CHICAGO – As physicians, clinical lab professionals and research scientists gathered at the American Association for Clinical Chemistry's (AACC; Washington) annual meeting to learn about new advances in diagnostics and disease management, a physician kicked off the meeting by challenging them to approach diagnostic thinking differently to avoid mistakes.
"If you dig very deep into the literature, about 80% to 85% of the time we make the correct diagnoses," said Jerome Groopman, MD, a physician at Harvard Medical School (Cambridge, Massachusetts) and a researcher delving into the basic mechanisms of cancer and AIDS. "If we had baseball hitters who were batting 800-850 it would be unimaginable. If 85% of the financial investment were winners, you'd make millions. But here it's different because the stakes are different ... people's lives. If 15% of all patients are not diagnosed correctly or have a significant delay of diagnosis, it causes harm to patients and that's a problem. It's a problem of a different dimension than what's in the Institute of Medicine (Washington) report of 1999, To Err is Human."
Groopman, a hematology and oncology specialist who has produced seminal work in the development of today's AIDS therapies such as AZT and 3TC, last year published a book, How Doctors Think, and on Sunday laid out his theories on how and why healthcare professionals fall into diagnostic traps.
He presented a series of examples such as this one: A woman in the midst of menopause visited four physicians saying she felt as though she was having periodic explosions in her body.
"When it's 4:45 p.m. on Friday afternoon and you've seen dozens of patients and in comes a jittery woman in the middle of menopause, all of your stereotypes and judgmental baggage goes boom," Groopman said. "Another wacko woman who can't handle menopause. That's the prejudice.
"Then she goes to a fifth doctor, an endocrinologist. She says 'I know I'm in menopause and it's no picnic. I'm a little kooky and high strung, but I'm telling you I feel explosions off and on at the weirdest times.' So instead of attributing this to a kooky woman with menopause, the doctor asked what else it could be."
It turns out the woman had a tumor that was occasionally rupturing.
"This fifth doctor probably saved her life," Groopman said.
Misdiagnoses aren't due to technical mistakes like getting the wrong lab values or X-rays, but because doctors fall into a thinking trap.
"The human mind is wired to take shortcuts," he said. "We overly weigh the first bit of data that we get. All clinicians at the bedside think probabilistically. But I don't know many people who walk around with calculators to calculate theories. It turns out the doctor's mind works like a magnet. It doesn't move in a linear way, strictly on history, exam, lab results and imaging – until the entire data set is in. Doctors think by pulling in bits of data from all directions like a magnet."
Another example: A doctor he knew was working on a Navajo reservation amidst an influenza epidemic.
"An elderly woman came in who didn't feel well," he said. "She's breathing quickly. Chest X-ray is clear. Maybe she's dehydrated? He draws blood and white count is a bit off, chloride and bicarb are funny. He assumes influenza/pneumonia."
As it turns out, the woman had aspirin toxicity.
"What was most available in his mind was flu because everyone had flu," Groopman said. "But the real punch line is that after residency this doctor has done a special fellowship in decision analysis. And he has written a paper on aspirin toxicity. In the real world, working under pressure, you take shortcuts. Sometimes they work, but maybe 15% of the time you fall into the trap."
The genesis of misdiagnosis starts with a theory called anchoring, based on the idea that healthcare professionals put too much emphasis on the first pieces of information received and grab a hold or anchor their thoughts. The next step down the misdiagnosis path is called availability, illustrated via the influenza example.
"Confirmation bias is very common," he said. "Once our mind has fixed on what we think is the answer, when we get contrary data, we rationalize it away and discount it. You look and cherry pick the data that confirm your hypothesis and you discount things that don't fit"
Finally, there's attribution, which is all about biases and cultural stereotypes, such as the woman with menopause and a tumor.
Despite the fact that clinical visit times with patients have collapsed to 12 minutes or less, Groopman said it's key for doctors to do a better job of listening, and try to avoid interrupting the patient within the first 18 seconds, which is typical. But beyond that, doctors – along with lab professionals – can incorporate a series of questions at the bedside or in the lab when making a diagnosis:
1. What else could it be? "This protects you against most of the sins of anchoring, availability and attribution. All of us want to grab ideas, make them fit and move on," he said.
2. Is there anything in the history of the patient, exam, lab results or procedures that have occurred that might contradict an initial conclusion? "This protects against confirmation bias," he said.
3. Could two things be going on at once? "Old theories say you should always look for one unifying hypothesis to say 'it's this.' That sounds great, but God didn't make us that way. Sometimes two things happen at once," he said.
Groopman pointed out that clinicians and lab professionals easily reach into computers and use information technology to aid with diagnostics, but "We have not reached into cognitive thinking about how we think, called metacognition. And I wanted to introduce to you to this whole new field of science. I was completely ignorant."
Big changes for diabetes diagnosis, monitoring
Major changes are afoot for both the diagnosis and monitoring of diabetes as research has shown there is a substantial need for improvements.
Experts in the glucose monitoring field sounded off at the meeting, with research showing that the accuracy of glucose meters – which can be wrong by up to 20% – is no longer acceptable because the imprecision can lead to dosing errors and the ensuing complications of hyperglycemia or hypoglycemia. Groups are pushing to up the standards for all meters to an error rate of 5% or less and for hospitals to stop using them, particularly in intensive care units (ICU).
Also, new guidelines coming from the National Academy of Clinical Biochemistry, a section of the AACC, recommend the use of hemoglobin A1c (HbA1c) to diagnose diabetes, replacing the glucose gold standard test.
"It's hard to know at what level of error patients are harmed by imprecise meters," David Bruns, MD, director of clinical chemistry and toxicology and associate director of Molecular Diagnostics Laboratory, University of Virginia (Charlottesville) told BB&T. "In the first study I published on this topic, all we could say was that as meter errors go up there are dose errors. But we didn't carry it further to see at what level it leads to hypoglycemia. To my knowledge, I'm not aware of other studies that have come out since our study."
That study, first examining the meter error rates via computer modeling, was published in Clinical Chemistry in 2001. Since then Bruns and his colleagues have continued to focus on meters, but he has been unable to attract funding to do further studies that compare the imprecision of meters.
Bruns led a session today at the AACC titled "How Accurate Are Glucose Meters? How Accurate Do They Need to Be?" His lecture is based on studies he and colleagues performed that reveal performance specifications established by the American Diabetes Association (ADA; Alexandria, Virginia) could lead to dosing errors in up to 23% of insulin doses, while devices that conform to Clinical Laboratory Improvement Amendments (CLIA) '88 standards are susceptible to errors in 16% to 45% of dosing events.
He re-emphasized the dire need for outcomes studies to determine if dosing errors pose serious threats to diabetics and whether current technical specifications are sufficient to match clinical need.
"Frankly, I've never found anybody interested in funding such a study," he said. "It's not like a drug study where you can spend tens of millions of dollars and, if a drug works, it can make billions of dollars. But the meter market is now in the billions of dollars and you'd think somebody would want to fund a trial."
Despite that, enough evidence exists to motivate the ADA and the FDA to take action.
The American Association of Clinical Endocrinologists (Jacksonville, Florida) has formally opened the matter with the FDA and the agency reported late last month that it is recommending reform of International Standard Organization (ISO) requirements for glucose meters.
Bruns said that he's particularly concerned with tight glucose control of critically ill patients in hospitals, as outlined in an upcoming paper he co-authored in Clinical Chemistry entitled "Tight glucose control in the intensive care unit: are glucose meters up to the task?" An earlier paper revealed such control protocol results in a 34% reduction in ICU mortality.
"How good are the meters that are on the market? Plus or minus 20% error rate was OK and it's been out there now for 20 years. Most of the newer ones hit 15%. Now the ADA wants 5%. The ISO wants something different," he said.
When asked if companies such as Medtronic Diabetes (Northridge, California) are putting the cart before the horse in developing a device that combines both a meter and a pump before meters have a better error rate, Bruns said accuracy can be improved with such devices because they automatically take glucose measurements more frequently than people do manually, reducing the chances of errors in dosing.
"If you have a system that can make measurements four times an hour then random error isn't such a big deal," he said. "In those cases, you can withstand the imprecision."
A1c – The new gold standard?
In addition to the evolution of glucose meters, another presentation at the AACC conference reported data from a new study from the International Expert Committee (IEC) that recommends the use of HbA1c rather than glucose for diagnosis of diabetes.
The test has been used for decades to measure long-term glycemic control in diabetics, but glucose was the primary test for diagnosing the disease.
The IEC included members appointed by the ADA, the European Association for the Study of Diabetes (D sseldorf, Germany) and the International Diabetes Federation (Brussels, Belgium).
"The report is being considered by the relevant committees of each of these groups," said David Sacks, MD, medical director of clinical chemistry, Brigham and Women's Hospital (Boston), who said, "Although the specific details of the recommendations to be proposed by these bodies have not been decided, it appears highly likely that all of the major clinical diabetes organizations will adopt HbA1c measurement for the diagnosis of diabetes."
Sacks authored an editorial to appear in the August issue Clinical Chemistry, titled "The Diagnosis of Diabetes is Changing: How Implementation of Hemoglobin A1c Will Impact Clinical Laboratories."
He moderated a symposium here this week called "Laboratory Medicine Practice Guidelines for Laboratory Analysis in Diabetes" where various speakers vetted criteria for screening and diagnosing diabetes and the ADA criteria for diagnosing diabetes in pregnant women.
Forensic toxicology, star status delays cause of death report
Can you bear another story about Michael Jackson? What about one with a forensic toxicology angle?
Millions of inquiring minds want to know about the pop star's cause of death. Toxicology experts at the meeting spelled out some of the reasons why it takes so long to determine the cause of death. With rapid point-of-care testing available, many people incorrectly assume that being a superstar is the primary delay in getting to the answer.
"In the hospital setting, a patient can come into the ER and present with specific signs and symptoms. Toxicology testing can be ordered and turned around quickly," said Barbarajean Magnani, MD, chief of clinical pathology and vice chair of pathology and laboratory medicine at Tufts Medical Center (Boston) and chair of the Toxicology Resource Committee at the College of American Pathologists (Northfield, Illinois).
"Point-of-care systems are usually screening tests that support the clinician's diagnosis," she said. "The type of testing in this case is forensic and is more involved."
Forensic toxicology is the post-mortem testing for lethal and/or foreign substances in body fluids and tissues, and it can be a very time-consuming process. Reasons for that slow process – which can take weeks – range from the extensive record keeping that is needed for forensic defensibility of the toxicology studies underlying a final report to the simple fact that some tests may be beyond the scope of expertise in the initial lab that received the specimens and those specimens then have to be referred to more specialized reference labs.
"Typically we screen the urine and heart blood for drugs," said Fred Apple, PhD, director and professor of lab medicine and pathology at the Hennepin County Medical Center (Minneapolis) as well as clinical toxicology consultant at the Hennepin Medical Examiner's Office. "We do 4,000 cases a year and 25% are toxicology related. We screen within a week or two and then talk about what was identified. Then the medical examiner meets with a toxicologist to decide what needs to be quantitated. We bring the specimens back to the lab. Usually we have a three-week turnaround. But we're a small operation. Eighty percent of the time we can use our technology in house – mass spectrometry, gas chromatography – to quantitate a drug and that may take three weeks."
Apple and Magnani led a presentation at the AACC designed to sneak a peak behind the curtains as medical examiners in California try to determine Jackson's death via toxicology studies.
Both pointed out that forensic testing demands a chain of custody so that for each step, from collection of the specimen to the final report, there is documentation of where the specimen has been and who handled it.
The precise process starts when toxicology reports are issued after a death investigation. But in certain circumstances, new information may become available and the toxicologist may have to test for additional drugs, or test other samples before the report is finalized.
"If it's a medical legal case that might go to court, they want the numbers," Magnani said. "Each individual drug may not have a high enough concentration, but many drugs are additive. Opioids, when mixed, can cause a higher concentration."
Apple said that if someone is found dead after filling four to five prescriptions – known therapeutic drugs that typically wouldn't have an adverse effect sometimes they mix drugs that might build up toxic levels. Or a person can take a therapeutic dosage but with another drug added in, it changes the effect and becomes toxic.
"We see a lot of polypharmacy in celebrities," Magnani said.
But why does the toxicology testing take so long? It's the details, quality control and reporting requirements. And that's assuming nothing runs afoul during testing in which case you have to start all over.
"The analytical part of the lab is not CSI [the TV drama]," she said. "Most of these labs are certified and you have run controls, it's not just about getting the numbers out. If one quality control is out on a run, you have to go back and start from ground zero."
Apple said the most progressive technologies currently in use by toxicologists are mass spectrometry and gas chromatography. And while these aren't new technologies, they have been refined and significantly reduced in size in the last several decades.
"We use bench-top instruments now," Apple said. "When I started off in 1980, our mass spectrometry machine was the size of room."
So given the current state of forensic toxicology testing, the fact that a final report has yet to be produced on Jackson's death is not surprising to these two toxicology experts.
"I suspect they've done their screening and they know what drugs are on board," Apple said. "They're probably sending it out to reference labs and making sure or maybe they got some report findings that were unexpected. Either way, they're being extra cautious because it's a high-profile case."
Ortho Clinical Diagnostics' 24/7 monitoring predicts lab problems
It's like a 911 call center on steroids for the diagnostics industry. But rather than responding to emergencies, it continuously tracks and predicts problems with lab instruments before they occur.
Ortho Clinical Diagnostics (OCD; Raritan, New Jersey) debuted its new Remote Monitoring Center (RMC; Rochester, New York) at AACC.
"Lab downtime is a very significant issue. Even if a lab is down just 1% of the time in a year, that could be hundreds of hours. I wouldn't want my family member to be unable to get appropriate test results in a timely manner," Troy Taylor, worldwide VP, Customer Technical Service at OCD told BB&T.
That and the cost of dispatching service engineers onsite were the genesis of the RMC. Onsite visits to remove a jammed slide or other problems great and small can be an expensive way to support technology. Since the RMC went live in February, about 1,600 instruments have been connected worldwide.
"E-connectivity for service and support has been around for a while and it's a nice engineering tool," Taylor said. "But the RMC automates the process and provides a predictive and proactive monitoring service."
As part of a demonstration of the RMC in the conference exhibition hall, computers showed how a two-way broadband Internet connection allows OCD to automatically monitor diagnostics tools and alert OCD staff if it appears something isn't right, up to a month before the device might actually break down.
But it's not all computer oriented or automated. A roomful of people operate the RMC fulltime in what is the first command center in Rochester. Other centers are set to open in Strasbourg, France and at an as-yet undetermined site somewhere in the Asia-Pacific region.
"We had all of this great data via e-connectivity already. But rather than being reactive to instrumentation, what if we could look ahead to alert labs earlier that something might go wrong or to provide reminders for service and maintenance?" Taylor said.
The RMC service is provided to OCD customers at no added cost and includes:
• Automatic two-way data exchange provides the ability to automatically send and receive data securely from the technical support team at the RMC.
• Assay and equipment predictive alerts are automatically sent to the technical support team when specific events or statistical trends have been observed.
• Remote diagnostic operation allows the technical support team to access the instrument remotely, with oversight from the instrument operator, so that technical challenges can be solved more efficiently.
• System upgrades can be performed so the OCD's Vitros Immunoassay Systems are running as efficiently as possible with the newest available technology.
It took four years and undisclosed investment to develop the predictive software that's at the core of RMC to offer this new service for free. Taylor said the return on investment to OCD is multifold, starting with reduced costs associated with onsite tech support as well as perceived value of a proactive support system to customers.
Every Vitros instrument that is securely connected to the RMC sends a stream of data, alerting customer technical service specialists and engineers to its operating status.
Taylor said it's a sea change from the old break-fix model to a new predict and maintain model.
While technical support specialists and engineers can perform remote system analysis, as well as monitor, review and upgrade system configuration, data and performance information, the RMC won't totally preclude onsite visits.
e-Connectivity technology is available to customers using Vitros systems, including: the Vitros 5600 Integrated System, Vitros 3600 Immunodiagnostic System, Vitros 5.1 and the Vitros ECi/ECiQ.
The Vitros 3600 is a high-capacity immunoassay system with a menu of 47 immunoassays, covering disease states such as cardiology, oncology, endocrinology, infectious disease, thyroid, metabolic conditions and anemia. The Vitros 5600 integrates clinical chemistry and immunoassay testing on one platform.
Taylor said the company intends to move the monitoring technology across a variety of other platforms from hematology equipment to lab inventory control.
"A lot of companies are doing monitoring," he said. "We're different because of predictive alerts. Nobody else is in this space."
In other news from Ortho Clinical Diagnostics, the FDA approved the Vitros Anti-HCV assay for use on the Vitros 5600 and 3600 systems.
Michael Samoszuk, MD, chief medical officer, OCD, said, "Clinicians can now test for hepatitis C as a part of their routine assay testing on the Vitros systems."
Results are available in 55 minutes.
Since the introduction of the Vitros systems in the last year, the company said it has released more than 100 assays.