Pattern recognition is the process by which a diagnosis is made based on physical clues or linkage identification. For example, a diagnosis of Down syndrome can be made by recognizing the physical findings that make up this genetic abnormality. Similarly, the diagnosis of Henoch-Schönlein purpura is immediately apparent if the rash has a characteristic pattern and distribution. Diagnosis by pattern identification requires familiarity with diseases through experience or study. The expression “the more you see, the more you know, and the more you know, the more you see” describes how pattern recognition develops.

Linkage identification is a form of pattern recognition. A diagnosis is based on history and physical or laboratory findings. For example, the finding of a micropenis and hypoglycemia in a neonate would result in a prompt diagnosis of congenital hypopituitarism. A history of bloody diarrhea in association with a white blood cell count demonstrating more band forms than mature polymorphonuclear leukocytes would result in an immediate diagnosis of Shigella gastroenteritis. Skill in linkage identification, like pattern recognition, is gained by observation and study. The seemingly intuitive diagnosis, often the hallmark of the older physician, is usually a result of linkage identification.

Sampling the universe refers to the inappropriate ordering of laboratory studies in hopes that an abnormality will appear that can result in a diagnosis. This is a diagnostic process to be decried. In the United States, approximately $30 billion per year are spent on laboratory tests representing approximately 3.5% of total health care spending, and another $2.8 billion per year are spent on chest roentgenography alone. Computed axial tomography (CAT) scans, MRI, and diagnostic ultrasounds are rapidly eclipsing the costs of laboratory studies and, according to many critics of the trend, are replacing the physical exam. An estimated 20% to 60% of these tests are unnecessary. If the estimates are accurate, a minimum of $6 to $18 billion per year are spent on procedures that do not aid in the diagnosis or treatment of illness. As the number of new tests available constantly increases, so will the amount spent on those that are inappropriately ordered. Laboratory tests should be obtained only to support a hypothesis. If the history and physical diagnosis do not suggest an underlying organic disorder, no rationale exists for ordering a battery of laboratory tests in an attempt to uncover an occult disease. As any practitioner who has been guilty of overdoing test ordering can vouch, one runs the risk of obtaining a questionable result on any number of laboratory evaluations that can then generate an additional mindless search for an answer to a diagnostically insignificant question. The evaluation of infants and children with failure to thrive is an example of this form of behavior. In a classic 1978 review of 2,607 laboratory studies performed on 185 patients with failure to thrive, Sills found that only 1.4% of the tests were of any positive diagnostic assistance, and all of them were specifically indicated by the history or physical examination. More recent studies on failure to thrive both confirm and uphold the observation that most patients with failure to thrive require very little laboratory evaluation.