Data scientists and medical clinicians look at things from different perspectives especially with respect to data. Data scientists focus on what is most common, while clinicians tend to concentrate on what is unique. In the real world of healthcare and hospital operations, this means data scientists will typically optimize for the most common type of patient, while clinicians instinctively optimize for the edge cases.
We call this “Conjoined Twin Syndrome.” Conjoined twins are very rare, but equally notable. A successful separation requires tremendous medical resources and planning – multiple teams of specialists working in multiple rooms for many hours. There’s also typically media coverage and press interviews. This is about as far away from a routine hernia repair as one can get. But even with the huge footprint of a single conjoined twin separation, the hernia repairs take up significantly more cumulative OR time annually than the one conjoined twin every decade.
Hospitals need to be designed for the most common type of patient. The edge cases are the stories we tell at the dinner table, but they don’t actually move the needle in terms of day to day operations. The majority of patients at a hospital are routine treatments with predictable care pathways and foreseeable outcomes. These are the patients we data scientists focus on and for whom we optimize. The edge cases, by definition, can only be dealt with on a case-by-case basis by thoughtful and experienced staff. These cases are too complex and unique to realistically invest in establishing a defined process that manages their throughput.
Conjoined twins are narratives, not data. As humans, we’re naturally wired to narratives and give them oversized significance. When Hospital IQ works with hospitals, we often hear narratives such as, “The ICU is always filled with behavioral patients.” When we actually drill into the data, we often see this happened once or twice, but certainly not “always.” However, this narrative persists because the impact on staff was sufficiently stressful to put everyone on high-alert whenever a single behavior patient is placed in the ICU. “The ICU is always filled with behavioral patients” is not patently untrue, but a more accurate statement would be, “There were two instances in the last 18 months where more than 25% of the ICU census was behavioral patients.” But which statement has more narrative appeal?
To be clear, data and narratives are not in competition. Organizational change comes about when narratives are shaped by data – supporting some aspects while refuting others until the two are aligned. Nobody can make a decision on data alone. Without a narrative, hospital data is just rows of patients in tabular format. The best decisions result when supported by a narrative that is both emotionally compelling and consistent with the facts. This is why hospitals need data scientists, and why data scientists alone don’t run hospitals.