References:


Diagnostic Errors, Acad Emerg Med, July 2002, Vol 9, No. 7, Gloria Kuhn, DO, PhD, gkuhn@med.wayne.edu


Overconfidence as a Cause of Diagnostic Error in Medicine, The American Journal of Medicine, (2008) Vol, 121 (SA), S2-S23, Eta S. Berner, EdD, and Mark L. Graber, MD, eberner@uab.edu


Overconfidence in Clinical Decision Making, The American Journal of Medicine, (2008) Vol, 121 (SA), S24-S29, Pat Croskerry, MD, PhD, xkerry@ns.sympatico.ca


Expanding Perspectives on Misdiagnosis, The American Journal of Medicine, (2008) Vol, 121 (SA), S30-S33, Beth Crandall, B, and Robert Wears, MD, MS, bcrandall@decisionmaking.com


Minimizing Diagnostic Error: The Importance of Follow-up and Feedback, The American Journal of Medicine, (2008) Vol, 121 (SA), S38-S42, Gordon D. Schiff, MD, gschiff@partners.org


Sidestepping Superstitious Learning, Ambiguity, and Other Roadblocks: Feedback Model of Diagnostic Problem Solving, The American Journal of Medicine, (2008) Vol, 121 (SA), S34-S37, Jenny W. Rudolph, PhD, JWRudolph@partners.org


Taking Steps Towards a Safer Future: Measures to Promote Timely and Accurate Medical Diagnosis, The American Journal of Medicine, (2008) Vol, 121 (SA), S43-S46, Mark Graber, MD, mark.graber@va.gov

Letter 2: THE ACTION THRESHOLD CONCEPT 

This time let's look at the diagnosis tool called the Action Threshold (AT) concept. AT is a point in medical decision-making that can help physicians decide when to treat, observe, or do a laboratory test or procedure.

How "AT" Works

Imagine a horizontal transparent cylindrical bar 10 inches long. It is calibrated from 1 to 10 with one-inch intervals. Inside this Action Threshold bar is a small red marble that can be moved from left to right and back. If the red marble is positioned at 5, it is read as 0.5 AT.

If the likelihood of an illness is to the right of the AT and it is not a serious illness, treatment can be started without any tests. If the likelihood of an illness is to the left of the AT or less than 0.5, two options are done: observe the patient or do a test to figure out what the illness is.

For purposes of a simple illustration, let's put the AT at 0.5.

Suppose that after the clinician takes the history and physical examination of the patient, he decides that the probability that the patient has a particular disease is at the right of 0.5 AT. If this disease is not generally considered serious like asthma, then he can justify starting treatment without any testing or work-up.

However, if the history and physical examination of the patient puts the probability of the likely condition to the left of the AT or below 0.5, then treatment is not justified and he has to do some testing to ascertain the condition of the patient or observe the patient without any tests but should come back if the symptoms persist or get worse.

A specific example would be a 2-year-old boy whose weight is only 18 pounds, but doesn't show any symptoms. Offhand, this is "failure to thrive." The probability of arriving at a diagnosis during the first visit is only 0.3 which is to the left of the AT of 0.5.

Suppose the patient is a feverish 6-year-old boy with petechiae at the legs and an enlarged spleen. With this combination of findings, the clinician will be considering a more serious illness and, therefore, needs to have the imaginary red marble at the 0.99 mark of the AT bar. This means that the clinician has to resort to more definitive diagnosis before starting treatment because the harm that the treatment can bring is considerable.

AT is a property of any treatment or recommendation for any condition or illness. AT is the likelihood of disease or condition above which -- or to the right of the 0.5 mark -- any treatment or recommendation, on average, provides more improvement than harm, and below which -- or to the left of the 0.5 mark -- will cause more harm than improvement.

"AT" In Sample Cases

When dealing with cancer, where the treatment like chemotherapy has significant serious side effects, the physician should be close to 100% certain of the diagnosis before treatment is started. In this instance, the AT should be near 1.0.

However, in cases like pneumonia, where the treatment itself does not pose great danger, the AT could be lower than 0.5. This would mean that a definite test, like chest X-ray, doesn’t need to be done since the treatment -- giving antibiotics -- is generally a safe bet.

In these two instances, the physician’s reasoning might go like this: “Since the risks in chemotherapy are high, I would not want to expose a healthy patient to this risky treatment. If the patient, indeed, has cancer, she would risk getting the serious side effects, but at least she would have an overall benefit from the chemotherapy. Therefore, because of the treatment's risks to a healthy individual, I need a high certainty on the diagnosis before undergoing treatment.

“In a case like pneumonia, suppose I am wrong with the diagnosis and I fail to treat, and it turns out to be pneumonia, serious complications can occur. If I treat a patient who actually does not have pneumonia, the antibiotic that I give will generally have no serious side effects. All things considered, treatment would be the best action to take, in which case I really do not need to bring up my AT near 0.9.”

The action threshold conditions in these two instances reflect how confident the physician is about the diagnosis before recommending treatment. The confidence level needed to be high in the cancer situation, and moderate in the pneumonia case.

In the daily practice of a clinician, making a diagnosis is “playing the odds.” Some patients are treated with medication or surgery for a particular illness that they do not actually have, while some patients with an illness are not treated at all. This happens because of the inherent qualities of the patient and the variety of manifestations of an illness. 

A common example is appendicitis. Some children with severe belly pain at the right lower side are sometimes operated on for acute appendicitis; when actually, the appendix is normal. On the other hand, some who are not operated on actually have appendicitis about to rupture a few days later, causing complications.

Many children are treated for recurrent bronchitis or pneumonia, when in fact their basic problem is asthma. Since the treatment of bronchitis or pneumonia is relatively safe, clinicians quickly prescribe it. 

Another example is frequent and prolonged colds or a viral infection. Many children are diagnosed to suffer from frequent or prolonged colds, when actually they have allergies to begin with.

In summary: 

If the likelihood of an illness is low, or to the left of the AT (Action Threshold), the clinician should not treat in most instances.

If the likelihood of an illness is high, or to the right of AT, the clinician is betting that the treatment is reasonable even without a test if the target illness is not serious.

If a test or imaging could change a clinician’s diagnostic confidence from either side of the AT, it has high diagnostic value, and therefore, must be requested.

If a test or imaging would not change a clinician’s diagnostic confidence at all, then it should not be requested or even discussed.

Sample Cases From Personal Experience

Last month, a third year medical student and I saw a 6-year-old boy with poor appetite, fever of 101, and fine scattered rashes at the trunk. The boy's throat was red with petechiae (blood spots) at the soft palate.

I explained to the medical student that the pre-test estimate of probability that the boy has scarlatina is high because the typical physical finding is consistent with strep throat.  I also knew that many children in the community are being treated for strep infection. Without doing a strep test I treated the boy with amoxicillin. In this instance, my pre-test estimate of probability of strep throat infection about 0.8 which is to the right of the 0.5 Action Threshold.

Early in the afternoon, we saw a 10-year-old girl with sore throat, no fever, and a history of exposure to someone with strep. However, the examination of the throat was not impressive – red throat, no exudates (white spot), and no petechiae. 

In this situation the pre-test estimate of probability of strep is to the left of the 0.5 AT. In other words, I was not convinced that the girl had a high chance of strep infection. So I did a strep screen test to see if there will be strep in her throat. I decided not to treat because my pre-test estimate of probability of strep is to the left of the action threshold or below 0.5. The girl did not have enough physical findings pointing to strep. The only thing in favor of strep is a history of exposure to the bug and red throat.

End of Letter 2 of 6

Letter 6: DIAGNOSTIC METHODS AND TRAPS TO AVOID

Methods of Diagnosis

We go to a physician for diagnosis and treatment. Between the two, diagnosis is more important because it leads to treatment. In most situations, physicians arrive at a provisional diagnosis within a few minutes of taking the clinical history.

About 85% of the diagnosis is found in the story of the patient's current complaint.  Except for skin problems, probably about 90% of the diagnosis is provided by physical examination. In a primary care practice, laboratory tests and X-rays are needed only in about 5% of the cases.

For example, it doesn't require a lot of brainpower to figure out that a 5-year-old girl experiencing pain in urination and fever has a urinary tract infection. Most mothers suspect this diagnosis even before seeing a physician. In this instance, the likelihood of arriving at a right diagnosis is about 0.9 or 90%.

However, a 2-year-old girl weighing 18 pounds, way below the 24-pound average for her age, will need extra effort on part of the physician to diagnose. This condition is medically labeled as "Failure To Thrive." The likelihood of getting the right diagnosis before laboratory tests are done is probably only about 0.2 or 20%. 

You might ask, "How do physicians arrive at a diagnosis?"

Most physicians use the "hypothetico-deductive" process. Some use "pattern diagnosis" especially in the diagnosis of skin or genetic disorder. Medical students use the "exhaustive" or "super-complete" history in getting into the differential diagnosis.

The easiest method of diagnosis is pattern diagnosis done by most skin specialist and geneticists. An example of this is Down’s Syndrome (Trisomy 21) and eczema in children. Even without the benefit of clinical history a child with Down’s syndrome and eczema can be diagnosed by just "one look." In pattern diagnosis, experience frequently rules.

Usually, medical students abandon the "exhaustive" or "super-complete" history method once they get into the residency or training program of the specialty of their choice.

By far, the leading method of getting a diagnosis is done by the "hypothetico-deductive" method. From the chief complaint and the first few minutes of the interview, the physician creates a short list of possible illnesses, maybe three or four different ones, that are highly likely based on his or her experience.

A few more targeted questions after the physical examination usually reduces the diagnosis into one or two possibilities. And in some instances, one or two laboratory tests will zero in on the diagnosis.  

Suppose your child is coughing for two weeks without a fever. You visit your primary care physician, who asks a few specific questions such as: Does your child cough after running? Does he cough more at night? Is there difficulty of breathing or wheezing after prolonged playing? Is there runny nose for longer than 10 days? Has the child had bronchitis or pneumonia before? Is there asthma in the family?

If most of your answers are "yes" to these questions, and the physical examination is normal, your physician will most likely give your child an inhaler for asthma.

Based on your physician's experience with hundreds of children who came in with similar symptoms and got better with an inhaler for asthma, he subconsciously estimates the pre-test probability of 0.8 (80%) or more in favor of asthma. Since the inhaler for asthma has very low downside or harm, even if administered to normal children, a physician can accept the probability of being wrong by 0.2 (20%).

Traps to Avoid in Making a Diagnosis

To make you aware of how clinicians can sometimes be wrong with their diagnosis, let me share a couple of case examples.

A clinician makes a diagnosis of a rare disease called Moya-moya in a 25-year-old woman with headaches and weakness of the right arm. The next time the clinician sees a patient with almost similar history and physical findings, there is a good chance he'll remember the patient with Moya-moya and consider it again.  But since it is a rare condition, his chance of being wrong is high in the second patient. This mistake is called “availability heuristic.”

Dr. Smith saw a 42-year-old man who is a gardener, and who had dental work done about four weeks before. He is in great health without any known risk factors for heart disease. Dr. Smith saw several splinter hemorrhages under the nails of both hands and a small ecchymosis on the sole of the left foot that reminded him of Janeway lesion. He heard a grade 2 pulmonic ejection murmur. Because of these findings Dr. Smith made a pre-test estimate of likelihood of Subacute Bacterial Endocarditis at about 3%. He ordered blood cultures.

Dr. Smith was misled by the typical or representative findings of splinter hemorrhages and ecchymosis in SBE. In fact, those findings are also found in trauma from gardening. If Dr. Smith looked at his reference book, the incidence of SBE is about 2 in 100,000 and it is almost always associated with fever (95%), anorexia (98%), and frequently in patients with risk factors. This error in diagnosis is called “representative heuristic.”

Being off by about 20-50% is often acceptable, but being off by a factor of 100 to 1000, as in the examples above, should be avoided. It is a product of flawed reasoning and failure to consider epidemiology.

Leonardo Leonidas, MD, FAAP

Assistant Clinical Professor in Pediatrics (Retired 2008)

Tufts University School of Medicine

Boston

End of CDM Letters of 6

Letter 5: HARM, BENEFIT, IMPROVEMENT, UTILITY, AND IMPACT 

Harm, Improvement, and Benefit

Patients visit clinicians because they want to become better. What patients really want is treatment or some kind of solution to relieve their chief complaint or health problem.

Any treatment, whether drugs or a change in lifestyle, has three possible results: benefit, harm, or improvement.

Antibiotics reduce symptoms and mortality for those with pneumococcal pneumonia. However, antibiotics, like other medications, have "harm" or side effects – such as diarrhea, rashes, or, in rare occasions, anaphylactic reactions -- when they're given to a patient with bacterial infection and a patient who is actually healthy.

Each “sick-visit” between a patient and a clinician is a “diagnosis-treatment” encounter with two important considerations: how much harm the treatment will cause if the patient was not sick, and how much disease improvement the treatment will bring if the patient was really sick.

We would, therefore, define "harm" as: the adverse or negative outcome of treatment that would occur in a patient without disease (in other words, the side effects in healthy individuals).

"Improvement" is defined as how much better a patient with disease becomes as a result of the medication or treatment, compared with what her condition would be without the treatment, disregarding the side effects of the treatment.

"Benefit" is how much better a patient with disease becomes as a result of the treatment, compared with what her condition would be without the treatment, after taking into account the side effects of the treatment.

You might be wondering why a physician would give antibiotics to a person who is without disease. They wouldn't if they knew that the person does not have a disease. However, many patients with colds, fever, and cough are often diagnosed to have bacterial pneumonia, when in reality, what they have is a viral infection. Sometimes also, a chest X-ray shows “pneumonia” but the cause of the pneumonia is not bacterial but viral.

Here's an example illustrating harm, benefit, and improvement:

Bestbiotic (fictional drug) is given to patients with serious bone infection, and it reduces mortality from 50% to 20%. This means 30 fewer deaths from bone infection for every 100 patients treated. Unfortunately, Bestbiotic causes severe anemia resulting in death in 10% of the patients who take it, meaning it kills 10 patients for every 100 patients treated. In this example, the "improvement" is 30 fewer deaths, the "harm" is 10 Bestbiotic-related deaths, and the resulting "benefit" is 30 minus 10, or 20 fewer deaths.

Utility

Each individual has a different concept of harm and improvement. So, ideally, clinicians should ask the patient what they consider harm and benefit in any treatment they recommend. If the clinicians don’t ask, they will never know.

For example, a 25-year-old software programmer is about to lose her big toe from a proposed surgery. Although she does not like the idea of losing a part of her body, it would not greatly affect her income and job performance. The harm in this situation is not significant.

Another 25-year-old, a professional ballet dancer, is faced with a similar surgery -- the loss of her big toe. In this situation, the harm is great because of her occupation.

Clinicians can help their patients make judgments by giving them facts and an outcome study. However, the final decision on what to do should be left to the patient. There is really no pure “right” or “wrong” in many medical recommendations. In Decision Analysis terminology, the judgment of “goodness” and “badness” of medical outcomes is called “utility.” Perfect health is the best outcome; permanent physical disfiguration and death are the worst results.

One of the duties of clinicians is to help their patients make decisions that will lead to more positive outcomes as determined by their values. Clinicians can inject the concept of utility as a relative measure of “goodness” and “badness” of a disease or treatment outcome relative to other outcomes from alternative recommendations.

Clinicians can give patients a utility score of 0 to 1, from bad to good. The score system will have no units. It is a mental concept that is easier for both the patient and the clinician to arrive at a “better” and even faster judgment. As a rule, when our mind frame has numbers, “better” decision seems easier to make.

Another way of looking at utility is to consider it a subjective measurement of the likelihood of an outcome.

For example: Anaphylaxis from penicillin leading to death is a bad outcome. In terms of utility score, it is 0. However, the chance of it happening is low, about 1:200,000. Considering these two ideas together, most will take penicillin for appropriate indication because of the slim chance of anaphylaxis, and because of the better chance of benefit, which is shortening the length of infection and reducing the risk of rheumatic heart disease in patients with strep throat.

When clinicians discuss the risks and benefits of treatment with their patients, they consider both utility and likelihood when judging a particular outcome in the decision-making. 

Impact

Unlike "utility," which is a subjective form of a bad-to-good measurement scale, "impact" simply reflects how much an outcome affects the patient compared to her not experiencing the outcome at all, regardless of the “badness” or “goodness” of the outcome.

Unlike utility, where lower is always worse and higher is always better, with "impact," the only consideration is how profoundly an outcome affects the patient. A happy event like a wedding, or an upsetting situation like a loved one being diagnosed with cancer, can result in a high level of “stress” despite the fact that one circumstance is pleasant and the other gloomy.

"Utility" refers to the subjective score from 0 to 1 of a given outcome or possible result of the treatment, while "impact" is the perception or feeling of living without the outcome compared with living with the outcome.

A severe adverse reaction from a medication can have a strong “impact” on the patient, as would an amazing cure from a serious disease. In both instances, an impact of 0.9 on a scale from 0 to 1 is possible. Think of an impact scale as an “importance score” for medical outcome.

At the end of the day, impact and utility give the same results in the final clinical decision.

In a nutshell, "harm" is the likelihood of the adverse or side effects multiplied by its "impact." "Improvement" is the likelihood of patient betterment multiplied by its "impact.”

End Letter 5 of 6

Letter 4: PROBABILITIES AND ODDS IN DIAGNOSIS

Definitions

"Probability" is defined as the likelihood of an Event or Index outcome expressed as a percentage of all outcomes. It includes the Index outcome itself plus the other Events or outcomes that may occur. The Event or Index outcome is counted twice and included in the numerator as well as in the denominator.

For example: In tossing a coin 100 times, the head may come out 50 times as well as the tail. The probability of the tail coming out is 50 out of 100 or 50%. The 100 times the coin is tossed includes both the Index outcome (head) itself as well as the other outcomes (tail).

"Odds" is another means of expressing likelihood. It compares the Index outcome on one side and all other outcomes (excluding the Index outcome) on the other side to come up with a ratio of the two. In this situation, odds count the Index outcome only once. Going back to the coin tossing exercise, odds would compare the 50 tails on one side with the 50 heads on the other side for a ratio of 50:50, or 1.

Using the concept of "odds" in medical decision analysis is important because most mathematical calculations are simpler when expressed in “odds form.” Using odds allows clinicians to do faster mental calculations, which would be difficult using probabilities because of its complex formulas.

To compare odds and probability: odds of 1 are the same as 50% probability. Any probability less than 50% has odds between 0 and 1. Probabilities higher than 50% can go from odds just over 1 up to infinity. Odds of 0.5 is the same as probability of around 33%, and odds of 1.5 is the same as probability of 60%. The odds of 99 are roughly the same as 99% probability. Above odds of 99 or probability of 99% is not worth considering because there are no odds equivalent to 100% probability.

Using Odds in Diagnosis

In the practice of medicine, it makes more sense to frame the thinking process in favor of "likelihood" rather than "certainty." To pursue certainty is impractical, extremely costly, and oftentimes dangerous to patients. Medical practice is a field of uncertainty as opposed to engineering or accounting where outcomes are accurately predicted. In medicine, outcomes are not assured -- the best way to predict an outcome is in terms of odds.

Here's an illustration of how odds and probabilities can be applied in diagnosis.

Let us mix 100 green and red marbles in a bucket. A win is if you pick a green marble. There are 51 green and 49 red marbles. In this scenario, you should bet on green marbles for an overall win, knowing that you will not win all the time. If there were 85 green and 15 red, you would still bet on green as the winner, of course, but you do not need to know beyond 51:49 to bet on green to win. So, it does not make sense to put more effort or cost to find out if it was 51:49 versus 85:15 in order to decide on which color to bet on.

This concept can be applied when getting a clinical history. One clinician can ask 10 questions to arrive at a reasonable diagnostic probability, but these 10 questions may not give him a 0.9 probability of hitting the mark. If another clinician could reduce the questions to five and arrive at a reasonable probability of just over 0.5, he would be as good as, if not better than, the clinician who asks 10 questions.

When asking questions it makes sense to have a structure for a particular chief complaint (CC).  For this particular CC, the same five or more questions can be asked every time.

Let us take the example of a 10-year-old boy with a CC of “Coughing.”

These are the structured questions that I ask all the time:

1. Is the cough longer than two weeks?

2. Is there no fever?

3. Is the coughing more after playing, running, exercise, or laughing?

4. Did he ever have bronchitis or pneumonia before?

5. Is there asthma, allergy, or hay fever in the family?

If there are at least three "Yes" answers to these five questions, I put asthma as the first probability. These five questions are the green marbles. The chance of being right is high and only five questions were asked. More questions can be asked, but the likelihood of asthma as the true diagnosis will not increase much.

Once a diagnostic likelihood of over 0.5 is made, after just a few questions, asking more questions will reduce the efficiency and effectiveness of the clinician. It is difficult to attain certainty with clinical history. As long as more than a likelihood of 0.5 in the clinical diagnosis is achieved, the clinician can be satisfied and treat depending on the severity of the target disease, the values of the patients or parents, or the “standard of practice” in the community.

Using Odds in Diagnostic Tests

Once the clinician has an estimate of the most likely diagnosis based on clinical history and physical findings, he decides on whether to treat or not. This decision depends on his estimate of the severity of the target condition or disease, the values of the patient or parent, and the standard of practice in the community.

If the clinician is considering an illness that is very serious like meningitis or a malignancy, he is obligated to do some laboratory tests before starting any form of treatment. However, if, from his estimate, the target condition is not serious at all and there is enough time to wait, he has two decisions to choose from: to observe or to do more laboratory tests. 

In most instances, observing and doing no tests is the better approach since the patient might get better while the clinician is observing. Or the patient may develop more symptoms that the clinician can use to request less or narrowed-down tests to arrive at a definite diagnosis.

Let us take this example of a 10-year-old girl who had belly pain for about six weeks. The clinician had done his history and physical examination and concluded that there were no serious conditions that require immediate attention. About two weeks later, the mother called and said that there is blood in the stool. This was the first time this symptom was noticed.

The clinician requested the mother to bring her daughter in again for follow up. After doing the routine history and pertinent physical examination, including a rectal exam, the clinician considered Crohn’s Disease as one of the top three possibilities in his differential diagnosis. He requested for CBC, ESR, and some blood tests that are sensitive to diagnose Crohn’s Disease. If any of these were positive, then the clinician would consider endoscopy for biopsy.

After the decision to observe the child after the initial visit, new symptoms appeared that helped the clinician limit the list of possibilities. Suppose that during the initial visit, he requested for a battery of tests, there would have been a good chance that all of the tests would be normal, which would not help at all. This would probably even complicate the problem if false positive tests, like a positive occult blood in the stool, or a WBC of 18,000 with normal hemoglobin and platelet, appeared in the results.

Now let's say that the patient or the parent wants the clinician to do some tests after he decided to just observe. There are two ways the clinician can deal with it. 

One is, he can explain to the parent, “At this time, with all the information from the story of your daughter and the physical examination, my best bet is there is no serious condition that we should worry about. But if the belly pain persists, gets worse, or there are new symptoms like blood in the stools, you can come back any time and we will immediately do some tests to figure out the cause.”

Another approach is to say: “We can do some tests now. However, sometimes certain tests complicate the problem. We might get a false positive result, which will make me pursue or order more unnecessary tests that could hurt your child even more, as compared to if we first observe for more symptoms. There's also the possibility that your child will get better while we are observing.”

End Letter 4 of 6

Letter 3: PRE- AND POST-TEST ESTIMATES OF LIKELIHOOD

Pre-Test Estimate of Likelihood

It's important that a physician, using his training, experience, and the epidemiology of the illness being considered, forms a pre-test estimate of disease likelihood in taking care of a patient.

If the pre-test likelihood of an illness being considered is high, and the physician is highly confident, and the illness is not serious, and it is to the right of Action Threshold (AT), treatment can be started without any tests needed. This decision-making process follows the AT concept we discussed previously. As in the AT bar, if the pre-test likelihood is to the right of the 0.5 mid-point, treatment is reasonable. 

Most of the time, clinicians make a diagnosis with very little doubt. In the clinical decision analysis world, this is called "high pre-test estimate of likelihood of the index disease."

This is best illustrated when a clinician makes a diagnosis of chicken pox or Down’s syndrome. Within a few seconds of seeing the patient, even without asking a single question, the clinician can make a diagnosis for which he is highly confident of being right.

However, most patients don't come with classic or typical physical findings or a typical clinical history. With these patients, clinicians are forced to ask many questions that will help bring the likelihood of a target illness as high as possible.

So, how do clinicians come up with a pre-test estimate? Depending on their attitude and thinking style, they arrive at a pre-test estimate by relying on experience, knowledge, judgment, their willingness to ask their superiors or colleagues, and the availability of time to do research using the Internet.

Right now, there are no clinical studies pertaining to how most clinicians organize their thinking process in making a pre-test estimates of likelihood of a target disease (TD), but in general, after listening to the chief complaint and present illness, the clinician first thinks of a list of about three to seven differential diagnosis.

As he continues to ask more questions, he tries to eliminate a few of the differentials until he's left with one or two considerations. He thinks of how specific the findings are for the TD, how many independent findings are seen in the TD, and what risk factors the patient has. Almost at the same time, the clinician's mind recalls his past experiences with similar patients that give him an intuitive “feeling” for the TD's likelihood.

By knowing the community and the epidemiology of a target disease (TD), clinicians can move fast in making pre-test estimates of likelihood.

For example, in the spring of 2003, here in Bangor, Maine, we had a tremendous number of children with strep throat, with petechiae at the soft palate, and typical scarlatina rashes. During that time, I treated a lot of children without even doing “rapid strep” test. If there were no increased number of children with strep in our community, I would be doing many “rapid strep” tests before giving an antibiotic. But since my pre-test estimate of the TD was high, I deleted testing. Once I no longer see numerous children with typical strep throat findings, I will start doing “rapid strep” tests again.

Sometimes, asking what the patients think they have can raise the pre-test estimate of likelihood of a TD. In one study of patients reporting that they thought they had “sinusitis,” more than 30% actually had sinusitis as proven by X-ray and sinus culture. With specific findings of sinusitis, the diagnosis leaped to 50%. (Ann Intern Med. 1992; 117:705-10, Clinical Evaluation For sinusitis: Making the diagnosis by history and physical examination. Williams, JW, et al.).

Most of the time, with just two or three symptoms gathered from the present history, clinicians can make a good pre-test estimate of likelihood. Even parents like you can have a high confidence level on what your child has.

Take for example a 5-year-old girl with pain on urination. Many times a mother would call my office with a complaint that her child has “urinary infection.” Indeed, in most cases the mother is right. A child with pain on urination, with or without fever, has a good chance of having Urinary Tract Infection.

Here are other examples:

Children with cough longer than three weeks without fever, has asthma until proven otherwise, especially if the cough is aggravated by running, playing, or laughing. 

Runny nose longer than 14 days with coughing more prominent at night has a good chance of sinusitis. 

A 7-year-old boy with poor appetite, pain at the right lower side of the belly, and more belly pain on jumping in place, has acute appendicitis until proven otherwise. 

Headache that is on and off, the severity is not increasing, longer than three months, and if there is family history of migraine, should be considered as migraine also.

A teenager with sore throat, tiredness for many days, and big “glands” under the jaw has a high chance of infectious mononucleosis.

A child with frequent bronchitis or recurrent pneumonia, with chronic diarrhea, and poor weight gain has cystic fibrosis until proven otherwise.

Brain tumor should be considered in any child with a headache for at least two weeks if there is vomiting and changes in behavior or personality. If there is blurring or double vision, brain tumor should be in the top list of differential diagnosis.

These shortcuts that experienced physician use is called “heuristics.” Even without the benefit of laboratory tests, there is a high probability of being right after asking just a few questions during the clinical interview.

Post-Test Estimate of Likelihood

Post-test likelihood (PTL) is estimated after the results of laboratory tests or procedures are known. In most instances, the post-test likelihood of discovering an illness will exceed the AT point or be to the right of it.

If PTL remains to the left of AT, and the patient is not getting better, more tests are needed. If the patient is getting better or at least not getting worse, observation is still reasonable.

End Letter 3 of 6

Letter 1: DECISION ANALYSIS AND DIFFERENTIAL DIAGNOSIS

Clinical Decision Analysis

The most common, often considered "standard," method of diagnosis practiced by most students, residents, and even experienced physicians, is the "random decision" or "intuitive thinking" method, which does not really follow any conscious methodology.

Clinical Decision Analysis or CDA, on the other hand, although not yet studied in a controlled trial showing positive outcome, is based on a step-by-step method of thinking that allows clinicians to advise their patients regarding the "best bet" or "best outcome" for any specific condition.

In the same manner that we should wear seat belts, install smoke detectors, recommend car seat for infants and young children, or tell parents to quit smoking, we should use clinical decision analysis in patient diagnosis. CDA keeps the interest of the patient always in the frontal lobes of the clinician.

The CDA method starts with treatment considerations such as harm and improvement, followed by testing and diagnosis with its likelihood ratio and pre-test estimation.

The reason for thinking of treatment first, then testing or diagnosis afterwards, is that decisions about tests can be properly made after the risks and benefits of a treatment have been critically dissected. With the knowledge and understanding of the risks and benefits, harm and improvement, the clinician and patient can discuss if a test is worthwhile doing.

The function of decision analysis, unlike scientific experiment that shows natural truth, is to help a decision maker choose the best option among many alternatives of treatment. Decision analysis does not reduce the uncertainty about the true nature of the patient’s illness, but rather it makes the choices more rational in light of uncertainty.

In the long run, if decision analysis is applied in most of the complicated patients, the probability of making a grave error is less compared to the usual "random" or “intuitive” method.

Differential Diagnosis

There are three components in making a diagnosis. These are Clinical History, Physical Examination, and Laboratory Tests and Imaging.

Among these three, Clinical History is the most important. Depending on the specialty of the physician, History contributes about 80-90% to the diagnosis. Except in Dermatology (Skin Problem), Physical Examination contributes only 10 to 15%.  In Dermatology, about 90% of the diagnosis is done by physical examination. Laboratory Tests and Imaging contributes only about 5 to 10% towards making a diagnosis.

A good Clinical History is dependent on a complete and practical Differential Diagnosis. Without a reasonable and complete differential diagnosis, it is difficult to ask good questions, both pertinent positive and negative ones.

The process is like this: Clinical History starts with a Chief Complaint (CC). The clinician then generates the pertinent questions that will reduce the number of considerations from the possible disease conditions in a complete Differential Diagnosis. The clinician designs the questions so that they rule in or out one or two disease conditions that is brought about the Chief Complaint.

For example: A three-year-old boy came in because of wheezing. With wheezing as the main symptom that prompted the parent to bring her child for examination, a clinician who does not have access to a computer or the Internet will rely on his memory and stock knowledge during the interview.

Most clinicians are capable of mentally listing only up to six or seven different causes of wheezing, like asthma, pneumonia, allergy, bronchitis, foreign body, cystic fibrosis, and bronchomalacia.

With this mental list, a clinician will ask questions such as: Is there a fever? How long has the wheezing been going on? Is your child coughing after running? How long does he cough? Is there a previous history of bronchitis or pneumonia? Is there chronic diarrhea? Is there family history of asthma or cystic fibrosis? When the coughing or wheezing started, did the child choke on a food or small toy? How is your child’s growth? 

Because the average "mental" differential diagnosis is limited to seven possible conditions, the clinician runs out of pertinent questions to ask.

Now, let's look at a clinician who has access to technology. Hearing "wheezing" as the chief complaint, he gets a handheld computer or logs on to the Internet, and types “wheezing” in a search box on Differential Diagnosis.

With just a few strokes on the keyboard, the modern clinician can easily bring up a listing of at least 35 causes of “wheezing” in children. 

From this electronic list, he can easily pinpoint the causes that are serious and more common in the community, or its epidemiology. With this list, he can easily generate pertinent questions.

In the case of the clinician who relies on brute mental recall power, there is a high probability that a not-so-common cause of wheezing, such as vascular ring or heart disease, will be missed during the first clinical encounter. The unfortunate result could be a serious diagnostic error if the right question is not asked.

The current “standard of practice” is 100% dependent on pure mental recall, which is prone to error because of the limitations of the human brain.

With the speedy advancement of technology, we should expect that within the next few years there would be a list of differential diagnosis that includes the pertinent questions to ask, as well as the diagnostic tests or imaging to order for any particular condition. This would be available online and linked to websites with the latest Evidence Based Medicine like Medline, Cohcrane Data Base, UpToDate, or Clinical Evidence. With the use of effective technology, diagnostic errors will certainly be reduced.

End of Letter 1 of 6.