VetInfo Digest November 2000
This Month:
Interpreting Test Results
Canine Leishmaniasis
New Surgery for Hip Dysplasia
Compounding Pharmacies
Administering Eye Medications
In our practice area, late fall and winter are usually very slow times. For some reason, probably having to do with Murphy's Law, there has been no slow down at all this year. At the same time, the VetInfo Digest subscriber list has been growing steadily. I am still hoping for some quiet time to help me catch up in the next couple of months but I apologize for the slower than usual response time to subscriber questions over the last month. I do not know how long this situation will continue but we will continue to answer every question, to the best of our ability and as fast as possible. All I can offer, if this is not good enough, is to refund the price of subscription. It continues to be our policy to offer refunds upon request, no questions asked.
Please be careful around Thanksgiving and Christmas when preparing large meals and inviting overly generous family and friends who will feed the pets too much. Every year we treat at least one or two pets for severe gastrointestinal distress from too much holiday foraging or for constipation from eating the turkey carcass. Let your guests know that feeding the pets too much really might make them ill.
Sensitivity, Specificity and Predictive Value -- Interpreting Laboratory Tests
It is possible to test for a wide variety of diseases at the present time. Many of these test procedures are highly accurate and others have limited usefulness. Even the tests that are very good have to be used appropriately in order to trust in the results, though.
On our web site, at http://www.vetinfo.com/catlab.html, we had the following question and answer:
Q: My kitten had a positive ELISA test today and I am going crazy thinking about what this means. My vet told me it does not look good. Please tell me what the chances are of a false positive. I can only hope but I can't bear the thought of losing my baby.
Julia
A: Dear Julia
The possibility of a false positive test would depend on what was being tested for. Feline leukemia virus tests are pretty accurate but probably have false positive results one to four times per 1000 tests. The feline leukemia tests are usually directly for viral antigen. Some kittens can probably fight off the virus completely and many others can suppress it sufficiently to keep it out of the bloodstream and not be affected by the virus unless they experience later immune suppression for some reason. It is always a good idea to retest 3 weeks or more after a positive feline leukemia test to see if the kitten has become negative. I think that the false positive rate is higher when using saliva or tears for testing than when blood is used.
Feline immunodeficiency virus tests are more likely to be false positive, especially in a kitten. This test is for antibody and maternal transfer of antibody can occur. This gives a positive test result and the kitten may not be infected.
Feline heartworm tests using an antigen detecting system have false positive rates similar to feline leukemia tests. Feline heartworm tests that detect antibody to the heartworms have an unknown false positive rate because no one really knows how many cats are exposed to heartworm larvae without ever becoming infected. A positive heartworm test in a cat should be evaluated in conjunction with the presence of clinical signs or other test results such as X-ray changes or ultrasonic identification of heartworms.
Your kitten probably had a feline leukemia test since it is done most frequently in kittens of these tests. Retest when your vet thinks it is best to do so and if the second test is positive consider having that confirmed with an alternative test such as IFA testing. If your kitten is still positive, you will have to make plans to deal with that situation.
Mike Richards, DVM
We got a note from a veterinary student about this exchange of information. Her note:
Dear Dr. Richards,
While researching for an epidemiological project on predictive value of FeLV tests, I came across your answer to someone named "Julia" who had concern over her cat's positive ELISA test (http://vetinfo.com/catlab.html). You told her that the probability of a false positive result on an FeLV test is about 1/1000 tests, which would be 0.1% rate of false positives. While this is approximately the correct rate of false NEGATIVE on an FeLV test, it is not anywhere close to the rate of false positives according to the papers I have read recently.
In the article "How Well do Serologic Tests Predict the Infection or Disease Status of Cats?", by Richard H. Jacobson, PhD, the author wrote:
"When a manufacturer claims, for instance, 97% specificity and 97% sensitivity for an in-house test kit, can be surmised that an incorrect result will be obtained only about 3 of 1000 times? The answer is.... No." "The terms sensitivity and specificity are often used erroneously to predict whether a positive or negative result will reflect the disease status of the animal. They do not... The prevalence of infection/disease also must be known or estimated..."
The article goes on to explain that if the incidence of disease is very low, then even an excellent test gives a high percentage of false positives. It lists a chart, and if you look at the predictive value for a test with 95% specificity and 99% sensitivity (such as the FeLV ELISA), and a 5% disease prevalence (which is even higher than the estimated 2% US rate for FeLV infection in the general cat population), then the predictive value of the test is only 51%. That is to say, about half of all positive results are false positives. The negative predictive value, i.e. the number of false negatives, is apparently much more resistant to the effects of disease prevalence and is likely to truly be a negative result. (This article was published in the JAVMA, Volume 199, No 10, Nov. 15, 1991).
Another article, "Interpreting Feline Leukemia Test Results", by Joseph Romatowski, which lists a similar chart specific for FeLV tests, said "Positive ELISA test results obtained in a screening program should be interpreted with caution, because a percentage --- approximately 72% -- of such as likely to be false positive results. On the other hand, routine screening is an excellent method for ruling out FeLV viremia, because a false negative result is likely to be obtained in only 1/1000 cats." (This article was in JAVMA, Vol 195, October 1, 1989).
Although they are somewhat dated, their statistical information holds true. Both articles illustrated their statistical calculations and I found the information both surprising and extremely useful. I thought you might, too. And Julia might well have a healthy cat!
Andrea Sparzo
DVM Class of 2002
ELISA stands for "enzyme linked immunosorbent assay". This is a test procedure in which antibodies or antigens are identified by being bound to enzymes that are linked to antibodies against the suspect antigen or antibody. IFA stands for "immunofluorescent antibody" testing and is a test in which a positive sample is identified by the presence of fluorescence.
It is very important to understand the values of test procedures in veterinary medicine. The advantages of proper testing for diseases and disorders are numerous, but many tests must be interpreted in correlation with the clinical signs and it is important to understand the reasons behind rises in levels of certain chemistry tests or the potential meaning of serum titer levels. To do this requires some understanding of the way that testing works, the definitions of test standards, the prevalence of a disease in a population and the presence or absence of clinical signs or disease exposure in the individual patient being tested.
The term "sensitivity" refers to the ability of a test to detect a disease or disorder. A test that has 99% sensitivity will detect the presence of a disease 99 out of 100 cases. Sensitivity is easy to obtain. To give you an extreme example, if a test always came up positive, no matter what the actual disease status of the patient was, it would be 100% sensitive -- it would detect every single case of a disease. Obviously, this would not be very useful, though.
The term specificity refers to the ability of a test to accurately detect a specific disease in a patient, or the "correctness" of the results. A test that is 97% specific will correctly identify a patient infected with a particular disease 97 out of 100 times.
The "predictive value" of the test is the likelihood that a positive test result is actually positive in the population being tested and is almost independent of the accuracy of the test itself.
So if a test is 99% sensitive and 97% specific, it would detect almost all of the patients positive for a disease, but 3 out of every 100 patients testing positive would have false positive test results.
In the case of feline leukemia test kits at the current time, the estimated sensitivity of the tests is a little better than 99% and the estimated specificity of two of the most commonly used tests are 98% (Witness FeLV tm) and 97% (CITE Snap FeLV tm). So the first thing to note in the answer I gave to Julia is that the actual false positive rate is more like 3 in 100 than 1 in 1000. This was an error in reporting the test result data. From that point, the situation becomes much harder to understand so I will try to give some examples of how tests might be used and what the results would mean, based on those uses.
It is hard to decide which is more important, high sensitivity or high specificity. Obviously the perfect situation is 100% sensitivity matched with 100% specificity, but there are no tests that I know of in veterinary medicine with this degree of accuracy. Almost all tests have some variance from 100% in either sensitivity or specificity. It is usually necessary to sacrifice accuracy in sensitivity in order to gain accuracy in specificity.
The important thing to remember about the two articles that Ms. Sparzo quoted from is that they are addressing the use of ELISA tests for feline leukemia virus used as screening procedures. A screening procedure is one in which many patients are tested in order to identify a few who may be infected with a particular disease. The three most common screening tests in veterinary medicine are probably feline leukemia virus testing, heartworm testing and fecal examinations. Due to the statement "My vet told me it does not look good," I was assuming that this was not a screening test and that the kitten was showing signs of disease, but that assumption could be wrong.
Fecal examinations probably have a sensitivity in the 70% range but have a very high specificity, since the presence of pathogenic worm eggs in the stool is virtually positive proof that there are adult worms producing the eggs in the patient. In most cases, it is possible to assume that a patient with worm eggs of a pathologic species of worm, found on fecal examination, is in fact infected with these worms. However, finding a negative stool sample does not rule out the presence of worms, since the test is not very sensitive to them. For this particular test, repeat samples may be necessary to try to detect worm eggs or intestinal parasites that are only intermittently present in the stool, since there is a high rate of false negative test results.
It is hard to determine sensitivity for ELISA or latex-aggultination based heartworm test kits. The sensitivity of the test varies based on the number of heartworms present in the dog's body. Overall, though, the sensitivity is probably greater than 95% with poor sensitivity for 2 or less heartworms and greater sensitivity as the number of worms increases. The specificity of heartworm test kits is very good, probably very close to 99.9%, making the false positive rate for these tests only about 1 in 1000 dogs.
If a test only gives false positive results in 2 out of 100 patients, how can there be a situation in which a positive result is wrong over 50% of the time? Suppose that an alien from some distant planet landed on earth and quickly transformed into human shape and then melted into the general population. On examining the spaceship the alien arrived in, it was discovered that he had plans to destroy the world in 100 days. By collecting samples from inside the spaceship a test is devised that will detect the alien and it has a specificity of 99.99999% -- it only has a false positive rate of 1 in 1 million tests. Everyone in the world is tested in order to detect the one alien. During this testing, despite the fact that the test itself is very good, with only one false positive in 1 million people, approximately 6100 humans will be falsely identified as aliens. This happens not because the test is inaccurate -- it is in fact extraordinarily accurate -- it happens because the population being tested does not harbor many aliens. If you look at the predictive value of this test in the population, it is less than 1% accurate, even if does detect the alien. To look at this from the other vantage point, suppose that a test is developed that will identify human beings, with similar specificity and sensitivity. In that situation, a positive test result, indicating that the patient tested was in fact human, would be extremely likely to be accurate, since there are over 6 billion humans and only 1 alien. In fact, just guessing that the person you are talking to was a human would also be extraordinarily accurate, as long as you stuck to a positive answer for the question.
Given the differences between sensitivity, specificity and predictive values for various test procedures, how easy is it to evaluate a positive test result on a feline leukemia or heartworm test?
There are veterinarians who believe that it is best to test every kitten for feline leukemia virus. In this testing situation, the likelihood of a positive feline leukemia test being a "true positive" is probably close to the 51% chance quoted in one of the articles above. On the other hand, if a kitten comes from a group situation, especially a high stress group situation, and has clinical signs of illness, such as fever, discharge from the nostrils, poor appetite, poor weight gain or other signs indicative of disease and this kitten has a positive FeLV test, the likelihood that the test is a false positive is much lower. In this situation it is reasonable to assume an accuracy rate of 80 to 90%, or better, simply because the patient is now in a higher risk population.
Heartworm testing on a regular basis is a routine part of veterinary practice. The specificity of heartworm tests is greater than the specificity for FeLV tests. In the Southeastern United States, the prevalence of heartworm disease in dogs that are not on heartworm prevention approaches 100% in some areas and probably averages at least 80%. In these areas, a positive heartworm test is very likely to be accurate. On the other hand, the prevalence of heartworm disease in the Northwest is very low -- probably less than 1 in 1000 dogs have heartworm disease in these areas. This makes the heartworm test much less likely to give an accurate result in these areas. If the test has a 1 in 1000 false positive rate and there are likely to be 10 positive dogs in a state with a dog population of 100,000, the test would only have a 10% positive predictive value. The math, broken down, works like this: there are 100,000 dogs. A false positive test result of 1 in 1000 would mean that there will be 100 false positive test results in this population every year (assuming yearly testing) while the actual infection rate has been shown to be 10 positive dogs. Assuming that all of the true positive dogs also test positive for the disease, then there will 110 dogs in the population that test positive but only 10 of these, or about 9%, are actually positive. This means that positive test results are wrong, in this population, about 91% of the time, even though it remains true that the test itself gives accurate results 99.9% of the time. This is the difference between actual test accuracy, as measured by sensitivity and specificity and the predictive value of the test procedure. The predictive value includes additional information that helps to define the value of the test in a particular population of patients.
This is a very confusing issue. In the case of feline leukemia testing, there is an alternative test, the IFA test, which has a much lower sensitivity (misses a lot of positive cases) but has a very high specificity (nearly 100% of cases reported as positive are in fact positive). This was the basis for the advice I gave Julia to retest in a few weeks and if the test remained positive to consider the IFA test before making any decisions. There are many diseases in which the screening test, though very accurate, has just enough lack of specificity to allow for major problems when testing large populations of pets. Having a second test that may be lower in sensitivity (misses a few positive cases) but very specific (doesn't have many mistakes in identifying the problem), is very helpful.
There are many other examples of this testing problem in veterinary medicine. It has been estimated that in a blood chemistry panel consisting of fifteen or so blood chemistry values, there is a very high chance that one of the values is off by at least 10%. For this reason, when test results do not seem to correlate with how the patient feels, or with the clinical signs associated with a disease, it is important to consider the possibility that they are not accurate.
This would all be just an interesting discussion on statistics if it were not for the fact that the results have very real effects on pets. The medications used in heartworm treatments are irritating to the tissues they are injected in, are costly and cause toxicity problems in some patients. Feline leukemia positive kittens are sometimes euthanized or returned to shelters or breeders where their chance of euthanasia is high. For this reason, a great deal of thought should be put into the value of testing procedures before they are done in situations in which the risk of a disease is low and there are no clinical signs or other strong indications that it is present. If testing is done, it is important to remember to evaluate the results based on the likelihood that the problem exists and to confirm test results in any situation in which they are questionable. It is aggravating to pay for testing several times but it may be the difference between life and death for some patients.
I appreciated Ms. Sparzo's letter, as it is important to present the whole picture, as well as that can be done, when discussing test results. Just because a test is highly accurate when evaluated in a laboratory situation does not mean that it will be used appropriately in the real world -- or that appropriate use won't still be misleading in some situations.
There is an old joke in which a person is noticed to be banging two sticks together while walking through the woods along the Appalachian Trail (or any other U.S. location). A companion finally asks what the person is banging the two sticks for. "To keep away elephants", is the reply. "There are no elephants here!" replies the companion. "See, it's working!" says the stick banger. To test for something that isn't likely to be there, even with very accurate testing, may be an example of exercising similar logic. With only a small percentage of testing inaccuracy, in a large population, it is possible to find a disease that really isn't there. For diseases such as feline leukemia and canine heartworm disease, in which there are potentially serious ramifications of a false positive test, it is best to be very careful in considering the overall picture when making decisions based on the results of a single test procedure.
Canine Leishmaniasis
Previously in the VetInfo Digest we have mentioned that leishmaniasis has been reported in the United States. At the present time, there have been cases of leishmaniasis in foxhounds from 21 states and also from southern Canada. This disease causes crusty skin lesions, weight loss, joint swelling, kidney failure, nose bleeds and swollen lymph nodes, although not all symptoms will occur in every patient.
Up until recently, all of the reported cases of leishmaniasis in the United States have been in foxhounds. In other countries this disease is spread by biting sand flies but in the U.S. the vector for the disease has not been clearly identified. Recently, there have been reports of positive tests for this disease in dogs that are not foxhounds and that have not associated with a foxhound in the recent past. While this disease is still very rare, we will try to update you on its spread and any information concerning control of the disease as time goes on. Foxhounds and dogs that might be exposed to foxhounds at field trials or shows have a higher risk of exposure and should be watched more closely for signs of leishmaniasis.
New Treatment for Canine Hip Dysplasia
A new surgical treatment for hip dysplasia has been developed by veterinarians at the University of Wisconsin. This surgery must be done in young puppies, usually between 16 and 20 months of age, who are already showing signs of hip dysplasia. This surgery is said to produce a similar effect to a triple pelvic osteotomy, in which a section of the pelvis containing the acetabulum (the socket of the hip joint) is rotated to provide better coverage of the ball of the hip, lessening future arthritic changes. The newer procedure uses radiosurgical techniques to cause arrest of growth at the pubic growth plate, which causes the angle of the remaining pelvic sections to change and produce a similar effect to a TPO surgery with much less trauma to the patient. While use of this procedure is not widespread at this time, it is likely to become more common if the puppies who have been treated so far do well over time.
Compounding Pharmacies
We have had several inquiries over the last couple of months about medication availability. There are certain medications which have gained acceptance in veterinary medicine which do not come in an FDA approved form or are simply not available due to manufacturer withdrawal or prolonged backorders. In other cases, the available form of a medication is difficult to administer and an alternative, such as a liquid or capsule form is desirable.
Potassium bromide is one example of a medication that is commonly used at this time but for which there is no FDA approved product. Another example is diethylstilbestrol (DES), which is used to control incontinence in female dogs. These medications may be compounded, or manufactured, by pharmacies equipped to handle the manufacturing process. Such pharmacies are referred to as compounding pharmacies. Producing medications used to be common practice in pharmacies but many are no longer equipped to do this. One of the ways in which independent pharmacies are competing with big chains is to offer compounding services, which has made this service available in most areas of the United States. For those areas without local compounding pharmacies, there are a number of pharmacies that will ship products directly to a consumer, or their veterinarian, on receipt of a prescription for the necessary medication.
I should note that some medications are easier to administer if they are put into capsules, to disguise bad taste or to combine multiple small pills. It is possible to buy empty gelatin capsules from many pharmacies and your vet can order them for you, as well. This really isn't compounding but it is a option that does make some difficult to administer medications a little easier to give.
Administering Eye Medications
Many of our clients have a very hard time administering eye drops or eye ointments. It helps a great deal when attempting to administer eye medications if you know that the upper eyelid is the one that has the most motion. Watch your dog or cat as he or she blinks and you will notice how much more motion there is in the upper lid than in the lower lid. When you need to administer an eye medication, placing the thumb of whichever hand you wish to use on the upper eyelid, right about the level of the bony rim of the eye socket and retracting the lid, works much better than trying to pull down on the lower eyelid. It does not matter if the eye drop or eye ointment is applied to the cornea (the clear part of the eye) or the sclera (the white part of the eye), as long as it is applied to the eye or on the inner surface of the eyelid.
When applying eye medications for corneal ulcers, in either dogs or cats, it is better if one drop of the medication is used than two drops, in most cases. This may not make sense at first, but studies of eye drop application show that two drops stimulate enough tear production that there is a bigger washout of the medication than when one drop is applied.
The best artificial tears for use in the eyes of pets, among the readily available brands, are Genteal (tm) drops, which contain a preservative that dissipates as soon as it is applied, making it less likely to cause eye irritation. These can be helpful in comforting eye irritation from herpes virus or from allergies. For dogs that have tear deficiency disorders, artificial tears are usually not adequate to control pain, discomfort and scarring of the cornea. Cyclosporin ointment ( Optimmune Rx) or 1% cyclosporin drops are the best medication for tear deficiency disorders in dogs.
Thank you for your support of the VetInfo web site and VetInfo Digest.
The VetInfo Digest is published by:TierCom, Inc.
P.O. Box 476
Cobbs Creek, VA 23035.804-725-5051, Fax 804-725-0149
email: mervet@inna.net or vetinfo@vetinfo.com
The opinions expressed in this newsletter are those of Michael Richards, DVM., author.
Copyright 2000, 2001 TierCom, Inc.
This page was last edited 06/21/04
Subscriber homepage |Backissues | Breed index | digestindex |Vetinfo | vetinfo4cats| vetinfo4dogs
| Canine Encyclopedia |
Feline Encyclopedia| VetInfo Digest |Links |Zoonotic infoPlease do not send anything in attachment form. We will not be able to open it due to security risks.
This page is authored by
Dr Michael Richards, DVM
and produced by TierCom, Inc.
Opinions expressed are those of Dr. Richards.
Designed and edited by Michal Justis
copyright ©1996,1997,1998,1999,2000,2001,2002,2003,2004- TierCom, Inc