I have a been a fully practicing PA-C for 2 months now in the field of outpatient internal medicine. It’s been quite an experience. Let’s talk.
Things are getting easier, slowly but surely. I hit a personal best of 12 patients in a single day (most providers see around 16) which I’m pretty proud of. And a good amount of these are my patients. They are either transferring from another office, needed a follow up after an ER visit and had no PCP, or just have haven’t seen a PCP in something like 30 years; they all needed a provider and guess who has wide open availability? For these patients, now when you look at the “PCP” area of Epic, it’s just my clueless grin staring back. Actually, my photo isn’t even loaded into Epic yet so it’s just a broken thumbnail with a question mark. Perhaps that’s more fitting for now.
I’m really leaning into primary care. The United States Preventive Services Task Force (USPSTF), which was once my sworn enemy in didactic year, has become my best ally. Every single A & B recommendation is printed out and pinned to my office wall. I’m screening my patients for HIV and hepatitis C until the cows come home. Sprinkle in a little zoster on the left arm, boom (and I’ll see you in 2-6 months for dose 2), let’s renew tetanus on the right, bam. Colonoscopy? Let’s get it on the books (and make sure you have a driver).
I am prevention.
Put the hoagie back in the box
It can be quite a jarring transition—from student to provider. Blood pressure medications that were once nothing but a multiple choice answer on an exam or a location in a memory palace (if you know, you know) are now sitting in a bag on the CVS pharmacy shelf with my name as the prescriber. That’s wild.
I was worried about the stress of work and how it would carry over into my real life. So far it’s been okay. I work half days on Thursdays (I go in at 1PM) and one particular morning when I was at home, I received a message from a cardiologist that a patient I had sent for an echocardiogram had a significant drop in their ejection fraction. I paused Con-Air which I had been watching and put down my Wawa hoagie that I was enjoying. That’s all the message said. No advice. “Your patient’s left ventricle is only ejecting half of what it’s supposed to; put down the turkey shorti and deal with it, kid.”
I woke up the other night at 1AM with my mind racing. I was trying in my dream to ask OpenEvidence how GFR is calculated if you only have one kidney. “Damn, that patient should probably be put on an ACE-I or an ARB. Was I supposed to do that? Will nephrology take care of that? What’s the difference between a urine-albumin creatinine ratio and a urine-protein creatinine ratio?” I typed all of this into OpenEvidence, got all of my answers, and went back to sleep.
Chasing Zebras
In medicine, a "zebra" is a slang term for an obscure, rare, or unexpected medical diagnosis. It comes from the classic medical school teaching: "When you hear hoofbeats, think horses, not zebras". It’s also rumored that the original title to the TV show House was “Chasing Zebras.”
Regardless, as a new provider, zebras constantly chase through your head. Could this patient who fell really have a spleen laceration? What do you mean you feel like you haven’t been emptying your bladder completely and you have back pain? Am I really going to send this patient to the ER for a stat MRI to evaluate for cauda equina syndrome? I’m learning a lot about the art of the gestalt. You sometimes just have to trust your gut that the dizziness isn’t a cerebellar stroke. Not everyone is dying or is presenting with that <1% condition that was on the PANCE.
There’s a certain comfort in primary care that you are the jack of all trades and the master of none. You can always refer to a specialist. It’s very, very hard, however to understand where primary care ends and specialties begin. I’d say that’s one of my greatest struggles, especially when certain specialties can’t see patients for over 6 months.
I wanted to shout out my physical/occupational/speech therapy colleagues. My goodness are they useful. Any sort of musculoskeletal pain? PT. Hit your head? Concussion PT. Feeling off-balance or have vertigo symptoms? Vestibular PT. Incontinence? Pelvic floor PT. Everything hurts? My friend, have you tried PT?
By the way, dizziness… good lord is that a challenging complaint. I built a SmartPhrase algorithm NASA would be proud of in order to help myself get to the bottom of one of the most vague yet potentially dangerous chief complaints of all time. I did a Dix-Hallpike test the other day to evaluate for vertigo. I started on the right… turning my patient’s head to 45 degrees and then swiftly laying them backward so their head hung off the table. I stood there, staring intently into my patient’s soul looking for the slightest nystagmus. “So how do you feel…?” “Um, fine,” she responded. I repeated the test on the left side, not expecting anything. As I brought her down, before I even had a chance to look at her eyes, she nearly jumped off the table. I actually had to embrace her to prevent her from falling.
“My god it actually worked,” I thought. One guess to where I referred her to.
What would you say… you do here?
One of our MAs came up to me on a day I only had a couple of patients scheduled and asked, like we were in Office Space: “So… what do you do every day?”
That’s an interesting question for me. I’ve had desk jobs in my past career where I was just clearly unengaged and there really wasn’t much to do at times. If someone would have asked me that question, I’d never have a great answer. When you haven’t yet found your passion, you spend time waiting for more work instead of being proactive.
But here, in my office, I spend every minute trying to get better. Whether it’s chart reviews for upcoming patients, comprehensive MyChart responses to a mildly abnormal lipid panel, printing out entire DynaMed pages, reviewing my school notes, pouring questions into OpenEvidence, or crafting SmartPhrases and patient education, I’m always doing something. So I showed her something I was working on. I built a “Lipid Profile Worksheet” that forces a provider (me) to not only look at each individual value of a lipid panel, but look at trends for each one. It’s easy to look at a snapshot of a normal lipid profile and quickly respond with a “Everything looks fine. Have a great day.” But I’ve picked up on concerning trends: a slowly rising LDL (even if it’s still in the normal range) or a steadily declining HDL that I’ve notified patients about. And in this worksheet I have action items that a patient can do to address each value. The worksheet which lives as an Epic SmartPhrase automatically pulls in the current lipid panel and the past 5 readings for each individual value. And I have SmartList drop downs that I can just fill out quickly whether the value is stable, or unstable and needs work. So it’s very straightforward and ensures that I provide every single patient the same level of detail.
Speaking of SmartPhrases (also called “dot” phrases), I’ve become a man obsessed. There’s an entire science to automation and shaving seconds off of workflows in Epic that over time can be the difference between going home on time, staying late and, even worse, staying late and then finishing charts at home (what we in the business call pajama time).
Keep calm and… dive into the weeds?
Also, speaking of school notes, I use them, a lot. I’ve talked about this in past blogs, but it’s so important to craft your study materials as something that you cannot only refer to in your clinical year, but something you can carry into clinical practice, which is exactly what I’m doing. The charts I made in school are pinned to the walls in my office. I’ll add to my school notes and fill out an entire profile for a medication as simple as Metamucil: Every FDA use and every off label use, dosing schedules, side effects, etc. Did you know that an FDA use for Metamucil is reduced risk of coronary heart disease? Then, I’ll pop into OpenEvidence and ask specific questions like: “When exactly in relation to a meal should this patient take this (that answer for psyllium depends on whether the patient wants to lose weight!)? When can we bring in Colace or Senna? What actually works and what doesn’t?” Did you know that docusate (Colace) is no more effective than placebo?
I’m learning that the nitty gritty details of patient education truly matter in primary care. Take iron for example. It’s common for us to recommend a patient take it with vitamin C to improve absorption, but there’s so much more. How many patients do you think take all of their medications in the morning with breakfast? Probably a good amount. But iron needs to be taken 30-60 minutes before a meal. Did you know that tea can reduce iron absorption by 90%. Ninety percent! Milk, yogurt, PPIs, multi-vitamins with calcium, and coffee also reduce absorption. How many patients do you think consume any combination of those things in the morning? How many patients are on a vitamin D supplement (that we recommended) that has calcium in it that is reducing the efficacy of the ferrous sulfate that we also recommended? I had a patient the other day call with advice on when and how to take her iron pill. I spent nearly 10 minutes walking her through all of the instructions. By the end of the conversation, she said: “Oh, wow, I would have broken all of these rules.”
Sell me this… colonoscopy
I’m learning that you need to be a bit of a salesperson in this role. When you have a new patient who has had zero bloodwork for decades, no colonoscopy, no mammogram, needs a hepatitis B, RSV, COVID, pneumococcal, tetanus, shingles, etc. vaccine, you have to pick and choose at that first visit. Oh and their blood pressure was also elevated in the office, so you have to start them on ambulatory monitoring. And their back hurts and they need PT. And there’s a wart on their foot. And their BMI is 35. There’s literally nobody saying I can’t see them in 2 weeks, or a month, or do a video visit tomorrow. An hour ago they had no healthcare, so I take comfort in the fact that I am at least starting them on a new path. If they’ve had masked hypertension for the last 30 years, they can wait a couple of weeks until we collect some ambulatory data.
I was explaining a mammogram to a patient who was quite hesitant to get one. I then explained the colonoscopy that she was also due for. “Hey, the mammogram doesn’t sound so bad now, does it?” I actually said this to her. We laughed about it. By the end of the visit, she agreed to get both.
The other thing too is understanding the level of responsibility as provider and where that responsibility ends. I have to understand that the patient-provider relationship is a partnership. I’ve evaluated the patient, made my recommendations, and provided education. It’s up to the patient to fill those scripts, monitor their blood pressure, and get that repeat BMP next week.
I had a new patient come in with his daughter during my second week. He hadn’t seen a provider in 20+ years. His blood pressure was 210/110 (my manual BP-taking skills are now off the charts). He was having headaches and dizziness (otherwise neurologically intact). Off to the ER he went. A couple of hours later I received a message from another PA in the ER: “Hey, I have your patient here… doing my best to get his BP down. What do you want to start him on when we discharge him?”
What do I want to start him on? I pushed aside my Wawa chicken bowl (I always seem to be consuming Wawa when I get these sort of messages; I don’t want to talk about it). A man I had met mere hours ago was now my responsibility, my patient. There are times I wish I could adorn my Epic profile with a bumper sticker that says “I’m new at this, please bear with me.”
I did some research on a good starting regimen and gave the PA my answer. I saw the patient 2 days later, then 2 weeks later, making blood pressure medication adjustments each time. I always included my at-home ambulatory blood pressure monitoring instructions (a SmartPhrase that I crafted of course) and we went over it at every visit. I finally had his pressure relatively controlled and could proceed with the preventative visit from which I had first met him. It was just such a great experience for me. His daughter was happy. He looked so much better. Myself, and the ER team potentially prevented this man from having a heart attack or a stroke.
“Okay now that your BP is controlled, about that colonoscopy you’re due for…”
OpenEvidence
Let’s talk about OpenEvidence. For the uninitiated, it’s an AI tool like ChatGPT, but for medicine. It’s stupid useful and provides links to medical journals and medical trials for every answer. I use it nearly every minute. Using literally any other resource like DynaMed or UpToDate is becoming less valuable. It even asks me questions back… “Hey you mentioned they had a CT scan a few years ago, what did that show?” It makes me feel better about not exhaustively reading medical journals, subscribing to newsletters, listening to podcasts, etc. I can ask questions in real time and it provides the latest guidelines.
You can even ask it “meta” questions. “Hey, what labs do PCPs usually order for no reason (TSH and CBC, by the way)? What are common pitfalls for new providers during a physical exam? What are some new trends for 2026 that I need to know about? Hey, can you cross reference this list of the top 25 most prescribed medications and tell me a patient pearl and a myth about each? Hey, I am attaching a screenshot of 15 medications this patient is on… any additive interactions here? Go.”
I don’t even know how I would find this information without AI. It just goes… “Yea here there was a RCT or a meta-analysis about that exact question… it was just published last month: here are the results.” It’s also incredibly helpful for patient education. “Hey, can you write me some patient instructions on evidence-based best practices for treating migraines?” Done.
It’s not perfect and definitely errs on the side of wanting to over-work up people, but I’d rather have that than the opposite. What’s really fascinating to me is that unlike other technologies like going from paper charts to an EMR, using AI is optional. But I don’t think it should be. Like any technology since the dawn of the human race, it’s met with resistance. If you’ve been practicing medicine for 30 years, why would you need something like OpenEvidence?
Here’s why.
I asked OpenEvidence the following question: “If there is a new guideline... for blood pressure, stroke, whatever, how long does it typically take for that to be discovered, agreed upon by committees and organizations, written, published, disseminated to health organizations and hospitals, communicated to clinicians, read by clinicians, and actually put into clinical practice for patients?”
The answer? Anywhere from two to fifteen years.
“A concrete example: tissue plasminogen activator (tPA) for acute ischemic stroke was FDA-approved in 1996 based on RCT evidence, yet by 2004 only 3–8.5% of eligible stroke patients were receiving it — an 8-year gap with still minimal adoption.”
I’ve been in healthcare for 15 years. I’ve seen how slow it takes to upgrade an enterprise’s PCs to the latest edition of Windows.
I apologize to the environment for my AI usage… but I have blood pressures to lower and patients to educate over here! I drive a hybrid; let’s call it even.
Wrapping Up
So yea, there’s where I’m at. Right now I’m constantly wavering between “Okay, I’m doing okay…” and “I’m so going to get fired for missing something huge.” But I think early on, that’s a healthy place to live.
I’ll see you in the next one.
For this post’s song, I think it’s fitting for an “AI song”. Naturally, AI is creeping into the two major aspects of my life: medicine and music. There’s an “artist?” under the name Rustic Poem Records who makes these twangy, americana, slow-burning wedding reception-esque pop-country covers of pop-punk/alternative songs. They are insanely good. There’s a cover of Sum 41’s In Too Deep, that you really just have to listen to. Ethical considerations of AI music are still developing, but I can objectively appreciate a great composition regardless.