Endodoc

In my last entry I discussed the rather contentious topic of how physicians order/should order laboratory tests.  I say “contentious” because lately there has been intense scrutiny by governmental agencies, physician specialty groups, and the like, to  cut down on the extraordinary health care costs in the U.S.  Anyway, just the other day, it was brought to my attention that I had not addressed the question of what to with the test result.  Last weekend, I was grading medical student tests (one of my least favorite pastimes).  At our medical school, the curriculum is so-called “problem-based.”  This means that for the most part, teaching is tied to problem-solving rather than to the classic approach of mostly lectures for the first 2 years and then, exposure to patients and their problems.  I think the curriculum is good but in my opinion, its one serious downfall is that even the tests are problem-based, that is, essay questions about a clinical problem.  So there I was last weekend, grading 300 long essays.   The details of the case are not important- the patient likely had a stroke.   The clinical documentation included a brief medical history, a physical examination, and results of the initial laboratory tests ordered in the emergency room.  The results for 2 of the many ordered tests (I believe most of the tests were ordered appropriately) were abnormal;  the serum sodium level was just below the lower limit of normal while the serum calcium level was just above the upper limit of normal.  I was fascinated by the different approaches students took in problem-solving the 2 minimally abnormal test results.  About 25% of the students (these were end of third year students who are  supposed to be pretty savvy) acknowledged the abnormal calcium and sodium results but noted that the results were barely out of the normal range and that they would probably just repeat the tests at some point and that these findings were not likely related to the patient’s primary problem.   A few of the students, never mentioned the test results while most of the rest gave the sodium and calcium results the kind of attention one would give an EKG that showed a serious cardiac arrhythmia- they wrote and wrote what the tests results might mean and what further testing they were going to do.  So how come, only about 25% of the students knew how to approach the sodium and calcium results correctly but the rest were way off course?  That’s what I want to discuss a bit in this entry.

What Does a “Normal Range” Mean?

I do not want to get all technical here, but what goes into establishing a normal range (really called ” reference interval) is rather complicated and is heavy into statistics.  I will try to keep it simple.  Basically, every laboratory is supposed to establish its own  reference interval for every laboratory test it performs.  Every pathologist and clinical chemist knows how this is supposed to be done.  One takes x number  of people (usually 50-100) who are presumably normal with respect to the test in question, and performs the test on them.  A mean value is calculated as is the normal range (usually the mean +/- 2 standard deviations or SD).  If the test values follow a “normal” distribution (you know the familiar bell-shaped curve), then the mean +/- 2 SD comprises about 95% of values in a healthy population.  In statistical terms, this means that out of 100 normal people, 5 individuals are likely to have their test results either above or below the 2 SD reference interval.  I told you this was complicated.  All you need to know is that people who are entirely normal have a reasonable chance of having a laboratory test result that is outside the reference interval.

Do All Laboratories Calculate Their Own  Reference Intervals?

Many laboratories actually do not calculate their own  reference intervals although as far as I know, all large commercial laboratories do this.  It is quite a bit of work and a small laboratory in a hospital or in doctor’s office is not going to want to mess with it.  Instead, they use either the  reference interval that comes in the packaging with the test kit they use or use some published reference interval.  Unfortunately, even if the laboratory does a great job in performing the test, the test results may not fit so well with the reference interval the laboratory uses.  For example, if the published serum free thyroxine level normal reference interval is 1.1 ng/dl +/- 0.5 (mean +/- 2 SD) but if the laboratory normal reference interval is actually 1.3 ng/dl +/- 0.5, many reports will be interpreted by the ordering physician as either normal or abnormal when the opposite is true.  Another important example is HbA1c, the test that quantifies average plasma glucose over the previous 2-3 months and is now widely used to screen for/diagnose diabetes or glucose intolerance.  A level 6.5% or greater is considered diagnostic for diabetes but only if the test is performed in a laboratory that has test results aligned (basically standardized) to those in the Diabetes Control and Complications Trial (DCCT) Reference Laboratory.  That approach obviates the need for the laboratory to establish its own reference interval (I still think it is a good idea for the laboratory to do that if it is at all practical).  I didn’t even mention the fact that for many laboratory tests there are age and sex differences in the  reference interval.  If this is not appreciated by the physician, serious errors in interpretation of the test result can be made.  For example, serum alkaline phosphatase is much higher in growing children than in adults.  In a child, a serum alkaline phosphatase of 300 U/L might be just fine, while in an adult it likely means either liver or bone disease.

Back to the Slightly Elevated Serum Calcium and the Slightly Low serum Sodium

In the “real” world, not the world of medical student tests, physicians must frequently figure out what do do with a test result that is outside the reference interval.  First, as I discussed in my earlier entry, no test should be ordered without a good reason.  Second, the test result must be reviewed and interpreted; some physicians are so busy that they often do not even look over the tests they have ordered or just scan them, looking for the test results listed as “out of range.”  Sometimes a test result is abnormal yet it falls just inside the reference interval.  Remember, just because a test result is outside the reference interval it is not always abnormal, that is, related to some medical problem.  In much the same way, just because a test result is inside the reference interval, that does not mean all is well.  Borderline low and high test results need to be scrutinized.  Maybe they mean something and maybe not.  That is more of the art and science of medicine.  How do medical students learn about these things?  I am not sure.  Based on the test I just graded, maybe we do not do such a good job in teaching them either how to order tests in an artful and scientific way and in figuring out what to do with the test results.

The Most Important Part of Sorting Out Test Results

In my opinion, arguably the most important part of interpreting laboratory test results is sharing the information with the patient and discussing what the results might or might not mean.  In this age of electronic medical records, e-mail, etc., I am amazed at how many otherwise excellent physicians do not have a systematic method of keeping track of what tests they have ordered for their patients and what the test results turn out to be.  Worst of all, they do not have a systematic way of getting in touch with their patients to let them know what the test results are and what it all means.  All patients deserve to be “in the loop.”

 

For years, I have found it fascinating that if two patients with virtually identical clinical circumstances go to two different physicians who have similar medical training, the patient management plans often differ considerably.  For example, let’s say that both patients are 14 year old females with histories of fatigue and weight gain over the previous 6 months.  Let’s also say that both physicians obtain excellent medical histories and perform equally excellent physical examinations and find no specific clues to the reason for the patients’ signs and symptoms.  Both physicians then consider the likely possibilities and formulate plans.  At this point, the physicians take somewhat different approaches.  The first physician, Dr. Jones  decides to order a few basic laboratory tests such as a blood count, urinalysis, and, perhaps, a thyroid profile (maybe a free T4 and TSH), planning to see the patient back for follow-up in 3-4 weeks.  The second physician, Dr. Smith,  also decides to order some laboratory tests; perhaps the same tests that the first physician ordered but in addition,  a fasting lipid profile, a pregnancy test, a metabolic panel (various tests of blood chemistries, liver function, etc), HbA1c (a test for diabetes mellitus), Serum insulin (to evaluated the patient for insulin resistance and, perhaps the metabolic syndrome, a 24-h urine for free cortisol (to rule out Cushing’s disease), an EKG and chest X-ray (to look for heart disease), and a nutrition consult.  The physician also plans to see the patient back in 3-4 weeks.  So, how is it that Dr. Jones orders a few basic studies,  while Dr. Smith orders many more tests (as well as a consultation).  Certainly, the costs for the visits differ significantly, by about $500-$1000.  Which physician is right?  Are they both right?

Do Physicians Order Unnecessary Laboratory Tests and Procedures?

There are endless data, especially in the past few years, to show that as a group, physicians order far too many laboratory tests and procedures.  The recent interest in this phenomenon is to a great extent related to efforts to bring some measure of control to U.S. health care costs.  For example, a recent report in the New York Times  addressed the question of laboratory tests and procedures, focusing on a report by 9 major medical specialty groups in which 45 tests and procedures that have no proven benefit to patients were  listed.   The newspaper article was entitled “Endless screenings don’t bring everlasting health,” and was written by Lisa Schwartz and Steven Woloshin.  There has also been a measure of pressure brought to the question of whether physicians who own a stake in a laboratory, order more tests (performed in the physician-owned laboratory) than those who have no financial interest in ordering the test other than, perhaps, a charge for interpreting the test result (see “Doctor’s stake in a lab affects biopsy rate, ” written by Mitchell, JM., in Health Affairs, April 2012).  To me, the most interesting data are those from a study carried out a few years ago in which the investigators studied whether physicians ordered fewer tests when the prices of the test were listed on the order form.  As it turned out, when the prices were listed, doctors actually ordered about 20% fewer tests.  Clearly they didn’t “need” those tests, but just wanted them.  That’s the good news.  The bad news is that after a year, when the prices were taken off the ordering form, the docs, reverted to their original ordering behavior.  Recent studies have found very similar results.  For example,  hospitalists at the Johns Hopkins Hospital in Baltimore studied how often common but expensive laboratory tests were ordered with or without the cost of the test listed (“Knowing costs affects lab test ordering“).  They found that over a 6-month period November 2009-May 2010, there was an average decrease of about $16,000 billing per test when costs were displayed on the order form compared with a 6-month period 1 year earlier.  This was just for one hospital.  The investigators, led by Dr. Leonard Feldman, concluded that much of the reason for the excessive test ordering was simply a lack of awareness of how much tests cost.  The investigators acknowledged that they did not have any data to show that patients’ outcomes were not affected adversely by stingy testing.   So, the excessive ordering is not always related to to either personal financial benefit or other factors such as medico-legal concerns.  What is it due to?  Beats me, but I think it is to a great extent a “reflex” action and one of convenience- if I order everything I might need right now, I will know what’s what sooner. That’s just my theory.  It might be mostly a question of personality- those docs who require immediate gratification vs. those who can tolerate waiting for answers?  We can gain some insights from a recent study entitled, “Why do physicians order unnecessary preoperative tests? A qualitative study,” written by SR Brown and J Brown (Family Medicine 2011;43:338-43).  the investigators interviewed 23 physicians and nurse administrators and found that the main reasons for ordering unnecessary tests were as follows: practice tradition, belief that physicians wanted the tests performed, concern about medicolegal issues, concern about possible surgical delays, and lack of awareness of evidence and guidelines.

How Should Physicians Order Laboratory Test and Procedures?

I do not believe most medical students and resident physicians are taught the “art” of ordering laboratory studies and procedures.  That gap in training is to some extent, related to the fact that many of the patients seen by students and resident physicians have either acute illnesses where time is of the essence or are patients for which reliable follow-up is quite uncertain.  So, it’s sort of “catch while catch can.”  In the real world, physicians are generally not in such a rush and they would do well to stage their testing unless the situation is urgent.  Why order every test you might need at the first encounter when starting with a few simple basic tests would suffice?  Sometimes, I hear that the blitz is performed because it causes less inconvenience for the patient- they won’t need to come back for a test that is needed based on the initial basic testing.  In my opinion, that argument is nonsense.  It certainly makes no economic sense.  If one is worried about the possible need for another venipuncture, extra blood can be drawn initially and stored, to be assayed as needed.  Laboratories do it all they time.  That’s what freezers are for.

Another side of the question of what to order, is the importance of explaining to the patient exactly what tests were ordered and why.  Many physicians just do not take the time to explain this to their patients.  Maybe, sometime the docs just don’t know exactly why they ordered what they ordered?

Finally, there is the question of letting the patient know what the tests showed and what it all means.  I am astonished how many physicians have no organized way to communicate test results with their patients.  It is really shameful.  Have any of them ever heard of e-mail or even the U.S. postal service?  In my view, if a physician takes the time and trouble (for the physician and patient) to order a laboratory test or procedure, the least the doc can do is get back to the patient with the test result and interpretation/plan in timely fashion.  Some physicians are super at this task and some are horrible.

The Future

It will be interesting to see how much pressure is put on physicians various “expert panels,” insurers, and governmental agencies (e.g., Medicare, Medicaid) to cut back on ordering tests?  It is already happening, particularly for various screening studies.  I do worry that some of our patients will suffer because of ill-advised recommendations based on epidemiological data (the kind that look at how many people we need to screen for this or that disease to save 1 life or to detect 1 cancer.  Good medical care is not always about maximum cost effectiveness.  But, there is no question that we as physicians could do better in the way we order laboratory tests and how we use the results to guide our therapies.

 

The Question

The other day I got into a discussion with a primary care physician (PCP) about a hypothetical patient.  We had been discussing a case history for an upcoming medical student examination.  The patient was a 40 year old female with a medical history suggesting hypothyroidism.  The physical examination revealed an enlarged thyroid gland.  The question was as follows: what thyroid function studies should be ordered?

The Answer

The PCP said that he would order total thyroxine (TT4), free thyroxine FT4), and thyroid-stimulating hormone (TSH).  I was a bit surprised by his answer and asked if he didn’t also want to order thyroid peroxidase (TPO) antibodies?  He replied something to the effect that in the “real world” docs don’t mess with that kind of stuff.  I became even more surprised and then asked him whether he thought it was important to know what one was treating.  He got a bit hostile and replied that if the TSH were high and the FT4 and TT4 levels low, he would have a diagnosis, primary hypothyroidism.  He explained that the cost of the TPO antibodies (about $50-$80) depending on the laboratory) was not worth the benefit.  My response was “hmmmm.”

What did I really think?

I did not agree with the PCP’s approach to the hypothetical patient but I did not go ballistic since there was some merit to his argument.  On the other hand, it is my opinion that testing for TPO antibodies in the patient described above is worth the modest extra expense.  First of all, as a general principle it is important to know what one is treating.  While primary hypothyroidism is a diagnosis, there are many different causes for the condition and the approach to treatment might well be dictated by the specific etiology.  In those parts of the world where iodine deficiency is not endemic, the most common reason by far for primary hypothyroidism is chronic lymphocytic thyroiditis (CLT) or Hashimoto’s thyroiditis as it is commonly called.  This is an autoimmune disorder which is highly prevalent in females, particularly those over 40 years of age.  As I have discussed in previous entries, the physical examination often offers clues to the diagnosis; in my experience, careful examination of the thyroid gland will reveal a small lymph node on the left, just above the thyroid isthmus.  This node is called a delphian node and its presence means the patient has either CLT or autoimmune hyperthyroidism or Graves disease.  I can’t remember if I have ever had a patient with a delphian node who did not have positive thyroid antibodies.

Anyway, the differential diagnosis of primary hypothyroidism includes CLT, goitrogens (mostly iodine-containing products), familial inborn errors of metabolism (genetic abnormalities of the various steps to synthesis of thyroid hormones or their degradation), gland dysplasia (e.g., hemithyroids), and other rather uncommon entities.  I find thyroid antibodies most helpful when the TSH and FT4 come back normal (I can’t think of any reason to order TT4 but that’s a topic for another time).  The question becomes why is the gland enlarged (here we are assuming that the physician is skilled at telling when a thyroid gland is enlarged rather than there just being a prominent fat ring around the neck)?  It is still likely that the patient has CLT or possibly a so-called simple colloid goiter an entity I don’t understand; I don’t even know if the disorder exists even though almost all textbooks that cover thyroid disorders list it in the differential diagnosis of goiters (a goiter is just another way of describing an enlarged thyroid gland).  If the patient has CLT and enlargement of the thyroid, many endocrinologists will recommend treatment with replacement doses of L-thyroxine to “put the gland at rest.”  There are some data suggesting that such treatment can prevent progressive destruction of the gland which can occur; a number of studies have shown that TPO antibodies are cytotoxic even though most of the inflammation in CLT is lymphocyte-mediated.

So, it’s not so simple deciding what laboratory tests to order or not to order.   I will return to my original argument that whenever possible it’s good to know what specific disorder is being treated with medications, or  maybe, even if the treatment is just observation.  But I can sort of see the PCP’s point of view, sort of.

A few weeks ago, I read an interesting piece in the New York Times about errors that a national reference laboratory had made in measuring Vitamin D levels.  Apparently the laboratory had recently discovered (or had been found out?) that for some time it had been reporting erroneous results for serum vitamin D levels, resulting in either falsely high or falsely low values.  The problem was traced to inaccurate standards used in the assay.  Anyway, the report got me to thinking about just how difficult it can be for a patient to have any confidence that a test his doctor orders will provide an accurate result, let alone be interpreted correctly by the doctor.  I do know a little bit about such things given my background as both a clinician who has ordered  lots of laboratory tests and as a former director of a large reference laboratory.

Obstacles to getting an accurate test result

Many patients and health care professionals assume that the mere ordering of a laboratory test is just about all the assurance they need that the test will be performed properly and that the result will provide useful clinical information.  Nothing could be further from the truth.  At every step of the way, from procurement of the specimen, be it blood or urine or whatever, to the actual test report, things can and do go wrong.  For example, if the specimen was supposed to be collected with the patient in the fasting state but the patient had recently eaten, right away the meaning of the test results should be questioned.  If the specimen is not collected, transported, and stored properly before analysis, serious errors can be introduced.  If the assay is not performed properly, the results may be erroneous.  If the standards used to normalize the test results are wrong, the results will be erroneous.  If the normal range used in the test report has been improperly constructed, the test results might be interpreted incorrectly.  If the test results reported for one patient are really those of a different patient, the results will be erroneous.  Finally, if the health professional does get an accurate test result but interprets the result incorrectly, as far as the patient is concerned, it’s almost as bad as an erroneous test result.  In summary, there are many places between the ordering of a test and the reporting of the test result that things might go wrong.

Don’t get me wrong, I believe that most laboratory tests are carried out with great skill and that the results are accurate.  But one should never assume so and wise clinicians know that test results must be interpreted in context; if a test result does not fit with the clinical history, the result should be viewed with suspicion or at least skepticism as to its clinical significance.

Improving the odds that test results will be accurate

There are standard procedures called quality control/quality assurance that all laboratories are supposed to carry out to optimize test results.  These might include running certified standards according to a specified protocol and participating in an external quality assurance program such as those carried out by the College of American of Pathologists or the NGSP (the organization that runs the standardization program for hemoglobin A1c, the test that is widely used in diabetes).  Still, in the end it comes down to the health professional who ordered the test, using clinical judgment to interpret the test result, always reserving the right to question whether the number on the test report fits with the clinical impression.