Summary:  A startling study published in the New England Journal of Medicine reveals that just as doctors in some towns are more aggressive in treating their patients, physicians in some places are more likely to send patients for tests, and to subspecialists. As a result, their patients are diagnosed with more diseases. Thus, if a Medicare patient who was living in Phoenix (and feeling perfectly healthy), moves to Miami, he may suddenly discover that he suffers from two or three chronic conditions.  
This creates a problem, not only for the patient (am I really sicker?), but for health care reformers who hope to pay hospitals and doctors more for higher quality care. To do that they have to adjust for risk: Providers caring for sicker patients should still be eligible for bonuses, even if their outcomes  aren’t as stellar as the results achieved by hospitals that treat more robust patients. But if the majority of Medicare patients in Miami have been diagnosed with one disease or another, does that mean that Medicare should pay Miami’s hospitals more than hospitals in Phoenix because their patients appear more vulnerable– at least on paper? The researchers conclude: “risk-adjustment is going to be tougher than we thought.”
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

At a dinner party in Manhattan, someone mentions the problems he has been having with his sinuses, and his doctor’s diagnosis. Since everyone at the table is over 40, his comment quickly leads to a lively discussion of back pain, rotator cuffs, high blood pressure, skin cancer, and diverticulitis. It seems that everyone in the room has been diagnosed with something. Finally, someone asks “Are we really that old? Can’t we talk about something else?” Everyone laughs and the conversation turns to politics.

I couldn’t help but recall that evening while reading an article in the May 12 New England Journal of Medicine (NEJM) titled “Regional Variations in Diagnostic Practices”  written by a team of investigators at Dartmouth.

Earlier work done by researchers at Dartmouth has shown that patients in some regions receive moretreatment than others. This newest study, written by Yunje Song, senior author Elliott Fisher, and colleagues, goes further, to suggest that patients in places such as Miami, New York or McAllen, Texas are more likely to be diagnosed in the first place. “Their doctors order more tests and refer more patients to sub-specialists than doctors in Atlanta, Phoenix or Jackson, Mississippi,” explains Jonathan Skinner, one of the co-authors, “and so they discover more disease.”

At first glance, this might suggest that physicians in Atlanta should devote more time to testing, so that their patients can enjoy the benefits of early detection. But when the Dartmouth team looked at survival rates, they found that in regions where there is more testing and more diagnosis, patients fare no better. This suggests that our enthusiasm for diagnostic imaging has led to “early detection” of what some doctors call “pseudo disease” –small breast lesions that will disappear, without every leading to breast cancer, or lower back pain that, with time, will vanish on its own, without any  need for surgery.

As Dartmouth’s Dr. Gilbert Welch points out, the explosion of medical testing has created an “epidemic of diagnosis:” “Exactly what are we doing to our children,” he asks, “when 40 percent of summer campers are on one or more chronic prescription medications?”

Given Regional Variation in Diagnosis, How Can Reformers Adjust For Risk?

Nevertheless, once the diagnosis is recorded in a patient’s medical record, it becomes a fact. The New Yorker who has been diagnosed with early stage prostate cancer now appears sicker than his cousin in Atlanta, who has never had a PSA test.  The New Yorker may never experience symptoms, but he is now listed as a cancer patient.

This poses an enormous problem for health care reformers who hope to adjust for risk when deciding how much to pay hospitals and doctors. Under the new legislation, Medicare plans to begin rewarding providers for the quality of the care they offer—paying more for better outcomes. But Medicare recognizes that some providers care for sicker patients, and in those cases, outcomes won’t be as good. Obviously, hospitals shouldn’t be penalized because they treat a  more vulnerable population. They, too, should be eligible for bonuses.

Reformers who plan to compare the effectiveness of various products and services, also need to adjust for risk. If a patient suffers from two or three diseases, this could explain why he is not responding to a new cancer drug.

Clearly, regional variations in diagnoses could wreak havoc with risk-adjustment. As the authors of the study note: “Risk adjustment is only as good as the information on which it is based. Current risk-adjustment methods depend on the diagnoses that are recorded by physicians in medical records  . . . or are coded by medical-records personnel and billing staff in hospital discharge abstracts and physician claims .  . .”  “If physicians have substantial and systematic differences in their diagnostic practices that are unrelated to the underlying health of their patients but are related to institutional or regional practice patterns, biases in risk adjustment will result.”                        

Testing the Hypothesis

To test their theory, investigators began by comparing trends in laboratory testing, imaging and diagnosis in different “hospital referral regions” (HRRs) across the nation. They grouped these regions into five quintiles based on the intensity of hospital and physician services in the HRRs, ranging from the 1st quintile (least aggressive) to the 5th (most aggressive).

They then identified 5,589,741 Medicare beneficiaries who received Medicare from 2001 to 2003, zeroing in on the 255,264 individuals in that group who moved sometime during those three years. Some migrated to a city in the 5th quintile where physicians do more tests; others moved to a place where diagnostic imaging is not as popular. The researchers asked two questions:  Would seniors be diagnosed with more diseases simply because their address had changed? What would happen to those who didn’t move?  (Researchers excluded beneficiaries who died during those three years because, wherever they moved, as their health deteriorated, they would have been likely to see more doctors and undergo more testing.)                                             

The Results of the Study

As the table below reveals, a change of place made all of the difference. 

Percent Increase in the Number of Major Diagnoses, Diagnostic Tests, and Imaging Services among Medicare Beneficiaries, According to the Intensity of Practice in the Hospital Referral Region (HRR) in Which They Lived.

The investigators had expected that, as Medicare patients aged, they would be diagnosed with more major diseases. And indeed, they were. The top line of the table shows that those who lived in a town like Phoenix (in the 1st quintile) and stayed put, nevertheless watched their diagnosis rate rise by 65 percent over the course of three years.

But what is striking is what happened to those who moved from a 1st quintile region to a 5th quintile city such as Miami. Over the same span, their diagnosis rate jumped by more than 100%.  (See the third line of the table). Suddenly they appeared much sicker than friends who stayed at home–at least according to their medical records. As the third and fourth columns show, their doctors ordered many more tests. Seek and ye shall find. 

“It’s not hard to understand how this happens,” says Skinner. “As Jack Wennberg [the father of the Dartmouth research] has said: ‘Older people are vast reservoirs of disease.’ It’s relatively easy to diagnose any Medicare patient if you look hard enough.”

What about patients who moved to an area where care is more conservative? The bottom two lines on the table show that those who moved from a 5th quintile town such as  McAllen, Texas to a town more like Atlanta or Jacksonville, Mississippi (both in the 1st quintile) appear healthier (at least on paper), after making the move. If they had stayed in a 5th quintile region, their diagnosis  rate would have risen by 55% over three years; those who moved to  a place where doctors are less assiduous  saw diagnosis rise by only about 43%.

Still, many Americans would find it hard to believe that Atlanta physicians aren’t short-changing their patients. Hundreds of millions of dollars are spent every year on hospital ads indoctrinating us to believe that more lab tests and earlier diagnosis will lengthen our lives.

And, in fact, sometimes this is true. But when Dartmouth’s researchers compared 1-year and 3-year rates of death after Medicare patients moved– adjusting for differences in age, sex, and race– they found no evidence of a survival benefit among those who moved to a region where doctors find more disease.

The only difference is that when the discussion turns to health problems at a dinner party, New Yorkers have more to brag about— the number of doctors they see, the number of diseases diagnosed, the number of treatments, the cost of all of the above…

More work needs to be done examining when and where more diagnostic imaging and lab tests will actually save lives. As the Dartmouth team acknowledges: “Although our study did not show a significantly higher rate of survival among beneficiaries who moved to regions with higher-intensity practices, this result should not be interpreted as implying that greater diagnostic intensity offers no benefits. Rather, it underscores the need for research to determine the specific clinical settings in which greater diagnostic intensity does — or does not — confer a benefit.”                

Implications for Reform

If, under reform, we try to pay for quality, and “payments to health care providers are based on risk-adjusted costs and risk-adjusted outcomes, those who diagnose more will have a double advantage,” Skinner observes. “Every time a hospital enters new diagnoses for its patients, reimbursements will go up because” Medicare will assume that it is more expensive for a hospital to care for patients suffering from more diseases. Meanwhile, in places such as Miami, doctors’ risk-adjusted outcomes will look better; even if their outcomes are only average, because it will appear that they are doing better than other doctors while treating much more vulnerable patients. “The temptation to diagnose more and more disease will be irresistible,” Skinner concludes.

The Dartmouth team estimates that variations in diagnoses could mean that “reimbursement rates would be as much as 19% higher in the high-intensity regions solely because of bias related to diagnostic practice. ..  .Alternatively,” in regions where doctors are not so quick to order a Cat-scan, “risk-adjustment models could fail to account for the difficulty of caring for truly high-risk patients or those whose care is made more difficult owing to challenges such as language barriers, poor health literacy, or lack of social support, encouraging some providers to avoid or stop providing care for such patients.”

“Risk-adjustment is going to be more difficult than we thought,” Skinner admits.  “The challenges are going to be to measure risk in more objective ways by focusing on more detailed measures.”

The Dartmouth investigators end their study by focusing on the problems that geographic variations in diagnoses pose for reformers trying to measure quality, and they offer some solutions. “These challenges could become more manageable as comprehensive electronic health records are implemented,” they write. They go on to suggests that when these electronic records are used to identify sicker patients, they should include not just clinical diagnoses, but “nonclinical factors that may predict a patient's lack of adherence to clinical advice (e.g., homelessness or poverty).”

This makes sense. When allocating bonuses, Medicare should recognize that low-income patients are harder to treat, not only because they suffer from more chronic diseases, but because a lack of education combined with a high-stress life makes it less likely that they will comply with their doctors’ recommendations. That said, providers who care for poorer patients should receive extra compensation only if treatment leads to some improvement in heath and survival rates, not simply because the physician correctly diagnoses more diseases.

Too often, a Medicaid patient’s health problems are recognized, but he or she still doesn’t receive recommended care. For example, the California Breast Cancer Research Program has found that while poorer women are more likely to be diagnosed with  breast cancer,  they are less likely to receive standard treatment. This is why, as I noted in a recent post, poor women are about twice as likely to die of breast cancer as wealthier women. They receive less medical care, and less sophisticated care, explains Otis Brawley, MD, associate director of the Winship Cancer Institute at Emory University. The same pattern holds true for other diseases. In general, even when they are on Medicaid, poorer patients face more hurdles in getting access to the right care at the right time.

The Dartmouth team also recommends that when adjusting for risk, reformers should focus on “clinical data that are less subject to bias that is due to differences in diagnostic practices.” Rather than simply assuming that all patients diagnosed with cancer are “at risk,” they should use “data that includes stage and grade of
the cancer.” For example, the vast majority of men who are told that they have early-stage prostate cancer will never experience symptoms.They are far healthier than patients diagnosed with a more aggressive cancer. Similarly, the researchers suggest, that “in the case of those with congestive heart failure, researchers should pay attention to ejection fraction” (the percentage of blood pumped out of the heart with each beat) which is an important measure of how serious the problem is.

Finally, they note, it’s always worth listening to the patient: “measures of health risks reported by patients (e.g., smoking and exercise patterns) and functional status (physical, social, and role function) could be incorporated in risk-adjustment models.”

I am glad that researchers at Dartmouth are taking a hard and honest look at the challenges that we will face now, more than three years before full-scale reform becomes a reality. As I have said in the past, mending our broken health care system will require constant “re-visioning,” as we learn more about how to assure that we are getting value for our health care dollars. My guess is that it will take a decade to restructure the system, changing how we pay for care, what we pay for, and how care is delivered.

But I also believe that we can “break the curve” of health care inflation in the next three or four years by beginning to squeeze the hazardous waste out of the system. As I have suggested in the past, much of that waste is easy to spot. Some of it takes the form of over-testing which then leads to an “epidemic of diagnosis.” But we cannot simply assume that all providers in a region where care tends to be more aggressive are diagnosing “pseudo-disease.” The Dartmouth team cautions that in order to assess benefit to the patient, Medicare will have to hone in on individual hospitals and doctor/hospital networks where patients seem to be undergoing an unusually high number of tests and procedures. Only then can Medicare begin to use financial carrots and sticks, on a hospital-by-hospital basis to encourage higher quality, less costly care. Take a look at the proposals for pilot projects in the reform legislation, and you will discover that this is precisely what Medicare plans to do.