To Find a Doctor, Mine the Data
Now that millions of consumers are surfing the Web to research their own medical symptoms, many are taking the next step: comparison-shopping online for hospitals and doctors.
This concept scares me. Data mining always is fraught with hazards. The mistakes made with outstanding clinical epidemiologic studies teach us the challenges of data mining.
We’ve got terrific measurements information in some areas, but in other areas we don’t have good measurements at all,” said Dr. Carolyn Clancy, director of the federal Agency for Healthcare Research and Quality, which is working to standardize the way health care data are reported.
At this point, there is much more quality-of-care information available about hospitals than about individual doctors, except in nine states including Florida, Pennsylvania and Wisconsin that make statistics available on the numbers of procedures that surgeons perform.
But even when we do have good data, we rarely have excellent data.
Specialists say that, so far, the sources of information are far from perfect. Data based on medical claims payments can be particularly sketchy and unreliable, said R. Heather Palmer, a professor of health policy and management at the Harvard School of Public Health. “The hope,” she said, “is that as we move toward electronic medical records, we will get data with more clinical detail.”
The data collecting companies make no claim of perfection. “In the absence of perfect information, we help equip people so they know what questions to ask to be smarter consumers of health care,” said Ann Mond Johnson, chief executive of Subimo, a five-year-old company based in Chicago.
Most specialists agree that improvements depend on achieving a standardized, computerized approach to collecting and assembling medical data at all levels, from big hospitals to small doctors’ offices – a long-range goal of the Bush administration.
And even when we do have standardized electronic medical records data, certain aspects of medicine have enough complexity that data mining may yield inaccurate results. There is an old acronym in computing – GIGO (garbage in, garbage out). Having spent much time this past year studying EMRs, I believe that they will do a good job on the common tasks. Data mining can tell you if a physician adheres to agreed upon quality indicators. Although unless we carefully design our data mining even these calculations could mislead.
Problems:
Example – 66 year old man is admitted with dyspnea. He has a history of a previous myocardial infarction, and has had a coronary stent placed. He has a long history of smoking but no known chronic lung disease. He is markedly overweight, and his wife states that he does snore loudly.
Physical exam suggests congestive heart failure, as does his chest X-ray. However, his BNP level is only modestly elevated. On the second day of hospitalization his echocardiogram shows normal left ventricular function, but markedly increased pulmonary artery pressures.
We take another look at his CXR and obtain a chest CT. The chest CT reveals that what we initially saw as pulmonary edema, was really lymphatic spread of lung cancer.
We treated him perfectly for the wrong disease initially. We did do the right confirmatory tests, and came to the right diagnosis.
If this patient had seen his physician as an outpatient, and received a diagnosis of CHF – we would look at whether he received an ACE inhibitor and a beta blocker. Yet, the key for this man was to make a correct diagnosis.
I value the clinical acumen which leads to finding the right diagnosis as a very valuable indicator of the excellent physician. Current quality scores do not include this feature
Quality is the current buzzword. Like all trends we skeptics must warn the profession and society of the potential externalities. Focusing too much on quality scores may encourage physicians to worry too much about quality check lists, and not enough about thinking through the entire clinical presentation.
Perhaps I am just an old codger. Perhaps I worry too much about scorecards. Perhaps I am right.
Related posts:
Related posts brought to you by Yet Another Related Posts Plugin.
6 Responses to Choosing your health care provider and hospital
David Toub
September 23rd, 2005 at 8:26 am
I found this article interesting. I was medical director for clinical informatics at a (now deceased) startup called DoctorQuality that developed systems to rate physicians and hospitals and provide this information to consumers. I think we were way ahead of our time, and there were many issues as you correctly point out with presenting this information to consumers. First, it’s hard enough to explain risk-adjusted mortality rates to our colleagues, let alone consumers. Also, there are many data streams that relate to quality: pt satisfaction, external physician recognition, raw data such as mortality and morbidity rates, etc. If one hospital has a risk-adjusted mortality rate that is 1% and another hospital’s rate is 1.1% but the second hospital performed better than expected and the first one performed worse than expected based on their patient mix, which hospital is better? Does this even have much meaning when the numbers are so close?
One of the biggest issues also relates to physician data. A lot of demographic data on the Web is garbage: one database provider we used had dead doctors, retired doctors, wrong specialties, etc. One very talented doctor was even listed under no less than 11 specialties: the world’s first gynecologist/urologist/chiropractor/orthopedic surgeon…you get the drift. Also, as relates to performance data, it is readily available (at least for MEDPAR) on the Web for most hospitals. However, performance data is very limited for physicians; some scattered data on CABG outcomes and a few other procedures is pretty much all that is out there.
You are correct that there is more than just numbers and quality check lists to rendering good medicine. However, it’s a start, and better than not focusing on quality. There are many things regarding patient safety that need to be improved, and if we don’t start doing it ourselves, it will be imposed upon us. Encouraging a quality-focused, patient-centered approach is better than doing nothing in this regard.
jb
September 23rd, 2005 at 6:53 pm
Another reason that this will be counterproductive is that it will exacerbate the already existing cherry picking that goes on, chiefly as a result of our legal and bureaucratic environment. Many docs already try to avoid keeping in their practice patients who are viewed as potentially litigious, or who have complex medical conditions that no amount of coding can result in a reasonable reimbursement for the time and energy spent in caring for. Add to this scenario a “report card” where our success rates will be publicly reported. No one will believe protestations that outcomes will be “curved” to account for degree of difficulty. Those of us who take care of the sick ones know that it is folly to try to quantitate degree of sickness for all but the simplest situations. There are reports out of NY state that cardiologists there, who already receive report cards of this type, are taking the patients deemed likely to do well for invasive procedures and treating the sicker ones, who have the most to benefit, with medication only. This is definitely a GIGO situation. Some things just can not be quantitated no matter how hard you try. They can and will come up with a number, but it will be meaningless, and docs will run their practices to come up with a good report card, to the detriment of some of their patients.
Meyerer Miller
September 29th, 2005 at 6:49 pm
I agree that numbers cannot tell the whole story on what hospital or doctor provides the best care, and I certainly wouldn’t trust just a list of stats. But from the patient perspective, how are you supposed to choose? The information that is out there is really rather skimpy.
You can ask a doctor about his/her qualifications, but you’ll have to pay for the appointment for each doctor you want to check out. I don’t begrudge them the fee, but I can only afford to check out so many doctors. The licensing and professional organizations don’t seem to have a lot of doctors listed or it’s not accessible to the patient, only other doctors. The pay sites have even fewer doctors in their databases.
As for hospitals, you can visit a hospital’s website, but of course that is going to paint the the best possible picture. If you’re lucky, there’s a hospital in your area that’s on some publication’s list of top hospitals, but if you’re in a rural area, then good luck figuring it out.
I’m having major abdominal surgery next month. How did I pick my doctor? A recommendation from a friend who used to work at USC medical school to start with. Then, I went to my appointment armed with questions I developed after reading the experiences of others on a yahoo news group and reading about my condition. The doctor answered the questions consistent with my own research and she listened attentively and explained well. I found very little other info on her through professional organizations besides published papers.
The hospital, I don’t get to choose. She only operates out of one for her private patients.
Don’t get me wrong, I’m confident in my doctor and the hospital has a good reputation, but most of my decision ends up being based on word-of-mouth and a 20-minute interview with the doctor. Some actual stats, even if somewhat skewed, would have been nice.
kate q
September 30th, 2005 at 2:01 pm
I also would like to know how we’re supposed to choose.
I almost never see doctors, largely because I don’t *have* one. Is this earache an infection? Is that sore throat strep? I don’t know; maybe if I ignore it, it will go away, and I won’t have to figure out where to find a list of doctors my insurance will allow, then try to decide based on two facts and an address which doctor might cover whatever it is I need, be accepting new patients, and be an otherwise good choice. I’m a shy introvert; I’d rather live with the earache than start making phone calls or, god forbid, actually go see people unnecessarily.
I suppose amateur hypochondriacs and the chronically/seriously sick know how to navigate the health care system, but for the rest of us…? Any doctor with a nice informative website has a huge advantage, as far as I’m concerned.
There’s a reason people tend to use the emergency room as the family doctor: it’s much more straightforward. I know hospitals hate that, but instead of trying to convince the consumer that he’s wrong to want what he wants, why not make it just as straightforward to see a more appropriate doctor…?
(I’ve noticed that dentists have started sending out advertising postcards explaining what they do and inviting new business. I like that very much.)
jb
October 2nd, 2005 at 11:12 am
MM and KateQ-
Thanks for your responses. I believe that we are in a situation in which perfect is the enemy of good. If you are looking at numbers, you may find the best doctor, but you also may find the doctor who is best at managing his numbers by treating only the easy patients. The doctor you want may be the one whose caseload includes the difficult cases, which will result in worse numbers, but it is the hard cases that we learn and improve from. As to the doctor who has a nice informative website, all that proves is that he, or someone he knows/hires, has good web skills. The doctor with a nice website is no more likely to be a superior physician than drinking Bud Lite is to get me a date with the Swedish bikini team. It’s all marketing.
There is no foolproof way to good medical care. The best way is to establish a trusting relationship with a primary physician (internist or family practice doc) when you are healthy, or in stable health. That may take some time and money, but it will be worth it. That way, when you need a referral to a specialist for surgery or other episodic care, you will have the benefit of her experience with other patients who have had simiar problems. It’s not foolproof, but it’s a lot better than waiting for reliable numerical ratings, which is a concept for the future, and always will be. If your problem is serious or urgent, your family doc will call the specialist and make sure he understands your situation. It will get you better care and more attention.
BTW, you can call any doctor’s office and get basic info from the receptionist for free, such as type of training, where she went to med school and residency, board certification, years in practice, etc. If the office receptionist will not give that over the phone, that’s a doc to avoid.
Most hospitals have websites that feature their docs, and give the basic info you will need. I would be surprised if your local hospital does not do this. Getting a referral from your family doc is a much better way to go.
Blue Cross of California
November 30th, 2005 at 4:33 pm
Great blog and information on choosing the perfect health provider for your needs. I hope we can work to build a better health care system. Health insurance is a major aspect to many.