Clinical implementation of SNOMED CT Charles Gutteridge Clinical

Clinical implementation of SNOMED CT Charles Gutteridge Clinical Engagement Lead Europe

A semantic life

Why do we need standardised vocabularies? Developing actionable insights Clinical characteristics Counting Patient level Prediction What will happen to me? Population health estimation Can we find a cause?

Developing a clinical vocabulary • Multidisciplinary • Easy navigation • Comprehensive • Computable NOT a List

Recording a patient history

Use of the SNOMED CT encoded Emergency Care Dataset in east London Charles Gutteridge Clinical Engagement Lead Europe Ben Bloom Emergency Care Physician Barts Health

CDS 010 Designed in 1970 s For when EDs were Casualties

ECDS

Chief complaint (low level)

Most frequent Chief Complaint terms

Moving ahead: NLP is ready to use

Implementing use of SNOMED CT Real world examples at Barts Health NHS Trust Kuala Lumpur SNOMED Expo 2019

Practical use of SNOMED Clinical Terms Clinical overview for point of care • Problem and diagnosis listing • Alerts and flags Population health interventions • Patient finding • Indexing • Analytics and reporting

Barts Health Exemplars Point of care • Learning disability flagging • Infection control management Population health intervention • Smoking cessation support • Care pathway management (COPD acute exacerbations)

Data Extraction and Clinician Led Analytics • Timely access to data • Rapid iteration of queries • Self-service analytics

Smoking Status and Smoking Cessation Recording and Reporting

Solution? Three-part process: (1) Data Collection (2) Data Extraction (3) Data Utilisation

Data Collection Smoking Status Selecting ‘YES’ auto added to the -generates a referral List to. Problem the Smoking as a. Clinic Cessation SNOMED Term

Why does that matter? Benefits of SNOMED …. . (1) Structured data – easy to extract (vs. data recorded as free text) (2) Highly granular – terms are more clinically accurate than ICD 10

ICD-10 Codes for smoking status • Short list of terms F 17% Mental and behavioural disorders due to use of tobacco O 99. 3 Mental disorders and diseases of the nervous system complicating pregnancy, childbirth, and the puerperium P 04. 2 Newborn affected by maternal use of tobacco P 96. 81 Exposure to environmental tobacco smoke in the perinatal period T 65. 2 Toxic effect of tobacco and nicotine Z 57. 31 Occupational exposure to environmental tobacco smoke Z 71. 6 Tobacco use counselling, not elsewhere classified Z 72 Tobacco use not otherwise specified (NOS) Z 77. 2 Contact with and exposure to environmental tobacco smoke Z 87. 8 History of nicotine dependence • Lacking granularity

Parent Term Child Terms

Further Child Terms

Why does granularity matter?

Addressing East London Needs Large South Asian community: • Tobacco chewing is common because tobacco is often added to paan • Paan = mixture of betel nut, herbs and spices +/- tobacco, wrapped in betel leaf and chewed • Chewing tobacco +/- betel is a major public health issue • risk of cardiovascular disease • risk of oral cancer • Adverse pregnancy outcomes • BME Stop Tobacco Project: 60% of Bangladeshi men & 50% of Bangladeshi women in TH chew tobacco • No ICD 10 codes to specifically indicate that a person chews tobacco • There is a SNOMED term for ‘Chews tobacco’ • Selecting ‘chews tobacco’ in the Power. Form adds that SNOMED term to the Problem List Accurate recording facilitates meaningful analytics that are specific to our local population

SNOMED for Analytics SNOMED Term Chews tobacco SNOMED Parent Term for Malignancy Enumerate the people who chew tobacco and have cancer (sub-group: oral cancer)

Data Extraction and Utilisation 120% 100% 80% Ward 4 E Ward 4 D 60% Ward 4 C 40% Ward 4 A Ward 4 B 20% 9 9 . 1 01 9 . 1 21. 01 . 1 14. 01 8. 1 07. 12 31. 24. 12 . 1 8 0%

Moving to the new world: the LHR • east London Patient Record (e. LPR) • IHE standards • SNOMED encoded lists • Free text • Multi-user platform • Discovery Data Service • Data data service for health and social care • SNOMED on FHIR • Publisher- subscriber model • Research access and consent model