Group Chap 35: Medical Claim Cost Trend Analysis Flashcards
Introduction
- Measurement of periodic trends is very important
- “Trend” or “trend rate” refers to rate of growth of incurred claim cost per member per month (PMPM)
- Can also refer to growth in sales, expenses, premium or other items -
Purposes of Trend Analytics
- Reveal insights into insurer’s current benefit and rate structure and identify emerging areas of concern
- Assist in setting future premium rates and/or budgets
Types of Trend
- Three Key Purposes for Medical Cost Trend
o Financial Reporting
o Pricing
o ExperienceAnalysis
1. Financial Reporting
a. Purpose of financial reporting – determine financial performance during a given period
b. Criteria for Trend Analysis in Financial Reporting
- Reporting is retrospective, but budgeting requires projected claims for the next few years
- Reporting done at enterprise level, as well as local levels
- If statutory basis, margin for adverse deviation is required. If GAAP basis, use best-estimate basis
c. 𝐼𝑛𝑐𝑢𝑟𝑟𝑒𝑑 𝐶𝑙𝑎𝑖𝑚𝑠 = 𝑁𝑒𝑡 𝑃𝑎𝑖𝑑 𝐶𝑙𝑎𝑖𝑚𝑠 + 𝑅𝑒𝑠𝑒𝑟𝑣𝑒𝑠 𝐸𝑛𝑑 𝑜𝑓 𝑌𝑒𝑎𝑟 − 𝑅𝑒𝑠𝑒𝑟𝑣𝑒𝑠 𝐵𝑒𝑔.𝑜𝑓 𝑌𝑒𝑎𝑟
- [Note – this formula has been removed from the most recent edition of the textbook, but I think
it’s important to know for basic understanding of how claims work and are computed]
d. Usually, general ledger reporting systems don’t have enough detail to perform a trend study, so you must rely on electronic data warehouse (EDW)
- EDW may only include payments for a specific claim and not aggregate amounts such as for a lawsuit or bonus payments
- Processing times between EDW and general ledger will vary (different cut-off and loading dates)
2. Pricing
a. Used to project experience to a new time period for premium rate setting or budgeting
b. Typically relies on data from EDW
c. Trends may be calculated on eligible, covered or net paid basis
- Eligible – billed charges before provider contracts or discounts are applied
▪ Good for looking at underlying patterns without effects of contracts and cost-sharing
- Allowed Charges / Covered charges – eligible claims after provider contracts/discounts but before member cost-sharing
▪ Gives info about underlying patterns including contracting but without member cost- sharing
- Net paid claims–bottomline, after contracting and member cost-sharing
▪ Provides overall net paid trend
d. Trends are very specific to a time period
3. Experience Analysis
a. A look at how experience for a specific block of business is changing over time
b. Based on retrospective claims
c. Techniques to incorporate unreported claims into analytics (based on reports from EDW)
Developing Pricing Trend - Component Method
- Component method – evaluate trend on each component and then aggregate
- Advantage – straightforward to evaluate and explain specific impacts
- Disadvantage – considerable resources required to work through each component
- Core Cost Trend
- Cost Utilization Trend
- One-time Changes
- Population Shifts
- Structural Changes
- Alternative Payment Models
- Leveraging
1. Core Cost Trend
a. Rate of increase in the covered cost per service (before adjustments for aging or one-time changes)
b. Key Pieces of Core Cost Trend
- Unit Cost Trend
▪ Measure change in cost of actual product year over year (similar to inflation measure)
- Severity
▪ Increase in intensity of treatment - E.g. moving from normal deliveries to caesarean deliveries
- Uses weighted classification systems to give weight per admission
- Mix of Services
▪ Could be distribution of inpatient v. outpatient v. professional fees or mix of providers used
▪ Very complex
▪ Sometimes may be a balancing item and determined by examining historical trends
c. Components are multiplicative: 𝑇𝑟𝑒𝑛𝑑 = 𝑈𝑛𝑖𝑡 𝑇𝑟𝑒𝑛𝑑 𝑥 𝑆𝑒𝑣𝑒𝑟𝑖𝑡𝑦 𝑇𝑟𝑒𝑛𝑑 𝑥 𝑀𝑖𝑥 𝑜𝑓 𝑆𝑒𝑟𝑣𝑖𝑐𝑒𝑠 𝑇𝑟𝑒𝑛𝑑
d. Most insurers adjust unit cost trend for product and area differences
e. Usually use overall book data for severity and mix trends
2. Cost Utilization Trend
a. Changes in utilization due to external forces such as pandemic shutdowns, economy, and number of
workdays
b. When economy is stable, can be predicted by econometric models
- In unstable times, factors like fear of layoffs may disrupt historical patterns
c. Utilization may vary by weekday and placement of holidays
- Monday is most frequent day for doctor visits
- Less hospital admissions on Fridays
- Less prescriptions filled on Wednesdays
3. One-Time Changes
a. Represents response to specific, one-time events based on specific, identifiable situations
b. Examples where trend is expected to return back to normal
- High flu seasons – goes up one year but likely to return to normal level next year
- Weather events – utilization decreases during a hurricane or large event, though a small increase is expected soon after
c. Examples where a sustained change may occur (i.e. won’t revert back to the “normal” level)
- Legislation – new mandated benefits cause increased costs
- Internal changes – changes in processing and procedures at insurer may affect cost trends
d. New technologies often have material impact on health care costs and trends
- Costs follow a “technology curve”
- Phases of Technology Curve
▪ Before Introduction
▪ 1. Start Up – utilization increases slowly, patients learn about it
▪ 2. Catch Up – highest utilization, emphasis on treating existing patients
▪ 3. Steady State – emphasis on newly diagnosed patients
▪ 4. Transition – new technology slowly replaced by another new technology or lower cost version
4. Population Shifts
a. Underlying population may shift as member leave and enter the group
b. Reflect expected change in costs due to population shifts not reflected in other parts of projection process
c. One may think adjustments not needed for most insured business because base rate is adjusted for age-gender factors to determine final premium
- Age-gender factors may be limited due to legislation (e.g. ACA 3:1 age rating limits)
- Pricing trend should reflect difference in premium increase vs true cost increase
d. If community-rated population is stable – core group of members will be in group next year, but one year older. Entire impact of aging should be reflected in trend calculation
- Populations aren’t always stable, thus other information must be reflected in calculations
e. Projecting trends for rapidly growing or new populations – most difficult situation
- 1. Project composition of new population
- 2. Estimate PMPM based on info and utilization for similar members
5. Structural Changes
a. If no structural changes – then must consider above items If there are structural changes – trend needs to reflect change in actuarial value (due to changes in coinsurance amounts) AND expected changes in insured’s behavior
- Structural changes – could include benefit changes, changes to clinical programs or network changes
- If benefits are reduced, people may use less of them (don’t want to pay as much)
- If benefit costs increase, healthy members may leave the plan or switch to a lower-cost plan, with
only higher risk/cost members staying in the plans
b. Benefit Rush-Hush-Crush
- When a big change is announced beforehand and affects utilization
- Rush – people rush to use benefits with the outgoing plan (often richer benefits)
- Hush – less usage in the next quarter because people already got their treatments and also have
uncertainty about how new plan works
- Crush – usage returns to normal, but the trend rate for that year is very high because it’s coming from an abnormally low prior year period (starting at “hush” levels)
- Common Examples
▪ Moving to Consumer Driven Health Plan – introduces high deductible plan, so consumers opt to receive certain services prior to this change
▪ Change in Insurer – creates air of uncertainty, so consumers use existing coverage while they still can
6. Alternative Payment Models
a. Fee-for-service (FFS) models criticized – no explicit provider accountability for outcomes, quality or efficiency
b. Alternatives to FFS – capitation and global budget models
- Capitation – providers are paid a capitation rate/fee for each member of a given population
- Global budgets – similar, but based on total population instead of individual
c. Alternative Payment Models (APMs) (aka value-based payment models)
- Becoming more sophisticated and prevalent – e.g. incentives and penalties
▪ Accountable Care Organizations (ACOs) – providers held accountable for population’s overall cost of care and share in savings if cost is below the target cost
d. Projected costs must reflect all expected claim payments, capitation, global budget amounts, incentives and penalties
7. Leveraging
a. Trend on allowed amounts usually different than trend on paid amount (after coinsurance and copays)
▪ E.g. Allowed cost increased 5%, copay stayed the same, paid trend of 6.7%
▪ Higher fixed cost sharing leads to greater leveraging impact
▪ This is called trend leveraging
b. Leveraging factor usually baked in to pricing trend, across a portfolio of plans
- Easy to determine if just one plan
- Harder with multiple plans – usually just compare ratio of net paid trends to allowed trends over time
Monitoring and Analyzing Trends
1. Insurers typically produce financial and pricing trends on monthly or quarterly basis
2. Questions to Consider
a. How accurate were original trend and PMPM estimates?
b. Which assumptions drive variation from expected to actual trend estimates?
c. How can process be modified to be more accurate?
d. What other factors (expected or unexpected) drove the trends?
3. Validating Prior Pricing Trend Estimates
a. First step is to determine difference in original estimate and current estimates of trend
b. Each component should be analyzed from period to period
- Substitute actual experience for projected experience
- Actual trends may be close in total, but this identifies differences in components as well
4. Analyzing Results
a. Going beyond population shifts and one-time changes to determine key drivers of costs and trends
b. Typical method is to analyze trend by service type (i.e. inpatient, outpatient, professional, mental health)
- Note that a look at a few periods will be more comprehensive
Conclusion
Component method of trend analysis provides valuable insight, but not every organization will have resources/data for this approach
- Common alternative – regression technique (e.g. Holt-Winter method – captures seasonality)
- Regression should be based on PMPM costs, normalized for key factors (such as contract changes and one-time changes)
