"Mortality differences between insured and uninsured accident victims were larger at teaching hospitals, perhaps because they have more resources and 'provide costly, life-saving care to the insured.' Hospitals that received larger fractions of their budgets from public sources also showed treatment differences and slightly larger mortality differences than the full sample."
As a group, people covered by health insurance generally are thought to spend about 40 percent more on health care than those who are not insured. Although this difference is much discussed, its effect on individual health remains controversial. Some argue that the additional money spent by those who have insurance does little to improve their overall health. Others point out that people who think they will be spending large amounts on medical care in the near future are more likely to buy health insurance. In this case, the insured spend more simply because they need more care.
In Health Insurance, Treatment and Outcomes: Using Auto Accidents as Health Shocks (NBER Working Paper No. 11099), author Joseph Doyle analyzes the effect of insurance status on health care spending by looking at the amounts spent on inpatient treatment for victims of severe automobile accidents in Wisconsin between 1992 and 1997. Using an unusually rich dataset linking police reports to inpatient hospital records, he finds that the spending differences are smaller than generally thought. People generally do not choose to be in severe auto accidents, so the estimate of only a 20 percent spending difference between the insured and the uninsured suggests that earlier estimates of the effect of health insurance on health spending indeed may have been inflated (by people who bought health insurance only if they knew they would need it.)
Facility charges for the uninsured were about 22 percent lower than for the insured, Doyle finds, and the uninsured received 20 percent fewer days of care. While the uninsured received more sutures and "alcohol and drug rehabilitation and detoxification," they received less of almost everything else: fewer spinal fusions, skeletal traction, fewer operations on organs like the brain, kidney, bladder, and large intestine, and less plastic surgery and skeletal traction.
Furthermore, the additional spending on insured victims reduces mortality. The author finds that "A ten percent increase in facility charges is associated with a reduction in the mortality rate of roughly 1.1 percentage points and the mortality rate is about 1.5 percentage points higher for the uninsured in a sample with a mean mortality of 3.8 percent."
But the type of hospital also matters, according to Doyle. Compared to other non-profit hospitals, religiously affiliated hospitals provide the insured and the uninsured with more equal treatment. And, there was no estimated difference in mortality in these hospitals. Mortality differences between insured and uninsured accident victims were larger at teaching hospitals, perhaps because they have more resources and "provide costly, life-saving care to the insured." Hospitals that received larger fractions of their budgets from public sources also showed treatment differences and slightly larger mortality differences than the full sample.
Lengths of stay and facility charges for the uninsured were about half those of Medicaid patients, and their mortality was 4.7 percent higher. It is possible that the higher charges reflect the fact that Medicaid recipients as a group are in poorer health than the uninsured and require more treatment to achieve the same result. Another explanation is that Medicaid reimbursement rules encourage more costly treatment. During the period under study, private insurers generally paid hospitals a flat fee for each case. Wisconsin Medicaid did not do this. It reimbursed hospitals additional amounts for their most costly "outlier" cases. This may have given hospitals an incentive to provide more costly treatment for Medicaid patients.
The finding that uninsured accident victims are 1.5 percentage points more likely to die than privately insured ones implies a 0.45 percentage point increase in the lifetime risk that an uninsured person will die in a severe automobile accident. The author estimates that "this is over five times the lifetime risk of dying due to complications or misadventures during surgery." But, as he points out, the money saved by not buying health insurance roughly compensates for the increased risk of death. If the value of a life isis $3,000,000, then reducing the lifetime risk of dying in an auto accident would be worth about $300 a year. At the time the data were collected, a health insurance policy with a $1,500 deductible cost about $300 for a 23 year-old-man. The author concludes that "the magnitudes of the benefits and costs of catastrophic insurance are roughly similar and that the differences in treatment and mortality are not unreasonably large.
-- Linda Gorman