Clinicians use a diagnostic process to guide their actions when faced with a medical problem. Hypothesis generation provides an efficient reliable framework to focus the clinical workup and come to the correct diagnosis and treatment plan.
A 12-year-old boy complained of pain at the umbilicus, beginning just after he started to urinate and lasting around 30 seconds post-voiding. There was no radiation of the pain, and he did not complain of dysuria. Though the patient noted painless retraction of umbilicus during urination for as long as he could remember, discomfort and pain had onset 3 days earlier. No suprapubic tenderness or abdominal masses were palpable; urine studies came back normal. An ultrasound examination of the abdomen revealed no evidence of a urachal cyst and neither did a voiding cystourethrogram.
Nonetheless, Pediatrician Dr. Peter Rowe and Pediatric Urologist Dr. John Gearhart both had persistent suspicions for the diagnosis and agreed that the child needed a cystoscopy. The study revealed a small opening at the dome of the bladder; this prompted surgical exploration to find a urachal remnant which reproduced symptoms of retraction.
Four approaches to reach diagnosis
Dr. Frank Oski, former chairman of pediatrics at the Johns Hopkins Children Center, wrote that clinicians typically use four approaches to reach a diagnosis:
1.) Pattern recognition
2.) Sampling the universe
3.) Clinical algorithms
4.) Hypothesis generation
Clinicians may use these approaches at various times depending on the patient characteristics and presenting problem.
One of the four approaches clinically excellent providers often use is hypothesis generation. Before entering the room to see the patient, clinicians should develop a differential diagnosis list based on the chief complaint. Then using history, physical exam, and diagnostic testing, check things off of the list leaving the most likely clinical possibilities (five or less is ideal).
Trusting the diagnostic process
Given the presenting complaint for this young patient the diagnosticians thought about the most common cause— that being a urachal cyst. Urine studies and imaging studies failed to uncover this diagnosis. Undeterred by this finding the clinicians were convinced that the presenting symptoms were caused by a urinary condition that could be identified and repaired. Trusting the diagnostic process, they considered the most common diagnosis but included in their differential list more rare possibilities. Based on this information and in sequential fashion they performed a more in-depth cystoscopy procedure. This revealed a urachal remnant that reproduced the symptoms. The bladder defect was repaired and symptoms abated.
Hypothesis testing provides the excellent clinician a framework to narrow the list of diagnostic possibilities, focus testing so as not to perform unwanted or unneeded tests, consider common diagnoses first but always keep an open mind for other possibilities when the data does not fit, and finally guide development of a reasoned plan for testing your hypotheses.