Disease-specific Factors

One reason for delayed diagnosis is the all-too-true aphorism, “diseases do not read the textbook.” For example, patients with endocarditis rarely have all or any of the classic physical findings associated with this disease. Some patients with periodic fever syndromes or other autoinflammatory diseases may not have fever, individuals with hyper–immunoglobulin D (IgD) syndrome may have normal levels of IgD, and hyperkalemia may be present in the classic hypokalemic form of Bartter syndrome. The great masqueraders of the past century—lupus, tuberculosis, and syphilis—still exist but have been joined by others, including hemophagocytic syndromes, periodic fever/autoinflammatory syndromes, neuroimmunologic diseases, and other autoimmune and immunodeficiency syndromes. In addition to having patients whose disease evolution and diagnostic evaluation have persisted for a prolonged time period, most large pediatric hospitals have cared for patients with URDs who have evaded a specific diagnosis in the early phase of their disease and who present acutely with a serious or catastrophic illness. In the authors’ experience, these acutely ill patients represent approximately 15% to 20% of patients; they require immediate consultation as well as aggressive and rapid diagnostic evaluation.

Patient-specific Factors

Families of children with URD may be viewed with certain negative biases. They are often anxious from the persistence of their child’s chronic undiagnosed medical illness. These families have often shared the diagnostic odyssey with multiple physicians and have reached a point where they are over-reporting the minutest details in hopes of discovering the clue or clues that will lead to a diagnosis. This may result in a diagnostician presented with an overwhelming amount of information, all of it seemingly equally important to the family, making it difficult to form an accurate representation of the child’s signs and symptoms. There may be some suspicion on the part of health care providers that the information is even factitious and an effort to force further diagnostic testing. Parents and families may be perceived as doctor-shopping as part of their effort to find answers to their questions. In addition, because of unsatisfactory previous interactions with the health care system, these families may not always speak well of their past experiences. An attribution error from this negative stereotype may lead to an inappropriate diagnosis of Munchausen syndrome by proxy (also known as medical child abuse or factitious disorder by proxy). Although it is essential to be aware of this possibility, the diagnosis of Munchausen syndrome by proxy in a child with a URD should be considered and made in a manner identical to all other diagnoses considered: after careful review of all available information; consideration of the specific pattern and details of the signs, symptoms, laboratory, imaging, and other data; and deliberate and thoughtful conversations with the patient, family members, and all involved health care providers. This diagnosis should not be made prematurely or simply because nothing else fits.

Physician-specific Factors: Issues in Clinical Judgment

Two important maxims in the evaluation of children with URDs are to expect the unexpected and to never say never. During the evaluation of many patients, well-intended and often senior experts in their field have made statements like, “I have never seen this symptom in the proposed disease,” “It cannot be this,” and, even mistakenly, “It must be this.” This appeal to authority often results in avoidable diagnostic errors. Despite evidence of convincing geographic clustering and the development of a preceding unusual rash in a majority of patients, the originally described patients with Lyme arthritis were most frequently misdiagnosed as having juvenile rheumatoid arthritis. Statements like, “In my experience” or “We see this,” are more authoritative than evidence based and eventually lead to a diagnostic momentum that excludes consideration of other disorders. Once a diagnosis is fixed in a medical record, in particularly a multiyear record, it is most often perpetuated, assumed to be correct, and difficult to question or rethink.

Appeals to authority can increase the tendency toward confirmation bias, in which the clinician seeks out only information that affirms a diagnosis and excludes facts that contradict the initial diagnosis. Atypical or new clinical findings may be a clue to the real diagnosis and should not be ignored. Overcoming confirmation bias is particularly problematic if a physician advocating for a particular diagnosis is a senior subspecialist recognized as an expert in the field, because it often prevents the entire team from considering other possible diseases. Even independent of appeals to authority, confirmation bias is detrimental to the diagnostic process because many URDs are atypical or uncommon manifestations of a common disease or are unrecognized manifestations of a rare disease. Always maintain a broad differential diagnosis and do not eliminate disorders from consideration solely because they initially do not seem to fit. On a busy hospital service or in a clinic setting, there may be a tendency to rule out a diagnosis too quickly. Once a diagnosis is incorrectly considered ruled out, it is difficult to rethink that diagnosis even in spite of evidence suggesting the excluded disease. Published clinical descriptions of diseases—in particular newly described diseases—tend to be biased in favor of the most commonly recognized manifestations and should be interpreted with caution. The true breadth of clinical manifestations of a specific disease may be much more extensive than suggested by descriptions in textbooks or in the literature. Recently, genomic testing has provided molecular evidence that many disorders exhibit greater clinical heterogeneity than previously appreciated. Always ask, “Do any of the features match what we know regarding a specific disease?” and “Could this be a condition I have seen before, but with a new or different clinical pattern?”

An attribution error in URDs occurs when minimizing signs, symptoms, or laboratory data. This is typified by the use of slight or slightly before describing an abnormality. Classically used by medical students who are unsure whether they hear a cardiac murmur or see a rash, when used by more experienced health care workers, the term, slight , may deemphasize an actual critical finding. Slight tachycardia should be stated as tachycardia if the heart rate is elevated above a normal range; the same is true for abnormal laboratory data. A sodium level of 150 mEq/L is elevated and abnormal; it may suggest a specific condition. These types of findings should not be overlooked in the child with an undiagnosed disease. Every symptom, every finding on the physical examination, and every discovery on laboratory testing and imaging should be considered potentially important clues that lead to a correct diagnosis.

Some children with URDs may appear seemingly healthy. Because many of the complaints of URDs may be vague or vexing, and because there is frequently considerable family anxiety regarding the lack of a diagnosis, the significance of the illness may not be appreciated or the parents may be thought of as over-reacting. This type of benign thinking is a type of affective error, where emotions may interfere with diagnostic judgment. “This child looks too well” or “Hoping for the good” or expecting that “Nothing really bad could be wrong with this child” clouds diagnostic decision making. Because most practitioners see common disorders on a day-to-day basis, they may miss the zebra among the horses. Because common diseases are common, a neutropenia may be assumed the result of viral suppression or drug induced, when a patient in fact has leukemia; a child presenting with tachypnea and fever initially diagnosed with pneumonia may have salicylate toxicity; or an athlete with prolonged knee pain initially diagnosed as a sprain may have osteogenic sarcoma. Conversely, significant family anxiety may lead to malignant thinking, another affective error in which a health care provider favors potential diagnoses with higher acuity or an increased risk for morbidity or mortality, simply as a function of that anxiety. Diagnosticians need to be self-aware of the tendency for these types of errors and to maintain the same impartial and broad differential diagnosis for a given set of signs and symptoms, even when a child appears generally well or in the context of significant family anxiety.

Perpetuating a diagnosis over time, whether due to appeals to authority, confirmation bias, attribution errors, benign thinking, or other affective errors, should be a cause to go back to the beginning of the illness and review all information available. Adolescent-onset behavioral and neurologic regression in a child originally diagnosed with autism may actually be a metabolic or neurodegenerative disorder. Recurrent episodes of swelling initially thought to be allergic reactions may eventually be diagnosed as episodes of hereditary angioedema or as nephrotic syndrome if the swelling is predominantly periorbital. Abdominal pain attributed to constipation but persisting despite evidence of normal stool output may in fact be porphyria, familial Mediterranean fever, lupus, or even malrotation with intermittent volvulus. Skin lesions that may appear eczematous or psoriasiform, when present in an unusual distribution restricted over the small joints of the hands, may be correctly diagnosed as associated with dermatomyositis.

Limitations in Diagnostic Testing

When evaluating children with URDs, it is important to recognize the limitations of diagnostic testing. Rarely is a diagnostic test 100% sensitive or specific, and as such, the positive and negative predictive values are often not ideal. Test results should not be considered dichotomously as ruling a particular condition either “in” or “out” but rather as increasing or decreasing the post-test probability of that condition in accordance with the positive or negative predictive value of that test as well as any other findings or results from laboratory or imaging studies. Laboratory error should be considered as well. There may be a tendency to believe the hard data of a particular test result more than believing eyes, ears, and hands and to trust that test results are more accurate and definitive than information gathered from history and physical examination. Finally, the absence of a positive laboratory test does not always disprove a diagnosis. For example, most but not all patients with lupus have a positive antinuclear antibody test; fewer have positive anti–double-stranded DNA titers. Only 80% to 90% of patients with endocarditis have positive blood cultures. The long QT syndrome (LQTS) has traditionally been diagnosed based on an age-specific length of the QT interval. There are more than 1000 genetic variants that account for approximately 80% of patients with LQTS. Nonetheless, 30% of patients with a gene mutation and resultant risk for arrhythmias do not demonstrate an abnormal QT interval on standard ECG testing.