'How to Conduct Clinical Research' Training Event CPFT Research Database CRIS Case Register Interactive Search CRATE - Clinical records anonymisation and text extraction Jonathan Lewis Nov 2017
Overview Access to clinical information is a challenge for R&D in CPFT Inefficient use of NHS clinical data for surveys, audits, epidemiological research Inefficient recruitment of patients to clinical trials, experimental medicine and other planned research studies Tightening of data protection rules regarding patient-identifiable data make traditional approaches less appropriate CRATE is a solution to some of these problems A software tool for anonymisation and searching of clinical records A recruitment module to allow more people to access research
CRATE Clinical Research And Text Extraction data flow schema Source data id: 123456 forename: JOHN surname: SMITH DOB: 3 Mar 1970 NHS#: 9876543210 address_1: 2 Narrow Lane consent_mode: green... id: 123456 letter: 24 Feb 2015 Dear Dr Colville, Re: JOHN SMITH, DOB 3 Mar 1970, NHS 9876543210 2 Narrow Lane, Fictionville I saw John today in clinic with his wife Sandra. He s on venlafaxine 225mg od. He feels his mood has been improving recently but his psoriasis is worse. ± via staging copy of database clinicians scrubber Anonymised research database A non-patient tables unchanged research_id: 385716 DOB: 1 Mar 1970 B research_id: 385716 letter: 24 Feb 2015 Dear Dr Colville, Re: XXX XXX, DOB XXX, NHS XXX XXX, XXX I saw XXX today in clinic with his wife YYY. He s on venlafaxine 225mg od. He feels his mood has been improving recently but his psoriasis is worse. external NLP tools database-to-nlp interface C Additional structured data from source text research_id: 385716 date: 24 Feb 2015 drug: venlafaxine dose: 225 units: mg frequency: od research_id: 385716 date: 24 Feb 2015 smoker: current cigs_per_day:? research_id: 385716 date: 24 Feb 2015 hamd: 6 researchers direct access via SQL CRATE front end website D
E Consent to contact patients about research id: 123456 research_id: 385716 consent_mode: green... Researcher finds a potentially suitable patient, without details that would identify them, and wonders if they would like to take part. Researcher is given no identifying details about the patient. Clinician Patient s choice (default is yellow if patient has not yet been asked) Patient Researcher may see the patient s clinical details and write to them with study information and an invitation to take part.
CRATE data Done CRS/CDL 159,505 RiO 160,818 n CPFT Firewall Awaiting anonymisation IAPT 51,355 CPFT-CRATE-P01 Awaiting transfer to CRATE SysOne DATIX 421,686 Physical server behind CPFT firewall Intel XEON 3GHz 24-core cpu 96GB RAM 9TB HDD Hospital Episode Statistics MS SQL Server 2014 GATE Potential CUH GP Finance
Feasibility Industry enquiry, elderly patients with Major Depressive Disorder Inclusion criteria Diagnosis F32 or F33 Year of diagnosis 2012 Age (as of 2012) >= 65 Exclusion criteria MMSE < 24 Diagnosed of diabetes - (E08 - E13) Co-morbidities, diagnosis of anything in F0*, F1*, F2*, F31*, F6*, G* All MDD diagnosis - 903 patients are found. For any year of diagnosis - 583 patients matched criteria. For the criteria above - 180 patients found.
Study approval process REC approval HRA approval Database oversight committee approval Feasibility assessment N N Y Epidemiological N Y Y Patient contact Y Y Y Notes Database overarching ethics approval covers study New rules for anonymised databases now mandate HRA approval via IRAS submission Database must be compatible with study's ethics approval. In particular, mention of database as potential recruitment route in ethics application. The R&D office can supply some standard text you can paste into IRAS/REC applications
Contact details and help Research.database@cpft.nhs.uk Access to CRATE Training Custom searches R&D@cpft.nhs.uk CPFT R&D approval Research passport/letter of access Application to ethics and HRA amendments to SSI