We are not yet entitled to good healthcare
We are so far from the world where perfect healthcare feels like an entitlement - a prerequisite for being human. Most of us still receive care as something granted, not owed. That mindset shapes how patients behave at the bedside in ways that doctors, especially junior ones, rarely fully appreciate.
I’ve recently occupied three roles in quick succession: provider of healthcare, supporter of a family member receiving it, and patient myself. I’ve never felt as vulnerable as in that last scenario. There is a thread of self-abandonment that runs through the human psyche when we become patients - a quiet sense that we are receiving something we have no birthright to. It is the kind of feeling that leaves a school principal waiting a full week before mustering the courage to ask the nurse for a toothbrush.
A lady with severe back pain can spend weeks after discharge without using stronger analgesia - immobilised, frustrated, exhausted - yet still have the presence of mind to ask the medical intern doing her review whether he has eaten lunch. Patients hold themselves to standards of decorum and gratitude that we, as clinicians, often fail to even register, let alone push back against.
I wonder how much of this is gendered. You wouldn’t necessarily expect it in a well-educated, culturally Western country, but the seeds of inequity are resilient. Women, in my experience so far, are more likely to defer, to minimise, to ask after others before themselves.
Trust, education, and the patient who won’t rock the boat
When a patient cannot trust her own body, she has no choice but to put that trust in the healthcare provider. And here, education is still very much a privilege. If she knew the answers, would she not guide her own care better than a stranger in a white coat could?
What role does the patient actually play in their own health?
I think of a lady in her seventies, on Capox - capecitabine and oxaliplatin - for colorectal cancer. She has never once told her oncologist about her substantially debilitating post-chemotherapy sensory ataxia. Why? She wants to get better. She wants to fix her cancer. The chemotherapy is important to her. She doesn’t want to rock the boat. She doesn’t want to be a chook.
And then, after a few minutes of silence, her voice shakes as she asks: “Is the cancer in my brain then?”
She has been carrying that fear, alone, for weeks. Not because the system couldn’t help her, but because the social contract of being a “good patient” is more powerful than the clinical contract of disclosure. We - the healthcare system - built that asymmetry. We benefit from it in efficiency. We pay for it in suffering.
The most important project of medicine
I’m increasingly convinced that medical suffering is ultimately a psychological phenomenon.
Patients die frequently with prostate cancer, not of it. The pathological process never reaches the threshold of affecting their phenomenological experience of living, and there is no psychological response to something that does not announce itself. We intuitively know this already - overinvestigation finds benign lesions, and those lesions only ever serve to worsen suffering. The patient would have been better off not knowing. Our entire screening discourse rests on the implicit axiom that suffering is correlated with, but not caused by, pathology.
If that is true, then the most important project of medicine is not what we think it is.
We aim, throughout training, to gain mastery over pathology. But perhaps the most important skill a healthcare provider can develop is mastery over the management of medical suffering. Pathology is the substrate. Suffering is the experience. We’ve built a curriculum, a research engine and a culture around the substrate, and have largely left the experience to chance - to the personality of whoever happens to walk into the room.
The therapeutic value of certainty
My father has had a few cardiac stents placed and lives with an ongoing mild occlusion of D3. He is hypersensitive to any chest or shoulder pain - which is unfortunate, because he also has cervical radiculopathy and frequently variable left-sided upper back and shoulder pain that has nothing to do with his heart. A rising sense of anxiety fills him with each new twinge.
The only phrase that consistently calms him is what his cardiologist told him in his mother tongue of Tamil: “It’s like a small dog - it barks a lot, but it doesn’t bite.”
That sentence has done more therapeutic work over the past few years than any of his medications. And yet, in the current medicolegal paradigm, it is the kind of sentence we are trained out of saying. We are taught to speak in probabilities and never to offer certainties. We hedge.
What good does it do to tell a health-anxious patient with a fistula that “yes, perhaps it could cause sepsis, but it’s unlikely” - when she has already found rare case reports and WebMD threads that corroborate her worst fears? In that moment, the therapeutic relationship has failed. The pathology has not changed, but the suffering has multiplied. We discharged the legal risk and inherited a clinical one - and the patient is the one who pays for it.
The uncomfortable conclusion
I have always found it interesting that Claude can terminate a conversation, or push back on a user’s framing, or make a decision on the user’s behalf. It makes me suspect that frontier LLMs may already be capable of outperforming doctors at the parts of medicine that actually matter most.
Medical knowledge is, almost certainly, better encoded and retrieved by a model than by any individual clinician. Given unbridled access to high-quality research, it is very likely better reasoned with too. The clinical decisions are better.
But that’s not the part I find most striking. If a model could build a stronger relationship with my lady with back pain - be infinitely patient with her questions, gently tease apart pain requirements that she habitually downplays, remain non-judgemental and compassionate at 2am as readily as at 10am - then I would replace every healthcare worker with it in a heartbeat.
That sentence is uncomfortable. It should be. It is meant to be.
If our discipline’s most important project really is the management of medical suffering, and if emotionally intelligent AI can do that better, more consistently, and more equitably than we currently do - then we have a duty to integrate it, not resist it. The bar is not whether AI can replicate the median clinician. The bar is whether it can finally give a school principal, on his fifth day, somebody patient enough to ask him whether he has a toothbrush.
That is what patients have been teaching me. They have been teaching me what medicine is actually for.