All Articles

Random Thoughts From April Part 1

Whew. What a month. Today is April 24th. There’s still a week left before the month ends, but this is actually the 55th day of March anyway and I have a lot to write about. For the reader, I don’t think this post has much value, it just answers a few questions. Today is writing for clarity. Here’s a table of contents:

  1. Anesthesia Operating Room Safety
  2. Thoughts on market fluidity (just in time versus stocked fulfillment)
  3. Dumb People
  4. Malignant
  5. Remembering Shit on Twitter

Here we go.

Anesthesia Operating Room Safety

There’s an interesting program at our school where undergraduates get to work on projects proposed by clinicians at the Hopkins Hospital. We’re in the midst of picking projects, and currently this is my top choice. To me, it’s interesting because you have to solve many problems all at once, with a harmonized high compliance solution. Let me elaborate.

Anesthesia medication errors account for $3.1 - $4.6 billion in costs for the U.S. healthcare system, affecting between 200,000 - 2,000,000 patients annually. Complications for patients include awake paralysis, high fever, and in roughly 1% of errors, death. Anesthesiologists suffer from the highest stress levels of all physicians, leading to 14% higher physician normalized suicide rate, and higher burnout, drug use rates, and overall mortality. While medication errors are thoroughly studied in medical literature, few interventions to solve the issue have been tried, and none have reached widespread usage.

The principal characteristic of this problem is heterogeneity across clinical workflows as well as committed errors. Sources of error include administering incorrect medication, dosing miscalculations, incorrect administration route, and omission of medication. The incidence of each has high variance across different study locations, but no type of error consistently accounts for greater than 50% of total errors during study. Common causes of error are also heterogeneous, including human centered errors such as forgetfulness or memory lapses, systemic errors such as work environment or understaffing, as well as technical issues such as unreadable fonts on medication or unlabeled syringes. While the risk of an error decreases with experience, it is still the case that experienced physicians commit errors at a dangerous level.

Current solutions offer both device and workflow based prevention strategies. The Intelliport™ Medication Management System integrates EHRs with automatic medication barcode scanning for safer medication delivery. Some hospitals have attempted universal color coding and labelling of syringes and medication ampoules. However, these solutions have low adoption rates and have yet to demonstrate a significant improvement in quality of care. There is therefore an opportunity to improve clinical workflows if a high compliance solution can be developed. Key design constraints will include compatibility with existing operating room procedures to ensure compliance and ease of use, as well as a human centered design approach in order to offload and automate aspects that contribute uncertainty and physician stress.

These were some initial thoughts gathered from literature review, but clearly its missing important things. Sure, literature reviews are great, but how are we supposed to really know how a space works (short of becoming an anesthesiologist). Shadowing is an option, but incomplete. With a rare(ish) problem with high heterogeneity, it will be difficult to get enough sample data. Steve Jobs tells us to figure out and deliver what customers want, before they know what they want, to never rely on market research. How do we even do this? Especially when it has to do with fixing mistakes in an already highly trained group of people? Ideation is hard. For Apple, it took a highly creative group of people and a lot of LSD. Lots of innovation comes from years of experience and experimentation (trial and error), lots of it is copying from other domains (nature inspired design for example), and lots of it inevitably must come from random burst of insight (Eureka or drugs). I’m very interested in how our team will solve this problem. Reading about the study of innovation has thus far been unimpressive to me. I am hoping that being forced to do it will give me some new ideas.

So how can we innovate? Copy or do drugs, but probably just put in the work to become an expert.

Market Fluidity

This is a very relevant question. In the U.S., distribution of medical supplies has been atrocious. People scream on Twitter, but unfortunately internet points don’t translate to shipping product. There has exceedingly been efforts to ‘raise awareness’ of shortages, or ‘connect’ decision makers and hospital systems through web apps, etc. Honestly, I don’t think that these provide much value. Even if they do, I think it’s a sign of a broken system. I don’t think technology any further than email or an easy to understand spreadsheet provides any excess value here but that is besides the point.

One problem seems to be that there is no central distribution system. Each state/city/hospital needs to procure their own supply. Each party will compete with each other, and while the first state who orders will get their shipment faster than if everyone ordered at once, on the whole the process is slower in a demand surge. Is this a failure of the federal government? With the USPS, there must be the infrastructure to deliver supplies effectively. As you can tell, I’m far from an expert…but does our system really rely on people to connect dots informally via Twitter pleas and cold phone calls?

Upon further thought, I reached the conclusion that this is a feature, albeit with weaknesses, of American capitalism. In normal circumstances, indeed it is individual hospitals who need to be responsible for refilling supplies. It would be a disaster if one had to deal with regional bureaucracy just to refill face masks. Upon even further thought, I realize that I learned this in 11th grade global politics class. This is simply the difference between a command economy and a market economy.

China and South Korea are able to centrally demand production of ventilators, face masks, and other resources. China mandated the production of an emergency hospital and got it done in 10 days. This is something that could never happen in the U.S.. On the other hand, outside of a global crisis, we’ve seen the clear advantages of market economies. North Korea looks like the only remaining pure command economy. China, South Korea, and the rest of East Asia have widely integrated and encouraged features of market economies to boost their economic growth.

I think the insight to draw here is that in countries with high dependence on market forces, those in charge of emergency services need to be extra careful, probably even subject to special protection. The fact is that government in the West moves at a snails pace compared to the East. We need to emphasize the importance of national stockpiling efforts. We already have this for oil. Turns out, we have a stockpile for medical supples also. We should support legislation to maintain and grow national stockpiles and emergency programs, especially because we live in a market economy. We should not be relying on Elon Musk to manufacture and deliver ventilators during crisis periods. It should be unacceptable for hospitals not to be able to afford stockpiling programs. CYA.

So why does the U.S. currently have distribution problem? We are somewhat ignorant of our weaknesses as a market economy, and haven’t invested enough into CYA strategies.

Dumb People

This one is contentious. Ketih Rabois inspired me to write this post. In the past 2-3 months, he has unfortunately been wrong repeatedly. The man wrote off Biden after Iowa, championed chloroquine as a Covid parachute, and says that with near 100% certainty, Covid-19 was manufactured in a lab in China. Don’t get me wrong, I think when it comes to business, the man is light years beyond me and 99.9% of people. Clearly, the problem is infiltrating an area in which you have no expertise, and assuming yourself as a thought leader.

It’s one thing if you are wrong. Usually, you’ll learn something and won’t make the same mistake, or if you do, you already lost your credibility. The real concern is when dumb people are right. The patzer wins because you lose on time, the gambler wins on an internet stock tip, someone undeserving gets the promotion because of connections, ulterior motives, or simply just a bad decision. What are we supposed to say to people who are technically right, but for the wrong reason. Also bad are those who piggyback off of experts, yet aren’t able to articulate how a conclusion was reached or study limitations. How do we reconcile this and get back to work?

For most, this is somewhat of a learned phenomena. At first, anything anyone says is insightful, it is very difficult to differentiate signal from the noise. There are no dumb people, opinions, or thoughts. Slowly, as we learn more about a field, we’re more able to discriminate between trash and value. The Dunning-Kruger effect is relevant here. Continuous growth comes from learning about areas where you can’t make this distinction.

This is a problem that needs to be fixed somehow. If we are unable to filter, posturing and surface level thought are rewarded. Buzzwords rule the day. Nuance and understanding fall to the wayside. Going straight to the meat of an issue without the small talk of unverifiable garbage is discouraged in discussion.

A couple answers to avoid the problem immediately pop up to mind. Stoicism, turning off media, and reflection. The first two have downsides. Sure, you’re shielding yourself from the dumb, but you’re also preventing yourself from the smart. If you block out the whole feed, you’re losing the noise but also the signal. Reflection seems the most promising to me. Looking bak on yourself is like eating your vegetables. You hate to do it, but it’s good for you. When I read this in 3 months, I’ll inevitably cringe. This sparks change.

So how do we refocus after getting riled up by dumb people? Look at what you did three months ago and sober up.


I read this book by Vinay Prasad yesterday. Why? Because I currently want to become a doctor, and I thought that this book would help me gain a clinical perspective on cancer drug prescription and regulation. Did it? Yes and no. I see the book as a nice summary of the major thoughts he presents during his Plenary Session Podcast. I didn’t find it groundbreaking by any stretch, if you have listened to a few of his podcasts, you could have more or less predicted what he was going to write. Still, I finished the book feeling that I had a more complete idea of his opinions, more than what I could have gained from listening to podcasts here and there.

Here, I wanted to briefly summarize his main points:

  1. Earlier prescription of medicine is not necessarily better, in fact it is sometimes worse.
  2. The pharmaceutical industry bribes physicians stakeholders in positions of power to influence marketing and drug approvals.
  3. Clinicians should be more focused on whether the drug has been shown to work in well designed clinical trials than biological plausibility.
  4. Surrogate endpoints are often poor predictors of meaningful clinical outcomes such as survival or quality of life.
  5. Phase 2 data cannot be used as an indicator of efficacy, as it is not randomized. Phase 3 Trials are efficacy trials, and we should consider a drugs performance only off of Phase 3 results
  6. We need to carefully consider study design elements such as crossover, arm sizes, blinding, representativeness of patient population, post-protocol therapies, censoring, and what constitutes a fair control arm (sham, placebo, etc).
  7. Good drugs needs to demonstrate single agent activity.
  8. It is easier to show improvement in the metastatic setting than the adjuvant setting
  9. Large companies will develop single compounds and try to gain approval in multiple different cancers. Smaller companies are more targeted (read: careful), designing novel compounds for specific cancers.
  10. Precision oncology is a developmental stage technology that benefits very few (2% of total patients) because so many are disqualified or not benefited compared to standard of care.
  11. Most of the time when people talk about drugs, they are overhyped. Modern drugs typically only net moderate benefits over standard of care.
  12. People on 3rd and 4th lines of therapy are healthier than typical patients. They are self selected to survive multiple lines of therapy and likely have indolent biology
  13. Be wary of meta analyses (SISO), historically controlled trials, observational studies, retrospective studies, single arm trials.

At this point, I’m not able to separate the signal from the noise. This is all fantastic information, and I’m not able to make an informed opinion. Consider this list a work in progress.

Random Twitter Bullshit

When compiling a list of what I wanted to write about, I realized that even with all the time I spent on Twitter, Reddit, and the general internet, I barely remember what I read day to day. Why does this happen?

The answer is pretty clear, it’s because I don’t integrate it into long term memory be ever thinking about it again (I’m talking mostly about recall, not recognition here). But why don’t I think about it ever again? A lot of times, I think that what I am reading is pretty insightful (An important fact, the results from a cool paper, or some salient advice). Is there some sort of app that helps you review the cool ideas that you read every once in a while? Sort of like an Anki for Twitter likes, or those Facebook memory videos that compiles them into one place.

What would I actually want in one of these products? Well, I’d like some way of integrating with the note takers that I use, and also social medias like Twitter and Reddit. If you could compile everything I upvote on Reddit, like on Twitter, highlight in Preview, or write down in an Evernote notebook, and show a mix of it to me once every 2 days or so, I’d be immensely happy. Please, someone, make this happen. Request for product.