Differential Diagnosis Of Common Complaints

How We Actually Approach A Bunch Of Different Symptoms

Differential Diagnosis Of Common Complaints isn't some grand scientific method that separates the amateurs from the professionals. It's mostly just the process of narrowing down what's wrong when someone walks in saying they've felt terrible for three days and you can't tell whether it's bacterial, viral, or just stress. I've spent more years doing this than I'd like to admit, and honestly the trick is less about being brilliant and more about not missing the obvious stuff because you're already thinking about the rare stuff. The standard way this gets taught is you take a complaint, list every possible cause, then systematically eliminate them until one is left standing. That's textbook. In reality you're rarely starting from zero because certain patterns repeat constantly. A headache in a 30-something is ten times more likely to be tension-type than a brain tumor, so you don't waste time ruling out the tumor first. You rule out the red flags and move on. What most beginners miss is the direction of thinking. You don't eliminate possibilities one by one like crossing items off a grocery list. You group them into "must not miss" and "probably not." The must-not-miss bucket gets your full attention immediately. Everything else is background noise until the urgent stuff is cleared. This flips the whole process. Instead of asking "what could this be?" you ask "what would kill this person if I sent them home?" The answer to that question is usually short.

I ran into a case last year where a patient came in with what looked like a straightforward migraine. Headache, photophobia, nausea, no focal neurological deficits. Classic presentation. The textbook differential pointed straight at migraine with aura. I sent them home with a triptan and told them to follow up if it didn't resolve. Three weeks later they came back because the headache had changed character, and this time I actually looked at the timing more carefully. They'd been on combined oral contraceptives for six years and had developed new-onset headaches at age thirty-four. That's a vascular risk factor combination you don't ignore. I ordered an MRI and it showed a cerebral venous sinus thrombosis. The initial presentation had been so typical for migraine that I almost let it slide. The workaround I use now is that any new headache pattern in someone over thirty who has additional risk factors gets imaging before I call it benign, regardless of how classic it looks.

Practical Steps That Actually Work

Start with the chief complaint and write it down in the patient's own words. "I feel like my chest is being squeezed" is a different data point than "my chest hurts." Then immediately separate the timeline: when did this start, what was happening right before, and has anything like this occurred before. The onset matters enormously. A sudden onset thunderclap headache is a completely different differential than a gradual pressure that built over weeks. Next you pull the history. Not a generic review of systems, but targeted questions based on what you're already considering. If you're thinking cardiac for chest pain, ask about exertion, radiation, associated sweating. If you're thinking pulmonary, ask about pleuritic component and recent immobilization. You're building a small working list of three to five possibilities and fishing for evidence that supports or refutes each one. This is where the common complaint trap hits hardest. Gastrointestinal complaints are the worst offenders. Heartburn, indigestion, epigastric discomfort — the differential for that region includes peptic ulcer disease, gastritis, biliary colic, pancreatitis, myocardial ischemia, and aortic dissection. All of them present with upper abdominal discomfort. The ones you don't want to miss aren't the ones with the ugliest name, they're the ones that look like the least bad option. Physical exam follows the same logic. You're not doing a comprehensive survey unless there's a reason. You're checking for the signs that would push you toward or away from your leading hypotheses. JVP for heart failure. Murphy's sign for cholecystitis. Calf circumference difference for DVT. These are low-effort, high-yield moves that take thirty seconds each and tell you more than a page of lab orders ever would.

When The Obvious Answer Is Wrong

I've seen this happen too many times to count. A patient presents with lower respiratory symptoms, crackles on exam, fever, and a consolidation on X-ray. Everyone agrees on pneumonia. Antibiotics are started. The patient doesn't improve after forty-eight hours. Something is off. The thing nobody wants to consider is that the original diagnosis was correct but incomplete, or that the whole thing was never pneumonia to begin with. In one case it turned out to be a pulmonary embolism with infarction mimicking pneumonia on imaging. The CBCT eventually sorted it out, but only after we'd exhausted the antibiotic route. The lesson isn't that you should always order a CT for pneumonia. It's that non-response to treatment is itself diagnostic data. When a common complaint doesn't get better with the common treatment, the differential just got a lot wider. Here's another counter-intuitive point: sometimes the absence of findings is more useful than their presence. A normal cardiac exam in someone with chest pain doesn't rule out coronary disease. Normal lung exam doesn't rule out early pneumonia. Normal neurological exam doesn't rule out a growing mass. Clinical examination has sensitivity limits that vary dramatically by condition and operator. Over-relying on a normal physical exam to exclude serious pathology is one of the most common errors I see in early-career clinicians.

Tools And Shortcuts

Decision support tools exist and they're useful if you use them as prompts rather than answers. The HEART score for chest pain, the Wells criteria for PE, the PERC rule — these are validated frameworks that help structure your thinking. But they have narrow applicability and they weren't designed to replace clinical judgment. A PERC-negative low-risk patient might genuinely not need a CT pulmonary angiogram, but a PERC rule doesn't account for everything. It works on populations, not individuals sitting in front of you. Lab work should be driven by the working differential, not by a screening impulse. A basic metabolic panel, CBC, and inflammatory markers are reasonable starting points for many complaints, but ordering every test in the book is how you generate noise. False positives cascade. An incidental elevated D-dimer leads to imaging, which leads to an incidental finding, which leads to a biopsy, which leads to anxiety and procedural risk for something that would never have caused symptoms. Threshold testing, where you only order diagnostics that would change your management, keeps you honest. I use a simple mental model for the common complaints that show up constantly: chest pain, headache, abdominal pain, shortness of breath, and fatigue. Each of these has a core set of dangerous causes that must be excluded before settling into a benign diagnosis. For chest pain that means acute coronary syndrome, pulmonary embolism, aortic dissection, and pneumothorax. For headache it's subarachnoid hemorrhage, meningitis, temporal arteritis, and mass lesion. For abdominal pain it's ischemic bowel, ruptured AAA, ectopic pregnancy, and perforated viscus. You run through that list first every time, regardless of how likely the benign causes seem.

Limitations You Need To Accept

Differential diagnosis as a method has real constraints. It assumes you have enough information to make reasonable probability estimates. In practice you often don't. Patients forget details. They minimize symptoms. They present late. The history you get is filtered through their perception, not objective reality. Physical exam findings overlap extensively across conditions. A fever and tachycardia appear in everything from simple viral illness to sepsis. Labs are imperfect. Imaging has false positives and incidentalomas. The whole enterprise is probabilistic, not deterministic. There are also situations where the method breaks down entirely. Atypical presentations in elderly patients are the classic example. A UTI in an older adult might present as confusion without urinary symptoms. Myocardial infarction might present as fatigue alone. The symptom repertoire shrinks and the usual patterns disappear. In these cases the differential diagnosis process needs to be broader and more tolerant of uncertainty than it is for younger, clearer-presenting patients. Resource availability is another hard constraint. In settings where CT scans aren't accessible or labs take days to return, you're making decisions with incomplete information more often than you'd like. The method still applies but the safety margin narrows. Follow-up becomes more important because you can't confirm your suspicions immediately. A plan that includes clear return precautions and a re-evaluation window is essential when you can't rule things out definitively on the first visit.

The best workaround for limited resources is structured observation rather than blind reassurance. Instead of sending someone home with "take acetaminophen and return if worse," you give them specific triggers and a time frame. Return if the pain becomes pleuritic, if you develop unilateral leg swelling, if the headache changes in character, if you can't keep fluids down for twelve hours. Specificity in return instructions reduces anxiety for the patient and gives you a safety net when you can't do exhaustive workup upfront.