Improved Surveillance of Bugs and Drugs that include patient data
Susan Hopkins Clinical Director Infection Services, Royal Free London Lead Healthcare Epidemiologist, AMR Programme, Public Health England
ESPAUR 2012
• CMO Report and book • Highlight scale of problem
2013 • AMR Strategy • Development of ESPAUR
2014 • 1st ESPAUR report: Resistance, Consumption and Stewardship • Development of Antibiotic Guardian; HPRUs; AST toolkits
2015 • 2nd ESPAUR report – UTI; granularity in prescribing; NHS Atlas; One Health UK • Quality Premium developed; National risk register; CPE surveillance
2016 • 3rd ESPAUR report; MDR displays; CPE ERS outputs; reduction in prescribing • Fingertips AMR local indicators; O’Neill report; CQUIN introduced; PPS
2017 • Focus on Big Ambitions: inappropriate prescribing and GNBSI • Big data development – patient level prescribing and AMR data linkage
ESPAUR 2010-2014: Year 2 Report
Data Linkage
DeepMind
Options for data access
Section 251 statue
Patient and their data
Minimise risk
Privacy
Maximise public benefit
Maintain public trust
Consent, studies
Pseudonymisation
Big Data or Big Headache
Need to protect patient confidentiality Maintain public trust Data protection Freedom of Information Information Governance Development of data linkage
techniques ‒ probabilistic versus deterministic
Principles of Data Protection Act
• Fair and Lawful • Purpose for processing • Adequate, Relevant and not excessive • Accuracy • Retention • Rights • Security • Transfer overseas
Working definition of pseudonymisation
• Technical process applied to identifiers which replaces them with pseudonyms
• Enables us to distinguish between individual without enabling that individual identified
• Either reversible or irreversible • Part of de-identification
Challenges in data linkage
• Modern cohorts/registries (NHS Number) • Huge amounts of data but different database structures/sizes • Major challenges when creating cross/cohort/platform analyses
• Semantic interoperability /data harmonisation issues • Original metadata - standards • Variable definitions from rad/lab results/GP(readcodes)/inpatient • Difficult to move very large and complex data
• Privacy protection essential • threat from re-identification scientists
• World-wide shortage of skills and expertise • No single person with all necessary skills • Build upon existing expertise, developments and investments
Data linkage and AMR Ecological Linkage
PHE Local AMR Indicators; Public Health Profiles; GP Profile
Individual linkage Mortality data : survival and cause of death GP and hospital activity: health service impact/comorbidity Laboratory and imaging systems: severity of condition/comorbidity Deprivation
Family/household linkage Impact on the wider family
Improve drug-bug surveillance
Increased coverage from NHS laboratories from 30% to 98% Increased daily reporting from 10% to 82% Increased automated reporting from 0% to 78%
Laboratory text files(preferred method)
SGSS LaboratoryData Import
Legacy CoSurv/Amsurv files
Bespoke Excel format (CDR and
AMR)
sFTP
Web Upload
Web Data Entry
Laboratory Person
SGSS Data
Warehouse
ReportingWeb
Business Intelligence Layer (BI)
Mining and Analysis
APIHPZone
External networks PHE network
BI Security layerRole based model based
on requirement / permission to view patient
level or aggregate data
SGSS Operational Database
Research in Progress Pharmacy systems
GPs, CCGs Hospitals
Deprivation & E. coli BSI rates, CCG, 2012/13-2015/16
~ 40% variation explained by deprivation
PHE mandatory surveillance team
Does antimicrobial stewardship explain the reduction in CDI?
Dingle et al. Lancet Infect Dis 2017 in press.
Mortality Number of deaths within 30 days of specimen collection by infection
Data Linkage Mandatory data and Mortality ONS More deaths within 30 days of E. coli BSI from S. aureus BSI and CDI combined. CFR highest for MSSA CFR E. coli = CDI
Mandatory Surveillance Team, PHE
Patients from LTCF have higher AMR in urinary tract isolates
Simple Graphics Data linkage, Postcode matching LTCF almost double trimethoprim and ciprofloxacin resistance in urines
Rosello, JAC, 2017
Factors associated with iGAS deaths in community onset cases
Data linkage for HES and SGSS, postcode matching, ONS mortality records
Saavedra Campos Epid Inf 2017
E. coli Resistance & Hospitals
Carbapenems
Gentamicin
Piperacillin/tazobactam Third generation cephalosporins
Fluoroquinolones
0.00%
5.00%
10.00%
15.00%
20.00%
25.00%
Fluoroquinolones 3rdg cephalosporins Piperacillin/tazobactam Gentamicin Carbapenems
Community onset 17.4% 9.3% 8.7% 8.6% 0.2%>2 Days 22.1% 14.4% 15.1% 13.5% 0.3%>7 Days 24.1% 15.9% 16.5% 14.8% 0.4%
E. coli BSI
Data Linkage 1. HES 2. ONS 3. HCAI DCS 4. SGSS
E. coli BSI
Data Linkage 1. HES 2. ONS 3. HCAI DCS 4. SGSS
Naylor BioRxIV 2017
Financial case for action
AMR Local Indicator
Sample Hospital with 464 patients in previous year: Excess costs = £605,000 Excess deaths = 60
https://improvement.nhs.uk/resources/preventing-gram-negative-bloodstream-infections/
E. coli bloodstream infections
Linkage of patient-level antibiotic prescriptions to laboratory reports
E. coli bacteraemia rate, England, 2011/12-Q2 2017/18 (PHE mandatory surveillance)
Incidence of Escherichia coli bacteraemia is still increasing
• For reports including risk factor data, about 50% of patients had a UTI reported as underlying source of infection
• Need to investigate prescribing and resistance pattern in the community
E. coli urine isolates susceptible or resistant to recommended UTI treatment options, England 2016
0%
20%
40%
60%
80%
100%
Acu
te
Com
mun
ity
Acu
te
Com
mun
ity
Acu
te
Com
mun
ity
Acu
te
Com
mun
ity
Acu
te
Com
mun
ity
Acu
te
Com
mun
ity
Nitrofurantoin Trimethoprim Co-Amoxiclav Quinolones* Mecillinam Fosfomycin
Prop
ortio
n E.
col
i urin
e sa
mpl
es
Not tested Susceptible Non-susceptible *99% ciprofloxacin
ESPAUR Report 2017
How appropriate are antibiotic prescriptions for UTI in the community?
Linkage of patient-level antibiotic prescriptions to laboratory reports
Linkage of patient-level antibiotic prescribing data to laboratory records
6 months before 1 Aug 2013 – 31 Jan 2014
6 months after 1 May 2014 – 31 Oct 2014
3 months 1 Feb – Apr 30 2014
PHE laboratory surveillance data All NHS numbers, all samples (e.g. urine, blood, sputum etc.)
NHS Business Services Authority Prescribing data
3 months 1 Feb 2014 – Apr 30 2014
Electronic prescribing system (EPS) only – captures about 20% of all prescriptions (does not include FP10 (green) forms that capture about 70% of all prescriptions)
Linkage of patient-level antibiotic prescriptions to laboratory reports
15,417 NHS no. 22,461 antibiotic prescriptions
(tablet only) AND bacterial isolate from URINE/KIDNEY
8,593 NHS no. 10,928 prescriptions
7,328 Trimethoprim prescriptions 3,600 Nitrofurantoin prescriptions
74% susceptibility to Trimethoprim
89% susceptibility to Nitrofurantoin 60% resistance to Trimethoprim
Trimethoprim prescription 3,000 +ve urine isolates identified in time relation of +/-
14 day prescription
Nitrofurantoin prescription 1,200 +ve urine isolates identified in time relation
of +/- 14 day prescription
*InitialoutputLinking English antibiotic prescribing and microbiology data
Linkage results
Antibiotic prescription combination Row
Oct '13 Oct '14Trim R T T T T T T T T T T T RNitro S R
Trim T T R T T T T T T T T T TNitro S
Trim T R T T T T T T T T T T TNitro S
Oct '13 Aug '14 Oct '14Trim R S SNitro S N N N N N N N N N N R S
Trim R T TNitro N S N N N
Jul '14Trim T T T RNitro N N S
Trim T RNitro N S N
Dec '13Trim S T TNitro S N N N
Dec '13Trim S T T TNitro R N N
Sep '14Trim T T T RNitro N N S
1 2 3 4 5 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Key: T Trimethoprim prescriptionN Nitrofurantoin prescriptionR Resistant urine isolateS Sensitive urine isolate
d)
c)
b)
a)
j)
i)
h)
g)
f)
e)
T T T T T T T T T T T
T N T N T
T N T N T
N N T T N
T N T T N
T N T T N
N N N N T N T
N N N N N N N N N N
T T T T T T T T T T T T
T T T T T T T T T T T T
6
T
AfterBefore February 2014 March 2014 April 2014
Example patients (n=10):Timing of prescription/testing
Linking English antibiotic prescribing and microbiology data Linkage of patient-level antibiotic prescriptions to laboratory reports
Antibiotic prescription combination Row
Oct '13 Oct '14Trim R T T T T T T T T T T T RNitro S R
Trim T T R T T T T T T T T T TNitro S
Trim T R T T T T T T T T T T TNitro S
Oct '13 Aug '14 Oct '14Trim R S SNitro S N N N N N N N N N N R S
Trim R T TNitro N S N N N
Jul '14Trim T T T RNitro N N S
Trim T RNitro N S N
Dec '13Trim S T TNitro S N N N
Dec '13Trim S T T TNitro R N N
Sep '14Trim T T T RNitro N N S
1 2 3 4 5 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Key: T Trimethoprim prescriptionN Nitrofurantoin prescriptionR Resistant urine isolateS Sensitive urine isolate
d)
c)
b)
a)
j)
i)
h)
g)
f)
e)
T T T T T T T T T T T
T N T N T
T N T N T
N N T T N
T N T T N
T N T T N
N N N N T N T
N N N N N N N N N N
T T T T T T T T T T T T
T T T T T T T T T T T T
6
T
AfterBefore February 2014 March 2014 April 2014
1,742 antibiotic prescriptions had a +MSU with AST in the 30 days prior to the prescription date
1,002 trimethoprim prescriptions: • 57% of isolates were susceptible to trimethoprim • 43% of isolates were resistant to trimethoprim
740 nitrofurantoin prescriptions: • 81% of isolates were susceptible to nitrofurantoin • 19% of isolates were resistant to nitrofurantoin
X
30 day window
Linking English antibiotic prescribing and microbiology data
Antibiotic susceptibility test result available prior to prescription
The future is bright
• Data linkage capabilities and projects increasing
• Capacity is a major issue
• Amount of work needed is often underestimated
• Public engagement and Privacy concerns
Acknowledgements Rebecca Guy, Katherine Henderson, Sarah Gerver, Emma Budd, Alex Bhattacharya,
Alicia Roseelo Dean Ironmonger, Richard Puleston, Anne-Marie O’Connell, Katie Hopkins, Rachel Freeman, Neil Woodford, Russell Hope, Graeme Rooney, David Ladenheim, Miroslava Mihalkova, Katherine Henderson, Chris Fuller, Peter Stephens, Mehdi Minaji, Diane Ashiru-Oredope, Berit Muller-Pebody, Alan Johnson
Members of the ESPAUR oversight group, PHE AMR delivery board and ARHAI
Health Protection Research Units, PHE staff, NHS BSA, Rx info, Quintiles IMS, NHS Digital, NHS Trusts, GPs, CCGs, DH, CMO, NHS England, NHS Improvement.
Top Related