How do doctors behave when some (but not all) of their patients are in managed care?
Today, I will review an interesting article by Glied and Zivin (2002) examining how physicians adjust their practice style when the HMO share of patients changes (note: the authors implicitly assume all HMOs pay physicians via capitation). First we will look at the three theoretical models the authors develop.
1. Excess capacity model. The excess capacity model assumes that HMOs seek to pay the marginal costs for any excess physician capacity. This is possible so long as HMOs make up a small percentage of the physicianâ€™s patient base. When the HMO makes up a large portion of the doctorâ€™s practice, the doctor will eventually refuse to accept HMO payments at marginal costs since, in the long run, this will not cover his/her fixed costs and the doctor will go out of business. Thus, practices with larger HMO shares will be paid higher reimbursement rates, which will lead to greater effort levels to managed care patients while leaving indemnity patients unaffected. Glied and Zivin predict that patterns of care between FFS and HMO patients will converge when HMO penetration increases within a practice.
2. Demand inducement model. If physicians are able to influence patient demand, the demand inducement will only occur for FFS patients. HMO patients cost the practice money and thus no demand inducement will occur for them. The authors thus posit that treatment intensity will increase for FFS patients as HMO penetration increases.
3. Fixed cost model. â€œâ€¦Fee-for-service patients receive variable cost effort that reflects marginal reimbursement rates for this effort. Any HMO patients seen in this practice, however, will receive lower variable cost effort to compensate for the â€œexcessiveâ€? fixed cost effort with which they are provided. Variable cost effort will be reduced to the point where total effort just satisfies the minimum service constraint. In a practice that treats mainly managed care patients, the optimal fixed cost investment will be smaller. HMO patients will then receive higher levels of variable cost effort than in FFS-dominated practices; again ensuring that total effort is large enough to satisfy the participation constraint.â€? In the empirical section of the paper, the fixed cost is the duration of the patient visit. The authors give convincing evidence that it is difficult for the doctor to schedule different visit durations for different patients based on their insurance holdings. The physician can, however, easily alter the visit intensity in terms of either effort, test ordering or medicine prescription. The authors predict that if the fixed cost model holds, treatment intensity of FFS and HMO patients will increase as HMO penetration increases with in a practice.
The data used is the 1993 to 1996 National Ambulatory Medical Care Survey.
1. Moving from 0% to 50% HMO penetration within a practice reduces the average visit duration (i.e.: fixed cost) by 1.3 minutes (7%).
2. Holding constant the share of HMO penetration, there is no change in visit duration based on the type of insurance the individual has. This lends some credibility that visit duration is a fixed cost.
3. The authors find more tests and medicine for all patients when there is a high HMO share.
4. Controlling for HMO penetration, FFS have more medicine prescribed and tests run than HMO patients.
5. The above evidence suggests that the fixed cost model most accurately describes reality.
Economic theory would predict a physician would provide different types of care based on the manner in which they patientâ€™s insurance compensate them. The authors conclude that physicians may be limited in their ability to adjust care on a patient by patient basis due to some real-world practicalities of operating an office (e.g.: visit duration is generally fixed). The authors do find that when doctors have a large HMO caseload, however, they change the way they do business altogether by reducing visit duration and increasing visit intensity.
Some problems with the study are the possibility of adverse selection. HMO patients may be healthier than FFS patients and this may be driving the difference. Glied and Zivin do try to control for this by incorporating diagnosis based variables into their regression. Also, the authors assume all HMOs pay physicians via capitation, but this is not true in reality. HMO compensation type often depends on the idiosyncrasies of each regional market. Physician selection may also be a problem; physician may choose to accept more or less HMO patients since they prefer to have a given practice style. Overall, however, the paper is clear and concise, and shows that measuring the impact of FFS versus capitation physician compensation is more complex in reality than economic theory would suggest.