I work in the field of injury epidemiology, and my institute does a lot of research on both drug abuse and poisoning. While I found this episode constructive overall, the repeated effort to distinguish between "overdose" and "drug poisoning" was semantic nonsense. The two terms are synonyms. The data do not distinguish between them becau…
I work in the field of injury epidemiology, and my institute does a lot of research on both drug abuse and poisoning. While I found this episode constructive overall, the repeated effort to distinguish between "overdose" and "drug poisoning" was semantic nonsense. The two terms are synonyms. The data do not distinguish between them because there is no distinction to be made.
But, if you grant the attempted definition, "The poisoning is someone does it to you," then the claim that the data do not distinguish between them falls apart. In the coding of death certificates, the underlying cause of death for a poisoning must be an external cause of injury, which always tells you the intent of the injury event - i.e., whether it is an accident, suicide, homicide, undetermined-intent, or a result of law enforcement or war. Similarly, in medical data, the ICD-10-CM codes used to code poisoning include both the substance and the intent in the same code, making it impossible to code one without the other. (The code for accidental poisoning by fentanyl is T40.411A. For assault by fentanyl, T40.413A.)
Bherwani faults the system for not providing real-time data, but that's not a realistic demand. Even under optimal conditions with adequate funding, it takes time for data to be coded and made available to users. In the age of HIPAA and privacy concerns, data providers are legally required to focus on keeping data out of the wrong hands and preventing misuse of data. Furthermore, data providers tend to be proprietary about their data. That means they are usually obsessed with quality control, but also that they want to maintain control over what even the most trusted data users can do with their data. And special interests (e.g., hospitals) restrict what you can do with their data (e.g., I am prohibited from investigating the relative safety of competing hospitals - this is something that the medical industry colludes to prevent anyone from knowing).
I work in the field of injury epidemiology, and my institute does a lot of research on both drug abuse and poisoning. While I found this episode constructive overall, the repeated effort to distinguish between "overdose" and "drug poisoning" was semantic nonsense. The two terms are synonyms. The data do not distinguish between them because there is no distinction to be made.
But, if you grant the attempted definition, "The poisoning is someone does it to you," then the claim that the data do not distinguish between them falls apart. In the coding of death certificates, the underlying cause of death for a poisoning must be an external cause of injury, which always tells you the intent of the injury event - i.e., whether it is an accident, suicide, homicide, undetermined-intent, or a result of law enforcement or war. Similarly, in medical data, the ICD-10-CM codes used to code poisoning include both the substance and the intent in the same code, making it impossible to code one without the other. (The code for accidental poisoning by fentanyl is T40.411A. For assault by fentanyl, T40.413A.)
Bherwani faults the system for not providing real-time data, but that's not a realistic demand. Even under optimal conditions with adequate funding, it takes time for data to be coded and made available to users. In the age of HIPAA and privacy concerns, data providers are legally required to focus on keeping data out of the wrong hands and preventing misuse of data. Furthermore, data providers tend to be proprietary about their data. That means they are usually obsessed with quality control, but also that they want to maintain control over what even the most trusted data users can do with their data. And special interests (e.g., hospitals) restrict what you can do with their data (e.g., I am prohibited from investigating the relative safety of competing hospitals - this is something that the medical industry colludes to prevent anyone from knowing).