Self-reported medication data

Process for recording self-reported medication

During visits to the assessment centre, participants use a touchscreen questionnaire to answer questions about common prescription and over-the-counter medications, dietary vitamins and other supplements (e.g. Medication for cholesterol, blood pressure or diabetes). This categorical self-reported medication data collected using the touchscreen is stored in Category 100045.  The touchscreen also asks the question “Do you regularly take any other PRESCRIPTION medications? Do not forget medications such as puffers or patches”. (Data Field 2492).  The specific questions asked can be seen in the Notes and/or Resources tab of each field within Category 100045.

After completing the touchscreen section of the assessment centre visit, participants proceed to a verbal interview with a trained member of staff (Resource 100235).  Participants who reported taking any of the following classes of medications: blood pressure lowering, cholesterol lowering, hormone-replacement therapy or oral contraceptive pills, will be asked for the name of the medication(s).  Similarly, if the participant reported regularly taking other prescription medications, then the interviewer will ask about and record these.  Only regular prescribed treatments that are actually taken by the participant should be recorded. Short term treatments e.g. a 1 week course of antibiotics medication that is prescribed, but not actually taken, should not be recorded.

To record the medications reported, the interviewer uses the Search facility to search a medication list identical to that in Data Coding 4. The list includes both generic and trade names and, in some cases, multiple doses and formulations for the same medication. For the purposes of UK Biobank, dose information was not captured and interviewers were instructed to select the lowest dose found on the list. In cases where medication reported by a participant was not found on the list then the interviewer recorded the name of the medication as free text. All medication recorded during the verbal interview is stored in Data Field 20003. Currently, any medication recorded as free text has not been processed further and appears as “Free-text entry, unable to be coded” in data field 20003.

Using the self-reported medication data

When using the self-reported medication data in Data Field 20003 researchers should be aware of the following points;

• The medication list (Data coding 4) has not been updated since 2010.  Newer medications (e.g. new oral anticoagulants) are therefore missing from the list and, when reported by participants, will currently be stored as “Free-text entry, unable to be coded”. We aim to update Data coding 4 and, where possible, re-code the free text entries to named medications in a future data release. Please refer to the Future data release timelines page for an update on when this will occur.
• The absence of medication data in Data Field 20003 may be because a verbal interview did not take place, for example due to time constraints.  Approximately 13% of first imaging visits and 23% of repeat imaging visits did not include a verbal interview.  These cases can be identified by the absence of value in Data Field 21831 (Verbal interview signoff timestamp).
• When entering the data, interviewers were instructed to take no account of dose and formulation and to select the lowest dose on the list. Despite this there are entries in Data Field 20003 for higher doses.
• Dates on which participants started and stopped taking the medication reported in Data Field 20003 is not captured.  This data should represent the medication that a participation was taking regularly at their assessment centre visit (the date of which can be found in Data Field 53), however it is subject to the anomalies of self-report.

Additional sources of medication data

In addition to the self-reported medication data captured using the touchscreen questionnaire and verbal interview, further information about medications is present in some online follow-up questionnaires (eg Categories 132, 1500 and208). The linked GP prescription records also contain medication data, although this is currently only available for 45% of the cohort and is censored to approximately 2016/2017, depending on the provider. The procedure codes within the inpatient data also contains information for some medications, though comprehensive data on inpatient medications is not currently available. For more specific information on sources of medication data, please see Resource 596 (https://biobank.ndph.ox.ac.uk/ukb/refer.cgi?id=596), particularly Appendix C: Information on medications.