Sec T Prop Score, Lvl Fund, Cred Form, & Canada Flashcards
Description of propensity score matching
1) Propensity score matching is a technique used for making a participant (intervention) group comparable to a nonparticipant group. It can control for observable variables like age, gender, and geography, but it does not control for important unobservable influences like willingness to change behavior
2) Each member in the participant group is matched with a member of the nonparticipant group based on propensity scores
3) The propensity score, p, is the probability that the member will be in the participant group
a) It is calculated using logistic regression based on that member’s values for the independent variables (like age and gender)
b) This process reduces a large number of variables to a single score
c) Members with similar scores can then be matched, even if they are not matched exactly on the independent variables
d) There should still be relatively close matches on those other variables
Steps for applying propensity score matching to a study
1) Run logistic regression to create a propensity score. Should consider as the independent variables any observable factors that may influence a person’s decision to participate in the program
a) The regression equation is ln[p/(1-p)] = alpha + beta*X + e
b) So propensity score, p = exp[alpha + beta*X] / (1 + exp[alpha + beta*X])
2) Use propensity scores to match each participant to a nonparticipant using one of the following techniques:
a) Nearest neighbor matching - the first member of the comparison population with the closest propensity score is selected, either w/ or w/out replacement
b) Caliper matching - a match is made if the member and match’s propensity scores are within a fixed distance
c) Mahalanobis metric matching - this metric is used to measure the dissimilarities between 2 vectors
d) Stratification matching - observations are stratified and then matched by stratum
3) Test the model for appropriateness and bias - testing for bias is difficult because the propensity score match only adjusts for observable variables. Models should be parsimonious (should use only the minimum # of variables necessary to achieve a stable model).
Comparison of propensity scoring and risk adjustment
Similarity: 1) Both reduce the effect of multiple risk factors (like age, sex, and diagnoses) to a single score, using multiple regression
Differences: 1) The propensity score is usually based on a wider range of independent variables. But the risk score will almost always take into account more detailed diagnosis variables
2) Risk adjustment uses the entire population, while propensity matching can result in many members of the population being discarded
Description of level funding products
1) Level funding is an ASO product with integrated stop-loss coverage. It allows groups to benefit from the advantages of self-funding, while limiting the disadvantages
2) The cost components of level funding products are:
a) An ASO fee to cover administrative and selling expenses
b) Aggregate stop-loss (ASL) coverage
c) Specific stop-loss (SSL) coverage
d) A paid claims fund held by the insurer to cover the group’s expected non-stop loss claim costs over the projection period
e) An IBNR fund to cover claims incurred during the projection period, but paid afterward
3) A 6th, unofficial component is an incurred claims cost projection, which is used to develop several of the other cost components
4) The paid claims fund is what allows the group to pay fixed monthly payments
a) The payment into the fund = the ASL corridor (or aggregate margin factor)* the group’s projected paid claims below the SSL deductible
b) This payment pre-funds the group’s maximum liability. Any amounts greater than this are covered by stop loss
c) If the group’s actual paid claims are below the maximum liability, the group will receive some portion of the paid claims fund’s surplus as a refund
Considerations when offering a level funding product
1) Insurers would want better-risk small groups to choose ACA products since these groups are very profitable under community rating. But these groups will seek lower-cost alternatives, so they could go to competitors who offer level funding policies
2) Insurers have not yet had much success with level funding products. But this should change once transitional policies go away, making level funding products the cheapest option for many groups
3) Level funding products are not easy to price, sell, and administer:
a) Insurers must have the skills to properly project the expected claim costs of individual small groups. They also must become familiar with stop-loss products
b) Insurers should retain legal expertise to understand the stop-loss regulations in their states and develop contracts appropriately
c) Selling stop-loss policies often requires the filing of rates and forms with state departments of insurance
4) Level funding products should be designed and priced to closely resemble the fully insured products they are replacing
5) Most small groups are not familiar with self-funding or stop loss, so insurers will need to help them and their brokers understand these products
Coverage options for groups that have kept transitional coverage
These groups must choose one of the following in 2018 once transitional relief epxires
1) Purchase ACA-compliant coverage
2) Drop coverage
3) Find some way to be defined as a large group
4) Enter into a self-funding arrangemen, such as ASO, ASO w/ stop loss, a minimum premium arrangement, or coverage administered by a third party administrator w/ or w/out stop-loss
Common features of the group medical insurance market
1) Premium rates are usually guaranteed for one year
2) Groups usually consist of employees (and their dependents) of a single company or a governmental unit, or members of a union
3) Healthy individuals tend to remain healthy and incur few claims, while many illnesses tend to last more than one year
4) Insurance coverage is written w/out individual health underwriting
5) Premium rates do not change during the contract year
6) Retrospective experience rating may be used for larger groups
7) Insurers have a manual rating system that includes many factors, like geographic location, age, and gender
8) Most insurers use the group’s own claim experience in projecting claim costs
a) For groups w/ less than 300 members, insurers typically blend the group’s own experience with a manual rate
b) This requires a credibility table or formula, which should not vary by too many factors and should be relatively easy to expain
9) Purchasers of group health insurance tend to be very knowledgeable about their benefits and will shop around for competitive premiums
10) The market for health insurance coverage is very price competitive
11) The policyholder is usually advised by consultant or broker who has little loyalty to the insurer
12) Competitive pressure exerts a downward price influence and reduces profits margins
13) Most covered individuals incur at least some claims over the course of the year
14) The distribution of medical claims by individual has a very high variance
Considerations for applying credibilty to the group medical insurance market
1) Many credibility models apply to products for which claims are rare, so they must model claim frequencies. But for group medical insurance most individuals have claims, so a reasonable simplifying assumption is that all individuals have claims
2) Individuals who have high claims in one year will tend to have above average claims in the next year. So the credibility of a group of only one member is significant
3) Because the market is competitive, insurers that use inappropriate credibility levels could experience significant losses
Basic credibilty formula for group medical insurance
This formula was created using a least squares credibility model
z = [k1 + (n-1)*k2] / [1 + (n-1)*k3]
1) n is the # of individuals in the group
2) k1 is the credibility of a group of 1 individual. It equals the regression coefficient of an individual’s claims in the current year based on the prior year. It is typically around 25%
3) k2 is the regression coefficient of claims for individuals in the current year based on the claims of others in the same group in the prior year. It is difficult to estimate, so it is often set equal to k3
4) k3 is a measure of how the claims of each individual are related to others within the same group. It is commonly set equal to 1%
Adjustments to the basic credibility formula for group medical insurance
1) Adjustment for employee turnover. Let p be the probability that an individual will stay in the group
z = [p*k1 + (n-p)*k2] / [1 + (n-1)*k3]
2) Adjustment for variance of members’ expected claims
a) Calculate an adjusted group size nprime = n*(mean^2)/(mean^2 + StdDev^2), where mean & StdDev are of the group’s age-sex factors
b) Then calculate the credibility by replacing n w/ nprime in the credibility formula
3) Adjustment for specific stop loss
a) Because of very high deductibles, k2 < k3
b) Let s = k2 / k3. Therefore, k2 in the credibility formula can be replaced w/ s*k3
c) s can be estimated as 100% - 10%*(attachemnt point / 50,000)
4) Formula for fractional years of experience, where f is the fraction of the year of experience and z1 equals z from the basic credibility formula:
zf = [f*z1] / [1 + (f-1)*z1]
5) Multiple years of experience - calculate a separate credibility percentage to apply to each year. The most recent year is year 1, the year before that is year 2, etc
a) For year 1, calculate z using the basic credibility formula
b) For years 2+: zt = (1 - SUM zr) * zt-1
Major physician remuneration models in Canada
These are the major ways in which physicians are paid for their services
1) Traditional FFS - the physician bills for each service provided. Each province establishes a schedule of benefits that lists the fees paid for the difference services
2) Enhanced FFS - most provinces offer bonuses and fee schedule enhancements for family physicians and some specialties. Enhancements include bonuses for chronic disease management, additional funding for treating special-needs populations, and an increase in fees in qualifying rural or remote areas
3) Alternative payment plans (APPs) - these methods are an alternative to traditional FFS payment
4) Salary - a regular payment which is specified in an employement contract. Remuneration is often done through “time-based payments”, such as annual salaries or hourly rates
Types of payments included in APPs
1) Fees for clinical services
2) Capitation
3) Time-based payments, such as hourly pay
4) Rewards for participation in specific clinical initiatives
5) Bonuses for achieving specific quality targets
6) Remuneration for administrative duties and costs
7) Financial contributions for medical information technology
8) Addtional payment types for academic physicians:
a) compensation for teaching
b) research funding
c) stipends for administrative duties
d) partial compensation or subsidies for staff, facilities, and equipment
Components that may be included in APPs that target primary care physicians
1) Patient-enrolled models - these models require providers to formally enroll patients in their practices and register their enrollment with the ministry of health (MoH)
2) Rostering - this is the process of enrolling patients and registering that enrollment with the MoH. Many APPs pay the physician a fee for the administrative work of rostering patients
3) FFS billing - payment is made based on the provincial fee schedule. An APP may incorporate a bonus top-up, such as an extra percentage of the fee for enrolled patients
4) Capitation payments - the physician receives a fixed payment for the comprehensive annual care of a rostered patient. The payment varies by age and gender
5) Shadow FFS billing - physicians who participate in a capitation APP must still submit FFS invoices for services provided to rostered patients. The MoH needs this information for evaluating patient access and utilization
6) Preventative care bonus - annual bonuses are offered for physicians who meet certain percentage targets for preventative health care
7) Comprehensive care management fee - a payment for the ongoing administrative work and upkeep that comprehensive family doctors do in addition to seeing their patients
8) Chronic disease management bonuses - an annual bonus for managing chronic diseases
9) New patient incentives - fixed bonuses to physicians who accept “orphaned” patients (i.e., those who do not have a family doctor) as new patients into their practices
10) Administrative fees - per-patient fees paid annually to help cover some of the administrative costs of meeting the accountability criteria of capitation APPs
11) Sessional fees - fees based on an hourly rate and paid for specific services. Many emergency departments offer physicians a guarantee sessional fee for working as the doctor on duty
12) Block funding - a guaranteed payment to provide medical services for patients in a specific location (typically a rural or remote area) for a defined interval of time
Description of Assuris in Canada
1) The goal os Assuris is to minimize the loss of benefits for members of life insurers that become insolvent
2) In general, insurance products are guaranteed up to 85% of benefit value, and deposit type products are guaranteed up to $100k. Assuris guarantees apply to life insurance, critical illness, health expense, disability income, LTC, annuities, and segregated funds
3) Assuris funding is provided by the member companies via a risk sensitive assessment system based on required capital. Every life insurer in Canada is legally required to be a member of Assuris
4) Assuris works closely with regulators to monitor insurers who are at risk of becoming insolvent
