Endodoc

Every medical student has heard the old adage (which, by the way, I can’t find in any anthology of famous quotations): “When you hear hoof-beats, think of horses, not zebras. ” What this means is that common medical disorders are more frequently encountered than rare ones. For example, if a patient comes in with a “runny nose,” a physician should think of an upper respiratory illness (i.e, a common cold) before considering the possibility of a cerebrospinal fluid leak. Unfortunately, many medical educators often spend much more time focusing on those zebras than the horses during medical school and residency training.

Our medical school has an excellent “problem-based learning approach” in which medical students take case history examinations at the end of each “learning block.” A recent examination was about a patient with gastrointestinal symptoms who turned out to have a very rare disorder (Whipple’s disease, if you are interested). I was a grader for the examination and enjoyed  one student’s discussion of his differential diagnosis in which he wrote something to the effect that “the diagnosis must be Whipple’s disease since it’s so rare and the type of diagnosis that might turn up on a problem-solving examination.” How right the student was. Even if he becomes a gastroenterologist, it is unlikely that he will ever encounter a case of Whipple’s disease.

How to tell a horse from a zebra

I’m not saying that it is always so easy to make diagnoses and that we should not consider uncommon disorders as part of differential diagnoses. Rather, that physicians should, if it’s not an immediate life or death situation, first focus on the most likely reasons for the patient’s signs and symptoms than on all of the possibilities, no matter how remote the likelihood. Such a “staged” approach can save both an incredible amount of money and patient discomfort, or even risk of complications from this or that diagnostic procedure or empiric therapy. More and more I am finding young physicians “throwing the kitchen sink at” a set of patient signs and symptoms rather than first thinking things through and maybe looking to see if there are any horses in the pasture.

In my next entry, we’ll come back to this question of common versus rare disorders as it relates to the diagnosis of congenital hypothyroidism.

Sarah D is an 8 year old female who had been in good health. She was successfully toilet-trained at age 3 years. Over the past 2 weeks, she had wet the bed on 4 separate occasions. There was no history of any day-time “accidents.” Sarah’s mother called the pediatrician’s office and scheduled a clinic visit.

The pediatrician, who had cared for Sarah since her birth, obtained a detailed medical history. In summary, the child had been quite healthy and the bed-wetting seemed to be the only problem. The “review of systems” revealed no history of excessive urination except for the night-time bed-wetting which is also called enuresis. There was no history of excessive thirst, fatigue, or any emotional problems that the patient’s mother could identify. There was no family history of members with childhood enuresis or with kidney diseases. The physical examination was unremarkable including a brief genital examination that revealed no evidence of inflammation except for slight erythema of the thighs, suggesting a mild “diaper dermatitis.”

So, what’s going on? What did the pediatrician think?

Sarah’s pediatrician scratched her head after obtaining the medical history and performing the physical examination. Clearly, the enuresis was not a normal variant- a finding that might be expected as part of the developmental process. Many normal children do take a long time to be consistently dry overnight and this is often genetic. But, to become completely dry for a number of years and then to start wetting the bed again was not normal. The physician’s first thought was to order some lab tests but she had the good sense to first put together a differential diagnosis.

She considered the following disorders as most likely: urinary tract infection or some other kidney or bladder problem; diabetes mellitus- a high blood sugar would lead to sugar in the urine which would “pull” water with it resulting in increased urination; psychogenic water drinking- a condition where an individual just consumes a large amount of liquid, as a habit or as the result of a psychological disorder; and diabetes insipidus- a disorder usually caused by a deficiency in the hypothalamic hormone vasopressin (stored in the pituitary gland), which controls water reabsorption by the kidneys; the kidneys filter the blood and if most of the water in the blood were not reabsorbed, it would mean rapid dehydration and death. Diabetes insipidus, or “DI” as it is commonly called can be caused by quite a number of medical conditions including head trauma, brain tumors, brain inflammatory processes, and for no apparent reason (in this instance it is called “idiopathic”). The pediatrician thought the most likely possibilities were a urinary tract infection or diabetes mellitus.

What did the pediatrician do?

After putting together a “rough” differential diagnosis, the pediatrician went back and obtained more medical history. Specifically, she found no history of head trauma, headaches, or other neurologic symptoms (findings that might suggest diabetes insipidus), no history of excessive fluid intake compared to other family members,and no history of urinary tract infections.

Next the pediatrician ordered a few simple laboratory studies including the following: routine urinalysis, blood chemistries to include electrolytes, BUN (a measure of a person’s state of hydration and kidney function). She also ordered a serum calcium because she remembered that a high serum calcium or a low serum potassium (obtained with the electrolytes) could mimic diabetes insipidus. She thought about but decided to wait on head imaging studies (e.g, MRI, CT scan) until the initial lab test results were in. A plasma glucose came with the blood chemistries but she knew that if the increased urination was from diabetes mellitus, the urinalysis would show glucose (normally, the urine is glucose-free).

What did the initial lab tests show?

The urinalysis was “clean,” with no evidence of a urinary tract infection or diabetes mellitus. The urine specific gravity (SpG) , a measure of the degree of concentration or dilution of the urine) was 1.002, a very low value, consistent with either normal kidney function, psychogenic water drinking, or diabetes insipidus. In DI the deficiency (or poor action) of vasopressin prevents the kidneys from concentrating the urine, excreting waste products while reabsorbing most of the water. A urine SpG of 1.010 is roughly equivalent to the concentration of blood. The maximum urine SpG is about 1.035 indicating kidney reabsorption of as much water as possible as well as plenty of vasopressin and normal kidney function. It would not be unusual for a normal person to have a urine SpG of 1.002 if he had been consuming a large amount of liquid.

The blood tests showed normal electrolytes except for a slightly high sodium (145 meq/L), slightly high BUN (25 mg/dL), normal calcium (9.3 mg/dL), and normal glucose (84 mg/dL).

What does all this mean and what should be done next?

The pediatrician reviewed the laboratory test results and concluded that Sarah might well have diabetes insipidus but definitely not diabetes mellitus or a urinary tract infection. The physician next sent me an e-mail to discuss the case and to ask if I would carry out the necessary studies to confirm the diagnosis and to initiate treatment. Thus, I scheduled the patient for admission to hospital for a water deprivation test. This test is exactly what its name implies- the patient is deprived of water until he shows an ability to concentrate the urine or that he definitely has diabetes insipidus. Since the goal of the water deprivation test is to induce some degree of dehydration, the test can be dangerous. Thus, I generally admit patients to hospital for the test. I generally admit the patient under a category called “23 hour observation status” which means the patient will not be charged for a hospital day if I can discharge them from hospital in less than 24 hours (I have no idea how many hospitals offer this option).

The water deprivation test

So, I admitted the patient to hospital late one afternoon. I obtained my own medical history and performed my own physical examination which added nothing to the pediatrician’s initial assessment. I then ordered baseline laboratory tests which included a urine specimen for SpG and osmolality (a more specific test for urine concentration- it ranges from about 200-1200 mOsm/kg water, roughly equivalent to urine SpG ranging from 1.001-1.035), blood chemistries, and serum osmolality (normal is 280-290 mOsm/kg water; a level of 300 or more without highly concentrated urine would confirm a diagnosis of diabetes insipidus). As part of the physical examination, I had a baseline body weight.

The test results showed about what the pediatrician had found; the serum BUN and serum sodium levels were borderline elevated. The serum osmolality was slightly elevated (300 mOsm/kg water) while the urine value was low (300 mOsm/kg water). Since the patient did not normally drink any liquids after going to bed, I restricted all food and drink after supper, at which time the water deprivation test formally began. The patient was to be supervised at all times. We repeated the body weight. We did not insert a urinary catheter even though we thought the patient might wet the bed- we suggested that she be encouraged to urinate in a collection bottle every 4 hours or so during the night (if we had been desperate to collect all urine, we would have inserted a urinary catheter). If the patient had given a history of drinking liquids throughout the night at home, we would have allowed this because of the risk of dehydration.

In the morning, the patient was feeling well but a bit thirsty. She had urinated (in the bottle) about 1000 ml , about a quart. Her body weight was down about 2 pounds (consistent with the urine loss). The urine SpG was 1.005 and the urine osmolality was 300 mOsm/kg water. Blood chemistries showed definitely elevated serum sodium and BUN levels. The serum osmolality was 310 mOsm/kg water. We had a diagnosis!

The next step

First, the patient was given access to water but asked to drink no more than 1 liter or so over the next hour. The patient was then given a subcutaneous injection of aqueous vasopressin, called pitressin. We continued to collect urine. Within 1 hour, the urine SpG had risen to 1.025 and the osmolality to almost 1200 mOsm/kg water. Thus, we learned that the patient not only had diabetes insipidus but was responsive to vasopressin. The test was discontinued. A synthetic form of vasopressin called d-desamino arginine vasopressin or dDAVP was prescribed to be given as one nasal squirt at bedtime (the medication dosage depends on the degree of vasopressin deficiency and weight but can range from 1 squirt per 24 h to as many as 3 or 4).

The next step

The patient responded well to the dDAVP and the enuresis did not recur. In retrospect, the parents noted that the Sarah drank much less liquid after starting the dDAVP nasal spray. Unfortunately, our work was not done; we needed to determine why Sarah had diabetes insipidus. Neurology and ophthalmology consultations were ordered. Head imaging studies were ordered (MRI and CT scan). To make a long story short, no specific cause for the DI could be identified and the disorder was classified as “idiopathic.” Idiopathic DI occurs in about 10% of childhood DI cases, at least based on the medical literature. In my experience, about 25% of cases fall into the idiopathic category. What does this mean? It means that we can’t explain why the patient has DI but that we need to monitor the patient long-term to be certain none of the rather nasty causes becomes evident. I try hard never to forget that one of my former students defined idiopathic as “the idiots just don’t know the pathology.” So right he was.

So?

This interesting “real-life” case teaches us several things. First, that primary care physicians are to key to timely diagnosis of medical conditions. Second, that knowledgeable primary care physicians can greatly facilitate the evaluation of patients with complex medical conditions. Finally, that step by step thoughtful evaluations solve medical diagnostic dilemmas as well or better than the “order as many tests as possible now and think later” approach used by some.

Recently, I have been asked more and more what the difference is between type 1 and type 2 diabetes and whether there is really a type 2.5. The questions have come not only from patients of mine and friends (who may or may not have diabetes) but also from a surprising number of attorneys. Why the attorneys? It’s because of the now well-known association between certain medications called “atypical antipsychotics” and the greatly increased risks for developing type 2 diabetes. There have been quite a number of lawsuits against manufacturers (both individual and class action) regarding the link between the medications and the development of diabetes, primarily related to the issue of insufficient warning to patients about the risks. There has also been a question of whether some manufacturers hid data from the FDA showing links between their drugs and the development of diabetes. I am not an attorney and have no opinion regarding these legal issues.

Back to the basics: What is diabetes and what is the difference between type 1 and type 2?

I have discussed these topics to some extent in earlier entries but it is worth revisiting the subject. You may want to search my archived articles- maybe I discussed these issues better earlier?

Diabetes mellitus or “sugar diabetes” (to be distinguished from diabetes insipidus, a condition related to poor control of water balance) is a group of disorders characterized by 3 basic features: insulin deficiency, which can be absolute (e.g., destruction of pancreatic beta cells) or relative (e.g, resistance to the release and/or action of insulin); hyperglycemia (high blood sugar levels) as a consequence of the insulin deficiency; and increased risks for the development of certain chronic complications of the eyes, kidneys, nerves, and cardiovascular system (heart and blood vessels).

It is worth adding that for all types of diabetes mellitus, the complications are caused by the same factors which include chronic hyperglycemia, hypertension, and abnormalities in blood lipids (e.g., hypercholesterolemia). It is also important to note that the complications are preventable, regardless of the diabetes type.

The confusing history of diabetes nomenclature

Early on (from the time of the ancient Greeks until the 1920s) diabetes mellitus was just diabetes mellitus. There was a form that mostly affected young children and was more or less fatal within weeks and a form that affected adults who were overweight, and was considered mild (I have no idea what “mild” meant). These 2 forms came to be called juvenile-onset diabetes and maturity-onset diabetes. These terms were widely used between about 1940-1970. Both types were considered to be the same disease with the same genetic basis, just different clinical presentations at different ages. In fact, that’s what I was taught in the 1960s.

Enter the age of semi-enlightenment

By the early 1970s, data were accumulating that the two forms of diabetes might be genetically distinct. There was strong support to rename the diabetes forms based on the degree of insulin deficiency. Thus came the terms “insulin-dependent diabetes” and “non-insulin-dependent diabetes,” or “IDDM” and “NIDDM.” This was a very bad idea and within a few years it became clear that if we were dealing with 2 different genetic disorders, defining them by whether or not the patient required insulin to maintain reasonable blood sugar levels (back then, we didn’t really even know how to quantify what was a “reasonable” blood sugar level) was illogical.

So, by the late 1970s, IDDM and NIDDM were scrapped and replaced by terms meant to reflect distinct genetic and etiologic differences between the diabetes types. The new terms were type 1 and type 2 diabetes. Type 1 was more or less synonymous with the old juvenile-onset diabetes, although the condition could present at any age, and type 2 with the old maturity-onset diabetes, although the condition could also present at any age (e.g., in obese teenagers). Type 1 diabetes is considered to be mostly (not all cases) autoimmune destruction of the pancreatic beta cells. Type 2 diabetes is considered to be a combination of insulin resistance and some true deficiency of insulin secretory capacity. Thus a person could have type 2 diabetes and require treatment with insulin, not just treatment with weight loss, diet manipuation, and so-called oral agents.

Type 2.5

Over time, it became clear that in some cases it was virtually impossible to classify a patient as having either type 1 or type 2 diabetes. For example, what if a 39 year old who was overweight presented with a two-week history of increased thirst and urination, had ketoacidosis (a sign of severe insulin deficiency) but no evidence of autoimmune destruction of the pancreas? And what if this patient had a strong family history of what would be considered typical type 2 diabetes? Is it type 1 or type 2 diabetes? This is where the idea of type 2.5 came about (some people use the term “double diabetes” and I’m sure there are lots of other terms floating about). Could a person have both type 1 and type 2 diabetes? Of course they could; type 2 diabetes is very common and there is no reason that a person who gets type 1 diabetes cannot also have the genes for type 2 diabetes, which would be much more likely to be expressed if the individual were overweight.

Is it important to determine which type of diabetes a person has?

In most instances (probably 95%) it is easy to classify a patient as having either type 1 or type 2 diabetes and to treat them accordingly. Occasionally, it’s virtually impossible to be certain which type of diabetes it is. But, in most instances, it doesn’t really matter. The treatment goals for types 1 and 2 (and type 2.5) are the same. Many with type 2 diabetes can achieve well-controlled diabetes (these days we do know how to quantify the degree of “control”) with weight loss, diet manipulation and, if necessary oral agents. Some patients will require insulin treatment.  We know that type 2 diabetes patients do have some degree of actual insulin deficiency which can worsen over time, particularly if the diabetes has not been well-controlled (there is a strong genetic difference between patients with type 2 diabetes who have a high degree of absolute insulin deficiency and those whose insulin deficiency is mostly insulin resistance). So, these days diabetes treatment is based on achieving therapeutic goals not what type of diabetes it may be. If a patient needs insulin to achieve treatment goals, they need insulin. It’s really that simple.

There are actually some reasons to know if a patient has type 1 or type 2 diabetes. For example, patients with type 1 diabetes, which is mostly an autoimmune disorder, are more likely to develop other autoimmune disorders such as chronic lymphocytic thyroiditis, pernicious anemia, and celiac disease- not so for patients with type 2 diabetes. Also, if a patient has clinical features suggesting both types 1 and 2, certain medications used exclusively in patient with type 2 diabetes might be of benefit (along with insulin). If, in the future medications are developed that can target directly the genetic defect or defects in patients with type 2 diabetes, it would be nice to know that the patient has type 2 diabetes (here I’m thinking about the insulin resistance over and above that due to the obesity, and the tendency for the insulin deficiency in patients with thype 2 diabetes to be progressive).

A case history

So here’s a real case of mine to demonstrate the complexities involved in determining whether a patient has type 1 or type 2 diabetes. The patient is a 22 year old male of Hispanic background who presented with a two-week history excessive thirst and urination and a 20 pound weight loss. The patient had been overweight, particularly after he was treated with an atypical antipsychotic medication for 2 months which was stopped about 3 months earlier. The initial laboratory tests showed typical findings for type 1 diabetes- plasma glucose 350 mg/dl, hemoglobin A1c 8.6% (the test that is an index of the blood gluocse level over the previous 3-4 months- normal is <6%), urine ketones were “strongly positive” (typically, urinary ketones which are at breakdown products, are found only in type 1 diabetes). Tests for islet-cell antibodies were negative.

The patient was treated for typical type 1 diabetes and responded well to insulin injections. Over time, the patient did well with his diabetes and hemoglobin A1c levels were always close to the normal range. The patient’s insulin requirements were lower than typical for age and weight; after 7 years of diabetes, the insulin dose had remained very low and hemoglobin A1c close to normal. His weight had remained fairly stable (within 2 weeks of initiation of therapy, the patient regained about half of the weight he had lost just prior to diagnosis.

As the treating physician (to be distinguished from a “medical expert”), I was recently asked by both a plaintiff attorney and a pharmaceutical company defense attorney if I thought the patient had type 1 or type 2 diabetes. My answer was “I don’t know.” At present, the patient has some clinical features that suggest type 2 diabetes; he is somewhat overweight, he has acanthosis nigricans (a skin condition associated with insulin resistance), and he is being managed with a very low dose of insulin (this is according to the attorneys since I have not been treating this patient for several years). The question the attorneys have is whether I thought the patient might have developed type 2 diabetes from the atypical antipsychotic medication he took prior to his diagnosis with diabetes. I have told both attorneys that I didn’t know but that the patient presented with typical type 1 diabetes. In fact, a history of weight loss prior to diagnosis, which was well documented in this patient, would have been more or less unheard of in a patient with type 2 diabetes- with weight loss, patients with type 2 diabetes become less insulin resistant and have lower blood sugar levels and fewer diabetes-type symptoms such as excessive thirst and urination. On the other hand, the patient’s present clinical status is much more typical of type 2 than tye 1 diabetes. The “negative” test for islet-cell antibodies at the time of diagnosis doesn’t help much- the test is not a great one and some cases of type 1 diabetes are not on a well demonstrated autoimmune basis, although most are. So, what do you think?

One final note- these days, treating the patient discussed above is much easier than answering the various legal questions. Does that mean it’s easier to be a doctor than a lawyer?

Background

Yesterday I had an entry about a study Medicare carried out in an effort to save money. I was more or less merciless in my criticism of the study- both the design and the validity of the conclusions. In short, it would be surprising to find a short-term cost-saving in patients with chronic medical conditions such as diabetes and heart disease by simply having nurses call the patients to encourage them to receive appropriate medical care. In fact, I would have predicted the initial costs would go up if the telephone calls worked since patients would likely use the health care system more.

I stand by that statement, but I should have added something about the health care benefits of telephone calls, e-mails, etc., to patients with chronic medical disorders.

Does patient care reinforcement help?

There is no question that frequent patient contact by health care providers who know the patients well can enhance medical care. Sometimes it’s just a quick call to see how a patient is doing with their new treatment plan. Sometimes it’s a call to find out if the patient is doing better with their care. Many physicians and other health care providers now use e-mail communication with patients to serve the same basic purpose as telephone calls. I have been quite surprised that quite a number of my patients who already use e-mail (about 90% of my patients have e-mail access and about 60% use e-mails quite regularly- it’s a whole new world out there), communicate regularly with me about this or that- sometimes it’s about their diabetes and sometimes it’s something exciting in their lives such as getting a new car, running a marathon, etc.).  I am particularly impressed by how many of my teenage patients send me e-mails regularly and I am convinced (not a  scientific study) some of my greatest “problem” patients have improved their care simply by the back and forth communication.  It’s rarely anything prophetic I tell them to do, it’s just knowing that I care.  Anyway, I just wanted to set the record straight- in general, I am a strong proponent of patient care reinforcement (positive reinforcement almost always works better than the negative type) which can be a powerful treatment tool, whether it’s done with a telephone call, an e-mail,a letter, or a brief follow-up clinic visit just to see how things are going

I was rather dismayed today after reading an article in the New York Times entitled ” Medicare Finds How Hard It Is To Save Money,” written by Reed Abelson (NYT April 7, 2008). The article summarized what seems to be Medicare’s failure to save money by having nurses call patients with chronic diseases (e.g., heart disease, diabetes). The study began in 2005 and included about 160,000 people. Medicare paid 8 companies about 360 million dollars to make the calls. So far, the results have not shown any significant cost savings to Medicare. There are no data yet (if ever) to show if the study subjects have better health outcomes.

Why was I so unhappy after reading the article?

First, I thought the article was well-written- that was not the problem. The problem seems to be the Medicare administration. It is hard to believe Medicare (presumably someone specific at Medicare?) thought that a few phone calls to patients with chronic diseases would save the system big money within only 2-3 years. I would have predicted that if the phone calls were effective in getting patients to seek more care, the short-term costs would go up. There are reams and reams of data from numerous studies that have already shown that “case management” of patients with chronic diseases saves money- lots of money. The way the money gets saved is by having expert teams manage the disorders (i.e., not just making telephone calls) which results in astonishing savings over time by preventing or slowing progression of disease. It takes time to achieve these types of cost savings. Let’s not forget what accounts for the majority of diabetes care costs. If you have read some of my earlier entries, you already know that the lions share of diabetes care costs are for management of the chronic complications, all of which are preventable by good care long-term. Didn’t the Medicare Program gurus look at the results of the Kaiser-Permanente Health Care Programs for chronic disease? Quite a long time ago, Kaiser-Permanente discovered that they could save gobs of money if they could treat their patients with chronic diseases (I think Kaiser has focused on diabetes, cancer, and heart disease, but I’m not certain- there may be other disorders included) in such a way that they could prevent the complications, which accounted for the major costs in treating the chronic diseases. For example, more than 50% of all Medicare patients on kidney dialysis have diabetes. Kaiser found that by setting up care teams that aggressively managed the patient care, health outcomes were dramatically improved (about 50% better than expected) and the cost savings to Kaiser were huge.

What should Medicare do?

Medicare would do well to emulate what Kaiser-Permanente and others are doing and forget about the phone calls. Phone calls are great to help us remember our hairdresser and dentist appointments, but Medicare could spend its money better. Also, Medicare should not get discouraged about efforts to save money- they just need to be patient (it takes time for chronic complications to develop in people with chronic diseases; 2-3 years would not be enough time even if their intervention was a good one (e.g. case management teams).