Computer algorithms don’t think like doctors

Too often, residents want you to address something, so they don’t have to — except for infectious problems where they putz around with antibiotics until lunchtime on Friday, then call ID.

For me, one example seemed rather routine: a diabetic with another medical illness. It wasn’t terribly well defined in the hospital records, but included atrial fibrillation and congestive failure at presentation. At day nine, with pressure from the DRG lady to discharge pronto, they figured that might be a good idea to have some predictable insulin dosing and reliable office follow-up, so they called me.

Nine days of hospitalization gives a lot of electronic record clutter, and most of the progress notes were the usual copy and paste of limited intellectual input on their part. Of note, the magnesium level never quite corrected. When I examined her, there was a Mg rider hanging, with the last measured level 1.7 mg/dl — a little low, but not dangerously so. I finished the exam, decided what to do with the insulin and went back to the hypomagnesemia. On presentation, she was in atrial fibrillation with a serum Mg 1.0, so nobody would look askance at the two IV infusions she received in the ICU. However, on day nine, she was receiving infusion #9 for a very borderline result. I went back through the lab testing and notes that were absolutely devoid of any search for cause or any discussion short of the orders for repetitive IV replacement. As I typed the consult in the computer room, I asked who was responsible for her care on the floor. The resident two screens down owned up, so I asked him about this. Well, the Mg was low, so he replaced it.

Well, does everyone need to be euboxic — a term that had pretty much disappeared from medical slang at about the time he was born? Of course not. And more importantly, if you do nothing to fix it and send her home with neither daily monitor or replacement, what did he think would happen to her? If the answer was nothing, she got at least six infusions too many. Excessive care is a variant of wrong. Thoughtless care sometimes goes beyond wrong to negligent.

Unless it’s an electrolyte problem, I’ve seen people get ten amps of D50 for low finger glucoses and normal sensorium in the absence of hypoglycemic agents only to find that the venous glucose done simultaneously was normal. If they really have a hypoglycemic disorder, they deserve diagnostic testing which starts with a bedside assessment. Even with a prolonged fast for insulinoma, the blood doesn’t get drawn in the absence of symptoms — even if the glucose reads low. If they do not generate enough capillary blood to give a proper measurement, their fingers and the pharmacy’s D50 supply should be spared.

I see two issues that are very common, and neither addressed well. We seem to teach by algorithm — if this; do that. The first event will probably get you by. The next one should arouse some suspicion, either to read the chart, see the patient, get a consult, or at least put on the thinking cap. The second failure may be lack of accountability. The resident sitting two screens over was one of four that had responsibility for her care over those nine days. There was an attending hospitalist too, a bystander for the days in the ICU and preoccupied with CHF and two resident teams to get the detail. Nor do we have the pharmacy as a safety net to intercept questionable care, outside of antibiotic use where certain automatic reviews take place. I would think in this day of computerization, 5 Mg or K-riders or 10 amps of D50 might be more easily identified by the pharmacy than by rotating residents and hospitalists. Those are patients who need a little more than just being processed through in the shortest length of stay for their assigned DRG.

One of the elements of internal medicine that attracted me as a student and remains 40 years later has been the analytical challenge. That may be the final deterrent of burnout. The inquisitive mind can probably still overcome the irritations of the medical computer and the functionaries you talk to at the pre-authorization desks, as neither of them thinks as well as a methodical clinician.

Richard Plotzker is an endocrinologist who blogs at Consult Maven.

Image credit:

View 1 Comments >

Most Popular

✓ Join 150,000+ subscribers
✓ Get KevinMD's most popular stories