Are Medical Tests Overused?

     There have recently been a slew of articles about the overuse of  medical testing. Special attention has been directed towards radiological studies (including in-office ultrasound) and OTC testing kits sold to consumers. However, none of these articles takes the patients' views and wishes into account.

     One philosophical question arises when the patient asks for a test, blood or radiological, which the physician knows will be negative. Is there anything ethically wrong in ordering this test to give the patient peace of mind, assuming that the physician explains the probability of a false negative test and the problems and possible morbidity that can follow from acting on a false positive test? I think not. In some cases a negative test (e.g. a cardiac stress test in a patient suffering chest pain from a panic attack) will permit the physician to direct the patient's attention to the underlying problem. This is especially true in irritable bowel syndrome, where it has been my experience that it is very difficult for the patient to agree that the symptoms are purely functional.

     The thought of some diseases strikes such terror in patients' minds that they insist on screening tests, even though no study has shown that a positive test leads to a clinical intervention that either saves or prolongs life. Screening for ovarian cancer with the CA-125 blood test plus a vaginal ultrasound immediately comes to mind. OTOH, abdominal aortic aneurysms do enlarge, rupture and kill the patient (as in the case of Albert Einstein who refused surgical treatment). Medicare will pay for one abdominal USG in a patient's lifetime if the first test is normal. "Cost effectiveness" would limit this study to the highest risk group: male smokers over 65, especially if they have hypertension. However, although the group  incidence published in an article in Lancet was 11%, I am certain that if a patient was told that he/she had a 5% chance of having a possibly fatal condition, then they would want the ultrasound.

     We do know how to treat many infectious diseases, so screening for tuberculosis, syphilis, AIDS, hepatitis B and Hepatitis C, all of which can be fatal if untreated, would seem to make sense, although there are no studies as to how often these tests should be offered if negative. One would think that screening for chronic kidney disease, which problem can lead to dialysis, would also make sense, but in the latest issue of the Annals of Internal Medicine the United States Public Service Task Force has come out against this. We can't even show that screening for diabetes saves lives, but we are on firm ground when we state that treating hypertension prevents strokes. Oddly enough, although untreated high blood pressure can cause heart failure, there is no clinical study that demonstrates that treating hypertension prevents heart failure. And although "everyone" agrees that a chest Xray is not needed in a non-smoker with no pulmonary symptoms, many anesthesiologists insist on a chest Xray before surgery if general anesthesia is to be used.

     Every female of child-bearing age who comes to the emergency room is given a urine test for pregnancy. Should the same be done at every visit to a physician's office? Or before any Xray, including dental ones?
The prevalence of arteriosclerotic coronary artery disease at autopsy in 40 year old men is 10%. Should every man of this age have a stress-echo cardiogram as well as a thallium stress test? What about smokers and insulin-dependent diabetics and morbidly obese patients at age 30?

     It makes clinical sense to do a complete blood count, because anemia can be treated and its cause found, and many leukemias can be cured. Similarly, deficiencies of vitamin B12, vitamin D and iron  can be tested for and treated, as can elevated mercury, uric acid, and  calcium levels as well as  low magnesium levels. Whether patients who grew up in a country where parasites are endemic or who visited such countries should be screened for parasites has never been properly addressed, and such patients are usually only tested when they evince symptoms. I assume the physician would obtain the patient's permission before testing for such drugs as cocaine, amphetamines, opiates and benzodiazapenes. Another debated practice is that of obtaining a baseline EKG at the first office visit: no patient ever has a completely "normal" EKG, and it can make clinical sense to know what the EKG looks like before the patient's first visit to the ER with chest pain. And annual EKG's can detect the occurrence of a silent MI, but no studies have, to my knowledge, addressed the utility of this practice, although an unchanged EKG does bring a degree of mental comfort to the patient.

     I have saved the thorniest problem for last. There are more and more OTC tests sold for patients to test themselves not only for a particular disease, but for risk factors for the disease. This is a real problem if the disease in question has no satisfactory treatment. And unless the false positive and false negative percentages are explicitly stated on the package along with the prevalence of the disease in the patient's population, he may well be needlessly worried or reassured. Clearly these tests will continue to be sold, but the patient should be firmly instructed by the pharmacist not to jump to any conclusions before discussing the results with his/her physician.

     Let me close with an anecdote from my practice. When you donate blood, the Red Cross routinely screens for AIDS among other diseases, and confirms a positive screen with a Western Blot test which is the  "gold standard". The WB is not used for screening because it is more expensive and labor intensive, much as syphilis is screened for with the VDRL and confirmed with the FTA. In the past, the flu virus for the vaccine was grown in human cell culture, and because of cross-viral antibodies, patients who received the flu vaccine would test positive for the AIDS virus on the screening test but negative on the Western Blot; this false positive existed for three to six weeks after the flu vaccine. I received a panicked phone call one December from a patient of mine who was a fraternity member at a Florida university where the fraternities were competing with one another in a blood bank drive. He told me that he and 25% of his brothers had tested positive on the screen, and what were they to do? Thankfully I was able to reassure him about the false positive once I had learned that they had all gotten the flu vaccine the previous week, and could tell him to wait for the Western Blot result. Of course, he could never again give blood, because although the Red Cross would officially tell them the Western Blot was negative and that they didn't have AIDS, they still would not risk using the blood.

     

How to Interpret Medical News and Tests

     There have always been and will always be news articles and TV news stories about the results of medical tests. Some of these articles will quote (usually incompletely and incorrectly) from a study published in a refereed journal, some will quote from a talk given at a medical conference, and some will be derived from an interview with a doctor whose research has come to their attention. One major problem, of course, is that the non-scientist  has no way to judge the medical significance of such news reports, let alone whether or not the reported work fulfills certain basic requirements of scientific rigor. In this blog I will present some guidelines to help such readers and viewers evaluate the news report or interview.

     Firstly, there are no real secrets in medical treatments. Everyone publishes their research, and, if it is significant, tries to get on Oprah's show. Failing that, the PR department of their hospital will issue news releases to local and national newspapers and magazines as well as to the TV media. All of us would like to become famous, respected by our peers, and rich, no matter what our field of endeavor, and keeping our successes secret is not the proper avenue to achieve this. For instance in the field of cancer, each hospital has its weekly tumor board meeting where interesting, challenging or puzzling cases and/or results are presented and discussed. Then in most large cities there is a monthly oncology get-together. There also are quarterly meetings around the country as well as an annual meeting, which meeting is always well-attended by the media.

     Secondly, be careful about placing too much credence in any single report of a talk or paper delivered at a medical meeting. Such a report may or may not be clinically accurate, but until it is submitted for consideration to a refereed journal and is published, it is just one doctor's opinion. In my prior field of physics we recently had  newspaper reports of an experiment which purported to show that neutrinos traveled faster than the speed of light, but a careful re-analysis of the data showed that they did not.

     Third, it is essential to understand and appreciate the difference between relative risk and absolute risk. Absolute risk compares you to the entire world, e.g. you have a 10% chance of going bald if you are a man. Relative risk compares your subset to others in your group: if you are an American male, you have twice the chance of going bald than if you are Chinese. Researchers and reporters use relative risk to over-emphasize their results, and sometimes to stampede you into a certain behavior:  the relative risk number is always larger than the absolute risk number because a subset (the denominator in risk calculations) is always a smaller number than the entire set of humans. If your chance of being hit by lightning is 1/2,000,000, or one in two million, and I can reduce your risk to 1/1,000,000, the absolute statement is that I reduced your absolute risk by 1/1,000,000 , or 0.0001%. OTOH I can say with equal validity, that I reduced your relative risk by one-half, or 50%. Always look to see the absolute number, because only that number has immediate relevance to your case.

   Fourth, in connection with the degree of risk protection, we have the "number needed to treat", or NNT. That is, how many patients, based on the statistics of the study, would have to be treated by the drug or intervention in question in order to get one positive result (i.e. cure one patient, diagnose one disease, save one life, etc.). And note that they NEVER quote the NNS, or the number needed to treat to have one patient get  a side effect, and usually the NNS is quite smaller than the NNT. Would you take a pill to prevent a heart attack if they had to treat 10 people to save one person from a heart attack? What about 100, or 1,000, or 10,000 or 100,000? And would you still agree to treatment if the NNS were 2, i.e. 50% of those taking the medicine or other treatment suffered a side effect?

     Fifth, those patients who enrolled in a study group are probably not typical of the average patient, and the results quoted for them may not apply to you. The study group patients are highly motivated,  have a strong interest in staying healthy, and are committed to seeing the doctor when scheduled. Also their reporting of side effects may not encompass all possible side effects if a question about the possible effect is not included on the questionnaire. A case in point is that  a few men who took Viagra reported seeing a bluish tint to their field of view, a reaction that certainly was not looked for until it was first reported.

     Then we have the case of honest error. Many clinical studies are not borne out when they are repeated by another researcher or research group. This is in part due to the fact that with commonly accepted statistical medical analysis there is at least a 5% chance that the beneficial conclusion was reached by error.
Or it may be simply that the second study group differed in some important way from the first. There are many, many cases of the New England Journal of Medicine, or Lancet publishing back-to-back articles which come to opposite conclusions about a seemingly identical clinical treatment of the same medical problem.

     You should also be aware of the placebo effect. Many studies have demonstrated a placebo effect of 30% across the board. In other words, taking a sugar pill or an injection of normal saline can produce a beneficial effect.
This may be related to the fact that almost all patients feel instantly better as soon as the doctor walks into the examination room wearing a white coat. The human brain is a marvelous organ, and the mind-body interaction is always occurring, either on an unconscious or subconscious level. Believing can sometimes make it so, but we don't know how to engender that believing on a consistent basis.

     Finally, distrust all medical advertisements about the wonderfulness of the doctor or clinic or hospital doing the advertising. If they were that wonderful, they would not need to advertise. I especially dislike the ad (hospital name deliberately omitted) "Cancer. Where you're treated first can make all the difference." The insidious suggestion is that if you do not go to them for treatment and your cancer spreads you only have yourself to blame!!!

Medical Tests for Patients

     This blog is a follow-up to a previous blog on medical testing. Here I will discuss the philosophy that underlies most, but not all medical testing. As you will learn, many tests have no clinical evidence of  their usefulness, but that  may not stop a doctor from ordering it or a patient from requesting it.

     Firstly, I never worry about whether or not a test is "cost-efficient". My moral commitment is to my patient, and not to the amount of money his/her medical care may cost. That is not my decision to make. In other words, I am making a medical decision for one person and not for the 300,000,000 people who dwell within our borders, and I am sure you would want your personal doctor to think and act in the same manner.

     Secondly, even if I think a test is not needed, and am fairly certain that the test will be negative, if the patient wants it I generally order it. To me, the fact that the patient will be relieved when the test comes back negative is generally sufficient reason to order it. Often the test is requested because a spouse suggested it, or a best friend came down with a particular disease, or it was discussed on TV or read about on the internet. For instance, for a while the magnesium content of red blood cells was thought to be related to chronic fatigue syndrome, so patients wanted this measured. I do, however, refuse to do tests suggested by The National Enquirer.

     As I have mentioned in a previous blog,  if a patient comes in requesting an HIV/AIDS test because he/she is starting a new sexual encounter, I refer them to the Red Cross or their nearest hospital. I explain that not only will the blood bank test the blood for AIDS for free (as well as for hepatitis and a host of other blood-borne diseases) but that they will also be helping their fellow citizens by their donation of a unit of blood. I also suggest that their future partner do the same.

     Although there is absolutely no evidence that a vaginal ultrasound or the blood test CA-125 can detect ovarian cancer early enough to save lives, I will never deny a woman's request for these tests, because a negative test sharply reduces their worry/concern about having this dread disease.

     I have a moral antipathy to the traveling ultrasound trucks that pull up to a nursing home and offer to test the residents for narrowing of their carotid or femoral arteries or an abdominal aortic aneurysm They tell the residents that they are giving them a half-price special because Medicare does not pay for these screening tests. But then they give the results to the patient and tell them to consult with their personal doctor about any abnormal results. To me, this is equivalent to abandoning the patient, and should be outlawed. Any doctor who orders or performs a test on a patient is morally (and should be legally) required to do all the necessary clinical follow-up of any abnormal results.

     The same holds true for the total body CT scan that looks for calcium in your brain, your lungs, your coronary arteries and your abdomen. Again the patients are offered a "discount" and are told to follow-up with either their personal physician, or a physician at the hospital where the CT was done. The patient is never informed about the percentage of false positive tests, or if detecting calcium in a particular organ does indeed lead to an intervention that saves lives.

     No  comment is necessary  about the PSA blood test for putative prostate cancer.

     Patients should be made aware that the "normal" range for a given blood test is usually the average value of same sex adults plus or minus two standard deviations. This means, assuming the values of the blood tests are scattered "normally" (i.e. in a Gaussian distribution) throughout the test population, then for almost  any blood test 5% of those tested will have an "abnormal" lab test, i.e. a value outside the "normal" range WITHOUT having a medical disease.
It is trivial to show that it then follows that in a panel of 25 blood tests, the average patient has a 50% chance (one out of two patients) of having an abnormal blood test. I have not even mentioned that there are racial differences in blood tests as well. For instance, without any disease, the average white blood cell count of white American adult women is 4.5, while the average white blood cell count of black American adult women is 3.5, just as the normal hemoglobin count for adult women is lower than that for adult men, and the normal ESR (sedimentation rate) is higher for women than for men, and teenagers have a "higher than normal" alkaline phosphatase because their bones are still growing. To repeat: you can have an "abnormal" blood test without having any illness at all, much in the same way that any plain Xray of the neck or lumbar spine of every adult 40 years old or older will always show "arthritis", even when the patient is pain-free.

     The public is also blissfully unaware that the cutoff value for certain blood tests is determined (and not unanimously) by a group of doctors in Washington, D.C. much as one year the American Psychiatric Association voted that homosexuality should no longer be considered as evidence of a mental disease.
When I was in medical school, the upper value of "normal" fasting glucose was set at 140. It has since been lowered to 120, then 110, and most recently 100, with patients between the "normal" level and a glucose of 200 have been labeled "pre-diabetic". This label has increased their life insurance and long-term care insurance premiums without any demonstrated improvement in their health.

     I will close by mentioning that not all clinical laboratories are equally skilled in measuring all lab tests (e.g. N-terminal parathormone), that some "normal" values for the same test are different for different labs, and that the first thing that should almost always be done with any abnormal blood test is to repeat it, rather than to automatically assume that the patient is ill.