New Health Affairs Study Suggests That Private Insurers Control Health Care Spending Better Than Medicare

Dec. 7, 2010, Bethesda, MD -- Whether Medicare or private insurance pays for health care appears to make a significant difference in health spending variation, according to a new study published in the December Health Affairs. The study, a follow-up to a highly publicized 2009 New Yorker article by Atul Gawande, shows that in two Texas cities, sharp differences in Medicare's per-capita health care spending were significantly diminished when private insurance paid the bill.   

The study, by authors Luisa Franzini, associate professor at the University of Texas Health Science Center at Houston (UTHealth); Osama Mikhail of the UTHealth; and Jonathan Skinner of Dartmouth College, examined 2008 claims data from Blue Cross Blue Shield of Texas, the state's largest commercial health insurer.

For the under-65 population insured by Blue Cross, total spending per-member-year in McAllen, Texas, was 7 percent lower than in El Paso, Texas.  By contrast, Atul Gawande's 2009 New Yorker article, which used data from the Dartmouth Atlas of Health Care on variations in Medicare spending, showed that per capita spending in McAllen was 86 percent higher than in El Paso.

Although the new study cannot explain definitively why variations in health care spending drop off dramatically under private coverage, the authors suggest that mechanisms for utilization review and management used by private insurers could play a prominent role.

"For a number of reasons, insurers generally are reluctant to intrude on medical decision-making," says lead study author Franzini.  "But the fact that these utilization management mechanisms exist may prompt some physicians who might otherwise overuse certain services to exercise more restraint."

The widespread regional variations in Medicare spending across the country have been well documented by researchers at the Dartmouth Institute for Health Policy and Clinical Practice, with which Skinner is also affiliated.  In his New Yorker article, Gawande visited Texas to understand the variations documented in the Dartmouth Atlas.  He attributed the nearly twofold spending difference to a change in McAllen during the mid-1990s when health care providers adopted a more pronounced "entrepreneurial spirit" and a "culture of money."   The authors of the Health Affairs study sought to determine whether those same health care providers would demonstrate similar patterns of care for people under age 65. They were able to obtain data from Blue Cross Blue Shield of Texas, making this study among the first published studies to explore whether the variations seen in Medicare spending hold true for the under-65 patient population.

The results were surprising, the study authors say.  Medicare spending in McAllen was 63 percent higher than in El Paso for inpatient care, 32 percent higher for outpatient care and 65 percent higher for Part B professional services.  The largest difference was for home health care:  McAllen was 4.63 times higher than the average in El Paso and 7.14 times higher than the national average. On the other hand, hospice spending in McAllen was just a quarter of the level in El Paso and the national average.

Medicare enrollees in McAllen were far more likely to be admitted to the hospital and to die in the hospital than they were in El Paso. They were also much more likely to be seen near the end of life by more than 10 physicians

But for the under-65 population, spending on professional and inpatient services was similar in both cities, and spending for outpatient services in McAllen was 31 percent less. Use of medical services was also similar or somewhat lower in McAllen compared to El Paso.  Inpatient admissions in McAllen were 84 percent of admissions in El Paso; professional and outpatient services in McAllen were 94 percent and 72 percent respectively of those in El Paso.

Comparisons for other types of services told a similar story, with one key exception:  the 51 percent higher use of professional services and spending among those under 25 in McAllen. Inpatient admissions for this group were 89 percent higher in McAllen, and per-patient inpatient spending was 117 percent higher in McAllen, roughly the same difference as the overall difference in utilization of Medicare services and spending.

The Health Affairs researchers explored several potential explanations for their findings.  Neither differences in health care prices nor population disease burden between the two cities accounted for these spending variations.
The most probable explanation, they speculate, has to do with which payers are better at controlling costs around what study author Mikhail calls the "grey zone of treatment" - areas where legitimate medical judgments can be quite variable.  Medicare exercises very little utilization management, whereas private insurers, such as Blue Cross Blue Shield of Texas, can be much more assertive about controlling service use.

For example, Blue Cross Blue Shield of Texas encourages members with costly "big ticket" conditions to participate in a disease management program. Other mechanisms encourage providers to practice evidence-based, cost-effective care. In addition, all elective inpatient admissions must be preauthorized. Counseling before admission and after discharge are used to establish postoperative goals and identify discharge planning needs.

Mikhail notes that health care needs and service use generally increase as the population ages, thereby expanding "the grey zone of treatment and opening the door to greater variation."  That could explain why private insurance spending for people ages 50-64 in McAllen is so much higher than in El Paso.

"Variation in and of itself isn't necessarily bad," Franzini said.  "Some variations may be justified on medical grounds.  But large variations may be a signal that something else is going on as well."

She added that she and her colleagues are currently working with Blue Cross Blue Shield of Texas to examine spending variations across the state.

A full table of contents from this issue is available here: