we generate through experience working with a population, mining data, or develop-ing expertise in a disease or technical area, we can better price, design, and deliver these new service programs, thereby reaping differential profits.

Taking on Financial Risk Accepting financial risk means, for example, that we offer a service pro-gram that includes both a “value-gen-erating component” (e.g. knowledge services and tools), as well as a diag-nostic testing component, designed to cover all testing needed for either a population or episode of care. The program would likely have two types of pricing components – the “at risk” aspect, either per patient (capitation) or per episode of care (“bundling”), and the value aspect, either a perfor-mance-based bonus or gain-sharing arrangement.



Figure 14: Progression of Coordinated Population Care Pathway offerings Along implementation Phases

PopulationManagement

Services

Core Pathway Services:• Informatics risk prediction• Individualized medicine• Utilization management• Quality and coordination

Example PathwayIntervention forEvolution and

Expansion Phases

Full PotentialSolutions

Take onFinancial Risk

Diabetes Solutionfor Bundled

Payment

DiagnosticsBenefits and

Value Program

2017-2022




233

imPLementAtion

In this section, we will provide high-level “planning guides” for implement-ing the pathway in terms of technol-ogy, medical practice, and business operations. We have divided the pathway implementation journey into three three-year phases – the “Launch Phase” (2014-2016), the “Evolution Phase” (2017-2019), and the “Expan-sion Phase” (2020-2022). As discussed above, the Evolution and Expansion phases will include discussions of pro-grams in which the practice might go at risk financially.

Given the numerous service applica-tions and permutations inherent to this pathway, we cannot provide an actual implementation guide in the context of this report. Instead, our ap-proach in this section will be to define a framework for developing the core categories of pathway services, and to include examples with more details to illustrate potential development paths. CAP is planning to support our efforts at implementing this pathway by of-fering a number of resources over the course of 2013.

Further, we’ll discuss operational and financial considerations during the launch phase.

Launch Phase 2014-2016 Planning guide The Launch Phase begins at the point at which we have initiated discussions with one or more provider organiza-

tions, and are committed to pursuing this pathway with some client in our local market.

We have also developed our strategic plan, as described above, and thus perhaps chosen a “specialty area” for our initial population manage-ment focus. This need not be the case, though, if we have not yet signed a contract and thus finalized our popula-tion provider partner.

We will discuss developing core cate-gories of pathway services broadly, but if we have chosen a particular area for specialization, then we may be able to just begin with services covering those applications. This really depends on the financial value potential of the specialty area selected, and whether it suffices for our partner ACO, at least for the first year or two.

Development of Core Pathway Services and ToolsThere are two primary strategies for implementing an initial set of services: an externally-focused approach that involves leveraging ACO capabilities and resources, or a more internally-fo-cused build, buy, or borrow approach.

Even smaller community pathology practices may gain entry into the population care practice pathway by leveraging the ACo’s technologies, resources, and capabilities, especially informatics and IT systems. This strategy



will be most successful for practices that have an existing relationship with a hospital within the ACO and are will-ing to expand their breadth and depth of services and capabilities.

In particular, ACOs should already have databases with diagnostic test-ing, and clinical and other data inte-grated. Many or most ACOs over the next few years should have advanced Heath IT (HIT) systems that integrate dif-ferent types of data, as well as strong informatics capabilities, because these form the backbone for population care and financial management.

By working with ACO experts and resources, even community practices with relatively little informatics expertise may be able to develop “informatics services.” For example:

• Instead of conducting data-min-ing ourselves, we may be able to submit queries to the informatics group and have the results report-ed back.

• The ACO informatics teams can help implement the most appro-priate EMR applications once we have decided on the content and functionality of the tool or service.

As with our other Pathways, we can take a build, buy, or borrow approach.

• Build: For example, we will cre-ate content to be implemented through an ACO’s EHR, EMR, or other HIT platforms. We will also build informatics applications by writing queries for others to run on

integrated ACO databases.

• Buy: Examples of components that can be bought include com-mercial software applications available for molecular testing de-cision support, such as McKesson’s “ClearOrders,” and data-mining packages. We may also retain consultants to develop offerings and train staff.

• Borrow: This option may involve forming a partnership with a spe-cialty lab or venture to deliver a program. The partnership could in-volve shared revenue and include a contract with knowledge trans-fer plans built in. A partnership could also be made with Aca-demic Medical Center research-ers to develop an offering. Such a contract could include providing them with IP and “royalties.”

Once you have selected an approach for developing services, the work of development begins. Here, we pro-vide a Framework to follow during the implementation of services. The Frame-work involves three primary steps:

• Research

• Development

• Delivery

This comprehensive multi-step ap-proach may not be necessary for all services. For some very specific and targeted services, the background research required may be minimal and the validation of an approach may



235

already by publicly available, for ex-ample.

This Framework process can also be applied to a group of Pathway services that fall into the same category, based on application, condition, or practice setting. Using this Framework, it is pos-sible to propose to an ACO one or more “programs of value creation" that target particular areas or issues, or that leverage particular technologies, such as informatics or molecular testing.

The research phase defines the value proposal for the service offering in terms of the outcome or quality im-provement and/or the cost savings the service will achieve. Creating a clear definition requires information gathering and synthesis to determine the technology and tools required for development of the service, including, for example, the appropriate molecu-lar and genomic tests involved, the informatics techniques and statistical algorithms to apply, the point-of-care instruments for trials, and the platforms to use for gathering data to assess the value-add.

“External” information gathering will support the design of the service or tool. This data gathering might involve a literature search, consultations with clinicians, pathologists, or other experts, and an assessment of the commercial market for similar tools and services. “Internal” research will be specific to the client (ACO or provider group), including an analysis of data, interviews with clinicians, an investigation of the

problem the service will help solve, and an assessment of existing solutions.

The research phase should also yield a model of the financial impact of the service that will supply a basic projec-tion for the intervention’s impact on clinical quality, outcomes, and cost savings, as well as a picture of the resources required to implement the service and an understanding of any capital investment required.

The Development phase focuses on the design of the offering and the creation of a prototype service along with plans to test it. This phase will likely begin by convening relevant clinicians who will help advise and co-develop the service or tool.

Several key questions must be an-swered while designing a new service or tool. How does clinical behavior need to change to create the offer-ing’s projected value? What interven-tion can impact that change? What is the content? For example, will the ser-vice or tool apply algorithms, prescrib-ing guidance, patients-at-risk reports, Dx guideline be integrated into existing guidelines, etc?

Implementation will initially involve the creation of a prototype and a trial, advised by clinician advisors, to ensure that the service is working as designed and meeting efficacy goals. The proto-type should include a mechanism for automatic collection of data required to support learning and continuous improvement of the service offering and to generate evidence of value to



support negotiations with clients.

During development, it is helpful to keep in mind these “Ten Command-ments,” as described by Bates and col-leagues, for effective computer deci-sion support applications:42

1) Speed is everything.

2) Anticipate needs and deliver in real time.

3) Fit into the user’s workflow.

4) Little things can make a big difference.

5) Recognize that physicians will strongly resist stopping.

6) Changing direction is easier than stopping.

7) Simple interventions work best.

8) Ask for additional information only when you really need it.

9) Monitor impact, get feedback, and respond.

10) Manage and maintain your knowledge-based systems.

The Delivery phase determines how the service will be integrated into prac-tice. For example, it may be delivered at the point of care as a computerized physician order entry (CPOE), an alert, or a readout on an EMR chart or dash-board. It may also be a tool delivered

through conferences or consulting services or some other mechanism.

If the service involves the implementa-tion of a new HIT tool, it is important to work with the ACO’s HIT support group and Informatics Directors to define how the tool will interface with existing tools and to plan integration.

If the service involves a new institu-tional or community program, several delivery channels must be worked out. The new service may require complete project plans, training of staff members by pathologists, the assignment by the ACO of nursing or other resources, the definition of a new function or service within a hospital or among community PCPs, or data collection from a com-munity (e.g., wireless collection for a group with a certain condition) to allow pathology to monitor or improve an existing initiative.

To understand how the Framework brings up different considerations for different types of services, we ap-ply it to the four general Application categories taken from the Pathway definition (See Figure 11 “Coordinated Population Care Practice Pathway Of-ferings Matrix”):

• “Risk Prediction” – using laboratory testing informatics to identify pat-terns that triage patients by risk for a disorder, or an outcome, com-plication, episode cost, etc.

42Bates DW, Kuperman GJ, Wang S, Gandhi T, Kittler A, Volk L, et al., "Ten commandments for effective clinical decision support: making the practice of evidence-based medicine a reality," J Am Med Inform Assoc., 2003;10(6):523–30



237

• “Individualized Medicine Guid-ance” – including molecular diag-nostic tools for antibiotic selection, programs for preventing avoidable ER admissions, and oncology tools.

• “Utilization Management” – in-cluding test-ordering guidance, blood product utilization guid-ance, and decision trees for anti-biotics selection.

• “Quality and Coordination” – in-cluding EMR apps for changing prescriptions or protocols for refer-rals, accurate glycemia control, and pathogen surveillance tools.

implementing the risk Prediction Pathway services CategoryRisk prediction is the use of informatics to predict risks among patient groups. Deliverable services include reports on patients at high risk of complications, worsening conditions, hospital admis-sions, or adverse events. Depending on the setting, risk prediction tools may provide recommendations for testing protocols, monitoring, referrals, or inter-vention guidelines.

Research is required to determine if the proposed service meets key criteria:

1) High-value delta: altering the clinical management of patients would result in a major change in value. The intervention can create substantial increases in outcomes and cost savings.

2) High priority for ACO or provider: the service targets a problem area for provider group, as per-

haps identified in Medicare per-formance metrics or by internal management.

3) The services falls within the prac-tice’s focus/specialty area.

The Research phase will vary depend-ing on the approach to risk prediction, whether the service is based on known biomarkers or on patterns discovered through data mining. For instance, if your practice has little to no informat-ics capabilities (yet), the biomarker ap-proach may be more appropriate.

To begin investigating validated bio-markers approach for a service, start by looking for existing clinical guide-lines or pathways for risk prediction for a target cohort or outcome. Deter-mine if the ACO is implementing these and if they include diagnostic tests or biomarkers. If so, and if the ACO has not yet implemented them, this is a service to include in a pathway pro-posal.

Supporting information that must be gathered will include details about how risk prediction can be advanced using diagnostic biomarkers, with pub-lished evidence of the utility of can-didate biomarkers. This research will likely involve a literature search and reports presented at conferences from specialty labs or academic medical centers.

In the more advanced, informatics-based data-mining pattern generation approach, statistical techniques are used to analyze databases contain-



ing diagnostic path and lab data, as well as clinical data (and ideally claims data, too). The objective is to identify diagnostic test patterns (e.g., a group of assays) that predict a particular outcome in a particular population co-hort, for example. This identified group of diagnostic tests, with threshold val-ues determined by statistical analysis of the database, is then implemented prospectively by monitoring the test results and signaling clinicians appro-priately.

Development of these services will involve design decisions, lab validation (as needed), and in some cases pilot testing.

Clinical collaboration is important for making key design decisions:

• To which clinicians/clinical setting/guideline is this risk intervention directed?

• At what point in clinical workflow should the clinician be informed of risk?

• What action should they take if there is a specific recommenda-tion?

Example actions:

– Switch therapeutics or start new therapeutic.

– Order more diagnostic testing.– More closely monitor patients.– Assign care management

resources.– Refer to another provider.

– Admit or do not admit to hospital.

• What are the best threshold val-ues for assays?

• What content should be included in communication to clinicians?

Design decisions also include selec-tion of source assay(s) for selected biomarker(s). Decisions will hinge on several considerations, such as wheth-er or not the test is already performed in the laboratory or if it must be sent out to an external lab. Kit availability will factor into decisions, as will cases in which the laboratory must develop new tests. Laboratory validation will also be required for new tests and potentially for send-outs or newly ad-opted kits.

If a biomarker is relatively new, untest-ed, or its clinical impact could be par-ticularly high risk, a short, observational pilot period during initial implementa-tion is recommended.

To create a business case that dem-onstrates the value generated by the service, the development state should include modeling that integrates pub-lished evidence with data from the ACO or hospital prototype.

To determine the most appropriate means of service delivery, the acute-ness of the setting and urgency of decision making must be considered. In some cases, the service may be de-livered as a consult via phone. While, in most cases of Risk Prediction, services



239

will be HIT based; we will not need to be creating software. Instead, work with the ACO’s informatics group and arrange to deliver content and specs to them. However, delivering risk pre-diction may require the implementa-tion of point-of-care diagnostic testing in hospital settings.

Example: Risk Prediction Pathway Services Category – Hyponatremia risk service

Based on evidence indicating strong predictive power of mild hyponatremia for falls and hip fractures, as well as a worse prognosis in certain conditions, we develop a service for hyponatre-mia monitoring in high-risk patients. In high-risk patients, including CHF, cir-rhosis, and renal failure, whenever the sodium level hits threshold (even mild low levels), EMR or EHR alert goes out to clinician explaining prognostic im-plications with recommendations for additional interventions, as agreed to with clinician collaboration during development.

• Inpatients with hyponatremia may receive alerts/messaging to be put on additional precautions for fall preventions.

• Patients on thiazide diuretics are identified as high risk (as indicated by EMR or pharmacy data). EMR messages that recommend that PCPs to test for sodium levels can be delivered.

Implementing the Individualized Medicine Guidance Services CategoryIndividualized Medicine Guidance ser-vices direct and optimize the use of mo-lecular diagnostics (MDx) and pharma-cogenetic testing (PGx) for therapeutic and other clinical decisions. Services and tools include HIT tools, guidelines, programs, and consultations.

Research into these services will pro-vide information that will help direct focus towards particular “problem areas” that have the potential for high-value improvement with targeted or individualized therapies. For example:

• Targeted antibiotics for infectious diseases in inpatient setting

• Pharmacogenetics for coagula-tion disorders

There is an extensive and rapidly ad-vancing body of literature in this area that provides support for increasing numbers of MDx and PGx test solutions. In addition, it is possible to collaborate with instrument vendors to receive state-of-the-art information on MDx point-of-care (POC) capabilities.

The development of individualized medicine guidance services should be done in collaboration with clinical colleagues who are experts in infec-tious disease and cardiology, as well as internists and PCPs. Development begins with identification of the value-generating components of the testing solutions first. Detailed design consider-ations for implementing the service in clinical care follow.



For example, select a test and instru-ment that can be implemented for targeted POC MDx for antibiotic selec-tion in pneumonia. According to pub-lished reports, targeted antibiotics can shorten the length of stays.

For inpatient therapeutic decisions, turn-around time must be quick, so tests will need to use POC MDx instru-ments. The components of the service will include sourcing, installation, train-ing, quality assurance, and perhaps results capture and monitoring. The service might also include pathologist consultations and inpatient “round-ing,” especially for infectious diseases.

Clinical guidelines for using MDx or PGx testing could also be incorporated into EMR of EHR or CPOE. Other HIT ap-plications to optimize test use might include the support of ordering and in-terpreting of results, recommendations for treatment, and links to protocols.

Delivery of the services will include more than offering a new test. Many, possibly even most, clinicians are not knowledgeable about the latest in MDx and PGx testing, so a major com-ponent of our services will need to be knowledge support and education. For instance:

• Clear directives on when to use the tests.

• How to interpret results.

• What actions to take given the results.

The best medium of delivery of knowl-

edge and education services will depend on the scope and setting of providers, as well as where and when in clinical workflow guidance takes place.

If the individualized medicine guid-ance services involve the introduction of new tests into routine care, e.g., PGx, or testing with new POC instru-ments, develop a proposal with a business case based on an economic model of value generation. If services include POC MDx testing, the delivery plan will need to implement a training program and a means for continuous collection of results for measurement of value generation.

Example: Individualized Medicine Guidance Services Category – PGx for clopidogrel in cardiac stenting settings

This could be implemented as either a preemptive program or as a reactive, point-of-care (POC) testing approach. CYP2C19 genotyping tests have be-come quite accessible, and interpre-tation of their results with respect to Clopidogrel dosage guidelines has a substantial evidence base in the litera-ture. In collaboration with cardiologists and interventionalists, developing pro-tocols for this testing in stenting settings should be feasible.

A preemptive testing program might genotype CYP2C19 alleles for ACO population patients at high risk for cardiovascular disease and store the results in the EMR. If and when these patients require stenting, or have an-other indication for Clopidogrel, the cli-



241

nician could refer to the genetic data to prescribe the right dose or an alter-native drug. Testing for CYP2C19 al-leles also provides information relevant to a number of other medications, including antidepressants, barbiturates, and proton pump inhibitors.

A reactive, POC approach requires testing at the time of stenting. POC PGx testing is conducted to guide the Clopidogrel dosage as above. The greatest challenge may be the turn-around time of the POC genotyping test, which may argue for the preemp-tive approach.

Vanderbilt University Medical Center recently published a paper supporting the use of preemptive PGx testing as superior to reactive testing based on a program they have implemented across their system.43 They base this recommendation on an analysis of the number of adverse events prevented, while accounting for the economics of the program. Their program includes CYP2C19 testing for Clopidogrel dosing support.

implementing the Utilization management services CategoryUtilization Management services op-timize utilization of various resources, therapeutics, and interventions. In terms of resources, services can en-sure that blood products and transfu-sions are utilized properly. In terms of interventions, services can assure that imaging test orders are not inappropri-

ate, particularly in cases of tests with high false positive rates.

Utilization Management services are not about managing the utilization of tests themselves. Rather, the services are based on the identification of inappropriate uses of services through analyses of test ordering patterns among providers. There is some value to be generated by optimizing test utilization, but it is relatively little com-pared to utilization management of other healthcare services since diag-nostic testing is only 3-5 percent of all healthcare spending. Also, focusing our value generating programs on reducing utilization of diagnostic test-ing signals to providers that we believe that is a good approach for finding value. By cost cutting through volume reduction, we are implying we are a cost center.

Research will help identify targets for utilization management services. To approach this investigation, first select a health care “resource,” meaning a service, intervention, procedure, condition, etc., that according to the ACO’s data has all of the following:

• High volume

• High costs

• High variability in usage

If analysis of usage reveals substantial inappropriate instances, with associ-ated avoidable costs and perhaps

43JS Schildcrout et al., “Optimizing Drug Outcomes Through Pharmacogenetics: A Case for Preemptive Genotyping,” Clinical Pharmacology & Therapeutics, Vol 92, No 2, August 2012



quality degradation, then this is a can-didate opportunity.

Next determine whether one or more diagnostic tests directly drive the us-age of this resource, and if those tests might also be inappropriately used or interpreted. If so, analyze whether there are wide variations in provider’s ordering patterns of those diagnostic tests. In this case, this is an excellent target for a utilization management high-value service.

A “standard” target is utilization man-agement of blood products. These programs are well developed and have been widely deployed. If the ACO/health care system does not have a robust utilization management program, this is good low hanging fruit. Protocols are available from other transfusion labs, online, and from pro-fessional societies. Consulting groups can also help implement turnkey pro-grams.

Once the target resource and diag-nostic tests are identified, develop-ment of a service intervention for changing clinician behavior begins. Key features to include in service de-sign include:

• Data regarding a clinician’s order-ing patterns versus their peers.

• An explanation for the optimal use of test, e.g. predictive power, interpretation of results, indications for ordering, etc.

• Economic consequences of inap-propriate ordering of test, such as resource overutilization.

• Potential alternative tests that may be more appropriate in a setting in which the test is being inappropriately ordered.

Delivery of utilization management services may include consultations or conferences to communicate pro-grams that add value through optimiz-ing resource utilization. One way to approach this is to introduce a con-ference series for clinicians in which each session discusses a different test or service. These communications can help communicate the message that diagnostic testing is the perfect “lever” for managing clinical resources.

Delivery will also involve implementa-tion at the point of care. Integration points might include alerts or messag-ing when ordering targeted tests in particular settings or integration into CPOE if test ordering is automated. The services might implement a test menu and other test-ordering messaging that is specific to that test and that explains the optimal utilization protocol.

Another delivery mechanism might include incorporation of test utilization considerations into an ACO clinical pathway or guideline.

Finally, the service might also provide clinicians with monthly reports of test or-dering versus peers with explanations of economic and value consequences.



243

Example: Utilization Management Services Category – CA 125 testing and imaging screening for ovarian cancer

An analysis of ultrasound imaging tests for ovarian cancers identified this as a high-cost area with low yield and high clinician variation. Further investiga-tion identified substantial differences in ordering patterns of CA 125 as a key driver of the variations in ovarian can-cer screening imaging tests.

Our service is designed to inform clini-cians at the point of care regarding the proper use of CA 125, as well as provide individualized reports on their ordering patterns of both CA 125 and ultrasound screening versus their peers. A cost accounting is also included.

Clinicians’ ordering patterns are tracked over time, and feedback regarding improvements or problems provided. Consultative interventions can be scheduled as appropriate.

implementing the Quality and Coordination services CategoryQuality and Coordination services comprise a variety of services that sup-port quality of care in different settings and coordination of care between providers. The services are designed to address a specific problem area, such as a high adverse event rate or poor adherence to a key clinical protocol, which has significant value-related consequences.

Research, such as reverse engineering of existing processes, can reveal qual-

ity “value gaps” where diagnostic test-ing and pathology can play a correc-tive role. Initially, in collaboration with ACO leadership and clinicians, we can use informatics analysis to begin to identify and design quality services. Ar-eas of initial investigation might include the following, which have proven ef-fective in the past:

• Pharmacy collaborations to im-prove prescribing data, patient adherence data, etc.

– For example, reducing adverse drug events due to poor moni-toring of drug levels, LFTs, etc.

• Closing the gaps in chronic dis-ease management, in terms of providers adhering to/implement-ing best practices.

– For example, tracking and analyzing various testing values associated with diabetes and co-morbidities, and applying established protocols to recom-mend/remind further testing to improve diabetes identification and management.

• Ensuring that point-of-care testing instruments that were adopted to drive important clinical decisions are being used correctly, inter-preted properly, and apply the best instrumentation.

– For example, hypoglycemic events and associated ex-tended stays in ICU, etc., have been found to be caused in part



by use of relatively inaccurate handheld blood glucose meters.

• Improving pathogen surveillance programs, a leading cause of avoidable costs in inpatient and community settings.

Quality and Coordination Services con-tent is developed in collaboration with clinicians, as with all pathway services. Because many of these services involve protocols or guidelines that are ACO wide, they also will likely need to be vet-ted by appropriate ACO committees.

Once a value-gap area has been identified, we can develop a solution, which might take a variety of forms. For instance,

• Alerts: Alerts the solution might be designed to send alerts to EMR reminders and decision support tools can be used to send mes-sages to PCPs to check LFTs or digoxin levels. If sequential lab tests suggest renal problems and the patient is on a drug that can cause renal injury (we know this from either connecting with phar-macy or EMR data analysis), a message can be sent to provider.

• Decision-Tools: Test results can be used to determine when a pro-vider should refer a patient to a specialist. For example, if HA1c levels are above a certain thresh-old for one year and the patient has not been started on insulin (we know this from either con-necting with pharmacy or EMR

data analysis), protocol states that the PCP should refer the patient to an endocrinologist.

• Instrument Management: A multi-component quality program for POC instruments could include selection of the most appropriate instrument for the indication’s per-formance requirements, econom-ics and user needs, distribution to the appropriate providers and set-tings, training and maintenance programs, quality assurance and proficiency, and the capture of instrument test results for reporting.

• Analysis: Informatics analyses of lab test results could help identify patients with chronic disease who are progressing or developing co-morbidities.

The best delivery platform can be determined in conjunction with our clinician collaborators. The optimal platform will be very specific to the intervention. If the best way to incor-porate content is via the EMR, options include CPOE menus and “alerts,” EMR dashboard tools, EMR reminders, and so on. Other options include consul-tations via phone, rounding, special sessions/webinars, conferences, and other educational outreach.

Example: Quality and Coordination Services Category – quality control program for POC glycemic control

We identify glycemic control in inpa-tients as an excellent “value-gap” opportunity, based on the National



245

Quality Forum’s inclusion of inpatient glycemic control as one of the 34 Safe Practices it recommends hospitals tackle. We determine that even with recent recommendations to switch from tight to moderate glucose con-trol, hypoglycemic events remain a problem in some populations, such as diabetics and surgical patients.

Based on our literature review, we determine that the brand of blood glucose meters (BGMs) used by our ACO hospitals has suboptimal sensitiv-ity for hypoglycemia. Furthermore, the BGM results are sensitive to hematocrit levels, introducing another source of unreliability. We calculate that with the insulin treatment protocols being used to control glucose levels in surgical and critical care patients, the degree of inaccuracy in these POC instruments could be accounting for a significant percent of hypoglycemic events.

While there are other BGMs that are relatively more accurate than the brand our hospital uses, we recom-mend hospital-grade POC blood analyzers. These instruments are more expensive, somewhat more difficult to use, and will require more nursing time, but based on our assessments, the accuracy and precision performance of the hospital-grade instruments is required to enable safe and effective glycemic control.

To convince providers, nurses, and ACO leadership that the switch is the right decision, we create an economic and quality model using projected ad-

verse event rates, and costs including nursing time. We also run a pilot study with one ICU using the blood analyzer and the other using BGMs. After the ACO has implemented the blood ana-lyzers, we launch training, proficiency, and quality assurance programs to ensure continued, optimal glycemic control performance.

operational and Business Planning The Launch Phase’s most important operational objective is to establish an ACO relationship (or comparable pro-vider organization managing a patient population). We should have a con-tractual agreement with that organiza-tion, and our goal should be to have that agreement include a value-based payment.

Securing such an agreement will first require getting “a seat at the ACO table.” This step starts with discussions with the ACO executive leadership to establish that our pathology laboratory group has a value proposition of sig-nificance. Making this case will require some type of evidence showing that our pathway interventions can pro-duce cost savings and quality improve-ments in whatever targeted areas we have chosen. This evidence can be a combination of peer-reviewed stud-ies and pre- versus post- observational data.

For services and applications that require development resources, there is a chicken-and-egg problem. We cannot generate the evidence un-til we have developed the pathway



interventions. In this case, we will need to create a business proposal focused on an investment case for resourcing these interventions. The return on in-vestment projections can be based on the literature or small pilot studies.

After convincing the ACO leadership that we have significant economic value to add, we move to negotia-tions regarding compensation. As an obviously critical step, we will want to retain an outside consultant with ex-pertise in both ACOs and pathology to help us negotiate.

As there are no market precedents yet to draw upon, we have the oppor-tunity to determine our own financial destiny here. First, we need to clarify that the Medicare ACOs continue to be paid by Medicare on a fee-for-ser-vice (FFS) basis, so our diagnostic test-ing FFS revenues will continue (at least in the early years), although they may decline due to utilization management by the ACO.

If our pathway interventions allow for discrete measurement, we have the option of using the first year to prove our case. One option is to ask for an attractive share of bonus payments, but to make those payments contin-gent on hitting target savings metrics. If the ACO is cautious and wants to evaluate our performance the first year, we might negotiate a service contract with a small, performance-based bonus.

To determine what percentage of the ACO’s shared-savings bonus we might negotiate, we can look at the approaches currently being used by ACOs. In a recent report44 on ACO payment allocation by actuarial firm Milliman, a variety of methodologies were described, including divisions based on volume of services, RVU counts, cost-per-episode targets, and responsibility by service line.

However, since each ACO determines its own payment allocation methodol-ogy and these are not generally made public, we may have to start with the metric used by the ACO we are work-ing with. Remember that our objective initially is simply to have our share of the value-based compensation even-tually be tied to the measurable value we deliver.

Practice CapabilitiesEventually, our group will need to have expertise in both molecular diagnostics and informatics to fully support these pathway services. However, to begin we need only be able to support the “specialty area” we selected.

There is one new function that we need to establish: measuring and man-aging the clinical value of our services. In fact, our ACO contract will likely contain performance metrics reflecting the value we deliver, so we will want to be tracking our progress against these measures.

44GN Berry and R Kipp, “ACO gain/loss sharing.” Milliman Healthcare Reform Briefing Paper, 2012



247

Efficient collection of evidence mea-suring our services’ value generation is a fundamental process and capability for this pathway and the new market-place. This likely will entail working with the ACO informatics group to create specific applications and registries for our programs, given that they have access to the integrated data we need for our analyses.

Financing and revenuesThe level of financing our group may need for the launch phase, if any, depends on many factors, so gener-alizations are impossible. These factors include:

• The practice’s starting resources and capabilities

• The practice’s relationship with the ACO, e.g., if there is an own-ership component, then financial resources can be provided by the ACO

• The size and needs of the ACO

There are a number of alternatives for obtaining funding for our Pathway pro-grams, ranging from private venture capital to advanced payments from the ACO. Depending on the use for the capital, bank loans or private debt is also an option.

Further, we can take the partnering approach and join forces with other pathologists or specialty labs using revenue shares to avoid the need for financing in the early years.

Finally, we can offset our needs for

financing at the start of our Pathway journey by trading off back-end, value-based payments for near-term, service contract revenues.

In terms of revenues for the Launch Phase, our practice will likely have a mixture of fee for service for diagnostic testing (unless we have proposed a capitated contract), Medicare Part B services, and a value-based share of the ACO’s gain-sharing component.

This market will evolve quickly. Our practice’s value-based revenues might be structured in myriad forms, and will likely evolve with experience. Similarly, ACOs are still experimenting with pay-ment methodologies, both for the organization and for sharing between providers. The most important funda-mental requirement is the develop-ment of a business case for the value we bring that is backed by evidence of quality improvement and cost sav-ings.

evolution Phase 2017-2019 Planning guide The Evolution Phase of the Coordinat-ed Population Care Services Pathway spans a period, approximately 2017 to 2019, when payment and care deliv-ery reform are projected to be firing on all cylinders in both private and gov-ernment sectors. One hallmark of this phase will involve an evolution in tar-get clients as well as service offerings.

Many demonstration projects being conducted over the next two years, especially evaluating bundled pay-ment programs, will be completed



around 2017. They are expected to show positive results. Even if Repub-licans were to repeal the Affordable Care Act, Medicare will still need to find new ways to cut spending.

Analysts expect new payment model programs like bundled payments that are proven to be effective in lowering spending to continue to be pursued by Medicare. In this scenario, for example, Medicare might be paying for some chronic disorders through bundled payments that cover all professional medical services, inpatient services, di-agnostic testing, therapeutics, ER visits, and procedures. Similarly, many indus-try analysts project that some form of ACO-type organization will be rela-tively common at least in the private sector during this time period.

What this distills down to at the pro-vider level is a greater share of com-pensation being value based rather than fee for service. Furthermore, this value-based compensation will likely be linked to some type of financial risk for that provider. What percentage of compensation will be value-based is difficult to predict. Some estimate as much as 50 percent by 2019-2020, oth-ers say 20 percent to 30 percent.

The implications are that this market evolution will drive a parallel matura-tion in the type and degree of de-mands for value-generation solutions from those who want to share the pie. As providers’ services become inte-grated, or “bundled” for payment and performance management pur-poses, and as they are forced to take

on more financial risk, they will expect other providers to take on risk, too, or to deliver substantial value if they want to share in the bundled payment pies.

Depending on how rapidly reform evolves, ancillary services like pathol-ogy and radiology may need to de-liver more potent solutions. They may also need to be willing to accept some financial risk. Otherwise the pressure to find savings through cutting costs in services like testing will dominate.

We project that for our "Coordinated Population Care Services Pathway" practices, this Evolution Phase may introduce new types of provider clients with bundled payment contracts, such as coordinated care provider organi-zations that may not be ACOs.

evolution Phase ProgramsWe will also want to introduce more advanced population management programs during the Evolution period. These new types of Pathway programs will likely be solution sets of services designed for bundled care manage-ment of a particular disorder, as shown in Figure 14.

The objective is to include a “full po-tential” program of services that ad-dress value gaps in clinical manage-ment of a given disorder across all types of settings, disease stages, co-morbidities, potential adverse events, quality issues, etc.

The population care “tool kit” we’ve been developing, including informatics skills, molecular expertise, quality ana-



249

lytics capabilities, and effective deliv-ery of diagnostic knowledge services, equips us perfectly for providing high-value, clinical “solution sets” for com-plex chronic disease management.

For example, in an economic model-ing analysis we conducted using the Prometheus PAC database of a com-mercially insured population, we found that by utilizing a number of different types of services, including molecu-lar and pharmacogenetic testing to guide therapeutics, quality control programs for point-of-care testing, and clinical decision support tools, patholo-

gists could reduce avoidable costs by a total of 30 percent.

In a bundled-payment program this would be a substantial contribution to value delivery. As previously noted, the Prometheus’ system of using claims data to quantify PACs (potentially avoidable complications or costs) is what has been adopted by Medicare for their upcoming bundling programs. The reduction in PAC costs is the lead-ing determinant of providers’ profit margins in the Prometheus bundled-payment system.



Let’s take a diabetes bundled care program as an example for how we might implement a full potential dia-betes solution. The provider organiza-tion accepting the bundling contract would look at their costs for diabetic patients historically to get a baseline assessment of where they have care and value gaps that must be man-aged – Are their ER visits too high? What are their 30-day readmissions for pneumonia? How is their population’s average HA1c level

compared to national averages?

Reviewing the major Prometheus PAC cost areas for diabetes will be im-portant. Some of the high-cost areas include ER visits for any cause, hospital admissions for co-morbidities, infec-tions, strokes, myocardial infarctions, sepsis, venous thromboemboli, and gastritis/ulcers.

We can now match our service tools to care management gaps for which our intervention could have an impact.

Figure 15: evolution Phase Pathway offering model for Providers managing Bundled Contracts

Full PotentialSolution forSupporting

DiseaseBundle

Quality andCoordination Services

IndividualizedMedicine Services

Risk Prediction Services




251


Figure 16: Potential service Components of Full Potential Pathway solution for Diabetes Bundle

BasicManagement

Ensure abnormaltests not “missed”

Test value protocolsfor referral to

endocrinologist

PGx test to increasepatient stainadherence(KIF6 from

Medco, Celera)

Biomarker prediction

panels for cardio,renal, coagulation,

etc.

Aggressive, precisetreatment for allother disorders –e.g. use PGx and

MDx in GERD,COPD, etc.

AvoidAdmissions, ERs

BetterManagementof Infections

AvoidAdverseEvents

Rapid, targetedtherapy guided bypoint-of-care MDx

for inpatientinfections,

including decubulcers, UTIs

Pathogensurveillance

program withfrequent

antibiograms –community PCPs

Considerpreemptive

pharmacogenetictesting of diabetics

for key genes

Quality programfor bedside/criticalcare glucoments

for hospitalglycemic control

As shown in Figure 16, we draw from all three of our Pathway service cat-egories to assemble a complete solu-tion. For example, from the “Risk Pre-diction” category, we have included biomarker prediction panels to help avoid diabetic admissions. From the “Individualized Medicine” category, we include a PGx test for patient statin adherence. Finally, from the “Quality and Coordination” category, the bed-side glucometers quality program has been included.

This multidimensional solution can de-liver potent value in terms of cost re-duction and outcomes improvement, so we will be well prepared to join our clinical providers as peers in generat-ing value in a diabetes care manage-ment bundle.

Our program will be structured so that we are taking on financial risk with the other providers, as we will want to be compensated as part of the service bundle. We may or may not want to provide the diagnostic tests them-selves at a capitated rate because the practice’s services are the value-generating component and should be compensated based on results.

Either way, we will want to negotiate a share of the bundle’s profits that cor-responds to the value our program generates.

There are a number of variations that can be implemented around this gen-eral model of a “solution set” of servic-es designed to target a disease or pro-cedure episode. Furthermore, during



this period other provider care models will likely emerge with still other popu-lation care management needs. Re-gardless, the basic theme of how we want to evolve our offerings by assem-bling them into integrated programs and solutions that take on financial risk holds across all these eventualities.

expansion Phase 2020-2022 Planning guide The Expansion Phase of the "Coordinat-ed Population Care Services Pathway" envisions that we have established our value-generating capabilities within at least one provider organization, and are prepared to expand to other cli-ents and services.

We assume that the number of coor-dinated care provider organizations and value-based payment programs will have proliferated widely by 2022, so that there should be plenty of mar-ket opportunities for us to pursue for growth.

Predicting market ChangesHow else might our health care envi-ronment have changed by 2020-2022?

In the informatics area, EMRs and EHRs will likely be almost ubiquitous, at least in coordinated care provider organi-zations. Furthermore, integrated da-tabases, containing lab, pharmacy, clinical, and claims data will probably be more commonly available. They will also likely be accessible to any group playing a major role within a popula-tion management organization, includ-ing pathologist groups.

This may mean that data-mining tech-niques, including those using lab test-ing values, will be utilized by many functions in a provider and/or payer organization. However, we will still play a critical role in informatics-driven services: that of the knowledge expert. Indeed, the number and complexity of diagnostic tests is expected to in-crease, not decline, and formulating intelligent data-mining queries de-pends on diagnostics expertise.

In the molecular and genomics area, next generation sequencing (NGS) for tumor profiling will likely be routine for at least some cancers, and point of care molecular testing to drive thera-peutic decisions will be increasingly common, especially in hospitals for infectious disease. Molecular point-of-care testing will require expert, routine oversight and quality control to ensure appropriate use and performance, given the clinical consequences.

The explosion in information and tech-nology in molecular and genomics testing in 2020 will require a dedicated medical professional to harness, man-age, and translate this knowledge into clinical care decision support tools. Given the services provide by our Coordinated Population Care Services Pathway, this professional should be a pathologist.

By 2020, physician shortages will begin to hit hard. Care resources will be or-ganized for efficiency and optimal utili-zation of scarce professional resources



253

A number of new models will likely emerge to cope. For instance, with pri-mary care physicians projected to be in especially short supply, some have suggested a model whereby diagnos-tic specialists and lower level provid-ers (e.g. physician assistants) can use protocol-driven testing algorithms to work-up initial complaints for a wide range of primary care presentations. Based on these results, the patient would be triaged to the appropriate physician. This diagnostic testing triage model is a model being developed by Richard Friedberg, MD, PhD, FCAP, chair of the Department of Pathology at Baystate Health and medical direc-tor of Baystate Reference Laboratories (Springfield, Mass.).45

Perhaps most significantly, we can project that the federal and state governments will be facing very serious healthcare budget problems by 2020. The pressures to cut reimbursement rates and benefits will be severe. It is difficult to predict what remedies will have already been implemented, and thus what more draconian measures will be needed during this period, but it is a good bet that demands for cost reductions will be high.

Given this ecosystem survey, how can we plan our "Coordinated Population Care Services Pathway" growth strat-egy? We do still believe it is a growth strategy, because as long as we can

deliver value – i.e., an element of cost savings – our services will be in de-mand in the marketplace.

Choosing a Long-term strategyAs always the strategy depends on your local market area. For example, if we are serving an ACO(s) in a rural area, then our best growth strategy is to maintain a broad set of services and capabilities so that we might be able to expand as the ACO grows, or serve an additional ACO.

On the other hand, if we are in a more crowded market situation, such as in a metro area with a number of ACOs and other types of coordinated care provider organizations, we may want to select an area for specialization. For example, we might focus on individu-alized medicine for population care management, and identify ACO cli-ents that have less need for informatics risk prediction, but greatly value indi-vidualized medicine programs.

Alternatively, developing expertise in value-generating programs for specific chronic disorders can be an excellent market growth strategy. First, chronic medical conditions comprise over half of Prometheus’ avoidable costs, so this is where the opportunity resides. Sec-ondly, there are a few disorders that contribute to a major share of these avoidable costs, including diabetes, COPD, and gastro-esophageal reflux disease. As bundling payment pro-

45G2 Intelligence interview with Richard Friedberg, MD, PhD, FCAP, Chair of the Department of Pathology at Baystate Health and medical director of Baystate Reference Laboratories (Springfield, Mass.) on September 11, 2012



grams become widely adopted for these conditions, which is likely at some point in the next decade or so, having expertise in generating value for one of these conditions will be a very at-tractive market proposition.

A final strategy for practices interested in taking on financial risk across a pa-tient population and innovating new models is to offer a new type of diag-nostics benefits program, one linked to value-based results. The fundamental diagnostics benefits program, which is another model proposed by Rich-ard Friedberg,46 is a benefits package whereby our pathology lab commits to providing all diagnostic testing for each patient in a defined population for a defined time period for a spe-cific price – i.e., a capitated rate. We would obtain a reinsurance policy on this contract to mitigate our risks.

The value results component adds a layer of Pathway services to the pro-gram, whereby we would utilize our array of population management services to maximize the value resulting from diagnostic testing in the popula-tion. In return, we would be compen-sated not just based on the capitated testing rate, but also based on value metrics for the population tied directly to our program.

There are a few elements to the ratio-nale behind such a program. First, it is probable that payers or some other industry entity will “carve out” payment of diagnostic testing services in this way. Whoever owns the risk and con-trols the benefits management for our services will also control profit margins and value-added service opportuni-ties. Thus, this program advances our control over our economic destiny and profits.

Secondly, this program offers another economic model for translating the experience and knowledge we’ve ac-cumulated along this Pathway into a stronger financial and market position.

There are therefore a number of av-enues we can take by the time 2020 arrives. Keys to success in the long run include:

• Accumulating experience deliver-ing value based on these three categories of services: risk predic-tion, individualized medicine, and quality and coordination; devel-oping expertise in informatics and molecular testing/genomics.

• Preparing an evidence base that supports the value our interven-tions can generate.

46G2 Intelligence interview with Richard Friedberg, MD, PhD, FCAP, Chair of the Department of Pathology at Baystate Health and medical director of Baystate Reference Laboratories (Springfield, Mass.) on September 11, 2012



255

ConCLUsionsFor ACOs, integrated delivery net-works, and medical homes to imple-ment population health management, there is a clear need for both a popu-lation-based view of care delivery as well as individualized, patient-centric care. Pathologists alone have the dual approach that marries “top down” population lab informatics applications that systematize care with “bottom up” patient-specific, molecular test-ing tools that individualize treatment guidelines. We also have particu-lar expertise in other areas critical to population health management, such as test analytics, instrumentation, and infectious disease.

The population health management model also demands novel approach-es to risk management and preven-tion. Because pathologists deliver the most potent of predictive tools in health care – in the form of individual-ized diagnostics and biomarkers – and because lab testing is performed fre-quently, pathologists are uniquely able to provide a steady supply of temporal predictive data to keep risk assess-ments fresh. Pathologists also have the expertise to utilize this data in powerful predictive applications that can pre-vent adverse events.

In most circumstances, a pathology group that was already contractually tied to either an ACO hospital or ACO physician organization (one in which a physician organization rather than

a hospital-led organization contracts with the payer) for diagnostic testing services will be the best candidate for this pathway. Because there are many different types of ACOs, there will be significant variability in how a pathol-ogy group approaches implementing this pathway.

The way forward will vary from practice to practice depending on the highest priorities of the ACO. Step one for any practice will involve working with their local ACOs or the groups with which they already have contracts to identify the highest priority needs and then re-sponding with services based on core capabilities that deliver clear value. Most practices will need to start with targeted offerings and then expand over time to add new capabilities, such as molecular testing expertise, and to develop additional services.

Compensation for these services will likely be based on value payments, such as ACO shared-savings bonuses. To demonstrate the value of the "Coordinated Population Care Pathway" to the ACO, we will need to develop a business case that docu-ments opportunities for cost savings and increased quality and allows us to negotiate a share of the ACO’s net savings.

The number of ACOs is growing and the ACO model is maturing, so ACOs are rapidly developing the solutions they need to implement population



health management. If pathologists want to play a larger role than data suppliers in this new business land-scape, they must act now and apply their skills and expertise to design the valuable services ACOs need.



L. eleanor J. herriman, mD, mBA is a pathologist executive with 15+ years experi-ence in the medical technology indus-try. She has worked across sectors with

both small ventures and corporations. Her market experience spans the clinical laboratory, in vitro diagnostics, e-health, biotech, and strategy con-sulting for Bain & Company. Other key accomplishments include growing a venture with a CLIA lab and FDA PMA approved computer-assisted cytology instrument from start-up to publicly traded company (IPO raised $100 million).

She is the leader of G2 Intelligence’s advisory services, which provides stra-

tegic consulting and custom market research and assessments for a range of clients, including pathology groups, clinical labs, health systems, hospitals, medical centers, diagnostic technol-ogy companies, and investors.

She is the author of three G2 reports: Health Care Market Reforms: Implica-tions and Prescriptions for Laboratories; How to Build a Molecular Testing Labo-ratory: Key Strategic & Operational Considerations; and Creating a Value-Driven Laboratory: Opportunities in the New Marketplace.

She earned her MD at Baylor College of Medicine, completed pathology residency at University of California San Francisco, and received an MBA from Harvard Business School with highest honors (Baker Scholar).

ABoUt the AUthor