Clinical guidelines are written for populations, but hospitalists treat individuals, and most of those individuals come with comorbidities that complicate the evidence. The patient with atrial fibrillation, recent falls, and cognitive decline doesn’t fit neatly into the case for anticoagulation. A postarthroplasty knee infection can pull infectious disease and orthopedics in opposite directions on the same morning. Recommendations diverge, specialists disagree, and someone still has to decide.

“All good physicians practice evidence-based medicine. But when patients fall between the lines and don’t fit into a particular category, that is where the art of medicine shines through,” said Precious Barnes, DO, a hospitalist at Skagit Regional Health in Mount Vernon, Washington.

Why Recommendations Diverge

Different societies write guidelines from different evidence bases, and the trial populations behind them often exclude the older, multimorbid patients hospitalists actually see. Each set of recommendations can make sense on its own and still pull against another when both apply to the same patient.

Anticoagulation guidance for atrial fibrillation bumps up against fall risk in geriatric care. The diuretic that pulls fluid off a patient with heart failure can also tank a marginal blood pressure. The gap between the patient a guideline imagined and the patient on the unit is where most of the conflict shows up.

“While there is almost always sound evidence and logic behind recommendations based on guidelines, sometimes patient-specific scenarios can pose specific limitations,” said Nagendra Gupta, MD, a hospitalist at Texas Health Arlington Memorial Hospital in Arlington, Texas.

He pointed to an older woman with atrial fibrillation living alone, with progressive cognitive decline and frequent falls. Stroke prophylaxis on paper. Bleeding risk in practice.

Frameworks for Real-Time Decisions

Hospitalists describe approaches that share core elements but differ in emphasis. Gupta works from the consultant outward, assessing the reasoning behind each recommendation before bringing the patient and family into the decision.

“It is extremely unlikely that a consultant would give a particular recommendation without any underlying basis. If the reasoning is sound, then involving the family in the decision-making always helps arrive at a consensus,” he said.

For Barnes, the work often starts with mechanism. She described a patient with heart failure who was severely volume overloaded but hypotensive, in a situation where the diuretic that would relieve the overload could also drop blood pressure further. The bedside nurse hesitated.

“I explained to her that I was aware but talked her through the physiology of why I chose that medication,” Barnes said. “Yes, this medication can cause severe low blood pressure, but if given in the right circumstances this medication will pull the fluid back into the intravascular system and essentially raise his blood pressure back to normal.”

A compromise half-dose was given. The pressure rose. The remainder followed.

Erhime Badejo, MD, MSN, a board-certified family medicine physician in Carbondale, Illinois, said her real-time framework leans on pattern recognition built through repetition.

“We run high-fidelity simulations over and over using different scenarios, different variations, so when something comes in, it’s not the first time we’ve seen it,” she said. “That’s what allows you to move quickly without feeling scattered.”

When Two Specialists Disagree

Gupta described a recent case that shows the pattern at its sharpest. A patient presented with redness and swelling around a knee joint following a recent arthroplasty. Infectious disease recommended re-exploration. The orthopedic surgeon who had performed the procedure believed the infection could be managed with antibiotics. Two specialists, two opposing recommendations, one patient.

Conservative treatment was started. Over the next several days, the patient did not improve, and inflammatory and infectious markers told the story. White blood cell count, procalcitonin, and C-reactive protein levels were all rising.

“Worsening infectious and inflammatory markers easily tipped the balance in this case, and the patient was eventually taken to the operating room for a joint washout,” Gupta said.

He framed his role in those moments as an advocate willing to push for a second opinion when trajectory and recommendation diverge.

“In situations where objectively the patient would need an intervention and the specialist disagrees, my responsibility as the attending physician is to advocate firmly for the patient, and to seek a second opinion in case of a disagreement,” he said.

Putting the Reasoning on the Record

Documentation, in Gupta’s practice, has a structure. He outlines the diagnosis, the recommendations, the reasoning for the alternative approach when one is chosen, and the discussion with the patient and family that produced the decision.

For the patient who declined anticoagulation, his note read: “Based on a CHAD-VASc score of 3, this patient would benefit from anticoagulation for stroke prophylaxis. However, in light of her advanced age and gradually declining cognition, associated with frequent falls, she remains at a much higher risk of adverse outcome due to the risk of hemorrhage in case she were to have a fall while on an anticoagulant.”

The note recorded that the medical necessity, the bleeding risk, and the alternative of physical therapy for gait stability were discussed with the patient and family. They declined anticoagulation at discharge.

“The key here is not to point fingers or play the blame game,” Gupta said.

Barnes carries similar instincts about written reasoning, drawn from a mentor early in her training.

“No one can fault your clinical judgment if you do something off label as long as you have sound documentation,” she said.

Where the System Pushes Back

Barnes recalled a patient who had recently started methadone for opioid use disorder, supplemented with fentanyl when his outpatient dose proved too low, and overdosed before his mother revived him with naloxone. Once admitted, he became aggressive as withdrawal returned. Barnes ordered a methadone dose calibrated to his clinical picture. The hospital pharmacist refused to release it, citing protocol and the absence of records confirming the prior outpatient dose.

“I had to explain to him that the reason for the protocol was because in the outpatient world we do not have control over the patient and their situation. In the hospital, I know exactly which medications he is taking, and I am able to get regular blood work to see how the medication is affecting his system,” she said.

The dose was eventually released, and the patient calmed.

“It takes a lot of people to run a hospital, and checks and balances and guidelines are necessary for patient safety, but when they delay quality care is when they become an issue,” Barnes said.

The Bedside Reality

The frameworks differ. The friction points repeat. The gap between guideline and patient stays the work.

“Cases like this are a reminder that what looks like a clinical dilemma is often a system issue showing up at the bedside, where the medicine points you in the right direction but execution depends on access, coordination, and adaptability in real time,” Badejo said.

No reported disclosures.