Last week, we talked about dominated choices on Healthcare.gov as this is one of my primary research aims at the moment. A dominated choice occurs when there are two choices (A,B) with a set of shared relevant attributes. Choice A is never worse than Choice B on all of the relevant attributes and on at least one attribute, Choice A is objectively better than Choice B. Most of the time on Healthcare.gov, this does not occur. When it does not occur, there are perfectly plausible reasons for someone to have a personal preference profile to pick either A or B. However when there is a dominated choice, people should only pick the superior, dominant choice, A. However, people routinely pick the inferior option of B.
So what are dimensions of comparison that can be seen on Healthcare.gov?
Insurer level observables and unobservables
Some insurers have a good reputation. Some insurers make it obvious that their business strategy is to screw over the beneficiary. Some insurers are easy to deal with and others run you around. Some insurers like to be explicit about their controls on costs by explicit benefit design while others will run people through pre-authorization, denial and appeal processes. Brand level decisions are important as they contain a massive amount of very valuable information that is constant within a particular insurer and between insurers.
A single insurer can offer multiple networks. Some networks are big. Some networks are small. Some networks can be fairly big and low cost while others can be fairly small and high cost. Networks can be optimized for healthy individuals or they can be optimized for individuals with high levels of medical need and complexity where the value proposition is on risk adjustment. Networks tend to vary between insurers but this variance may or may not be important to individuals. Individuals care if their doc is in network, and if the local hospitals and potentially high prestige hospitals are in network. I would be fundamentally indifferent between two networks if they are virtually identical except for the inclusion/exclusion of one hospital and a dozen docs in the Tennessee River Valley Authority drainage basin.
There are four basic plan types; HMO, POS, EPO, PPO. These are answers to two sets of questions: is a gatekeeper required and are there out of network benefits available?
|Primary Care Provider Gatekeeper|
|Out of Network Benefit||Yes||POS||PPO|
There are gradients of restrictions between insurers. Not all HMOs are hyper restrictive. When I worked at UPMC Health Plan, the HMO was operationally not too different from an EPO. Some HMOs are strict about pre-authorization requirements.
What drugs are covered? At what cost-sharing? How does pre-authorization work?
How much do you pay out each month?
What does it cost to go to a doctor? What does it cost to have a knee replaced? What does it cost to have a baby? This will almost always vary between insurers. It will often vary within insurers. Given plausible differences in individual expected health outcomes, some people will find a low or no deductible plan with a huge out of pocket limit to be better than a medium deductible plan with the out of pocket limit being the deductible.
These are some of the obvious differences that people need to think through to assess if they are in a dominated plan. In most states, strictly and transparently dominated plans are unlikely to be common. There is enough cost-sharing design space to create minimally different plans that could conceivably give useful variation to someone. California, and a few other states, with standardized plan designs and active buyer marketplaces reduce the within and between insurer variations that allows us to see strictly and transparently dominated plan choice. In most other states, we are likely to only get highly probably fuzzy domination.