Automated detection of LASA medication errors


Look-alike/sound-alike (LASA) medications – also referred to as sound-alike, look-alike drugs (SALAD) (1) — have been a thorn in the side of healthcare professionals for as long as I’ve been a pharmacist.

Many solutions to the LASA problem have been proposed, including Tall Man Lettering (2), physical separation of look-alike drugs, printing of both brand and generic names on packaging and storage bins, use of colorful warning labels, and so on and so forth. The problem with all these solutions is that they involve humans. Working in acute care pharmacy has taught me over and over again that as long as humans are involved there will be errors.

Technologies can help. Automated carousel technology and robotics can help manage physical separation of the medications and eliminate visual bias generated by human eyes. Bar-code scanning can certainly aid in identifying medications correctly. Bar-codes don’t care that medications have similar names, they’re either right or wrong.

In a recent article by Rash-Foanio (3) et al. the authors use an algorithm to flag potential errors from LASA drugs when an order meets the following criteria:

  1. a medication order is not justified by a diagnosis documented in the patient’s record
  2. another medication whose orthographic similarity to the index drug exceeds a specified threshold exists
  3. the latter drug has an indication that matches an active documented diagnosis.

In the study the authors perform a retrospective analysis to identify errors that involved cyclosporine and cycloserine. The algorithm wasn’t perfect. Sixteen orders involving unique patients were found. Additional chart review of the errors discovered that 5 (31%) identified by the algorithm did not involve a medication error, i.e. the intended medication was correct. However, the algorithm correctly identified 11 (69%) LASA errors.

While it may not catch all LASA errors, it seems that EHRs should give AI and some deep learning serious considerations for items like this. Preemptively catching greater than 50% of LASA errors is better than catching zero. (5)


  1. I came out of pharmacy school having learned the phrase “sound-alike, look-alike drugs” (SALAD). At some point it changed to look-alike/sound-alike (LASA). Not sure when, how, or why it changed, but them’s the breaks. Just go with it. Adapt or die, I suppose.
  2. I’ve never been a fan of tall-man lettering, and it isn’t even clear that it works to reduce errors.
  3. Rash-Foanio, Christine et al. “Automated Detection Of Look-Alike/Sound-Alike Medication Errors”. American Journal of Health-System Pharmacy7 (2017): 521-527. Web.
  4. Kondrak, Grzegorz, and Bonnie Dorr. “Automatic Identification Of Confusable Drug Names”. Artificial Intelligence in Medicine1 (2006): 29-42. Web. 28 Apr. 2017.
  5. Honestly, one of the simplest things we can do is force providers to select an indication when ordering LASA medications. That simple act has the potential to significantly reduce these errors.

So I bought a Chromebook 

I’ve had a Chromebook for a while now. My wonderful wife got me the original Samsung Chromebook in early 2012 for my birthday, I think. The device has been a trusted couch companion ever since. I mostly use it to check email, watch YouTube videos, surf the internet, and so on. It has a small 11.6-inch screen and limited horsepower. The little guy was never meant to be used for any heavy lifting. I have a Windows 10 laptop for that.

A few months ago I got the itch to buy another machine. I would dearly love to have a Microsoft Surface Book or Lenovo ThinkPad Yoga. It is my personal belief that these are the two best machines on the market today. With that said, they’re expensive, and honestly, they’re overkill for what I do. I mostly do research on projects and create content, typically in the form of documents, spreadsheet, blog post, and presentations. My presentations can get pretty gnarly in terms of size and media content, but still, a Core-i7 with 16 GB of RAM just feels extreme for my needs. So why not a Chromebook? Yeah, why not, indeed.

So I set about making a list of things that I wanted from a Chromebook. Two of the most important items were:

  1. A 14-inch or larger screen. I’ve historically preferred smaller screens, especially between 12-13-inch. However, I’ve been thinking about trying a laptop with a larger screen for a while. And what better way to do it than with an inexpensive Chromebook?
  2. Long battery life. This is a no-brainer.

After the two items above, my list grew to include:

  • Under $500. Let’s face it, it’s hard to spend more than $500 on a Chromebook knowing that for around $700 I could get a really nice Windows 10 tablet.
  • Good keyboard. I do a lot of typing, so whatever machine I buy has to have a good keyboard.
  • Backlit keyboard. This seems to be a polarizing topic. I feel that it’s important, at least for me. I do a lot of early morning and late night surfing. I don’t always have a great light source, so having a backlit keyboard is nice.
  • At least 32 GB of local storage. Chromebooks are notoriously bad when it comes to local storage. I can’t see buying something with less than 32 GB of storage.
  • Touchscreen. Once you’ve used a laptop with a touchscreen, it’s hard to go back. It would be especially handy should my Chromebook support Android apps.
  • Expandable storage.  Any acceptable way to increase storage, not only an SD card slot. As long as there’s a free USB port, I’m good.

I was pretty excited when I saw the Samsung Chromebook Plus and the Asus Chromebook Flip C302 hit the market. Both have touchscreens and offer a 2-in-1 configuration. The Samsung pulled at me hard, but besides having only a 12.3-inch screen, it was also missing a backlit keyboard, and was at the very edge of my $500 mark. For reasons unknown, the Flip C302 doesn’t capture my interest. It has a backlit keyboard and better storage options than the Samsung, but meh.

After reading a lot of reviews and looking through countless “best Chromebooks of 2017” lists, I came to the conclusion that a 14-inch Chromebook with a backlit keyboard just wasn’t in the cards. I ended up going with the Acer Chromebook 14, in gold. I got a great price at Amazon.  It lacks almost everything on my list, but it does have a 14-inch screen and remarkable battery life.

The Acer Chromebook 14 has an all metal build, a good but not great 14-inch screen, a good keyboard, solid battery life, and it’s snappy. It’s also easy on the eyes. When up close and personal, no one would mistake it for a $2000 premium laptop, but it’s still a handsome device.

I’ve been using my new Chromebook for about two weeks. So far things have been pretty good, but I definitely have some minor quibbles.

There are several keys missing from the Chromebook keyboard. For example, there are no dedicated ‘delete’, ‘home’ or ‘end’ keys. I’m a content creator, so I use those keys a lot. Fortunately, there are Chromebook equivalents, like ‘alt’ + ‘backspace’ = ‘delete’, or ‘search’ + ‘right arrow’ = end of line, etc. I keep a copy of the Chromebook shortcuts handy. Fortunately, you can also remap certain keys. For example, remapping the relatively useless ‘search’ key to CAPS LOCK. Seriously, who doesn’t need to type IN ALL CAPS ONCE IN A WHILE. I’m slowly starting to get used to the keyboard function and layout, but it hasn’t been a smooth transition.

The “right-click” on Chromebooks is not a right-click, at all. In fact, the right click is a two finger tap. This tiny difference is giving me fits. Who knew that I right-clicked so often?

Everything being in a browser tab will take some getting used to. I’ve been using the Windows OS for more than 20 years. I’m comfortable with a bunch of little windows on my screen, not just a single browser with multiple tabs. Changing some of the web apps to “open as window” has helped. I did this for Google Docs and Google Music. It has nothing to do with functionality, but it makes my screen more familiar, which ultimately makes me feel better. Don’t judge me!

I’ve always considered myself a heavy user of web-based apps. I spend most of my time in the Chrome browser flipping through websites, reading literature, using Gmail/Google Calendar/Google Keep/Google+, using the web versions of Twitter and Facebook, and so on. Working in “the web” is great, except when it’s not. There are some web applications that simply don’t measure up to their desktop counterparts.

Here are some examples:

Evernote – I’ve been trying to extricate myself from Evernote for quite some time, but I haven’t found a suitable replacement. The application is too valuable to the way I work, so for now it stays. Unfortunately, the Windows 10 desktop version of Evernote is significantly better than the web version. Converting back to the “old” version of the web app has helped, but it’s still not the same.

Microsoft Office vs. Google Docs – As a content creator, I use Microsoft Office 365, a lot. I create documents, blog posts, spreadsheets with graphs, and presentations. For the most part, replacing the web versions of Word with Google Docs hasn’t been an issue. For example, I’m using the online version of Office 365, Word specifically, to write this post. I used Google Docs for Sunday’s post. Both worked fine. I prefer Word, but Docs is usable. Replacing Excel with Sheets is ok, but I’m struggling with graphs. Google Slides is another story. Slides is way behind Microsoft PowerPoint in my opinion. The desktop version of PowerPoint is much better than the web app, and the web app is still much better than Google Slide. This might be a deal-breaker for me. I’ve found a few workarounds for PowerPoint Online, but honestly, I don’t like workarounds. It’s not all terrible. I do like the auto-save functionality when using Google Docs or Office 365 Online. Every change I make is automatically saved. No more losing 20 minutes worth of work because I was too lazy to hit ‘CTRL S’.

Zotero – I’m a pharmacist. I read literature. I’m also a digital pack rat. I collect and store lots and lots of journal articles, whitepapers, presentations, and so on. I refer to them when composing blogs or papers and when building presentations. Using a reference management tool is a must. In my case, that reference manager is Zotero. I collect all my references in Zotero. It pulls complete bibliographies from digital object identifiers (doi’s), allows me to link to full article PDFs stored in OneDrive, and gives me the option to attach notes to each entry. New projects – articles, blog posts, presentations – get their own folder. Each time I reference a journal article for the project, the reference gets copied into the project folder. When the project is complete, I simply print a bibliography of all the references used. It works great. Fortunately, Zotero has a nice desktop application with great functionality and it’s free. Unfortunately, Zotero’s web app is an abomination. I don’t know how it’s possible to have such a great little desktop app and completely ignore “the web”. Of all the things that have given me trouble, this is the biggest. I don’t know if I can get past this one. The inability to use Zotero on the web has had a negative impact on the way I work.

Notepad – Yep, one of the simplest applications on the planet. I use it a lot. Sometimes I’ll use it for nothing more than a quick, disposable note. Other times I’ll use it as a quick and dirty way to eliminate all formatting associated with a bit of text, i.e. copy the text and paste it into notepad; this is a great trick when I’ve tried some crazy formatting and can’t seem to back out. Notepad is always open on my Windows 10 desktop. While there are lots of note-taking apps for Chrome, I haven’t found an equivalent. For the time being, I’m using Google Keep. It’s not the same, but it works.

In many ways, the Acer Chromebook 14 meets my expectations: it has a solid build, it has a nice 14-inch screen, good keyboard, the integration with my Android smartphone (Samsung Note 5) and Google Home is nice, it has ridiculously good battery life – I’m getting ten plus hours per charge — and it’s nice to look at. In other ways, the Acer Chromebook 14 has failed to meet my expectations: lack of a backlit keyboard, Android apps not yet available even though the device itself was introduced over a year ago, and lack of solid web-based apps. In all fairness, that last item isn’t the Chromebook’s fault, but it makes a big difference.

Overall, I find that the Acer Chromebook 14 is a great machine. I’m sure it could be a Windows 10 laptop replacement for someone, but probably not for me. At the moment, I’m still on the fence about keeping it. At sub $300, it might be worth keeping around. But then again, a new Windows 10 tablet would be nice.

EHRs are an untapped, but almost impossible to use, health resource

We’re all familiar with the promise of “big data” in healthcare. Crud, I’m a huge fan of using data. I think the amount of information inside an EHR has the potential to do a lot of wonderful things, not only for healthcare in general but specifically for a pharmacist. How many kinetic consults have been done by hand, tracked manually, and refined by voodoo magic? Thousands, I can assure you. The number of things pharmacists still do manually is staggering. “Monitoring” should no longer involve rummaging through charts — electronic or otherwise — looking for tidbits of information that need to be “fixed”. The days of dosing medications like vancomycin, warfarin, phenytoin, and aminoglycosides — just to name a few — should be long gone. We can contemplate building a Hyperloop, but we can’t figure out how to get someone’s INR to a therapeutic level within five days? Seriously, think about that for a second.

FierceHealthcare: “For public health agencies, tapping into EHR data could augment the costly and time-consuming process of surveys….Data analytics has emerged as a key tool for providers to target high-risk populations with chronic conditions, although some have argued that health IT systems are still ill-equipped to adequately manage population health.” There’s the crux of the matter, data is valuable, but it’s tough to get. I’ve only recently started to request specific data from the EHR to look at some things I find interesting. Unfortunately, I’ve run into roadblocks. Apparently, the data inside an EHR — at least inside this particular EHR — isn’t easy to retrieve. At least that’s what I’ve been told. How hard can it be? Dude, just dump the raw data somewhere and I’ll build the queries myself. Again, apparently not that easy. 

In a nutshell, all patient data, from demographics and notes to labs and medication use should be easily accessible to anyone with appropriate credentials, i.e. a pharmacist that works for the hospital where an EHR is used, for example. Only when we, as healthcare professionals, can access data at will, and use that data to answer questions, will EHRs become valuable to patient care. At present, EHRs are full of potentially valuable information that no one can get. It’s like having a savings account that only allows deposits, no withdrawals. The balance might look great, but what do you do when you need a little money and the bank says “sorry, there’s no way to take your money out”? Throw in the fact that EHRs are a usability nightmare and you realize that we have a long way to go.

A cautionary tale for healthcare. A lesson for pharmacy.

A couple of days ago I read an interesting piece by John Halamka over at In it, Halamka shares two stories about purchasing parts for some projects. One part required a lot of legwork and took a month to secure. The other required a quick Amazon search and less that 24 hours to obtain.

“It’s clear [the people running the company have] been asleep since 1985. When it comes time to replace the windows in my buildings, I would never consider Marvin Windows products, since it’s clear they care more about preserving an ancient business model and less about their customers’ modern expectations and experiences”. I feel that this accurately describes today’s pharmacy model.

The pharmacy practice I returned to after nearly a 10-year absence appears to have been frozen in time. Yes, things have changed here and there, like the introduction of EHRs, but the distribution model feels the same as the one I left around 2006. It’s a model that relies heavily on humans to perform manual tasks, which isn’t terrible per se. Unlike others, Amazon, for example, there has been a failure in pharmacy to introduce and integrate automation and technology into the distribution process. While pharmacy continues to rely heavily on people and very little technology, Amazon has designed a system that uses people plus strategically chosen technology to improve efficiency and accuracy.

In today’s pharmacy model, medication orders are manually pulled, stocked and counted. Manual “double” and “triple” checks are used in place of simple technologies like barcode scanning and robotics; often leading to slips and lapses. Medications are frequently lost due to the inability to accurately track them once they leave the pharmacy. This in turn results in medications being re-dispensed and often given late. Stickers identifying medications as high-risk, high-alert, STAT, sound-alike-look-alike, “note dose”, “rectal use only”, “neuromuscular blocker”, and a slew of others, decorate the shelves turning them into a Jackson Pollock painting.  Buyers still walk the shelves and manually evaluate stock levels. Paper logs are filled with information that is often redundant and available elsewhere in digital form. Legacy software sits on desktop computers while the cloud begs to be used. “Mobile pharmacy” means the pharmacist is walking around. If you didn’t know better, you’d think I was joking. I’m not.

In the article, Halamka states that “The lesson learned is that in the near future, healthcare organizations that offer an Amazon approach will displace this [sic] which do not.” I wonder if this applies to the pharmacy as well as healthcare organizations. I think not. Time has shown that pharmacy is an often forgotten part of the healthcare system. Pharmacists get promoted to move up and out of the pharmacy. No one ever moves “down” into the pharmacy. As the saying goes, it’s a good place to be from.

Halamka describes the progressive nature of Beth Israel Deaconess Medical Center (BIDMC). It’s impressive, to be sure. But even so, I’ve never heard of the BIDMC pharmacy. I’m sure it’s a fine pharmacy, but it’s never come up in any conversation or discussion of “the best” or the “most progressive”. I hear about BIDMC and Halamka all the time. Why not the pharmacy? Is the department as forward thinking as the rest of BIDMC? I don’t know. Perhaps someday I’ll find out.

Companies that provide technology to pharmacies are partly to blame. Their products often lack proper functionality and are woefully inadequate for the tasks at hand. With that said, it’s unfair to hold them responsible for our lack of progress. After all, pharmacy administrators are cheapskates, unless of course, they can make a lot of money – 340B, specialty pharmacy, etc – and then they’re all over it. But for the average, desperately needed technology, those in charge are unwilling to invest even when the technology has demonstrated improved patient safety and workflow.

It’s difficult for me to take a Pharmacy Director seriously when they say “we don’t have the money” only to take a stroll through the pharmacy and see inefficiency compounded with bloated drug budgets and out-of-control overtime numbers. Fixing even one of those would free up enough money to implement any number of additional technologies.

More often than not, I find that I’m speaking to myself when attempting to engage others in this conversation. Most want to discuss “clinical pharmacy” and “patient safety”(1) initiatives, all the while missing the bigger picture in favor of the minutia. It’s this clear lack of strong leadership and forward thinking(2) that will continue to plague pharmacy for the foreseeable future.

In the best interest of healthcare systems everywhere, I think it’s time to turn all pharmacy operations over to Amazon. I jest, but only a little.


  • Strange that many pharmacists don’t see the connection between using automation and technology and patient safety. True story.
  • There’s a whole nother blog post in there.

Thoughts on my time with DoseEdge Pharmacy Workflow Manager 

Yesterday, I had a great opportunity to spend the day using DoseEdge in its native environment. That is to say, I staffed at a facility that was using DoseEdge in its cleanroom to prepare CSPs, including patient-specific products as well as stock bags, TPNs, etc. I’ve used DoseEdge for brief periods in the past. I’ve also spent time with the engineers that have worked on the product, product managers, sales people, marketing folks, and various other Baxter employees at conferences like ASHP midyear. But, this is only the second time that I have worked an entire  “staff pharmacist” shift using the product. And as one might imagine, using a system in its native environment can often provide a new perspective.

DoseEdge System at Boston Children’s Hospital

I’ve written about DoseEdge, and systems like it, extensively on this site over the years.  A quick search of revealed several articles mentioning DoseEdge, dating back to 2010. I’ve also written about DoseEdge and similar products extensively elsewhere.

With over 300 installs in the U.S., DoseEdge is one of the most popular IV workflow management systems on the market and is still the most widely implemented product of its kind in the U.S. It’s a very good system, certainly in my top five. There are several things that I really like about the system, such as how it controls labels, its ability to track partial vials, and the fact that it talks. Seriously, it speaks to the user. Pretty cool.

However, there’s always room for improvement. For example, after using the system, I found that I don’t really care for the user interface (UI). I found it to be rather unintuitive and a bit clumsy. The UI is stuck somewhere between a legacy system and a modern web-based system. It’s not good. Too many clicks, things in weird places, naming conventions that simply don’t make sense in my mind, just to name a few. With that said, it’s still quite usable, and honestly, it’s likely as good as any other UI on any other product that I’ve used in the pharmacy.

The other thing that hit me yesterday was just how terrible the process of using images is to verify the accuracy of compounded products. I don’t care for it. As good as the images are – and they’re quite good in DoseEdge – there are still shadows in strange places that make reading syringe volumes difficult at times. This is especially true for small volumes. There were times yesterday when I simply made educated guesses to the exact volume and assumed that the volume was accurate, as I couldn’t quite see the exact location of the plunger. Don’t get me wrong, this is still way better than the syringe-pullback method, but image-assisted verification isn’t optimal. I would have liked to have had the gravimetric option available to me yesterday. DoseEgde offers gravimetrics, although it isn’t widely used.

So, good system, but not perfect. Better than the manual process, but room for improvement. Imaging better than pullback, but not great.

Does charge-on-chart hurt or help medication chain of custody?

Historically, hospital pharmacies have used a charge-on-dispense (COD) model for medications. The model charges the patient for a medication when it is dispensed from the pharmacy and credits the medication if it’s returned to the pharmacy unused. Simple, but labor-intensive. The model itself has been around for a long time.

The introduction of electronic health records (EHRs) and electronic medication administration records (eMARs) has pushed the COD model aside in favor of the charge-on-chart (COC) model; sometimes referred to as “charge on administration” (COA). In the COC model, the patient isn’t charged for a medication when it is dispensed from the pharmacy because the charge is captured when the medication is scanned by the nurse and administered to the patient. When the nurse scans the medication, the information is captured by the eMAR and charted, hence the name. There are several benefits to this model, including no need for the pharmacy to credit medications that go unused. Unused medications are simply returned to the pharmacy and folded back into the inventory.

Put simply, the COC model eliminates the need for pharmacies to charge and credit medications as they are dispensed and returned to the pharmacy. But here’s a little untoward side effect of the COC model, it eliminates much of the pharmacy audit trail for medication movement into and out of the pharmacy.

The old COD model wasn’t perfect, and there were plenty of discrepancies, but I wonder if the COC model has created even less transparency regarding inventory reconciliation and the movement of medications throughout the hospital.

Inventory management systems like AutoPharm from Talyst and Pyxis Pharmogistics from Carefusion should, in theory, give pharmacies real-time inventory numbers. But the promise of these systems has fallen short. Both utilize barcode scanning to track inventory, which unfortunately requires humans to be diligent when scanning items in and out of inventory. Human laziness usually prevails, and numbers are frequently inaccurate.

Medication tracking systems are available from a couple of companies, but also utilize barcode scanning, thus fall prey to the same weakness mentioned above. These systems also fall short when following medications throughout the medication distribution process as they typically stop as soon as the medication is delivered to the nursing unit, i.e. they don’t track the return of the medication.

Track and trace regulation, which will require serialized barcodes and tracking from manufacturer to patient, could potentially help with this issue. However, that process has the same weaknesses as those mentioned above, namely human intervention.

RFID technology would surely be better than barcode scanning, except that RFID tags are too costly for use on all medications and drug manufacturers are nowhere near ready to do anything like this.

Currently, the only medications that receive enough scrutiny in a pharmacy to determine location and quantity at any given moment are controlled substances, i.e. morphine, fentanyl, oxycodone, and so on. And this falls short on some level once the medication leaves direct oversight of the pharmacy.*

It’s interesting to think that as much time as we spend managing inventory in a hospital pharmacy, we still have a long way to go.


*This includes leaving the pharmacy itself as well as storage devices like automated dispensing cabinets (ADC). When a medication leaves the ADC we assume it has been administered to the patient once it has been charted. We cannot confirm this, however. For all we know, the healthcare provider that removed the medication and documented the administration, simply put it in their pocket and walked out with it. You never know for certain.