We’ve built a healthcare system defined by specialization. This can be one of our greatest strengths and also one of our most persistent blind spots. Unfortunately, these specialist and subspecialist silos are creating more complex care for our patients and often causing delays in diagnosis and treatment, as testing can be redundant and records not easily transferable.

However, patients don’t experience disease in silos. Yet we evaluate them that way.

As specialists, we are trained to filter symptoms through the lens of our domain expertise. That’s efficient when we’re right. It becomes limiting when we’re not. And in complex patients, particularly those with chronic autoimmune diseases, the margin for diagnostic error widens in subtle but consequential ways.

One of the most underutilized tools in closing that gap is not a new diagnostic modality or biomarker. It’s the patient. More specifically, it’s a patient who has been explicitly taught when, and how, to advocate for themselves when the clinical trajectory doesn’t make sense.

Once a patient carries a diagnosis like lupus, that label follows them into every room.

Fatigue? It’s your lupus.

Toe pain?It’s a lupus flare.

Shortness of breath? Probably your lupus.

Sometimes those are the appropriate answers. But more often it’s probabilistic reasoning at work. When this occurs over time, probability can quietly harden into assumption, and the assumption is that the patient’s complaint is not “my area of expertise.”

We don’t talk enough about diagnostic momentum: the tendency for prior conclusions to propagate forward without being actively re-examined. In a fragmented system, each specialist may be making a locally reasonable decision without stepping back to reassess the global picture. This happens more often when patients have more than one organ-specific complaint or concern.

I recently saw a long-term patient with well-controlled systemic lupus erythematosus who developed progressive shortness of breath. This was not a persistent flare. Her labs were stable, actually improving from a disease control standpoint. She was improving in other areas with her medication, yet continued to see me, expressing a feeling of panic when she would lay down. She felt like she was suffocating unless she slept sitting up. These symptoms did not align with her usual disease pattern. From a rheumatology standpoint, something didn’t fit.

She did exactly what we would expect: She followed referrals. Pulmonology evaluation was unremarkable. No effusion or serositis. Steroids did not change the way she felt, other than exacerbating her feeling of shortness of breath. She was noted to have a hiatal hernia, but it was not felt to be contributing to her symptoms. ENT evaluation was normal, and she was empirically treated for reflux. This did not change her symptoms.

At each step, there was reassurance, but no resolution. And consistently, she reported to me at every follow-up visit that she was told her lupus was the rationale for the way she felt.

That’s where many patients stop. Not because they’ve actually improved, but because they’ve reached the perceived boundary of what medicine can explain.

She didn’t stop. She pushed. We pushed for a second opinion. She continued to question the attribution of her symptoms to lupus — not out of defiance, but because it didn’t match her lived experience of her disease. We talked extensively about what lupus meant from an immune standpoint and the things we should be seeing if her lupus was not controlled. I adjusted her medications in the event that I was incorrect, and she still did not feel better.

Ultimately, she saw a second gastroenterologist who pursued further evaluation. Advanced testing revealed an esophageal tumor causing positional airway compression. When the tumor was removed, her symptoms resolved.

Her lupus was never the problem. The problem was that the limitation of our testing did not show a mass on imaging in the traditional format. The problem was that she needed “eyes” in her esophagus to see what the actual problem was. She needed someone to look and not to blame her symptoms on her medical history, but to see a new problem for something unique.

At her last visit with me she was feeling better than she had in years. She was feeling better because her lupus was not the only thing causing her symptoms. And it was not “all in her head.”

So how can we learn from patients like these and change our own thoughts about symptoms when a patient presents as a consult?

Every clinician involved made reasonable decisions based on the data in front of them. That’s precisely the issue. Our system is optimized for point-in-time decision-making, not for longitudinal pattern recognition across specialties. When no one owns the full trajectory, subtle diagnostic mismatches can persist longer than they ideally should.

The question is not whether each step was defensible. The question is: Why did it take so long to pivot?

We spend years training physicians to recognize red flags. We spend almost no time teaching patients the same skill set. Instead, we inadvertently train them to accept reassurance as a stopping point, even when their symptoms persist. If we want to reduce diagnostic delay in complex cases, we need to normalize something we rarely practice: patient-driven escalation. That means giving patients clear criteria for when “this isn’t working” should trigger re-evaluation.

What This Looks Like in Practice

This is not about encouraging unnecessary utilization. It’s about precision in follow-up. We can replace false closure with contingency language.

Instead of: “Everything looks normal,” we can try, “Your testing today is reassuring, but it doesn’t fully explain your symptoms. If this doesn’t improve, we need to keep going.” This subtle shift preserves diagnostic openness and follow-through.

We can define failure of treatment explicitly. Most patients don’t know when persistence becomes a signal.

Concrete messages can look like: “If this hasn’t improved in four to six weeks, that’s not expected — we need to reassess.” Or, “If the medication doesn’t change anything, that’s useful information. It means we may be asking the wrong question.” These skills help teach patients how to interpret nonresponse as data and not defeat.

We can proactively normalize second opinions. We should be the ones saying: “If this doesn’t get better, I’d like you to get another perspective.” Second opinions should not be viewed as a threat to clinical authority; they are a hedge against complexity. We can learn from them and catch that scenario quicker when it comes around the next time in a different patient.

We can teach patients to trust pattern deviation. Patients with chronic disease are highly calibrated to their baseline. When they say, “This feels different,” that should carry weight. We can reinforce this messaging with reassurance. “Even with lupus, new or different symptoms deserve a fresh workup.” Repeating labs or imaging is not a failure or wasteful. More data lends to increased probability of a correct diagnosis. Showing patients that you are willing to look at something again also demonstrates that you trust what they are telling you is happening to their bodies.

We can frame advocacy as clinical data, not friction. Patients often hesitate to push back because they don’t want to be perceived as difficult. We can remove that barrier and work with them rather than against them. “If something doesn’t make sense to you, I want to hear it. That helps me take better care of you.” When advocacy is reframed as collaboration, engagement increases.

Why This Matters Beyond Our Specialties

This is not just about one patient or one diagnosis. When patients are empowered to re-engage the system appropriately we can decrease diagnostic latency, reduce errors and unnecessary testing, improve collaboration across specialties and primary care, and develop more trusted and dynamic referral networks. Even in large health systems where patients are referred to a colleague “down the hall,” we can improve the clinical questions we are asking each other.

We often think of patient advocacy as something we do for patients. In reality, one of the highest-leverage moves we can make is teaching patients how to advocate within the system when we’re not in the room. We aren’t present for most of their clinical journey, their day-to-day struggles. They are the only constant variable and we can trust that they understand the way their bodies feel.

Where To Go From Here

The goal of medicine is not to be right on the first pass. The goal is to converge on the right answer as efficiently as possible. That requires recognizing when the current working diagnosis is no longer serving the patient. And sometimes, the person best positioned to recognize that inflection point is not the clinician, but the patient. If only we can teach them how to have that voice.

My patient didn’t improve because of a single test or a single specialist. She got better because she refused to accept an explanation that didn’t fit; and because, somewhere along the way, she learned that she was allowed to keep asking.

We should make that lesson explicit. Because “it’s just your disease” should never be the end of the conversation.

When do you invite a patient to push back on your clinical reasoning? Share in the comments.