This post also appeared on KevinMD.

Software has opinions. No, I’m not talking about opinions on the next presidential election or opinions about flossing before or after brushing. Software has opinions about how data should be displayed, opinions about users’ comfort with the mouse, even, in some cases, opinions about what you should have for dinner (see your local on-demand food ordering service).

We tend to view software as a tool that is either good or bad. Good when it lets us do what we want with as little frustration as possible and bad when it doesn’t. Maybe we should be a little nicer to software.

In 7 years of medical training, I’ve discernibly used, to varying degrees, between 5-10 different electronic medical record systems, not to mention different versions of the same one or the dozens (hundreds?) of “hidden” layers of information systems. They all did effectively the same things: results viewing, documentation, order entry, and maybe a few other things that were less important. Each of them expressed different opinions:

“I think most recent results should be on the left.”

“I think it should be easy to copy-forward yesterday’s note.”

“I think medication orders should list as needed and standing medications separately.”

These examples are simple, but they can have substantial implications for workflow. For example, there is a body of research demonstrating that progress notes in systems where it’s easy to copy-forward yesterday’s note look meaningfully different from those that don’t. The workflows are also very different. If you’ve been practicing on one workflow for a nontrivial amount of time, having to change to another one is going to seem like a very bad idea.

Sometimes we agree with the software’s opinions and like it. Sometimes we don’t and don’t. Of course, software’s opinions are really the opinions of the designers and thus I arrive at the tautology that all software is designed. Of course, everyone knows that.

My call to action is simple. We need to 1) recognize that software has opinions and 2) design software that agrees with our opinions and our values. This is particularly true for the electronic medical record which has an outsized influence on the quality of care, patient safety, and physician satisfaction. In practice this is hard because individual providers have justifiably different workflows. The answer will not be accommodating everyone’s style (having no opinion can be just as frustrating) nor will it be forcing everyone to do the same thing. Instead, it will be figuring out what our opinions are, which ones we ought to share, and using those opinions to inform the design of future versions of the medical record.

In that spirit, I would love to move away from the thinking “this EMR is [good/bad]” or even “this EMR is [well-designed/poorly-designed].” Remember: EMRs have feelings, too. We should be saying “this EMR is opinionated that…” My hope is that this will elevate the conversation from one of complaint to one that draws focus to what features and design principles are important and maybe get us to the EMR we all want sooner.