How to Write Riveting Assessments

September 02, 2018
by Mark David Siegel

Hi everyone,

The "assessment" at the close of your presentations is your time to shine, your chance to highlight salient data, show what you know, and display your clinical reasoning.  The best assessments guide your listeners, focus your thinking, and point towards effective diagnostic and treatment plans.  

Well-done assessments are consistently structured, beginning with a summary, followed by laying out an approach, creating a differential diagnosis, and ranking the possibilities.

  1. Summarize: Here you should highlight the key features of the history, physical, and diagnostic evaluation.  You should only include information that will contribute to the patient’s diagnosis and management.  Inexperienced presenters often include extraneous detail, for example reminding us that a patient with gout also has irritable bowel syndrome.  Good summaries synthesize.  You should transform the information you’ve already presented into pithy, memorable statements.  For example, if you told us in the HPI that the patient is a “94-year-old man with coronary artery disease, s/p placement of three drug eluting stents on ASA and clopidogrel, now presenting with 3 episodes of hematemesis and 5 episodes of melena,” you can summarize as follows: “In summary, this is an elderly man with CAD on dual antiplatelet therapy, presenting with a large GI bleed.”
  2. Lay out an approach: Here you should outline your diagnostic plan, usually calling upon a “script” or favored approach to confronting a paradigmatic problem.  The first step—often more difficult than we acknowledge—is to correctly identify the problem needing your attention.  Sometimes the problem is the one the patient identifies in the Chief Concern (“crushing substernal chest pain”) but sometimes it will be a crucial finding on your physical exam (a surgical abdomen) or in your diagnostic workup (severe hypercalcemia).  Patients may present with more than one problem needing attention.  Sometime multiple problems will be obviously related and trigger a script in your mind (fever and leukocytosis); in these cases, you can combine the problems into a single issue to be addressed.  However, be careful not to combine problems that may be unrelated or don’t create an obvious script (mouth ulcers and shortness of breath).  In these cases, it’s usually best to address each problem individually, recognizing that they may or may not lead to a unifying diagnosis later.  Here are some familiar scripts. “In this patient with severe thrombocytopenia, we can consider decreased production, destruction, and sequestration.”  “In this patient with AKI, we can consider pre-renal, intra-renal, and post-renal causes.”  Sometimes we use mnemonics: “In this patient with a gap acidosis, we can consider ‘MUDPILES’: methyl alcohol, uremia, DKA, propylene glycol, isoniazid, lactic acidosis, ethylene glycol, and salicylates.  Sometimes we use anatomy: “In this patient with right lower quadrant pain, we can consider disorders in the abdominal wall, appendix, ureter, ovary, colon, and blood vessels, as well as referred neuropathic pain.” Sometimes we consider disease categories: “Causes of weight loss could include infection, inflammation, malnutrition, metabolic causes, and cancer.”  At this point, if the data allow, you may choose to focus on one of the sub-categories: “In this patient with hypercapnia, we can consider diagnoses caused by increased CO2 production, increased dead space, and/or hypoventilation.  However, since the patient presented with a low respiratory rate, we should focus on hypoventilation, which may be caused by depression of the respiratory drive—for example due to drug overdoses, brain stem disorders, metabolic abnormalities, excessive oxygen supplementation, and congenital abnormalities—as well as respiratory muscle weakness and fatigue.” Similarly, if a patient has fever, you can consider infectious and noninfectious causes.  But if the patient has been sick for one day with rigors, cough, sputum, dyspnea, leukocytosis, and a bandemia, you should limit your attention to infectious causes unless there is a compelling reason to do otherwise.
  3. Create a differential diagnosis: Here you should create a list of specific diagnoses.  Your list should be long enough to include all diseases that warrant attention but not so long that it includes disorders that would only be distracting if you mention them.  For example, if you admit a patient with chest pain to the CCU, focus on disorders that deserve your attention, such as MI, pericarditis, pulmonary embolism, esophagitis, and musculoskeletal causes.  You don’t need to mention zoster in this patient population unless you have compelling data (“the EKG is normal, troponins are negative, and the patient has a dermatomal rash”).  Be sure, however, to include “can’t miss” diseases, even if they’re unlikely.  For example, in a patient presenting with traumatic neck pain, you should consider a cervical spine injury since missing it could prove devastating. 
  4. Rank the possibilities: Here you should use your clinical reasoning to address each diagnosis in your differential, ranking them according to likelihood.  In general, you should consider each possibility as mutually exclusive. Even though it’s possible for patients to present with hemorrhoidal and diverticular bleeding, it’s far more likely for them to have hemorrhoidal or diverticular bleeding.  Mathematically, the probability of each diagnosis should add up to 100%, so the more likely one diagnosis is, the less likely the others must be.  If you think a patient probably has pneumonia (because of the fever, cough, sputum, and infiltrates), then PE, CHF, and noninfectious pneumonitis must be less likely.  Assuming you plan to pursue further diagnostic testing, this is where you assign “pretest probabilities.” These probabilities should be based on a number of factors including epidemiology (i.e., host factors and prevalence of a disease in the community), what you know about a disease (“that’s how rabies presents”), and data from your history and physical, and the diagnostic tests you’ve already done.  At the end of summer in Connecticut, a patient presenting with fever and a rash should make you consider Lyme Disease, but if you’re in Arizona, you’d assign greater weight to diseases you would see there, like coccidioidomycosis.  Your clinical reasoning should refer back to the salient information you highlighted in your summary, like the cyclical fever (malaria), love of hiking (tick-borne illness), distended neck veins (CHF), or the elevated CPK (rhabdomyolysis).  

Your goal is to identify diagnoses that warrant your attention, either because they’re likely or because you don’t want to miss them. For example, “In this patient with a history of alcohol use disorder, right upper quadrant pain, jaundice, and elevated transaminases, I am most worried about alcoholic hepatitis, particularly since the AST is higher than the ALT.  Viral, vascular, autoimmune, and genetic causes are much less likely.  However, since the patient has a history of suicidal ideation, we shouldn’t completely rule out toxic causes, particularly acetaminophen overdose, since that would require specific treatment.” 

So in summary, the assessment is your opportunity to bring your presentations to a compelling conclusion.  Follow a consistent, structured format: 1) summarize, 2) lay out an approach, 3) create a differential diagnosis, and 4) rank the possibilities.  Do it this way every time, and your assessments will become smooth, cogent, efficient, automatic, and useful.  Effective assessments will serve you, your colleagues, and your patients well.

Enjoy your Sunday and Labor Day, everyone.  I’m going to a feis, where Francesca is dancing, and then spending the afternoon at Quassy Amusement and Water Park.

See you soon,

Mark

Submitted by Mark David Siegel on September 03, 2018