Electronic medical records, hospitals and doctors - Comment on Clinical Information Technologies and Intpatient Outcomes"

Comment on “Clinical Information Technologies and Inpatient Outcomes,” Archives of Internal Medicine, “January 26, and Commonwealth Fund Response to Article

The Devil is in the Details.
Common Saying

I have been asked to comment on an article in the Archives of Internal Medicine supported by a grant from the Commonwealth Fund, and the Fund’s response to the article. I shall do so now.

The Study

Five Texas researchers, supported by a grant from the Commonwealth Fund, conducted a cross-sectional study of 72 acute general Texas hospitals located in 10 geographically dispersed metropolitan areas: Abilene, Austin, Dallas, El Paso, Houston, Laredo, Lubbock, McAllen, and San Antonio.

Researchers used a questionnaire called the Clinical Information Technology Assessment tool, which was sent to a randomized cross-section of Texas doctors who admitted to the hospitals. The questionnaire measures the hospital’s level of automation based on physician interactions with the hospitals information system.
Questions concern test results. automated notes and records, order entry, and clinical decision support. Each hospital was given a score based on physician responses to these questions. Thirty one (31) hospitals were stricken from the study when the researchers did not receive 5 randomly sampled physician responses. Thus, 41 hospitals were considered responders and 31 were not.

Next all hospitals were studied for inpatient mortality, complications, coasts and length of stay among patients older than 50 years admitted between December 1, 2005, and May 30, 2006 on outcomes for all patients, but specifically for those with myocardial infarction, heart failure, coronary bypass, and pneumonia. The hospitals were categorized by number of beds, urban vs. rural, teaching status, and census tract.

Synopsis

The study of hospitals found those with automated clinical information had fewer complications, lower mortality rates, and lower costs. For all medical conditions, those with high automation scores had a 15% decrease in fatal hospitalizations, a 9% decrease in odds of death for myocardial infarction, a 55% decrease in deaths from coronary bypass, and a 16% in complications for all patients.

Hospital Savings

The decrease in mean adjusted hospital savings per hospitalization was as follows.
• All patients, $538
• Patients with myocardial infarction, $225
• Patients with heart failure, $555
• Patients with coronary bypass,$1043
• Patients with pneumonia

Responders Compared to Nonresponders

As I pondered these impressive results, I asked myself; compared to what? The answer is that the researchers were comparing the responder hospitals, those with 5 or more responding physicians to the nonresponder hospitals, those with less than 5 responding physicians. Under the subheading Characteristics of Study Hospitals appeared this paragraph.

“We received 5 or more physician responses for 41 of 72 targeted hospitals (58% response rate; mean number of responses, 9. There were not significant differences in hospital ownership, operating margin, total margin, safety net status, or information operating expenses between responding and no responding hospital. However, responding hospitals tended to be larger (mean number of beds, 402 vs. 216 for no responders, and more academic (15%) vs 0 teaching hospitals.”

Skepticism

The difference in hospital size set off an alarm in my head. It is a devilish detail that ought to be expounded upon. An article in the Journal of Medical Systems “Organizational and Environmental Determinants of Hospital EMR Adoption (volume 31, October 2007) indicates four factors determining speed of adoption include environment uncertainty, type of system affiliation, size, and urban directness. Barriers to adoption include small size, distant from urban centers, lack of system affiliation, and low environmental incentives to change.

I am skeptical because of lack of mention of the socioeconomic status of the population served. In a state like Texas, which leads the nation in the percentage of uninsured (18.5%), with record numbers of newly arrived legal and illegal immigrants, a high level of unpaid hospital debts, a large Hispanic population, many of whom are health illiterate; drug-related violence along its borders, socioeconomic factors are important, and vary greatly from a border town like Laredo to an affluent university town like Austin. Surely these factors delay needed care, prompt greater ER admissions, and result in higher death rates, more complications, and higher costs to treat more complicated advanced cases.

I am reminded of a recent interview I conducted with Richard “Buz” Cooper, MD, a professor of medicine at the University of Pennsylvania and a principal in The Leonard Davis Health Care Economics Institute at Penn. Cooper is critical of a Dartmouth Institute study comparing the “efficiency” – i.e. costs, outcomes, and care variations – between such prestigious medical centers as the Mayo Clinic, The Cleveland Clinic, UCLA, Massachusetts General and Johns Hopkins.

Here are Cooper’s comments in response to my question of the fallacies of comparing variations among academic medical centers.

“A good example is the Dartmouth study of academic medical centers. You find that one group of academic hospitals provide more care than another group. The Dartmouth folks say that Mayo is more “efficient” in resources used per patient or in number of doctors devoted per unit of patient care than in LA, Philadelphia, Miami, Chicago, and New York City. But the so-called “inefficient” hospitals are all in dense urban centers, while “efficient” hospitals are all in smaller cities, often college towns liked Madison, Wisconsin or Columbia, Missouri, or in places like Rochester, Minnesota, where Mayo is located. Rochester is 90% Caucasian with low poverty. But in fact, Mayo is the most resource intensive center in the upper Midwest. Among peer institutions in similar socio-demographic environments, Mayo actually uses more resources. But you can’t compare Mayo to Los Angeles, where only 30% of the population is non-Hispanic white and where you have tremendous pockets of poverty. “

“The Dartmouth group doesn’t acknowledge the fact that there are enormous social differences between populations served by academic hospitals in various cities and even in the same city, where patients distribute in a non-random way. If you ignore these fundamental considerations, you can make the numbers fit the preconceived notion that there is more spending where there are more doctors and doctors cause the spending. You can “prove” that it’s the fault of specialists. But in the movie of “The King and I,” Yul Brunner, the King of Siam, tells his son to watch out for “people who try to prove that what is not so is so.” And they do. But that’s because they ignore the complexities of social structure and get it backwards. Doctors go to where they are needed, and the needs in urban centers are huge. “

The Current State of Hospital EMR Adoption

I am troubled that physician answers to a questionnaire like the Clinical Information Technology Assessment Tool are taken as Gospel that accurately reflects the state of EMR adoption by hospitals. It is widely known that only about 15% of hospitals have adopted EMR technologies compared to roughly the same percentage of doctors, that hospital EMRs rarely communicate with physician office EMRs, that the hospital cost of software for writing interactive interfaces to doctors who may have different EMRs is prohibitive, that despite Health and Human Services moves in 2006 to assure financing of 85% of the software for physician EMRs for up to $15,000 per physicians initially then up to $40,000 over 4 years, and the removal of such EMR barriers as Stark laws banning self referrals and the IRS kickback laws, that hospitals have been reluctant to finance physician EMRs, partly because of lack of enthusiasm on the part of doctors, who must still bear the burden of hardware and maintenance costs?

The Commonwealth Fund to the Archives Article

Whereas I am skeptical about the Texas multihospital study, the Commonwealth Fund regards it as a landmark study. In its comment on the study, the Fund says,”Hospitals using Health IT provide better care at lower costs. When physicians use health information technology to its full potential, the result is fewer deaths fewer complications, and lower costs, according to the first study to directly measure physicians’ use of IT in a hospital setting. In a Perspectives on Health Reform essay, David Blumenthal, MD, director of the Institute for Health Policy at the Massachusetts General Hospital, writhe that the federal government can help provides overcome the financial, technical, and logistical obstacles to adoption of Health IT. And Fund President Karen Davis calls health IT a critical comment in health reform, when implemented alongside payment reform and an overall commitment to performance improvement.” Or so it seems to people in the Policy World, who are sometimes removed as to what transpires on the ground.