Electronic Patient data

Question 1

What types of studies is EHR used for? (5)

Answer 1

1. Prevalence/incidence rates
2. Aetiological factors (risk factors and psychological outcomes relationship)
3. Outcome studies: follow-up studies to determine specific outcome
4. Service/intervention evaluation: comparison of different service approaches
5. Prediction models: develop/ validate diagnostic/ prognostic models of an outcome (e.g. psychosis)

Question 2

Advantagues of EHR (7)

Answer 2

1. Real-world data
2. Reduces burden on researcher and patients (no need to interview/collect samples)
3. Can obtain data from large patient samples (entire country)
4. Anonymisation means that ethical issues often reduced
5. Utilises rich data collected by clinical teams that contains valuable information about patient care/treatment
6. Can be linked to a wide range of other sources
7. Standardised approach that should mean studies are replicable across research groups

Question 3

Disadvantagues of EHR (5)

Answer 3

1. You can only access data that has been routinely collected by the clinical team; not suitable for more nuanced variables not typically ascertained by clinical team (e.g., childhood trauma)
2. Quality of data you can extract depends entirely on service – can be unreliable/invalid even for standard outcomes (e.g., ward stays, demographics)
3. Although researcher burden reduced, not always easy option as manual coding often required; novel variables must be validated
4. Consistency/ease across countries unclear
5. Patients may be concerned about data linkage

Question 4

Natural Language Processing (NLP)

Answer 4

Is a text mining method which facilitates extraction of information from text documents

Allows analysis of large volumes of text data to quickly identify useful clinical information

Can be done with machine learning