Find adverse events in patient support calls and notes
Reads patient support calls, notes and emails for possible adverse events and the report elements present. It never judges seriousness or causality.
Try it on this example
Channel (patient support call, medical information, market research, email): Patient support programme call
Products covered (brand and generic names): Veltrinab (tralumab) 150 mg pre-filled pen
Note, email, chat or transcript
- Does the text describe a real health problem in a person exposed to one of the products?Real event in an exposed person100%
- Which special situation does the text describe, if any?Medication error100%
- Does the text report a problem with the physical product, its packaging or its device?Yes97%
- Does the text refer to a specific person who had the problem?Yes99%
- Does the text come from, or name, a person who could be contacted for follow-up?Yes86%
- Is a product from the products field named or clearly meant as something the person took or used?Yes99%
- Does the text mention death, a life-threatening event, a hospital admission, lasting disability or a birth defect?No86%
- Does the text say the event or complaint has already been passed to drug safety or the quality team?No90%
These are real answers stored from one run on this example.
The prism behind it
Find adverse events in patient support calls and notes
Fields
- Channel (patient support call, medical information, market research, email)
- Products covered (brand and generic names)
- Note, email, chat or transcript
Context
Working definitions of a pharmaceutical company's drug safety team, for text from channels outside drug safety: patient support programme calls and notes, medical information enquiries, market research interviews, customer service emails, social media posts and sales representatives' notes. Each text is read here so that anything drug safety must see reaches the safety intake queue the same day. Names and contact details may have been removed before the text was sent, and placeholders such as [PATIENT] or [PHONE] may appear. Nothing here decides whether an event is serious, whether the product caused it, whether it is expected, or whether a case is valid or reportable: drug safety staff decide all of that. Adverse event: any unwanted medical occurrence in a person who has taken, used or been exposed to a product in the products field, whether or not anyone blames the product. It includes symptoms, side effects, illness, injury, abnormal test results, a visit to a doctor, clinic or hospital, and death. Lay and casual descriptions count, such as "been dizzy since I started it" or "my lips went funny". Special situations are collected even when no harm is described: exposure during pregnancy or breastfeeding; use in children or older people; overdose, misuse or abuse; a medication error, such as a wrong dose, a wrong product, or a missed or incomplete dose because of how the product was handled or a device problem; lack of effect; use for a condition, age group or dose the text says is outside the product's approved use; and exposure of a carer or worker who handles the product. Product quality complaint: a report of a problem with the physical product, such as its packaging, label, a pen or device fault, how a tablet looks or smells, contamination, or a suspected fake. A product complaint can come with or without an adverse event. Minimum information for a report: an identifiable patient, an identifiable reporter, a suspect product and an event. Drug safety follows up whatever is missing; a text that lacks some of them still goes to drug safety.
Questions
Does the text describe a real health problem in a person exposed to one of the products? Choice
Read the text, the products field and the definition of an adverse event in the context. Choose the option that fits. If the text has a real problem in an exposed person and also a general question, choose the real problem. The threshold for this answer is set low on purpose.
Which special situation does the text describe, if any? Choice
Read the text and the list of special situations in the context. Choose the special situation the text describes. If it describes more than one, choose the one it says most about. A special situation counts whether or not any harm is described.
Does the text report a problem with the physical product, its packaging or its device? Yes / No
Read the text and the definition of a product quality complaint in the context. Does it report a problem with the physical product: packaging, a label, a pen or device fault, how a tablet looks or smells, contamination, or a suspected fake? Yes: The text reports at least one problem with the physical product. No: The text reports no problem with the physical product. A complaint about price, delivery or service does not count.
Does the text refer to a specific person who had the problem? Yes / No
Read the text. Does it refer to a specific person who had the health problem or special situation, such as the writer, "my mother", a patient placeholder, or "a patient of mine in her sixties", rather than people in general? A removed name still counts when the text is about one person. Yes: The text refers to a specific person. No: The text refers to no specific person, or describes no problem.
Does the text come from, or name, a person who could be contacted for follow-up? Yes / No
Read the text and the source field. Does it come from, or name, a specific person who reported the information and could be contacted again: the caller, the writer, or a health professional named or described by role? A placeholder where a name or number was removed still counts. Yes: There is a specific reporter who could be followed up. No: The reporter is anonymous or cannot be told from the text.
Is a product from the products field named or clearly meant as something the person took or used? Yes / No
Read the text and the products field. Is a listed product named, or clearly meant, as something the person took, used or was exposed to? In a programme or enquiry about one product, "the pen" or "the injections" counts. Yes: A listed product is named or clearly meant. No: No listed product can be told from the text.
Does the text mention death, a life-threatening event, a hospital admission, lasting disability or a birth defect? Yes / No
Read the text. Does it mention death, a life-threatening event, admission to hospital or a longer hospital stay, lasting disability, a birth defect, or the reporter calling the event serious? This sets priority in the queue only; it is not a seriousness assessment. A visit to a walk-in clinic, urgent care or an emergency department where the person was sent home the same day does not count on its own. Yes: The text mentions at least one of these. No: The text mentions none of these.
Does the text say the event or complaint has already been passed to drug safety or the quality team? Yes / No
Read the text. Does it say that the event, special situation or complaint has already been reported to drug safety or the quality team, or that an adverse event or complaint form has been filled in? A promise to pass it on later does not count. Code checks the safety database whatever the answer. Yes: The text says it has already been passed on or a form was filed. No: The text says nothing of the kind, or only says it will be passed on.
Lens columns
adverse_event, adverse_event_probability, special_situation, special_situation_probability, product_complaint, product_complaint_probability, identifiable_patient, identifiable_patient_probability, identifiable_reporter, identifiable_reporter_probability, suspect_product_named, suspect_product_named_probability, seriousness_words, seriousness_words_probability, forwarding_recorded, forwarding_recorded_probability
Run it on your own text
Add this prism in the app, change any question, and test it on a file of your own.