Natural language processing application to identify covert administration of medicines: development and pilot audit.

Aims and method: The covert administration of medicines is associated with multiple legal and ethical issues. We aimed to develop a natural language processing (NLP) methodology to identify instances of covert administration from electronic mental health records. We used this NLP method to pilot an audit of the use of covert administration.

Results: We developed a method that was able to identify covert administration through free-text searching with a precision of 72%. Pilot audit results showed that 95% of patients receiving covert administration (n = 41/43) had evidence of a completed mental capacity assessment and best interests meeting. Pharmacy was contacted for information about administration for 77% of patients.

Clinical implications: We demonstrate a simple, readily deployable NLP method that has potential wider applicability to other areas. This method also has potential to be applied via real-time health record processing to prompt and facilitate active monitoring of covert administration of medicines.