Picture of Anthony M DiGiorgio, DO, MHA, FAANS

Anthony M DiGiorgio, DO, MHA, FAANS

Economic Concepts & Neurosurgery Transaction Costs

As the CANS newsletter moves into a new direction, so will my monthly column.  I’m going to talk more about economic concepts and how they relate to the practice of neurosurgery.  To take that on, let’s talk about the CANS meeting and Kevin Schulman’s talk on Artificial Intelligence (AI), one of my favorite subjects.

Schulman’s article on AI, which he mentions in his talk, is actually about one of fascinating economic topic: transaction costs.  There’s an entire podcast on transaction costs (highly recommend).  As Adam Smith aptly stated, “The real price of everything is the toil and trouble of acquiring it.” 

Transaction costs are a huge deal in medicine.  Many would argue that they are the main driver of healthcare costs in America. 

In a basic sense, transaction costs include the time and effort it takes in making an economic exchange.  If a consumer goes to the store and buys a widget for $100, that consumer also spends resources (time, energy, transportation, effort) on acquiring that widget, so the true cost is more than $100.  The producer of that widget still only receives the $100.  If the producer sells the widget on Amazon and offers free shipping, the price could be $110 and the consumer would still buy it because the overall transaction costs are decreased.  It’s why Amazon is so successful.  The producer can collect more and the consumer has reduced transaction costs.

Transaction costs are comprised of triangulation (buyer and seller finding one another), transfer (getting the good or service from seller to buyer), and trust (assurance that buyer and seller both receive what is expected).  We can already see how healthcare has massive transaction costs when compared to that Amazon example, and most of it comes from the “trust” leg of the transaction cost triad.  The whole reason we have complex billing,

coding, claims adjudication, prior authorization and quality metrics is for trust, especially when we recognize that the “buyer” in most medical transactions is a third-party payer. 

When that third-party payer became the US government, the complexity skyrocketed.  Schulman bemoans this complexity in his article, arguing that the complexity must be reduced before AI can relieve us of our administrative burdens.  The complexity he mentions, such as the 599,204 unique codes to describe healthcare services, is not a product of transaction costs in a free market.  It’s the product of the US government creating an administrative pricing scheme for services it purchases.  When the government figured it still didn’t trust what it was purchasing, it added in more layers of transaction costs in the form of value-based purchasing.  It even outsources much of the transaction costs to private entities in the form of Medicare Advantage. 

Schulman is right.  The complexity must be eliminated while being mindful of reducing transaction costs, not adding to them.  More burdensome regulations that force standardization run the risk of creating compliance costs and erecting barriers to entry.  Instead, reducing regulations, such as decoupling Medicare from administrative pricing, would go a long way to reducing complexity. 

However, it’s unlikely Congress will massively reform Medicare any time soon.  Therefore, one of the only ways we have to reduce transaction costs in the complex system is with AI.  Large language models can already perform billing and coding, extracting ICD-10 and CPT codes from notes, and place orders.  That could shave a few clicks off each patient encounter, reducing transaction costs.  AI has the potential to upend the quality metric industrial complex, eliminating all those burdensome coding queries.