VetInfo Digest                October 2001

 

This Month:

Algorithms (Diagnostic Plans)

Why use an algorithm?

What does an algorithm

look like?

What goes wrong with

diagnostic plans?

Do you need a diagnosis?

Can you read the map?

Thunderstorm Anxiety

 

This Month's Note:

I had some difficulty concentrating on answering questions last month. I know that many of you had similar feelings as we all watched the horror that was occurring in New York and in my state, Virginia. Perhaps the most difficult thing for many of us has been watching and not being able to help in any concrete, meaningful way. Some of our subscribers live in Manhattan and others were able to help in several ways with the rescue efforts following the attack on the World Trade Centers. The rest of us can only support this effort through our contributions. If you have been waiting to see how you can help the displaced pets and working dogs who are part of the tragedy, the best thing may be to contribute money to the organizations who are already in place in New York and providing shelter and aid for these dogs. These are three web addresses that should be reliable and are of organizations that are involved in the relief effort:

www.nsal.org

www.suffolkspca.org

www.libertyunites.com/aspca.adp

Thank you to all who have already provided support, money and supplies for the relief effort.

 

Algorithms

Last month I wrote several long lists of symptoms and potential causes for them. These lists can be very confusing without some sort of plan to sort out the probable causes for a symptom from the less probable causes. One way to do this is through the use of algorithms.

What is an Algorithm?

An algorithm is a diagnostic plan broken down into small steps, each with a "yes/no or true/false" answer. If the steps are followed in order and "yes" and "no" or "true" and "false" are answered correctly, the algorithm should be able to lead the practitioner from the symptoms exhibited to the disease that is causing them. Since algorithms have been written for most of the diagnostic challenges in veterinary medicine, it seems like making a diagnosis should be possible in almost all cases. However, this doesn't actually work out to be true. I will try to present some examples of algorithms for common problems in veterinary medicine and to explore why they work most of the time and why they don't work some of the time.

Algorithms are really well made for computer aided diagnostic efforts. In theory, by writing a good set of questions, it should be possible to come up with a diagnostic plan that a computer can follow to the logical conclusion, which would be the diagnosis. Computer aided diagnosis has not caught on in a big way in medicine, yet, but it is getting a little closer to reality. Some of the original computer studies were pretty interesting, because the computer tended to do better than most physicians in arriving at the diagnosis, when input was adequate, but was much less efficient (more diagnostic steps were often necessary) and was not as good as about 20% of the physicians whose case records were studied. There are probably a number of reasons for this, but a big reason is that almost every algorithm has at least one step in which "true" and "false" are not the right answers to the question -- where there has to be some interpretation of "mostly true" or "mostly false". This is where practical experience is invaluable and where computers and inexperienced diagnosticians both fail more than an experienced medical practitioner.

What does an algorithm look like?

This is an example of an algorithm for the clinical sign of vomiting:

1) Is there a known toxin exposure or medication being administered that might cause vomiting? If Yes, discontinue the medication (when possible) or treat for the toxin. If No, go to step 2:

2) Are there parasites present on a fecal examination that might cause vomiting? If Yes, treat for these and possibly go to step 3. If No, go to step 3:

3) Are the results of a blood panel normal? If yes go to step 4. If no, go to step 5:

4) Are there signs of a neurologic disorder, especially peripheral vestibular syndrome? If yes, treat this. If no, go to step 4a:

    4a: Is there evidence of pharyngitis? If yes, treat this. If no, go to step 4b:

    4b: Is there reason to suspect megaesophagus? If yes, confirm with radiographs. If radiographs are negative or answer is NO, go to 4c:

    4c: Treat for gastroenteritis until the problem resolves or other signs help indicate a new diagnostic path to follow.

5) Are the BUN and creatinine elevated on the lab panel? If YES, begin workup for kidney disease. If NO, go to 5a:

    5a: Is the glucose elevated? If yes, test for ketones in urine and if present treat for ketoacidotic diabetes. If No, go to step 5b:

    5b: Is the ALT, alkaline phosphatase or bilirubin elevated? If yes, start workup for hepatitis, including bile acid testing to try to rule out hepatic encephalopathy. If not elevated, go to step 5c

    5c: Are the BUN, eosinophils, lymphocytes and possibly potassium elevated? If so, test for hypoadrenocorticism. If not, go to step 5d:

    5d: Is the amylase or lipase level elevated? If yes, start workup for pancreatitis. If NO, go to step 6:

6) Is the white blood cell count elevated? If NO go to step 7, if YES go to step 6a:

    6a: Are X-rays suggestive of peritonitis? If YES, start workup for this. If NO, go to step 6b:

    6b: Is this an intact female patient? If Yes, go to 6c: If no, go to 6d:

    6c: Are X-rays suggestive of uterine enlargment? If yes, start workup for pyometra. If no, go to step 6d:

    6d: Is this an intact male patient? If Yes, go to 6e, if NO go to 6f:

    6e: Is there pus in the urine or an exudate of pus from the penis? If YES treat for prostatitis. If no, go to step 6f:

    6f: Is there increased sucrose absorption? If yes, look for gastrointestinal ulceration. If no, recheck the blood panel and start the process again.

There are several things that can go wrong with this algorithm and it misses one of the major causes of apparent vomiting in cats, which is coughing that resembles vomiting. Trying to rule out coughing is an easy step to add, but requires a judgment call, opening the door for errors.

The first place that this algorithm has problems is in the initial step, are toxins known to be present? It is nice if it is a known problem, but there can certainly be toxin exposure that the pet owner is unaware of. Since the treatment is specific for the toxin, in most cases, this is a good question to ask but it may be misleading.

The second step, ruling out parasites, also has problems. Many practitioners will skip the step of looking for parasites through a fecal examination. Sometimes this happens because there is insufficient stool available in the rectum to obtain a fecal sample when the owner has not brought one with them to the clinic. Other times it just seems very unlikely that parasites are present and so this step is skipped. After all, a little house dog with no exposure to the outside doesn't seem like a likely candidate for vomiting induced by whipworms and roundworms are such an infrequent cause of vomiting in older cats that routine fecal samples are one of the first places that costs are cut when a pet has the symptom of vomiting and the owner has financial constraints. Most of the time this works out OK, because parasites aren't a common cause of vomiting. However, in a few cases skipping this test results in a much more expensive diagnostic workup that can't provide an accurate diagnosis.

There is a big temptation among veterinarians to jump to conclusions, or at least to try a "trial and error" approach to diagnosis prior to running a general blood panel and white blood cell count. Gastroenteritis is probably the most common cause of vomiting and it often responds to very simple treatment, including simply withholding food for 24 to 48 hours in cats or up to 72 hours in dogs. Since this is inexpensive and many pet owners need an inexpensive approach to the diagnosis, the blood panel is skipped and treatment for gastroenteritis initiated. Often, this approach seems very reasonable to me. However, it does stop the diagnostic process and there are times when that will allow a more serious problem time to become worse. At this stage you may have to appear interested in reaching a diagnosis to stimulate your veterinarian to do a blood panel and blood cell count. Otherwise, he or she may assume you'd rather pay less and play the odds that your pet will be OK.

Once in a while there will be a physical exam finding that really points in a particular direction, such as a greatly enlarged prostate on rectal examination or a fever highly suggestive of an infection, along with obvious uterine enlargement. These types of things may also lead to a short-cut in the diagnostic process. For many patients and clients this works out to be a good thing, but there are times when the physical exam findings lead a veterinarian astray. The bad thing is that it is often difficult to get back on track in these cases as a veterinarian becomes "vested" in a particular diagnosis and has a hard time giving up on it. It is reasonable to ask that lab values be checked, or rechecked, at any time in a diagnostic process. Your vet may think this is unnecessary but usually it is possible to get your vet to run the lab work by pointing out that it is your money and that the test is unlikely to cause harm.

The costs of lab work start to elevate as you move through the diagnostic process. The inability to pay for a complete diagnostic work up is one of the most common causes of algorithms failing. It doesn't help to know with certainty that the next logical step is to take X-rays, if there is no money to pay for X-rays. At that point, it may be best for the pet for the practitioner to make his or her best guess and put the money that might have been spent on diagnosis into money that will be spent on treatment. It is enormously frustrating to know that a diagnosis is possible but to be restrained from obtaining it by the costs. Many times I wish that I could afford to pay for all the testing necessary to make a diagnosis when pet owners can't. Unfortunately, there is a definite limit to my monetary resources, as well. So we have to pick and chose carefully when to give a client assistance by paying for lab work out of practice funds. This failure in the diagnostic process is frustrating for both the veterinarian and client.

For some reason, many of my clients seem to assume that pets are less complex organisms than humans and that the diagnostic process should somehow be easier. I had two elderly ladies in my waiting room one day who were apparently partially deaf. They were discussing one of the women's health loudly enough that the conversation was easily audible in our exam room.

One woman said to the other one, "they have had my husband in the hospital all week running tests. I think they are up to their 23rd test now but they hope to have a diagnosis by tomorrow." Shortly after that, my receptionist showed her into the examination room.

I examined her dog carefully, weighed him and took his temperature. My receptionist had failed to write the reason for the visit on the travel sheet, so I asked what had prompted the visit.

"My dog is sick. What is wrong with him?"

I said that I didn't know but would be glad to start the diagnostic tests necessary to find out. The woman would have none of this. "What do you mean you don't know what's wrong with my dog? Aren't you a real vet?"

I mumbled something about how I thought I might be able to reach a diagnosis with less than 23 tests but that I would need to some sort of lab work if she really wanted to know what was wrong with her pet. She didn't. She left in a huff with the parting remark that we shouldn't send a bill since we were obviously incompetent. This is perhaps an extreme example, but I have lots of clients who stop the diagnostic process while waiting to decide if a particular test is really necessary or if I am just trying to rip them off. If your pet is very ill, it is best to go with suggested testing and then try to sort out the necessity later. If your pet is not so ill you may or may not need to rush into further testing. Give your vet a break, though. If you won't allow testing, it isn't reasonable to expect a firm diagnosis.

Pet insurance might be a good investment if you wish to be able to do all the diagnostic testing necessary to come to the most certain conclusion possible. This is especially helpful if you can afford to keep up with insurance but do not have access to large amounts of ready cash on an instant's notice.

The diagnostic process for vomiting doesn't include too many complex steps but it does have a couple of steps that might be hard to arrange for in some localities or in even in most veterinary clinics. Many clinics have in-house laboratories that are not capable of providing complete electrolyte tests (sodium, potassium, chloride, calcium, phosphorous). This complicates recognition of a couple of disorders but has the largest impact on the diagnosis of hypoadrenocorticism, which is an uncommon cause of vomiting but is important to recognize in young dogs because it is a life threatening illness.

Many algorithms contain at least one step that is difficult, or impossible, for the average veterinary clinic to perform. In the vomiting example, this test is the sucrose absorption test. To run this test, sucrose is administered orally and then urine collected for the next hour. This will probably require urinary catheterization. After urine is collected for one hour it is then tested for sucrose content. Most general practitioners have a hard time catheterizing female patients and don't have someone available just to watch the patient to ensure that the urine collection process isn't interrupted by the dog's actions. Worse, to the best of my knowledge, this test is not actually available through commercial laboratories at the present time. So why include it in the algorithm? It wasn't very important in the vomiting algorithm, but in many other cases, a test that is of limited availability or that requires extraordinary precision is a really important diagnostic step if an algorithm is to succeed in providing a pathway to an accurate diagnosis. This is one of the weak links in the development process for algorithms. Since most vets just skip these really difficult steps and proceed with their best guess as to what they would have shown, the diagnostic process often breaks down.

One of the problems with algorithms is that they get very complicated if all possible factors are included. A good example of this would be causes of vomiting that are age related. Even when the signs really point strongly towards prostatitis, this would be really unusual in a dog less than six months of age, for instance. Adding steps to factor in breed differences, species differences, differences based on the patient's sex (including whether the patient is spayed or neutered) would make the process much longer.

Laboratory errors can have a major impact on the outcome of a diagnostic process. A really striking example of this, which actually happened in our practice, is when the lab mixes up lab samples. We nearly treated a dog for diabetes once when the lab sent us blood results that actually belonged to a different dog. Fortunately, they recognized their error and called to say that the blood sample had been swapped with another dog's and the results were not correct before we initiated treatment. Our patient's increased drinking and urinating were due to kidney disease. Administering insulin could have been life threatening in this dog. More often, there is simply a variation of one lab value that throws off the diagnostic process. When lab results just don't fit the rest of the diagnostic process or when they just don't seem to fit the patient, it is best to consider rechecking the lab work.

While space limits prevent me from putting too many algorithms in this newsletter, I would like to include one more, which is a little more complex and a little less dependent on test results that are obtained from objective machines such as blood chemistry and blood count analyzers.

Veterinarians are often presented with itchy pets. Pruritis (itchiness) presents a unique problem for the veterinarian who is trying to follow a diagnostic plan. Most clients want the itch to stop right away and most veterinarians are aware of the fact that they can almost always stop pruritis by using corticosteroids. There are many times when I am standing across the exam table from a client who I know is going to leave and not come back if I don't provide nearly instant relief from the itching for their pet. Sometimes I know this because they have come to me shortly after visiting another veterinarian who tried to do the right thing by working through an itch control algorithm without making an immediate effort to stop the itching. I know at these times that if I cave in and use corticosteroids I am making the diagnostic process much harder. However, I also know that if I prove I can stop the itching if it is absolutely necessary the client might come back later and allow me to work through my itch control algorithm and to reach a firm diagnosis that will help over the long run. Follow the steps in the following abbreviated itch control algorithm and you can see where many clients lack the patience, or the determination, to reach a diagnosis.

The first step in an itch control process is to examine the pet for fleas. If fleas are present or it looks like they have been on the dog or cat recently, the best first step is to eliminate them and then wait to see if the itching stops. If it does, you are done. If not, go to step 1 in the diagnostic process:

1) Is there hair loss? If YES go to step 2. If not, go to step 3.

2) Do skin scrapings from hairless areas show mange mites? If YES treat for these. If NO, go to step 3.

3) Treat for mange. Did the itching stop? If YES, you are done. If NO, go to step 4)

4) Test for fungal infection (ringworm). If POSITIVE treat for ringworm. If NEGATIVE go to step 5. Note that the test takes up to three weeks since it is a culture test. During this time it is usually a good idea to go to step 5, anyway.

5) Administer appropriate antibiotics for at least 3 weeks to rule out a bacterial pyoderma. If this helps, it may be necessary to continue treatment for months. If it doesn't help, go to step 6.

6) Is the itching seasonal? If yes, go to step 7. If no, go to step 8.

7) Test for atopy (allergies from inhaling or contacting allergens such as pollens). Skin testing is still considered to be the best test but blood tests are improving and are reasonable to use if careful attention is paid to how likely a positive result is to be real in the area in which the dog lives. If allergies are present, treat for them. If allergies do not appear to be the cause, go to step 8.

8) Test for food allergy. It is necessary to feed a hypoallergenic diet and absolutely nothing else (without your vet's approval) for the entire period (usually 6 to 8 weeks) in which the food trail lasts. Not doing this can result in a failure to make an accurate diagnosis. If food allergies are present, treat for them. If not, go to step 9.

9) Test for immune mediated disease. This is easier said than done. Skin biopsies can be helpful in ruling in or ruling out this problem, as can lab work such as ANA testing. If this seems to be the problem, treat for it. If not, go to step 10.

10) Test for hormonal diseases. The most common ones associated with itching are hyperadrenocorticism (usually dogs over 9 years of age, uncommon in cats) and hypothyroidism (usually dogs over 6 years of age, uncommon in cats not being treated for hyperthyroidism). If hormonal diseases are present, treat for them. If not, go to step 11.

11) Treat for the itchiness without a diagnosis or start with step 1 again, skipping only those steps in which earlier test results seem to be certain.

It is pretty easy to write up a diagnostic algorithm for skin disease. It is much harder to ever be sure that it is followed and that useful information is generated by the testing procedures. In our practice, it is often hard for us to get past step 1. We have a number of clients who refuse to use the new and vastly superior flea control products because of the cost. For single pet owners and owners of just two or three pets, the refusal to use these products is often much more costly and much more irritating for the pet than to go ahead and use them. For multiple pet households flea control using Frontline (tm), Advantage (tm), Revolution (tm), or Sentinel (tm) or Program (tm) in conjunction with Capstar (tm) can get prohibitively expensive. This is sad, because these homes need it most. However, the bottom line for the diagnosis of the itchy pet is that if fleas can not be eliminated it is difficult to write an algorithm that has any chance of routinely succeeding in leading to a firm diagnosis for the cause of the itching.

There are many other pitfalls in the diagnostic process. It is still worth using algorithms to help in the diagnostic process, though. Having a diagnostic plan is like having a map on a road trip. You might be able to find your way without one but reading the map can save a lot of wrong turns.

I have used text based algorithms in this document so that I could explain some of the steps as we went along. This is how most algorithms are actually presented in the literature:

Algorithm for Elevated Liver Enzymes:

 

Thunderstorm anxiety

Dogs with severe thunderstorm anxiety are one of the most difficult behavioral problems for pet owners. Truly frightened dogs may tear up the house, injure themselves, run away in a panic or engage in other fear based behaviors that are extremely difficult to live with. We have had client's pets who jumped out of closed windows or through screen doors only to run panicked to a neighbor's house and then jump through the screen door there to get into the house. Cats sometimes exhibit fearful behavior in similar situations but usually try to hide and be still, which is less worrisome to their owners and this limits the number of cats who are brought in for treatment of this condition.

Recently we attended a continuing education seminar at which Dr. Sharon Crowell-Davis from the University of Georgia discussed animal behavior problems. She has a special interest in thunderstorm anxiety. She feels that it is worth attempting to decondition pets to the sounds of thunderstorms, since that is often the easiest effect of the storm to mimic, using thunderstorm recordings. However, this doesn't really reproduce all the things that happen in storms, such as high winds, drops in barometric pressure, rainfall, lightening and darkening of the sky. For this reason, just desensitizing the pet to the sounds of the storm may not stop the anxiety.

Most behavioral problems really require behavioral modification in conjunction with medications. Thunderstorm anxiety is probably best treated with both approaches, but for the truly panicked dog, it can be worthwhile to use medications even if behavioral modification is not possible. Dr. Crowell-Davis says that the best medication to control the panic attack associated with storms is alprazolam (Xanax Rx). The dosage for dogs is 0.02 to 0.04mg/kg every 4 to 12 hours and the dosage for cats is 0.125mg/cat every 12 hours. For dogs who consistently exhibit fearful behavior with storms, it may be best to use a medication such as amitriptyline (Elavil Rx) on a continuous basis and then supplement with alprazolam when storms are suspected. It is best to give this medication when storms are anticipated so that the pet is medicated when the storm arrives, but it can be helpful even after a storm starts, according to Dr. Crowell-Davis.

It is also important to avoid encouraging your dog's fearful behaviors, if possible. Do not console your pet, allow it to jump into your lap for comfort or inadvertently encourage other similar behaviors. It is possible to reinforce your dog's fears or to even reward the dog for fearful behavior by consoling it. This is especially important the first few times that a dog shows signs of fear in a storm. Let it find a place to hide or to sit near you if that helps, but do not facilitate the fearful response.

 

Please keep us in mind if you move, or if you change e-mail addresses. Every month a few of the VetInfo Digests are returned or we receive e-mail notification of a failure to reach a subscriber's mailbox. It is always possible to find the VetInfo Digest online but we really want you to get your individual copy, as well. Let us know your new address, postal or e-mail, so that we may continue to send the VetInfo Digest. If we can not reach an e-mail address for two months in a row we try to send it by post. However, some subscribers do not fill in a postal address when they subscribe. Conversely, we have some postal addresses without e-mail addresses, which can also be a problem. If you would like to send updated contact information, please send it to mervet@inna.net.

 

Thanks for Your Support!

The VetInfo Digest is published by TierCom, Inc., P.O. Box 476, Cobbs Creek, VA 23035. The opinions expressed in this newsletter are those of Michael Richards, DVM., author. Please send e-mail for Dr. Richards to mervet@inna.net

Copyright 2001, TierCom, Inc.

This page was last edited  06/20/04

 Subscriber homepage |Backissues  | Breed index | digestindex |

Vetinfo | vetinfo4cats| vetinfo4dogs 

| Canine  Encyclopedia |
  Feline Encyclopedia| VetInfo Digest |Links |Zoonotic info

Please 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