1
Q
Reasons why plausibility factors may be a poor validator of care management savings calculations
Plausibility factors = disease-specific admissions per 1000 in the program year / disease-specific admissions per 1000 in the baseline year
Intent is to validate savings calc by demonstrating that actual utilization is reduced by the intervention.
A
- Calculated based on entire condition-specific population so they don’t exclude members who are not eligible for or managed by the program
- Calculated based only on admissions and ER visits for primary diagnoses, which represents a very small % of all admissions/costs)
- Don’t account for changes in population
- Don’t account for risk profile of a population
- Don’t account for volatility in admission rates
- Don’t account for existing trends in population
2
Q
Types of care management methods
A
- Pre-authorization
- Concurrent review - monitoring care while member is still receiving care in hospital/nursing home
- Case management - health care professional coordinates care of patient w/ serious illness
- Demand management - passive form of informational intervention, often over phone
- Disease management - focus on chronic conditions
- Specialty case management - care manager w/ expertise in particular area
- Population health management - statistical tools used to ID potential high-cost patients who could benefit from voluntary intervention program
- Patient-centered medical home - physical responsible for coordinating all patient care
- ACO - network of doctors and hospitals share responsibility for providing patient care
- Non-traditional provider interventions and care settings - e.g., pharmacists
- Gaps in care and quality improvement programs
- Telehealth, telemedicine
- Bundled payment initiatives
3
Q
Characteristics of chronic conditions that make them suitable for DM programs
A
- Once contracted the disease remains with patient for rest of life
- Disease is often manageable w/ combo of drug therapy and lifestyle changes
- Patients can take responsibility for their own conditions
- Average annual cost is high enough to warrant spending resources to manage condition
- Expected cost of non-adherence is high
4
Q
Principles for establishing a PCMH
A
- Personal physician
- Physician-directed medical practice - team of individuals taking responsibility for patient’s ongoing care
- Whole person orientation - arranging care w/ other qualified professionals
- Care coordinated and integrated across all elements of health care system and community
- Quality and safety - includes patient-centered outcomes, evidence-based medicine, and continuous quality improvement
- Enhanced access through open scheduling, expanded hours, and E-visits
- Reimbursement structure to support and encourage this model of care
5
Q
Ways in which ACOs are expected to generate savings
A
- Implementing care coordination to manage the care of patients who need additional services
- Reducing the need for tests via access to integrated medical records and consistent management by physician
- Developing a network of efficient providers for referrals and limiting use of less efficient/more $$ providers
- Focusing on quality –> fewer unnecessary nervices
- Emphasizing preventive services
6
Q
Types of interventions conducted by pharmacists
A
- Drug utilization review - substitute lower cost alternatives, require prior auth
- Medication Therapy Management (MTM) - required by Part D plans; good for those with multiple chronic conditions, taking multiple Part D drugs, annual costs > $4k for covered drugs
- Pharmacist-delivered care management programs - often focus on drug adherence, measured in two ways:
a) Medication possession ratio = # of days supply in patient’s possession / # of days during measurement period during which patient could have had drug
b) Proportion of days covered = # of days of coverage / total # of days in measurement period
7
Q
Types of clinics that can be used to provide basic health care
A
- Retail convenient care clinics - staffed by NPs
- Employer worksite clinics
- Urgent care clinics - freestanding centers offering full range of ambulatory services
- Federally qualified health centers (FQHCs) - designated by gov’t to provide health care to underserved and uninsured
8
Q
Benefits of being designated an FQHC
A
- Reimbursement for services provided under Medicare and Medicaid
- Medical malpractice coverage
- Eligibility to purchase meds for outpatients at reduced cost
- Access to Nat’l Health Service Corps
- Access to Vaccine for Children Program
- Eligibility for various other federal grants
9
Q
Areas where actuaries can be involved with care management programs
A
- Economics of care management programs - help w/ understanding relationship between program inputs and outputs
- Risk adjustment and predictive modeling
a) Predictive modeling - used to ID candidates for intervention programs
b) Risk adjustment used to assess outcomes - Financial outcomes evaluation - help in achieving comparability between reference and intervention population
10
Q
Principles for measuring results of care management programs
A
- Reference pop
- Equivalence - reference pop should be equivalent to intervention pop
- Consistent statistics - same stat should be measured in same way in reference and intervention populations
- Appropriate measurement - avoid extraneous/irrelevant variables
- Exposure - exposure group must be clearly defined and all members who meet definition should be included in appropriate group
- Reconcile results - reconcile outcomes of small pop with those of entire health plan (“plausibility analysis”)
11
Q
Issues that affect DM evaluations for chronic populations
A
- Regression to the mean - high % of high cost patients in one period will not be high cost in next period b/c of one-time events
- Identifying patients - due to regression to mean, may not be appropriate to use patients’ past data as comparison group. Common alternative is to use population approach (entire pop)
- Establishing uniform risk measure for comparability - objective, consistent definitions should be used to identify candidates for care management program (ensures equivalence)
- Patient selection bias - results with volunteers
- Patient drop outs - may also create bias
- General vs specific population - some interventions are performed on extremely small pop, so some methodologies are inappropriate for measuring results
12
Q
Considerations when using claims data for evaluating DM programs
A
- Fixed time periods - 1 year may be too short for outcomes evaluation
- Enrollment issues/eligibility - timeliness of enrollment and disenrollment should be factored into study
- Claims run-out - due to claims lag, results may not be known for up to 2 years after program begins
- Outlier claims - may distort results
- Special problems with claims data - some members are miscategorized
13
Q
Risk factors for care management studies
A
- Demo variables
- Exclusionary conditions that exclude certain members (members might not be good candidate for care management)
- Exclusionary conditions that exclude certain claims - exclude claims for conditions that DM does not try to affect (e.g., maternity)
- Persistency - understand terms under which member may enter or leave group
- Chronic prevalence and risk classification - % of individuals in a pop with the condition
- Severity of illness
- Contactability
- Operational issues - # of eligible members, # of chronics identified/enrolled, grad rates, meths used
14
Q
Components of care management value chain process
A
- Data warehousing - integrate mbr/claims data, ID conditions
- Predictive modeling - apply models to determine members to target for interventions
- Intervention development - develop campaigns to deliver interventions to target pops
- Outreach and enrollment
- Member coaching and assessment
- Outcomes assessment - clinical/financial/operational
15
Q
Reasons for measuring health care quality
A
- Improving health of population
- Monitoring services rendered
- Evaluating outcomes
- Shaping provider behavior
- Meeting reqs of gov’t regs, biz partners, accreditation agencies
16
Q
Challenges when using codes to measure quality
A
- Codes don’t give complete picture of care provided (no quality info)
- Coding errors and fraud are prevalent
- Electronic medical records may contain wrong or missing diagnoses
- Source of coded data affects interpretation
- Codes can only indicate if care was provided, not if patient complied with doctor’s orders
17
Q
Organizations that measure health care quality in US
A
- Nat’l Quality Forum (NQF) - lead responsibility for determining which measures should be recognized as nat’l standards
- Agency for Healthcare Research and Quality - developed quality indicators which use hospital data to highlight concerns and ID areas for investigation
- Joint Commission - primary accrediting body for hospitals, nursing homes, etc
- CMS - develops measures of quality
- Nat’l Committee for Quality Assurance - develops quality standards for various health care orgs; develops HEDIS measures
- Hospital Quality Alliance - develops performance measures of hospital care
- Measures Applications Partnership - ID best performance measures for specific applications
- American Medical Association Physician Consortium for Performance Improvement - developed evidence-based performance measurement sets
18
Q
Categories for measuring health care quality
A
- Structure - resources and org arrangements are in place to deliver care
- Process - appropriate physician and other provider activities are carried out
- Outcomes - results
19
Q
Methodologies for assessing progress of clinical quality initiatives
A
- Percentage compliance = # of times service was provided / # of times provider could have performed service (not counting patients whose conditions preclude them from treatment)
- A vs E performance - consider comparability between intervention and comparison populations (risk and case mix)
- Performance against benchmark - provider’s performance is compared to benchmark to determine efficiency
20
Q
Possible reasons why DM studies show improved clinical outcomes but not cost savings
A
- Measurement of financial outcomes is not stable enough to detect positive outcomes
- Programs are either not focused on financial outcomes or not structured to optimize financial outcomes
- Program sponsors do not understand economics of DM programs and therefore do not optimize programs for financial return
- Improvements in quality of care do not always lead to lower costs
(Factors to help resolve contradiction include better understanding of economics of DM programs, more rigorous measurement of financial outcomes, and reconciliation between program savings/overall claim costs/cost increase trends)
21
Q
Financial measures for DM programs
A
- ROI (usually gross)
a) net ROI = (gross savings - cost) / cost
b) gross ROI = gross savings / cost
c) program costs generally include direct costs (e.g., salarise), indirect costs of supporting activities, mgmt costs, overhead costs, and setup costs
d) gross savings come from decreased utilization as result of DM program - Total savings - may be more useful since represents dollar savings for plan
a) Average savings = total savings net of program cost divided by total pop
b) Marginal savings per chronic member = increase in savings (net of costs) due to intervention on marginal pop, divided by # of members in marginal pop
22
Q
Key metrics in design of DM programs
A
- # and risk-intensity of members to be targeted - # must be large enough to produce savings that offset implementation costs, but not so large that marginal costs exceed marginal savings
- Types of interventions to be used in program - mail, automated dialing, etc
- # of nurses and other staff needed
- Methodology for contacting and enrolling members
- Rules for integrating program w/ rest of care management system
- Timing and # of contacts, enrollments, and interventions
- Predicted behavior of target pop if there were no intervention, and predicted effectiveness of intervention at modifying that behavior
23
Q
Components of Risk Management Economic Model
A
- Prevalence of different chronic diseases
- Cost of chronic disease
- Payer risk - most savings for plan will come when plan is at financial risk for all of patient’s costs
- Targeting and risk - members should be prioritized based on probability of experiencing targeted event
- Estimated cost of targeted event
- Contact rate
- Engagement/enrollment rate
- Member restratification rates - initial risk rank will be restratified after nurse assesses member’s risk
24
Q
Description of opportunity analysis
A
- Definition - data-driven analytical process that extends traditional predictive modeling by matching opportunities w/i pop to care management programs and services
- Is retrospective - looks at past data to ID pockets of opportunity
- Is applied prospectively - current members meeting opportunity pop profile can be included in program
25
Components for designing a care management program using opportunity analysis
1. Analytics - members are segmented by medical conditions into subpops amenable to different types of interventions; util data is compared to benchmark to highlight areas with most potential for UM savings
2. Searching the evidence base for knowledge of what works and what doesn't
a) Lit review is done to find programs that are efficacious (evidence can be trusted), cost-effective (ben > cost), and generalizable to pop to be managed
b) 3 step approach is used: search for relevant publications, assess quality of evidence, determine generalizability
3. Weighing the economics
a) Population is risk ranked using predictive model which determines expected cost for each person
b) Cost is compared to person's cost w/o intervention to determine savings
c) Savings is compared to cost of intervention to determine at what point in risk ranking is it economically feasible to intervene
26
Steps for implementing a care management program using opportunity analysis
1. Develop predictive model to populate risk distribution
2. Establish production analysis and reporting unit, and develop necessary reports
3. Determine likely # of care managers required
4. Develop budget for program, accounting for all req'd resources
5. Hire and train care managers to conduct interventions and manage patients
6. Develop plan, including estimates of #s of patients identified and engaged
7. Roll out intervention and enroll patients
8. Operate program, track outcomes, modify as necessary
27
Reasons for using opportunity analysis for identifying patients for care management interventions
1. Studies have shown that clinicians are not particularly good at IDing high risk patients
2. Economics of program planning can't be ignored in system with limited resources
3. Structured approach is important for understanding which subpops are amenable to intervention and likely value of that intervention
4. Structured financial model provides framework against which actual outcomes may be compared, IDing areas where program needs to be corrected/improved
28
Considerations when evaluating results of DM studies
1. Has measurement been performed according to valid methodology?
2. How has methodology been applied in practice?
3. Are results arithmetically correct?
29
Requirements for care management methodology to be valid
1. Familiarity - purchaser must be familiar with meth or able to grasp it quickly
2. Ease of replication and auditability - meth must be documented so it can be replicated by another
3. Results upon applying meth must be consistent with client's savings expectations, and be plausible
4. Results should be stable over time and between clients
5. Meth must be practical (possible to implement cost-effectively)
6. Inherent validity - lack of obvious bias
7. Scientific rigor
8. Market acceptance - how method is perceived in market
9. Application - how meth is applied in practice
30
Types of methodologies for estimating care management savings
1. Control group methods - attempt to match study subjects w/ other subjects not part of the study
2. Non-control group methods - population methods that do not use control groups
3. Statistical methods - use purely statistical techniques, rather than constructing an explicit reference pop
31
Control group methods for estimating care management savings
1. Randomized - compares equivalent samples drawn randomly from same pop (preferred method)
2. Geographic - compares equivalent pops in two different locations
3. Temporal - compares equivalent samples drawn from same pop before and after intervention program
4. Product control methodology - compares samples drawn from same pop at same point in time, but differentiates between members who have different products
5. "Patient as own control"
6. Participant vs non-participant studies - experience of those who volunteer is compared to experience of those who choose not to participate (selection bias)
32
Non-control group methods for estimating care management savings
1. Services avoided methods - savings = estimated cost of service requested through pre-auth minus actual cost after intervention
2. Clinical improvement methods - change in clinical measure is observed and resulting improved health and reduced utilization is estimated from outside studies
33
Statistical methods for estimating care management savings
1. Time-series methods - curve is fit to data over time and divergence from this best-fit line can be observed once intervention is applied
2. Regression discontinuity - line is fitted to data that relates pre- and post-intervention experience
3. Benchmark methods - values of certain key stats are compared between pop being managed and some benchmark pop
34
Description of propensity scoring
1. Propensity score matching is a technique used for making a participant (intervention) group comparable to a nonpar group
2. Each member in participant group is matched with member of nonpar group based on propensity scores
3. Propensity score, p, is probability that member will be in participant group
a) Calc'd using logistic regression based on member's values for independent variables (such as age, gender)
b) Process reduces large # of variables to single score
c) Members with similar scores can then be matched, even if they are not matched exactly on independent variables
d) Should still be relatively close matches on other variables
35
Steps for applying propensity score matching to a study
1. Run logistic regression to create propensity score, considering as independent variables any factors that may influence person's decision to participate in program
2. Use propensity scores to match each participant to a nonpar using one of the following techniques:
a) Nearest neighbor matching - 1st member of comparison pop w/ closest propensity score is selected, either w/ or w/o replacement
b) Caliper matching - match is made if member's and match's propensity scores are w/i fixed distance
c) Mahalanobis metric matching - metric is used to measure the dissimilarity between two vectors
d) Stratification matching - observations are stratified and then matched by stratum
3. Test model for appropriateness and bias - difficult b/c propensity score match only adjusts for observable variables
36
Comparison of propensity scoring and risk adjustment
Similarity
1. Both reduce effect of multiple risk factors (age, sex, diagnoses, etc) to a single score, using multiple regression
Differences
1. Propensity score is usually based on wider range of independent variables, but risk score will almost always take into account more detailed diagnosis variables
2. Risk adjustment uses entire pop, while propensity matching can result in many members of pop being discarded
37
Formulas for calculating DM program savings
Based on actuarially-adjusted historical control design (temporal control group method)
1. Savings = [PY util rate per chronic mbr * (1 + non-chronic trend) - Act util rate per chronic mbr] * Chronic members * cost per service
2. Savings PMPM = savings / MMOS
38
Reasons a member may be excluded from DM program
1. Member class is not receptive to DM
2. Member is candidate for program administered by another vendor (e.g., mental health)
3. Pattern of member's claims is subject to sharp discontinuity which can distort trend calc
4. Member's claims are significant which could introduce noise or dominate group
39
Conditions that would exclude member from DM program
1. ESRD - DM may delay cost but can't reduce or postpone cost
2. Transplants - high claims until transplant, then stable
3. HIV, AIDS, mental health - privacy issues
4. Institutionalized - may not be reachable, may not benefit
5. Members w/ cat claims - not manageable by DM program, often subject to management by another program
6. Members who are eligible for other management programs
40
Challenges when calculating DM savings (when using actuarially-adjusted historical control design)
1. Applying proper trend rate - trend of non-chronic pop is typically used b/c chronic trends are impacted by DM efforts; trend must be adjusted for average risk of pop
2. Demonstrating equivalence between baseline and measurement periods - must account for change in mix of new, continuing, and terminating members and any changes in conditions and co-morbidities; can be done by re-weighting claim costs that are used in savings calc
41
Methods for calculating trend to use in evaluating a DM program
1. Group-specific trend - most savings calcs use trends based on employer's non-chronic pop, but this trend is subject to random fluctuations even for large groups
2. Population trend - to reduce random fluctuation, a very large trend source should be considered (like plan's trend for all groups combined)
3. Truncated trend - reduce random fluctuations by truncating claims at $50k; problem - savings for larger claims don't get counted
4. Utilization trend - calculate reduction in admissions instead of reduction in claim costs ---> credibility being achieved at much smaller group sizes
a) Claim cost savings can be calc'd if reasonable cost/admission can be estimated
b) Doesn't count savings that result from reducing length of stay
42
Approaches and interventions for ACOs to optimize care and achieve performance targets
1. Care redesign to improve delivery and coordination of care - e.g., establishing PCMHs, improving transitions in care
2. Care management of patients w/ costly, complex needs - individualized approach to ID and address unmet needs
3. Patient and family engagement and patient activation initiatives - ID personal goals for lifestyle changes, educate about treatment options
4. Integrated data and analytics - to ID patients who could benefit
5. Supportive payment models and financial incentives - capitation, shared savings, etc
43
Benefits to stakeholders of using standardized measures for assessing employee health management programs
1. Employers and benefit managers - core metrics can facilitate comparison, provide basis for developing vendor performance metrics; employers can use data to ID gaps in programs
2. Benefits consultants - standard set of metrics may lead to reliable comparative data for making vendor recommendations, negotiating performance standards
3. Health management program managers - core metrics will provide data to fine tune health enhancement interventions
4. Accrediting orgs - metrics can be used to evaluate vendor/health plan compliance
5. National health management policymakers - core metrics will facilitate development of recommendations
6. Employee health management services vendors - core metrics will create level playing field for competitors and encourage product improvements
7. Employee health management participants - will benefit from product improvements stemming from competition
44
Steps taken to develop standard set of measures for EHM programs
1. Review literature to discover what metrics are currently used to measure performance of EHM programs
2. Obtain guidance and advice from SMEs
3. ID and/or develop recommended measures
4. Review work with key stakeholders to obtain feedback and consensus
5. Release work through conference presentations, publication, other channels
45
Steps of EHM value chain (help in understanding how EHM programs add value)
1. Assess all individuals in population to ID opportunities to maintain or improve health
2. Engage individuals with programs and tools through which they can address these opportunities
3. Continue engagement long enough for them to acquire and sustain healthy behaviors
4. Sustained effective engagement will result in preventing or reducing lifestyle-related risk factors
5. Sustained healthy behaviors and clinical outcomes will result in fewer ER visits, hospitalizations, procedures
6. Fewer of these lead to medical, absenteeism, workers comp, and disability cost savings
7. Improved employee productivity and performance contribute to improved financial outcomes
(First 5 are plausibility metrics)
46
Recommended measures for assessing EHM programs
1. Financial outcomes
2. Health impact (physical, mental, emotional, healthy behaviors, overall health measures)
3. Participation (telephonic, web-based, in person)
4. Satisfaction (patient, client)
5. Organizational support (org support elements, employee-perceived level of organizational support, leaders-perceived organizational support)
6. Productivity and performance
7. Value on investment (comparison of investment vs outcomes)
47
Organizational support elements for EHM programs (deliberate steps the employer can take to create an environment that supports health and well-being)
1. Company-stated health values
2. Health-related policies
3. Supportive environment
4. Organizational structure
5. Leadership support
6. Resources and strategies
7. Employee involvement
8. Rewards and recognition
48
Leading indicators of savings for EHM programs (indicate during 1st year whether a program is likely headed for savings later on)
1. Identification, stratification, and targeting (outreach)
2. Program enrollment and use of tools
3. Continuing engagement or program completion
4. Behavior change
5. Behavior maintenance
6. Processes of care
7. Medication adherence
8. Achieving clinical targets
9. Patient activation
10. Satisfaction with EHM
11. Well-being
49
Lagging indicators of savings for EHM programs (often improve after sustained high performance on leading indicators)
1. Functional status
2. Quality of life and well-being
3. Absenteeism and presenteeism
4. Morbidity (ER, hospital, procedures)
5. Healthcare claims cost
50
Recommended financial metrics for EHM programs
1. Directly-monetized claim savings metrics
a) Cost trend compared w/ industry peers (compares trend to peers w/o EHM)
b) Adjusted-expected compared to actual cost trend
c) Chronic (observed) vs non-chronic (expected) trend - used for DM
d) Cost or trend comparison of program participants vs nonpars - after neutralizing impact of non-EHM differences
e) Comparison w/ matched controls in a non-exposed pop - compares costs of members who meet criteria for EHM targeting in employer pop w/ members who meet criteria in a comparison pop that does not have EHM
2. Monetized impact on utilization that is potentially preventable by EHM
3. Financial impact based on model that links to what occurred during program and characteristics of program participants
4. Reduction or prevention of lifestyle-related health risk factors - relates to published evidence on economics of preventing/reducing such factors
51
EHM value prop
1. Identify opportunities to:
a) Improve (or maintain) health
b) Mitigate or eliminate current risks or avoid future risks
2. Address these opportunities with effective programs and tools to improve pop health status, improve productivity, and lower health-related costs
52
Considerations in choosing whether to use modeled or measured savings for EHM calculations
(Modeled are estimated by multiplying factors from published studies by util reductions or other results of EHM program)
(Measured are estimated by comparing actual claims to what claims would have been w/o EHM)
1. Measured savings are not accurate for small pops, so models should be used for them (common cutoff = 25k members)
2. Measured savings calculations require fully-adjudicated claims data, but savings models require only data typically generated through program such as demos, diseases, etc
3. Measured savings are usually calculated annually, while modeled savings are calc'd whenever
4. Measured savings inherently incorporated organization's specific data, but modeled savings must incorporate this data to be as accurate
5. Measured savings are validated/audited by 3rd party, while modeled savings are based on published evidence or studies
GH Advanced
flashcards
Decks in class (4)
# Cards
