MedicineLakex

Finding a match: how successful complex care programs indentify patients

CALIFORNIA HEALTHCARE FOUNDATION Finding a Match: How Successful Complex
Care Programs Identify Patients
Meet Mr. D, age 40. Over the past year, he saw his A critical step for CCM programs is to select and primary care physician 23 times and missed 10 other prioritize high-risk individuals for whom care man- appointments. He phoned his primary care clinic agement interventions might improve care and 22 times. He made 21 emergency department vis- reduce costs. This report provides practical guidance its and was admitted to the hospital three times for on selecting such patients, based on approaches issues that could have been treated by his primary identified during in-depth interviews of leaders at care physician. Affected by fetal alcohol syndrome, 20 leading CCM programs. (See Appendix B.) The developmental delay, and obesity, Mr. D has a his- research focuses on programs in which CCM teams tory of high blood pressure, diabetes, asthma, and work closely with primary care teams — such pro- chronic low-back pain. He is a victim of childhood grams are sometimes referred to as "primary-care abuse, and suffers from post-traumatic stress disor- der, anxiety, and bipolar disorder. He lives in a public housing development, is unemployed, and is insured The details of Mr. D's profile illustrate the complexity by Medicaid. of patient selection. The obvious reasons to select him for complex care management are his high uti- The challenges of caring for high-need, high-cost lization of services and high risk for poor outcomes. patients such as Mr. D. have led increasingly to the On the other hand, if a particular CCM program is use of specially trained complex care management not equipped to address his behavioral health issues (CCM) teams that identify and engage high-risk sufficiently to reduce utilization, Mr. D might not be patients, assess their needs, and intervene rapidly to an appropriate candidate. The approach that a pro- changes in their health status. These programs use a gram takes to select patients for their program can variety of approaches, but their common goal is to mean the difference between its success and failure.
improve coordination of care and quality of life and reduce unnecessary use of services.1- 6 Issue Brief
Selecting Patients for
Successful CCM programs align the selected popu- There are two key aspects to patient identifica- lation, the planned interventions, and the outcomes tion: (1) predicting risk in specific patients along the CCM Programs
of interest by performing three tasks: dimensions or outcomes of interest, such as reduc-ing inpatient admission or recurrent, unnecessary ED 1. Specify, prioritize, and agree on the outcomes of
visits; and (2) predicting care sensitivity — the likeli- Before selecting patients, successful CCM programs think carefully about what they interest and the time frame in which outcomes are trying to achieve (the desired outcomes) hood that a particular high-risk patient will respond should be achieved. For example, a high-priority as well as the program design (what resources the to the care management intervention. A number outcome might be to achieve savings and/or a organization or its partners can bring to bear). The of patient identification approaches are available return on investment at three years.
desired outcomes and program design determine to CCM programs,7 including quantitative, qualita- the optimal target population and patient selection 2. Identify a sufficiently high-risk and care-sensitive
tive, and hybrid models. The reported advantages approach. For example, a predictor for total costs target population in which the outcomes can be and disadvantages to the different approaches are at three years may be less relevant if the outcome summarized in Table  1. Appendix C provides an of interest is return on investment at six months. overview of approaches used at each of the 20 pro- Similarly, a precise selection algorithm for patients 3. Match the planned staffing/resources and inter-
grams interviewed for this report. with risk for admissions from behavioral health ventions to the target population to achieve the problems is suboptimal if the intervention cannot desired outcomes, building on existing services leverage behavioral health resources to successfully to fill care gaps.
engage and manage patients with such issues.
Table 1. Patient Selection in CCM Programs: Advantages/Disadvantages of Different Approaches, continued
Use claims data (or sometimes $$ Well-validated for identifying $$ Is not thorough – it accounts for a low percentage of the variability in costs, utiliza- billing data) to predict differ- a subset of high-risk patients tion, and other outcomes ent future outcomes (e.g., total (particularly patients at high risk $$ Does not account for some factors that are important for risk-stratifying patients medical expenditures, acute for future costs) (e.g., recent admissions that are not in claims, or identifying poorly activated care utilization events, etc.) Some provide risk scores $$ May not adequately identify psychosocially complex patients within clinical groups (e.g., condition categories) and $$ Depends on completeness of data (e.g., claims data) others provide a cumula- $$ Lack of continuous claims data because of frequent disenrollment may reduce tive risk score that takes into utility and precision of predictive modeling account multiple risk factors $$ Cumulative risk scores for medical expenditures do not guide specific interventions $$ Dependence on recent acute care utilization leads to over-identification of some low- or moderate-cost individuals with recent episodic high-cost events $$ The highest-risk individuals may not always be the most responsive to care management interventions (e.g., patients with metastatic cancer) California HealthCare Foundation Table 1. Patient Selection in CCM Programs: Advantages/Disadvantages of Different Approaches, continued
Use prior acute care utilization $$ When real-time data are available, $$ Misses high-risk patients who do not have prior use of acute care services utilization focused
to identify high-risk patients identifies a high-risk population $$ Prior utilization/cost is not the cause of future utilization/cost — and it does not at a time of significant need and help identify mutable factors that drive it opportunity for impact Use claims/billing data $$ Widely available $$ May not adequately identify patients at high risk for utilization/costs condition- or
(ICD-9 codes), internal data $$ May be more easily received by $$ Not all patients with high-risk conditions are at high risk — requires risk warehouse data (problem lists, health care providers because it stratification within condition categories medication data), or pharmacy gives them obvious targets for data to identify patients by high-risk conditions or medications Uses a weighted instrument $$ Takes advantage of the strengths $$ Complex and time-consuming to create multifactorial risk
(often developed internally) of different approaches $$ Final risk assessment can be hard to interpret that may draw from any of the $$ Brings data together from above sources to assign risk $$ Poor data quality or conceptualization can lead to an inaccurate risk assignment scores for patient selection $$ Allows inclusion and weighting of qualitative measures Referral by clinician Uses clinician or CCM team
$$ Clinicians prefer to have the ability $$ Clinician referral identifies patients who are challenging to manage, but not or CCM team, or
clinical assessment skills to to refer their patients into CCM necessarily those at high risk for future utilization/costs or those who might be identify high-risk patients responsive to care management interventions $$ Clinicians may have the most $$ Patients who are less activated and more vulnerable (and often the highest risk) complete picture of the patient, may not self-refer including clinical, behavioral health, and environmental and socioeconomic risk factors $$ Patient self-referral identifies motivated patients with higher self-efficacy, who demonstrate readiness for CCM Hybrid: quantitative Sequentially uses combinations $$ May be most reliable approach
$$ More complex to implement — clinician review is particularly time-consuming of quantitative and qualitative for selecting high-risk patients $$ Hybrid approaches with qualitative gates depend on the clinicians' understanding who are also responsive to care of the program goals and interventions and their ability to select the right patients management interventions Can have a qualitative gate — which may not always be easy to achieve (e.g., clinicians decide final $$ Takes advantage of the strengths selection) or a quantitative of different approaches gate (e.g., a quantitative threshold is used to decide final selection) Finding a Match: How Successful Complex Care Programs Identify Patients and prior cost or utilization. These models most often Quantitative protocols include: predict total cost of care, partial costs (e.g., inpatient A Commercial claims-based risk-prediction ost quantitative risk-prediction approaches vs. outpatient expenditures), and health care utiliza- tools (e.g., Hierarchical Conditions Categories used by CCM programs are borne out of tion (e.g., hospital admissions, readmissions, and [HCC],8 Clinical Risk Groups [CRGs],9 Impact the science of claims-based risk adjust- ED visits). Table 2 provides a summary of the data Pro®,10 Milliman Advanced Risk Adjuster ment, employing a combination of age, diagnoses sources as well as some reported advantages and [MARA],11 MEDai12) (gleaned through coded diagnoses or medications), Table 2. Data Sources Used by CCM Programs: Advantages/Disadvantages
All bills submitted by $$ Provides a complete picture of patients' billed interactions $$ Data gaps exist when patients are previously insured by another physicians and health with the health care system when available including expendi- payer, move between payers, or lose insurance coverage care systems, and ture data, medication fill data on prescribed medications, and $$ Can take months for claims data to become complete ("claims lag") payments made by services, procedures, and tests provided $$ Diagnosis data are dependent on coding and are incomplete (significantly, at times), inconsistent, and unstable $$ Limited clinical information and socio-demographic information Internal
Billing, clinical (e.g., $$ Data are available more quickly, in real time $$ Does not capture patient interactions with the health care system electronic medical outside of the network available in the data warehouse $$ Diagnosis data are more complete and stable in electronic record data, vital signs, $$ Data structure is more complex, and data come from numerous laboratory, imaging sources, requiring reconciliation and validation medication/prescription $$ "Problem list" data allow identification of prior procedures/data), appointment/ $$ Data linkage (e.g., making sure data from multiple sources belong registration data, and to the same patient) can be challenging $$ Clinical data (e.g., laboratory data) may allow screening for disease registries and non-billed, non-problem list diagnoses and provide more richness regarding severity of illness or control of conditions $$ Clinical data may be more complete – include non-prescription medications, detailed information on clinical encounters that are amenable to chart review and natural language processing $$ Medical records data also increasingly include both coded and non-coded qualitative data (see next row) Surveys delivered by $$ Provides data not available in quantitative datasets that may $$ Not widely available at the population level payers or delivery be strong risk predictors, such as screening for medical and $$ Expensive and time-consuming to collect systems (e.g., health risk psychosocial risk factors $$ Patient responses are often unstable $$ May be useful for risk stratification within high-risk populations $$ Survey response rates may be low, reducing utility Prescription fill data $$ When complete, may be the most accurate source of medication $$ Does not capture medication refills outside of the pharmacy system California HealthCare Foundation A Acute care utilization thresholds based on Due to these and other limitations, half of CCM patient factors that drive future utilization and cost prior utilization programs in this study do not use claims-based risk- within their population and incorporate them in prediction models at all. Instead, many programs their selection approach. When patient selection A Chronic disease count thresholds, or inclusion use criteria and set thresholds based on prior acute is focused on high-risk diagnoses, programs often based on specific high-risk chronic diseases care utilization or specific high-risk conditions. Most stratify within conditions to find higher-risk subsets.
pulled from electronic data warehouses or programs find prior acute care utilization to be a particularly useful proxy for future risk. However, Programs frequently develop their own quantita- A Internally developed risk-prediction tools David Labby, chief medical officer of Health Share of tive tools using multi-factorial risk assessments (e.g., Aetna,13 Network Health,14 Veterans Oregon, cautioned that "utilization is an effect, not (see box), which combine different quantitative Affairs Care Assessment Need [CAN] score15) a cause." Therefore CCM programs strive to identify approaches into one measure or risk score. For In general, published risk-prediction models make predictions on total medical expenditures after accounting for only a quarter to a third of the fac- AtlantiCare Special Care Center — Patient Referral Form
tors that lead to future expenditure16,17 and do not perform well within the highest-risk subgroups that are the targets of CCM. Thus, even the best available Congestive heart failure (heart pump failed tools frequently fail to identify many future highest- to function with excessive fluid retention) cost patients. Additional data, such as clinical and Coronary artery disease/stroke survey data, when added to traditional claims-based risk predictors, may improve predictive models.18,19 Cardiovascular disease — other heart disease (abnormal rhythm) However, clinical status often changes quickly, and survey data are expensive and time-consuming to Chronic anti-coagulant (coumadin therapy) Diabetes mellitus (high blood sugar) Obesity (height and weight) Furthermore, many programs reported that the Kidney disease or current serum creatinine >2 Mental illness (depression, anxiety, etc.) claims-based risk prediction models they used too (estimated glomular filtration rate <60); placed on meds for kidneys or on dialysis often identified patients who were not going to be high risk moving forward (e.g., time-limited, episodic Chronic obstructive pulmonary disease ≥2 Hospitalizations/ED visits in past 12 months high-cost episodes such as high-risk pregnancy), Asthma/on maintenance meds Taking ≥5 chronic prescription meds so they developed processes for removing such (chronic/daily meds) (other than pain meds) patients. A few programs prefer risk-prediction prod- ucts that present risk within categories (e.g., Clinical Risk Groups risk-stratify patients within clinical condi- No primary doctor tion categories) because these give CCM teams a clearer clinical target for intervention compared to To qualify, patient must have at least one from Column A and total of 6 points: Columns A + B = • tools that simply provide an overall risk score. Finding a Match: How Successful Complex Care Programs Identify Patients example, programs might combine some of the relatively low-risk, anxious patient who is perceived following: claims-based risk scores, prior utilization, to be complex because of frequent phone calls and high-risk diagnosis or medication groups, laboratory office visits. Other physician referrals might identify data and other indicators of disease severity, socio- As with quantitative approaches, best prac- tices in applying the various qualitative patients with intractable heroin addiction that they economic indicators, and other risk factors. Some methods have not been systematically are unable to manage when the program does not multi-factorial risk assessments also incorporate data described in the literature. Clinician input may help have sufficient resources to address addiction issues. from focused or comprehensive qualitative assess- identify high-risk and care-sensitive patients, but the Programs worked to improve the quality of the refer- ments such as depression screening, self-rated data are inconsistent.20-22 Qualitative approaches rals by providing clinicians and teams with clear health or functional status, measures of health lit- include patient self-referral and clinician assessment criteria for referral and educating and re-educating eracy, patient activation or engagement, or clinician them on the program goals and interventions. assessments. Combining multiple approaches allows programs to take a more comprehensive approach Because many of the highest-risk patients do not and use all available data resources, leveraging their proactively seek support, it is rare for CCM programs strengths to select the best population for their to depend on patient self-referral. On the other hand, clinicians frequently request the ability to refer The majority of programs in this study use hybrid approaches that take advantage of patients to the programs. Interviewees felt that cli- both quantitative and qualitative strategies to nician referral was a good way to engage primary select patients. care clinicians in the program. Primary care teams can be particularly effective at identifying aggra- Most often, CCM programs use a quantitative vating factors such as subtle gaps in care, social approach to generate a list of high-risk patients isolation, psychosocial issues, or recent changes in for clinical review and assessment by primary care clinical status not yet captured in available data. A teams, alone or working together with the CCM primary care clinician's rich knowledge of the patient teams. Joint clinical reviews occur on a one-on-one also enables identification of mitigating factors such basis or in regular meetings or "huddles" specifically as the presence of an active caregiver in the home designated to select patients for the program. or comprehensive care team already in place (e.g., comprehensive cancer center). Such aggravating Quantitative approaches are easier to apply to and mitigating factors are not easily gleaned from broad populations and allow programs to narrow other data sources. the list for qualitative clinical review to a high-risk subgroup. During the clinical assessment, reviewers However, interviewees noted that, because referral work to both exclude patients who are not high-risk decisions are frequently based on subjective clinical over time and select ones likely to be care-sensitive. assessment, clinicians tend to select patients who Typical exclusions might be: require the most effort to manage — a group that is not always the highest risk or likely to respond to A Patients who are not sufficiently high-risk, the planned care management interventions. For including those who have a time-limited, example, a primary care clinician might select a high-cost episode that has resolved — such California HealthCare Foundation as cancer treatment, a high-risk pregnancy, have a comprehensive care team (e.g., com- in order to select the right patients and optimize this surgery, or trauma event.
prehensive cancer center or diabetes center).
approach. Many programs provide clinical reviewers with criteria to guide selection and frequently edu- A Patients who have their care needs met, such Because these assessments are subjective, clinical cate and re-educate them on program goals and as those with a strong network of family care- reviewers must clearly understand the program goals givers, those in assisted living, or those who and what the CCM program does and does not offer Finally, because clinical review is time-consuming, a few programs train team members (e.g., program leader, nurse, or medical assistant) to select patients, Hybrid Approaches: Cambridge Health Alliance Triage Scoring Protocol
or to narrow the list for clinical review by using chart The Cambridge Health Alliance (CHA), a Massachusetts safety-net delivery system, identifies high-risk pa- review to obtain a richer clinical picture of the patient. tients using three claims-based risk predictors for different payer populations: $ Top 10% of risk for future medical expenditures using the Network Health Managed Medicaid Predictive Some programs flip the hybrid model and apply (MMP) prospective risk score (Network Health Managed Medicaid patients) "quantitative gates" to self-referred patients or $ Top 10% of risk for future medical expenditures using the Diagnostic Cost Groups (DxCG) risk score those referred by primary care or other care teams. (non-Network Health Managed Medicaid patients) Patients are selected for the program only if they $ All patients with >75% probability of admissions in one year using the ImpactPRO risk score (Medicare meet specific quantitative inclusion criteria. See box Pioneer Accountable Care Organization). To this group, they add all patients with total medical expen- for examples of different hybrid approaches.
ditures in the past year at least 10 times the CHA population average and those with eight or more ED visits or three or more admissions in the past year. These patients are selected in monthly batches and presented to primary care teams for qualitative clini- Care Sensitivity
cal assessment. It is designed to be a "5-minute task" during a monthly meeting in which the primary care Identifying care-sensitive patients is a major selec- teams (including the primary care clinician, team nurse, medical assistant, and front-desk staff person) meet tion challenge because it is not always clear who with the CCM nurse and review five newly identified high-risk patients per month (60 per year). will respond to care management interventions and there is no generalizable approach; care-sensitiv- The primary care team reviews the lists guided by a patient summary that includes: (1) name/age; (2) ED visits and admissions stratified by in-system and out-of-system; (3) reason for inclusion on the list (e.g., high ity depends on the clinical focus of the particular risk score, high inpatient/ED utilization, high-risk diagnosis, etc.). The instrument used in the review asks: (1) CCM program. Most interviewees felt that qualita- Would you be surprised if this patient were admitted to the hospital in the next six months? (2) Would the tive assessment is necessary to select care-sensitive patient engage in care management? (3) Does this patient have an unmet medical need or care gap that the CCM team could help with? (4) Where should the patient be referred for additional support (e.g., CCM program or other care management program)? Approaches to identifying care-sensitive patients Data from these assessments are then combined with clinical data on acute care utilization, control of high- include the following: risk chronic medical conditions or co-morbid behavioral health issues, polypharmacy, primary care engage- A Excluding patients with needs that the program ment, social risk factors, and poor or declining functional status. A triage scoring system determines selec- is not sufficiently resourced to address. For tion for the CCM program. CHA also allows primary care team referral. Primary care clinicians complete the example, reviewers might exclude a patient same qualitative instrument, and patients undergo the same quantitative assessment using the triage tool.
with a significant personality disorder or serious Finding a Match: How Successful Complex Care Programs Identify Patients mental illness if the program does not have suf- A Identifying "windows of opportunities" or Another aspect of care sensitivity is delivering the ficient behavioral health expertise. patients at high-risk times, such as hospital- right intervention with the right intensity at the right to-home or ED-to-home care transitions. time for the patient. Almost all programs depend A Identifying patients with specific care gaps or Interviewees noted that during such high-risk on ongoing informal reassessments of the patients' barriers to care that the CCM team can address. time periods, patients may also be more recep- health status by CCM team members to dial up or Some programs focus on patients for whom tive to help from CCM team members. Real-time down the intervention and outreach intensity to gaps in care are most likely, looking for individu- data on hospital or ED utilization must be avail- match changing patient needs. For example, a CCM als with multiple chronic conditions or specific able to apply this approach. A few programs also intervention might be triggered by a hospital dis- high-risk conditions that the team has had suc- track changes in patient risk status over time and charge, worsening shortness of breath, or a patient cess addressing. Or they identify patients with attempt to identify those with rising risk scores. who is newly homeless. Programs often discuss poverty or addressable barriers to access. What a patients in routine case conferences or huddles to specific program deems addressable varies. For re-assess and re-stratify them. example, many programs exclude patients with conditions that are cared for by specialty-based comprehensive care teams, such as dialysis patients and those undergoing chemotherapy, A Hybrid Approach: Iora Health Adds "Worry Score" to the Mix
because they feel the CCM team is unlikely Iora Health, a primary care delivery organization, runs two different CCM models at its sites. The first is an to significantly augment the care provided by ambulatory intensive care unit model (AICU),* in which the entire practice cares for only high-risk patients. specialty-based comprehensive care teams It uses a health risk assessment consisting of 10 weighted questions, with patients scoring 3 or higher ac- already in place. However, one program noted cepted into the AICU.
that their dialysis patients were frequently admit- At other Iora sites, all patients are risk stratified to identify a subset for more intensive management by their ted for volume overload or dehydration and multidisciplinary primary care team. The protocol starts with a risk predictor, the Milliman Adjusted Risk cancer patients for side effects of chemotherapy Score, focusing on the top decile of risk for future cost. The results are paired with an internally developed that were not addressed by their specialty care risk assessment, called the "Worry Score," to identify the highest-risk patients. teams. They felt they could address these gaps The Worry Score has a scale from 0 to 10 that is generated by adding and subtracting points based on risk with the right team, anticipatory guidance, and factors. The score takes into account diagnoses and control of chronic conditions, recent acute care utiliza- management and worked with specialty teams to tion, and a list of modifiers including smoking status, age, and socioeconomic risk factors. For example, a diabetic man might start with a score of 5, then go to a 7 because his most recent hemoglo- A Identifying patients who are at risk for or have bin A1c is greater than 9; then to a 9 because he was admitted to the hospital in the past six months; and experienced care coordination issues. For finally to a 10 because he is a smoker. Patients with a score of 10 would be discussed daily in team meetings example, some programs include patients with or huddles; those with lower scores are discussed at specified less-frequent intervals. multiple specialist interactions, poor engage- The Worry Score uses data from claims, EMRs, data warehouses, and surveys. Because they can run continu- ment with primary care manifested by frequent ously, the scores reflect changes in clinical status in real time, enabling primary care teams to continuously missed visits or absence of visits, or frequent ED visits or hospitalizations for ambulatory care sensitive conditions.
*Redesigning primary care for breakthrough in health insurance affordability. Mercer Human Resource Consulting. August, 2005.
California HealthCare Foundation One critical ongoing assessment that programs stakeholders about cost drivers and look at data The studied programs vary in the frequency with employ is patient readiness to engage with CCM to identify both cost drivers and issues they could which they select populations. Some do this at fixed teams. Successful teams use motivational interview- address through interventions. From this information intervals such as annually or monthly, while others ing approaches to assess readiness and reduce they develop a quantitative identification approach. take a continuous approach. Programs that use fixed intervention intensity when patients demonstrate intervals face the challenge of performing clinical low readiness. Only a few programs use formal or Some programs favor simpler quantitative inclusion review and program enrollment in batches; review- automated risk-stratification approaches to scale up criteria, while others use multiple factors to triangu- ing a large list of patients requires a concentrated and down the intensity of care and outreach. See late and get the most comprehensive view of risk time commitment that takes away from care manage- box about the Iora Health Worry Score, page 8. within their population. In either case, interviewees ment activities during the review. On the other hand, recommended choosing an algorithm that is well the continuous approach requires dedicated analytic Developing and Applying received by the primary care teams, as their ongo- support and access to continuous data streams.
ing engagement in the process is critical to program A few interviewees raised concerns that overly proto- colized selection or risk-stratification activities could nterviewees stressed that the process used to develop the patient selection algorithm and achieve alignment between population, inter- vention, and outcome can be as important as the Multi-Stakeholder Process at Denver Health
final algorithm itself. Engaging clinicians and other Denver Health, an integrated safety-net system, designed their selection approach as part of a CMS Health stakeholders early and frequently was a key recom- Care Innovation Award grant. Their purpose was to create complex care management goals based on mendation from many interviewees. achieving the triple aim of improved health and health care at reduced costs. The grant's charge included developing an innovative patient selection and risk-stratification approach, then studying and improving it.
Programs generally start with the best possible Denver Health's process was transparent and inclusive. They brought together stakeholders including clini- approach from their available data and use continu- cians, health service researchers and analysts, administrators (e.g., finance people), and CCM team leaders ous quality improvement to refine their strategy over to develop the risk-stratification algorithm. time. For example, Denver Health (see box) follows They use Clinical Risk Groups (CRGs) to divide their population into four risk tiers, then apply patient-specific patients who are included and excluded by clinicians risk factors (e.g., recent acute care utilization, high-risk diagnosis) to move people into higher tiers. Identifi- to hone the clinical review criteria, and use this infor- cation of the highest-risk patients typically considers tier status, intervention-specific criteria (e.g., substance mation to provide decision support to clinicians as use disorder for focused intervention), as well as clinical review by primary care clinicians. Automatically they select patients for the program.
generated lists of high-risk patients are reviewed by clinicians to remove patients who are not a good fit for the program. Some programs select and prioritize their targeted The analytic team created a tool by which CRG categories and other data can be manipulated in real time outcomes, then engage stakeholders in discus- during meetings to re-tier patient populations based on suggestions from the group. Patients can be sions about drivers for poor outcomes and review grouped and re-grouped until consensus is reached. The quantitative algorithm is reviewed approximately data to identify a target population. For example, every six months. Patients who were "misclassified" in lower tiers or those excluded by providers that ended if a program's primary goal is to reduce health up in high-spending tiers are assessed to identify opportunities to improve the approach.
care expenditures, they may have discussions with Finding a Match: How Successful Complex Care Programs Identify Patients distract teams from the clinical assessments and patient selection approaches and care-sensitivity Clearly there are opportunities to greatly improve care provision that is critical to successful CCM. For factors. Improving approaches to risk stratify and the value of health care delivery by improving our example, the time required for a nurse care manager classify high-risk patients is also critical. This process, ability to identify high-risk patients, risk stratify to routinely and formally re-stratify their population which some call "segmentation," organizes patients them, and define subgroups that are amenable to into risk tiers would decrease the time they have to into groups for which programs can design effec- intervention. This issue brief provides guidance, but deliver care management interventions. CCM pro- tive interventions. In particular, it will be important patient selection and risk-stratification approaches grams try to balance the need for rigor and protocols to move beyond chronic condition-based groupings are evolving rapidly, so we will need to continue to with the need to develop patient relationships and and identify high-risk patient subgroups that may be share learning and evaluate different approaches.
do the work of care management. defined by utilization patterns and specific care gaps or barriers. The rise of big data 23,24 and improved qual- ity and access to electronic data repositories Seven Take-Aways from Successful CCM Programs
(e.g., health information exchanges) that combine data from numerous sources bring great 1. The patient selection approach should align with desired outcomes and planned care management
opportunities to improve the identification of high- interventions. Know what the CCM program does and does not do.
risk patients. Data sources that provide psychosocial 2. Quantitative approaches to risk prediction can be applied to large populations, but are subject to
predictors, patient-reported outcomes, functional data quality issues, too often identify patients with time-limited increases in risk, and may not provide assessment, and survey data will expand variables specific information required to guide interventions.
available for use in predictive modeling. Examples 3. Using large clinical and non-health-care related datasets may enhance a program's predictive capabili-
include: integrating data from government and com- ties. This could enable use of multiple data sources to triangulate risk and multiple algorithms to get at munity organizations, payer- and purchaser-collected different risk populations.
health risk assessments, patient reported outcomes 4. Asking clinicians to select patients may better identify care-sensitive patients, but this approach may be
collected through patient portals or mobile phones, hard to administer and may not identify the highest-risk patients.
and continuous data from remote monitoring tools.25 Implementation of ICD-10 also brings potential 5. Hybrid approaches take advantage of the strengths of both the quantitative and qualitative approach-
advantages with more granular data on the sever- es, but may be complex to administer.
ity of conditions, but it will require programs to 6. High-risk populations can be stratified into subgroups likely to benefit from specific interventions. How-
adapt current risk-prediction approaches and will be ever, this remains largely an individualized, clinical review process rather than an automated one.
subject to issues (e.g., coding) faced during the tran- 7. The process used to develop the patient selection approach is as important as the algorithm itself.
sition from ICD-9 to ICD-10.
Engage multidisciplinary stakeholders and use continuous quality improvement to refine the approach.
Researchers and implementers will need to work together to identify strong predictors within this growing dataset and improve quantitative risk-stratification approaches while also studying hybrid California HealthCare Foundation About the Authors Clemens S. Hong, MD, MPH, is a practicing primary The authors would like to thank the Commonwealth care general internist and health services researcher Fund (especially Melinda Abrams) for funding a at Massachusetts General Hospital (MGH), and report on complex care management, "Treating co-founder of Anansi Health. His research focuses High-Need, High-Cost Patients: What Makes for broadly on improving primary care delivery to vul- a Successful Care Management Program," which nerable populations, with a focus on the integration laid the foundation for this report. Also, the authors of community health workers into primary care are deeply grateful to the program leaders and teams, risk stratification and identification of com- interviewees who shared their time and expertise plex, high-risk patients in primary care, and primary care-integrated complex care management of high-cost Medicaid and dual-eligible patients. About the FoundationThe California HealthCare Foundation works as a Andrew S. Hwang is a student at Tufts University catalyst to fulfill the promise of better health care School of Medicine. He will graduate in 2015 with for all Californians. We support ideas and innova- MD and MPH degrees. He is currently applying for tions that improve quality, increase efficiency, and residency programs in internal medicine. His research lower the costs of care. For more information, visit focuses on the characteristics of persistent, frequent users of the emergency department and the asso-ciation between outpatient visit no-shows and risk 2015 California HealthCare Foundation for poor preventive screening, chronic disease, and acute care utilization outcomes. Timothy G. Ferris, MD, MPH, is a general internist and pediatrician and senior vice president of popula-tion health management at Massachusetts General Hospital and Partners HealthCare. He is also an asso-ciate professor of medicine and pediatrics at Harvard Medical School. As the principal investigator for a six-year Medicare demonstration program and as the leader of the Partners Pioneer Accountable Care Organization in Boston, he leads programs designed to systematically improve patient care. Finding a Match: How Successful Complex Care Programs Identify Patients 1. Hong CS, Siegel AL, Ferris TG. Caring for High-need, 11. Milliman Advanced Risk Adjusters. Milliman. Accessed 23. Roski J, Bo-Linn GW, Andrews TA. "Creating Value in High-cost Patients: What Makes for a Successful Care Health Care through Big Data: Opportunities and Policy Management Program? The Commonwealth Fund. 12. How MEDai Can Help Predictive Modeling. MEDai. Inc. Implications." Health Affairs. 2014;33(7):1115-1122.
August 2014.
24. Bates DW, Saria S, Ohno-Machado L, Shah A, Escobar G. 2. Bodenheimer T, Berry-Millet R. Care Management of 13. Aetna Integrated Health and Disability. Aetna, Inc. 2004. "Big Data in Health Care: Using Analytics to Identify and Patients with Complex Health Needs. Robert Wood Manage High-risk and High-cost Patients. Health Affairs. Johnson Foundation Research Synthesis Report No. 19. December 2009.
14. Yeracaris P. Community Engagement: Targeting High-risk Members to Enhance Care Management Effectiveness 25. Nguyen OK, Chan CV, Makam A, Stieglitz H, 3. Bodenheimer T. "Coordinating Care — A Perilous [PDF]. Retrieved January 14, 2015 from: Amarasingham R. "Envisioning a Social-Health Journey through the Health Care System." N Engl J Med. Information Exchange as a Platform to Support a Patient- Centered Medical Neighborhood: A Feasibility Study." 15. Fihn S, Box T. Care Assessment Need (CAN) Score and 4. Ayanian JZ. "The Elusive Quest for Quality and Cost J Gen Intern Med. 2015;30(1):60-67.
the Patient Care Assessment System (PCAS): Tools for Savings in the Medicare Program." JAMA. 2009; Care Management. Veterans Health Administration. June 2013. Accessed September 29, 2014, 5. Brown R, Peikes D, Chen A, Ng J, Schore J, Soh C. The Evaluation of the Medicare Coordinated Care 16. Winkelman R, Mehmud S. A Comparative Analysis of Demonstration: Findings for the First Two Years. Claims-based Tools for Health Risk Assessment. Society Princeton, NJ: Mathematic Policy Research Inc. of Actuaries. April 2007.
17. Cumming RB, Cameron BA. A Comparative Analysis of 6. Goodwin N, Sonola L, Thiel V, Kodner D (2013). Claims-based Methods of Health Risk Assessment for Co-ordinated Care for People with Complex Chronic Commercial Populations. Society of Actuaries. May 2002.
Conditions. The King's Fund. October 2013. 18. Stam PJ, van Vilet RC, van de Ven WP. "Diagnostic, 7. Levine SH, Adams J, Attaway K, Dorr DA, Leung M, Pharmacy-based, and Self-reported Health Measures in Popescu B, Rich J. Predicting the Financial Risks of Risk Equalization Models." Med Care. 2010;48(5):448-57.
Seriously Ill Patients. California HealthCare Foundation. December 2011.
19. Perrin NA, Stiefel M, Mosen DM, Bauck A, Shuster E, Eirks EM. "Self-reported Health and Functional Status 8. Pope G, Ellis R, Ash A, et al. Diagnostic Cost Group Information Improves Prediction of Inpatient Admissions Hierarchical Condition Category Models for Medicare and Costs." Am J Manag Care. 2011;17(12):e472-478.
Risk Adjustment. Health Care Financing Administration. December 2000.
20. Lynn J, Schall MW, Milne C, Nolan KM, Kabcenell A. "Quality Improvements in End of Life Care: Insights 9. Hughes JS, Averill RF, Eisenhandler J, et al. "Clinical from Two Collaboratives." Jt Comm J Qual Improv. Risk Groups (CRGs): A Classification System for Risk- adjusted Capitation Based Payment and Health Care Management." Med Care. 2004;42(1):81-90.
21. Hasan O, Meltzer DO, Shaykevich SA, et al. "Hospital Readmission in General Medicine Patients: A Prediction 10. Impact Pro® Health Care Analytics for Care Management. Model." J Gen Intern Med. 2010;25(3):211-219.
Optum, Inc. Accessed September 29, 2014, 22. Freund T, Mahler C, Erler A, et al. "Identification of Patients Likely to Benefit from Care Management Programs." Am J Manag Care. 2011;17(5):345-352.
California HealthCare Foundation Appendix A: List of Interviewees
Aetna's Medicare Advantage Provider Collaboration Program Dorothy D. Briggs, RN, CCM AtlantiCare Special Care Center Sandy Festa, LCSW Lois Van Abel, MBA, RN Cambridge Health Alliance Complex Care Management Program Camden Coalition of Healthcare Providers CareOregon Health Resilience Program (working on behalf of Health Share of Oregon) David Labby, MD, PhD Community Care of North Carolina Carlos T. Jackson, PhD Denver Health Complex Care Program Dan Brewer Tracy Johnson, PhD Geisinger ProvenHealth Navigator Joann Sciandra, RN Geriatric Resources for the Assessment and Care of Elders (GRACE) Steven Counsell, MD Health Quality Partners Ken Coburn, MD, MPH Benjamin Berk, MDRushika Fernandopulle, MD, MPP Kaiser Permanente Northwest Complex Care Program Karen Carter Yvonne Zhou, PhD Massachusetts General Hospital Intensive Care Management Program Christine Vogeli, PhD San Francisco Health Plan Complex Care Program Maria C. Raven, MD, MPH, MSc Southcentral Foundation Melissa K. Merrick, LCSW, CDC ILaZell Hammons, RN, BSN Sutter Care Coordination Program Daren Giberson, RN Paul Herman Veteran's Affairs Palo Alto Complex Care Program Donna Zulman, MD, MS West County Health Centers Complex Care Program Jason Cunningham, DO Finding a Match: How Successful Complex Care Programs Identify Patients Appendix B. About This Study
The aim of this study was to identify best practices in patient selection and risk stratification from successful primary care-integrated complex care management programs. It followed a Commonwealth Fund-supported study of 18 primary care-integrated complex care management programs,* starting with 11 programs from the original 18 that had the most developed approaches to patient selection. Snowball sampling was then used to identify another nine programs for interview. Each site received at least two email invitations to participate in the study. Participants chose a key informant with knowledge of their patient selection and risk-stratification approach. The authors assessed each program using semi-structured key-informant interviews and review of published manuscripts and program materials obtained from each of the sites. At least one 45- to 60-minute, semi-struc-tured interview was conducted with each key informant. Additional interviews were performed, as necessary, to obtain further clarification and detail. Four broad study domains were assessed during the interviews: 1. Patient identification and selection approach
2. Approaches to identifying care-sensitive patients
3. Patient risk-stratification approach
4. Lessons learned and recommendations for other programs
*Hong CS, Siegel AL, Ferris TG. Caring for High-need, High-cost Patients: What Makes for a Successful Care Management Program? The Commonwealth Fund. August 2014.
California HealthCare Foundation Appendix C. Program Overview
pAtient selection DAtA sources useD For pAtient selection Aetna's Medicare Payer
1) Risk predictor – internal Aetna algorithm 1) Clinician referral Advantage
2) Multifactorial risk assessment – based on health 2) Patient self-referral 2) Health risk assess- Provider
risk assessment at enrollment (Medicare) 3) Referral from other 3) Other – system-generated triggers Aetna care manage-ment programs AtlantiCare
Multifactorial risk assessment – weighted scoring 1) CCM team chart Special Care
insured entities, system with points for aggravating/mitigating review using a risk 2) Electronic medical 2) Clinician referral 3) Patient self-referral Bellin Health
Utilization – LACE* tool/readmissions 1) CCM team chart quantitative with 2) Electronic medical 2) Clinician referral fee-for-service, Medicaid, 3) Health risk assess- Cambridge
1) Risk predictors – Network Health Managed 1) Clinician clinical Health Alliance
quantitative with Medicaid Predictive (MMP)‡ prospective risk score 2) Electronic medical Complex Care
qualitative gate (top 5% of Network Health patients), Diagnostic 2) Referral from PCPs Management
Cost Groups (DxCG)§ risk score (Medicaid), with quantitative ImpactPRO15 risk score (>75% probability of admis- sion in 1 year in Medicare Pioneer Accountable Care Organization patients)2) Cost – total medical expenditure 10x average for population3) Utilization – 8+ emergency department (ED) visits OR 3+ admissions in past year 4) Multifactorial risk assessment based on mixed quantitative and qualitative measures – applied to patients identified by criteria 1-3 or through referral *LACE – an evidence-based risk score calculated from Length of stay, Acute admission, Comorbidity (Charlson score), Emergency room visits – van Walraven C, Dhalla IA, Bell C, et al. "Derivation and Validation of an Index
to Predict Early Death or Unplanned Readmission after Discharge from Hospital to the Community." CMAJ. 2010;182(6):551-557.
†Quantitative gate – patient undergoes a final quantitative screening for entry into the program.
‡Yeracaris P. Community Engagement: Targeting High-risk Members to Enhance Care Management Effectiveness [PDF]. Retrieved January 14, 2015 from: .
§Ellis RP, Pope GC, Iezzoni LI, et al. "Diagnosis-Based Risk Adjustment for Medicare Capitation Payments." Health Care Financing Review 1996;17(3):101-128.
California HealthCare Foundation pAtient selection DAtA sources useD For pAtient selection Camden Coalition Regional care
1) Utilization – 2+ admissions in past 6 months 1) Electronic medical of Healthcare
eligible, Medicare quantitative with 2) Chronic disease – 2+ conditions review using a risk 3) Other – 3+ criteria from checklist CareOregon
Utilization – 6+ ED visits AND/OR 1 non-obstetrics 1) Clinician/CCM team 1) Claims Health Resilience
quantitative with (OB) hospitalizations in the last year clinical assessment 2) Acute care utiliza- Program (working
2) Clinician referral on behalf of Health Community Care
1) Risk Predictor – Care Management Impactability CCM team clinical of North Carolina
quantitative with Score# – top 2% based on impactability assessment – care 2) Retail pharmacy 2) Risk Predictor – Clinical Risk Groups** – top 2% manager review for of highest-yield hospitalized patients for transi- a rural, multi-payer 3) Statewide, hospital Denver Health
1) Mutifactorial risk assessment – combines Clinician clinical 1) Claims (managed 21st Century
quantitative with Risk Predictor (Clinical Risk Groups**), utilization, Care – Intensive
and diagnoses – CRG places population into 2) Internal data Outpatient Clinic
4 tiers, then patient-specific triggers (e.g., recent warehouse – Internal & High-risk Care
hospitalization, ED visits) promote patients into billing, registries higher-risk tiers Medicaid Fee-for-Service, uninsured Geisinger
1) Risk Predictor – MedAI‡‡ (highest-risk groups and 1) Clinician/CCM team 1) Claims quantitative with patients moving up 2+ risk groups), Predicted Risk 2) Internal data 2) Utilization – 3+ ED visits or 1+ admissions in 2) Clinician referral warehouse – disease 3) Referral from care registry, pharmacy 3) Cost – >$50K/year annual spend management inter- 4) Diagnoses – heart failure, chronic obstructive vention, inpatient 3) Electronic medical pulmonary disease, and end-stage kidney disease case management, home health, and 4) Health risk assess- the medical neigh- 5) All hospital discharges in Medicare patients or ments – self-insured borhood (e.g., local those meeting the following criteria: 55+ years and purchaser groups old, hospital length of stay 5+ days, diagnoses of resource centers) high-risk cancer, heart failure, chronic obstructive pulmonary disease, end-stage kidney disease. #Personal communication with Carlos T. Jackson, PhD, on January 15, 2015.
**Hughes JS, Averill RF, Eisenhandler J, et al. "Clinical Risk Groups (CRGs): A Classification System for Risk-Adjusted Capitation-Based Payment and Health Care Management." Med Care. 2004;42(1):81-90.
††Denver Health's 21st Century Care project is supported by the Department of Health and Human Services, Centers for Medicare & Medicaid Services, Contract Number 1C1CMS331064. The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.
‡‡How MEDai Can Help Predictive Modeling. MEDai, Inc. Web. Retrieved September 29, 2014.
California HealthCare Foundation pAtient selection DAtA sources useD For pAtient selection Geriatric
1) Age/Utilization Clinician referral Resources for
delivery system, eligible (including referral 2) Electronic medical Assessment and
from primary care Care of Elders
3) High-risk diagnoses physician, nurse 4) Multifactorial risk assessment care manager, and inpatient geriatric consult team) Guided Care
Risk Predictor - Hierarchical Condition Categories score§§ (top 25%) Health Quality
1) Risk Predictors – Aetna proprietary risk predictor Clinician/CCM team (for Aetna Medicare Advantage)## clinical assessment – 2) Hospital billing data quantitative with 2) Diagnoses – heart failure, coronary artery disease, chronic obstructive pulmonary disease, and diabetes3) Utilization – 1+ admissions in one year (Medicare only) Intensive manage- Integrated
1) Cost – Ranked in Top 5% by total medical expen- 1) Veteran's Affairs ment Patient
Decision Support Aligned Care
2) Risk Predictor – Care Assessment Need (CAN) System (DSS)††† Team (ImPACT),
Risk Score*** – Top 5% based on 1-year risk of 2) Regional data Veterans Affairs
warehouse (RDW)‡‡‡ Palo Alto
secondary insur-ance Multifactorial risk assessment – 4-tiers – combines 1) Clinician/CCM team 1) Claims Permanente
quantitative with risk predictor (Hierarchical Condition Categories§§ clinical assessment 2) Internal data Northwest Team
6 month likelihood of hospitalization, major chronic of highest-risk (tier 4) warehouse – disease Based Care
disease indicators, and selected clinical data and registries, hospital 2) Clinician referral acute care utilization from tier 2 and 3 3) Electronic medical record §§Pope G, Ellis R, Ash A, et al. Diagnostic Cost Group Hierarchical Condition Category Models for Medicare Risk Adjustment. Health Care Financing Administration. December 2000.
##Preston S. Aetna PhilaSUG Meeting PULSE AIM [PDF]. Retrieved January 14, 2015 fr.
***Wang L, Porter B, Maynard C, et al. "Predicting Risk of Hospitalization or Death Among Patients Receiving Primary Care in the Veterans Health Administration." Med Care. 2013;51(4):368-373.
†††Kraft MR, Hynes DM. "Decision Support within the Veterans Health Administration." Stud Health Technol Inform. 2006;122:100-104.
‡‡‡ VA Informatics and Computing Infrastructure. Accessed January 16, 2015 fr California HealthCare Foundation pAtient selection DAtA sources useD For pAtient selection Risk Predictor – ImpactPRO§§§ risk score (for total Clinician/CCM team General Hospital
quantitative with medical expenditures – 10+ automatic inclu- clinical assessment Integrated Care
sion, risk score 2-10 included if they have specific 2) Electronic medical Management
combinations of chronic conditions and utilization record for qualitative – lower-risk conditions require higher-risk utilization criteria for inclusion) San Francisco
Utilization – 1 non-OB admission plus 5+ ED visits 1) CCM team chart Health Plan
quantitative with OR 2+ admissions OR 6+ ED visits in prior 12 2) Electronic medical CareSupport
2) Referrals from health plan utilization management nurse or of Public Health Care Coordination Database feeds Utilization – 2+ admissions OR 1+ admission plus 2+ 1) Clinician clinical 1) Electronic medical quantitative with ED visits OR 4+ ED visits OR 2+ specialty visits OR 12+ ambulatory care visits in one year 2) Clinician referral 2) Enterprise data 3) Patient self-referral Sutter Care
Multifactorial risk assessment – based on 25 items 1) Clinical assess- 1) Electronic medical quantitative with ment by clinician qualitative gate### 2) Hospital data system – for acute care 2) Clinician referral West County
1) Cost – total medical expenditures (top 100 1) Clinician clinical Health Center
eligible, Medicaid quantitative with 2) Electronic medical Complex Care
2) Modified LACE**** (transitions of care) 2) Clinician referral 2) Clinician referral §§§Optum. Impact Pro® health care analytics for care management. Web. Retrieved January 14, 2015 from ###Qualitative gate – patient undergoes a final qualitative screening for entry into the program.
****Modified LACE – Kreilkamp, R. Application of the LACE Risk Assessment Tool at Chinese Hospital [PDF]. Retrieved January 14, 2015 fr. Published 2011. Accessed September 25, 2012.
California HealthCare Foundation

Source: http://allh.us/qr3e