Using Primary Care Data for ‘Real Time’ Health Protection Surveillance Gillian Smith on behalf of primary care surveillance team.

Using Primary Care Data for ‘Real Time’ Health Protection Surveillance

Gillian Smith on behalf of primary care surveillance team

UK work

Birmingham – all the ‘B’s

‘Bullring’

‘Bilbo Baggins’

‘Baltis’

Outline

Why do we need real time surveillance?

What primary care systems could we use?

Two examples of primary care systems used in the UK:

NHS DIRECT

QRESEARCH

How have these systems been used in heat related illness?

Discussion about way forward

Why do we need health protection surveillance in primary care?

Clinical diagnoses made by GPs (and symptoms reported to NHS Direct) offer the ability to systematically monitor a variety of syndromes/symptoms which could give early warning of a health protection issue (microbiological/chemical/radiological)

To monitor milder illnesses which may not present to hospitals

To monitor illnesses for which laboratory specimens not taken ( e.g. influenza,chicken pox)

Assess burden of of infection in primary care

To assess impact of health protection policies in primary care

To reassure that there has not been an increase in symptoms in the community

Overview of surveillance systems in UK

• Pre primary care – based on reported symptoms ( NHS Direct)

•‘Spotter schemes’ based on consultations (e.g. RCGP)

•GP databases ( e.g. GPRD and QRESEARCH)

Why ‘real time’ surveillance?

Why ‘real time’ surveillance?

Why ‘real time’ surveillance?

Why ‘real time’ surveillance?

Principles of work

In collaboration with colleagues working in primary care

Work on syndromes which may be related to infection, chemicals or environmental causes therefore across the HPA

Focus on areas where primary care data can provide ‘added value’ to the overall picture

Background to NHS Direct

NHS Direct is a telephone health help line.

Open 24 hours a day, 365 days a year

Aims to provide the public with health

advice and information

First introduced in 1998, providing a

service in 3 pilot areas

Now service covers the whole of England and Wales

22 sites covering 2.5 million people each

Aim

The aim is to identify an increase in symptoms that may be caused by the deliberate release of a biological or chemical agent, OR MORE COMMON INFECTIONS/ HEALTH PROTECTION ISSUES

How the project works

NHS Direct site 1

NHS Direct site…..

NHS Direct site...

NHS Direct site 22

NHS Direct National Operations Centre

Health Protection AgencyWest Midlands

Regional Surveillance Unit

NHS Direct sites

Health Protection

Units

Departmentof Health

Health Protection Agency

NHS DirectNational Team

Other agencies

PCTs/SHAs

Data collection

Analyses

Dissemination

Data collection

Call data on 10 ‘key’ symptoms/syndromes are transferred each week day from the 22 call centres (covering all of England & Wales) to the Health Protection Agency at West Midlands

Cold/flu Cough Fever Difficulty breathing

Vomiting Diarrhoea

Double vision Eye problems Lumps Rash

Heat stroke monitored during summer months

Call data are broken down by site, symptom, age-group and call outcome

Further details of individual calls including postcodes can be requested if needed

Analysis - control charts

Control charts constructed for 6 algorithms for 10 sites (major urban centres).

Model constructed assuming the Poisson distribution using the Gamma distribution to account for over-dispersion.

Incorporate a bank holiday and seasonal factor with a day factor and time trend factor included if required.

99.5% prediction limits are calculated for each day. The prediction limit for the "future" is based on the average number of total calls to-date.

99.5% upper prediction limits are also constructed for the remaining sites and algorithms (where control charts are not available) using standard formula (bank holiday/day/time trend factors not included).

Manchester NHS Direct

0

2

4

6

8

10

12

14

16

Oct

-01

No

v-01

Dec

-01

Jan

-02

Feb

-02

Mar

-02

Ap

r-02

May

-02

Jun

-02

Jul-0

2

Au

g-0

2

Sep

-02

Oct

-02

No

v-02

Dec

-02

Jan

-03

Feb

-03

Mar

-03

Ap

r-03

May

-03

Jun

-03

Jul-0

3

Au

g-0

3

Sep

-03

Oct

-03

Pro

po

rtio

n (%

)

proportion of fever calls (%) 99.5% upper prediction limit

Proportion of NHS Direct fever calls (5-15 year age group) by region

Figure1 - NHS Direct fever calls (5-14 year olds) as a proportion of total calls for this age group (2005-2006) [7 day moving average]

0

5

10

15

20

25

22-O

ct-0

529

-Oct

-05

5-N

ov-0

512

-Nov

-05

19-N

ov-0

526

-Nov

-05

3-D

ec-0

510

-Dec

-05

17-D

ec-0

524

-Dec

-05

31-D

ec-0

5

7-Ja

n-06

14-J

an-0

621

-Jan

-06

28-J

an-0

64-

Feb

-06

11-F

eb-0

618

-Feb

-06

25-F

eb-0

64-

Mar

-06

11-M

ar-0

618

-Mar

-06

25-M

ar-0

61-

Apr

-06

Pro

port

ion

of c

alls

(%

)

London West MidlandsNorth West North EastWales East MidlandsSouth West South EastEast England and Wales

Benefits of daily call data

• Provided an early indication of a rise in illness• Surveillance of children not necessarily visiting

GPs or hospitals• Response to media - confirmation of regional

levels• Daily reporting when needed• Ongoing monitoring

NHS Direct callers self-sampling study 2004/05

Aims of study to answer the following questions:

1) Can self-sampling by callers to NHS Direct give early warning of an increase in influenza activity in the community?

2) Does self-sampling by callers to NHS Direct provide added value over existing surveillance systems for influenza?

3) What are the implications of introducing self–sampling for influenza testing to NHS Direct callers and to the NHS Direct service?

Sampling: 2004/2005

November 2004 – February 2005: NHS Direct callers (>15 yrs) in Hampshire, West Midlands and South Yorkshire reporting ‘cold/flu’ were recruited by NHS Direct nurses

Each caller was sent specimen kits containing 2 nasal swabs, viral transport medium, instructions and information, packaging and a pre-paid reply envelope

Callers were asked to return kits to the HPA virus lab & samples were tested by multiplex PCR for Influenza and RSV viruses, and if (+) cultured for viable virus isolation

Results were sent back to national influenza surveillance team within the HPA CfI and to the NHS Direct callers

Sampling method

Results

Response rate: 294 kits sent out - 142 samples (48%) returned

Positivity rate: 23 of 142 samples (16.2%) tested positive for influenza viruses. Positivity was 30-40% during peak weeks

Influenza: 23 samples

Influenza A (H1N1): 3

Influenza A (H3N2): 16

Influenza B: 4

RSV: 8 samples

Positive samples included the 2nd community sample of influenza A (H1N1), 4th of influenza A (H3N2) and 1st influenza B sample received by the ERNVL during the 2004/2005 influenza season.

7 of 141 callers reported minor problems in taking swabs 

Average time from NHS Direct call to result was 7 days

Conclusions from sampling project

• Self-sampling by NHS Direct callers provided early warning of influenza circulating in the community and detect multiple strains of the virus

• The added value of the scheme may lie in providing early warning of influenza rather than ongoing surveillance throughout the entire season

• Despite spending on average a week in transit the samples provided good viability for antigenic characterization (virus growth by culture) as well as for molecular detection

• Proof of concept that the NHS Direct community sampling tool can provide high quality and timely samples

HPA/Nottingham University Surveillance Project

HPA/Nottingham University Surveillance Project: What is QRESEARCH?

•A non-profit making, nationally representative sample of volunteering general practices (500) who use EMIS general practice computer systems

•Includes data on 8.2 million patients (4 m current)

•Some practices have up to 10 years of historical data on the database

•Patients and practices within the database are completely anonymous – no personal/practice identifiers are recorded

•QRESEARCH ‘added value’ lies in the ability to link to prescribing data and to undertake analyses using socio-economic data (Census, Townsend, IMDS, Rurality score etc) uploaded to the patient records

GP PRACTICE CLINICAL COMPUTER

QRESEARCH TEAM IN NOTTINGHAM

QRESEARCH FILESERVER IN NOTTINGHAM

EMIS FILESERVER

GENERAL PRACTICE TEAM

Activation of QRESEARCH

Recruitment and informed consent

Download of all coded data and then daily downloads

Episodic data transfers

Extraction of subsets for researchers & morbidity analyses

HPA/Nottingham University Surveillance Project

HPA/Nottingham University Surveillance Project: What is it?

•In April 2004 received funding from HPA for a two year pilot project - project now extended to March 2008 and funding provided for daily data provision ( influenza indicators)

•Aim to investigate the feasibility of providing weekly surveillance data and using the QRESEARCH database for a number of strategic projects

HPA/Nottingham University Surveillance Project: routine outputs

•Pilot Weekly Bulletin launched on 11th November 2004 with three ‘key’ indicators – influenza-like illness; influenza-like illness with anti-virals prescribed; vomiting

•Each week the bulletin includes a ‘key messages’ section, highlighting changes to the indicators

•Gradually extended indicators

•Widened distribution of bulletin in July 2005

•Evaluation of usage by HPA and NHS to be conducted in 2006/07

Key indicators – monitored weekly (could be monitored daily)

•Influenza-like illness

•Pneumonia

•Severe asthma

•Wheeze

•Vomiting

•Diarrhoea

•Gastroenteritis

•Mumps

•Measles

•Pertussis

•Heat stroke

•Influenza-like illness with anti-virals prescribed

•Uptake of influenza vaccine

•Proportion of those under 5 years with diarrhoea prescribed re-hydration therapy

•Impetigo and the proportion treated with fusidic acid

The weekly bulletin

• Weekly bulletins are produced every week including over Christmas and New Year

• Includes data for the previous week for a set of key indicators which are presented at UK, country, region and SHA level

Insert Image

HPA/Nottingham University Pilot Surveillance Project: routine outputs

UK GP consultation rate (per 100,000) for Influenza-like illness, 2005/6

Source: QRESEARCH weekly outputs

0

10

20

30

40

50

60

Rat

e pe

r 10

0000

26 28 30 32 34 36 38 40 42 44 46 48 50 52 2 4 6 8 10 12 14 16 18 20 22 24

Week number

UK 2005/2006 UK 2004/2005

Copyright QR 2003-2006(Database version 7)(Data source QRESEARCH - 15 Mar 2006)

all patients all agesInfluenza like illness rate per 100000 population in the UK

HPA/Nottingham University Pilot Surveillance Project: routine outputs

UK GP consultation rate (per 100,000) for Influenza-like illness with anti-virals prescribed, 2005/6

0

.5

1

1.5

2

2.5

3

3.5

4

4.5

5

Ra

te p

er

10

00

00

26 28 30 32 34 36 38 40 42 44 46 48 50 52 2 4 6 8 10 12 14 16 18 20 22 24

Week number

North East

North West

Yorks. & Humber.

East Midlands

West Midlands

East of England

London

South East

South West

Copyright QR 2003-2006(Database version 7)(Data source QRESEARCH - 15 Mar 2006)

all patients all agesInfluenza like illness with Anti-Virals rate per 100000 population in 2005/2006 by region

HPA/Nottingham University Pilot Surveillance Project: routine outputs

GP consultation rate for influenza-like illness by Region (rate per 100,000) 2005/6

Source: QRESEARCH weekly outputs

0

10

20

30

40

50

60

Rat

e pe

r 10

0000

26 28 30 32 34 36 38 40 42 44 46 48 50 52 2 4 6 8 10 12 14 16 18 20 22 24

Week number

North East

North West

Yorks. & Humber.

East Midlands

West Midlands

East of England

London

South East

South West

copyright QRESEARCH 2003-6(QFLU database version 2)

all patients all agesInfluenza like illness rate per 100000 population in 2005/2006 by region

Buncefield incident

Able to provide information

on consultations to SHA

level (and PCT)

Able to monitor range of conditions (as per bulletin)

‘Switched on’ daily reporting and piloted extended flu data set

No unusual increase in respiratory or complaints e.g. asthma etc.

Able to provide data ‘real time ‘ to incident team

Pandemic influenza planning

Data required daily – now have capability to do this for weekdays (can be ‘switched on’ in two days)

Can provide data to PCT level

Extended dataset for influenza related conditions and prescribing (QFlu available on 17 million patients)

QFlu has practices from every SHA in England and from 292 out of 303 PCTs

Can provide data on UK ( 30% overall - though under- represented in Scotland)

included in bulletin in last few weeks

So what has happened as a result of your work?

EARLY WARNING AND TRACKING OF RISES IN LLNESSS IN THE COMMUNITY

Part of influenza surveillance system and DH ‘heat health watch project’ (exploring utility for environmental monitoring)

Ability to detect sudden rise in symptoms/syndromes in the community, both at a national, regional level and now local level ( e.g. norovirus, influenza like illness)

Able to provide weekly uptake estimates for adult vaccinations

Only national daily surveillance system in UK and only national surveillance system using a health help-line in the world

PROVIDING REASSURANCE DURING TIMES OF PERCIVED HIGH RISK

e.g. Ricin incident , London bombings and Buncefield – data used by incident teams and quoted publicly by HPA

MONITORING OF HEALTH PROTECTION POLICIES

e.g. use of antibiotics post SMAC report - increase in three day courses (compared to longer course) of trimethoprim for UTIs)

Summary – So what has happened as a result of your work? (contd.)

EXERCISES

Daily data used regularly in Exercises ( e.g. heat wave, bioterrorist and influenza) and provided only consistent daily data feed for Exercise United Endeavour

PANDEMIC INFLUENZA PLANNING

Helped in stopping proliferation of local ‘spotter practice’ schemes ( as part of influenza pandemic planning) to monitor FLI - will be provided in consistent and standard way

Able to provide information on where anti-virals being used for FLI

NHS Direct can provide an alternative source of specimens if needed (from patients directly)

Provided daily data for modellers

PEER REVIEWED PUBLICATIONS

August 2003 Heat-wave

Lasted 2 weeks in France

4th hottest summer on record in UK

Brogdale near Faversham (Kent) recorded England's highest temp 38.5 °C on 10 Aug

Previous record 37.1°C Cheltenham, 3 Aug 90

32 °C was exceeded on three consecutive days between 4 and 6 August and then on five consecutive days between 8 and 12 August, somewhere in the UK

But in 1976, temperatures exceeded 32 °C (90 °F), somewhere in the UK, on 15 consecutive days starting 23 June.

Temperature distribution across

Europe on 10 August 2003 at 1500hrs

British Summer Time

Heatwave plan for England and Wales - Summer 2004

“By the time a heatwave starts the window of opportunity for effective action is very short indeed…”

“The HPA, in collaboration with NHS Direct, will refine mechanisms for the surveillance of increased heat-related illness with the aim of being able to provide daily real-time reports to the Department of Health.”

“These will provide a source of intelligence on (a) how severe the effects are, and (b) how well services are responding.”

DH - Heatwave Plan - 2004 [data:NHS Direct & Met Office]

Daily proportion of 'heat/sun stroke' calls to 6 NHS Direct sites ~ Summer 2003

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1-Ju

l-03

8-Ju

l-03

15-J

ul-0

3

22-J

ul-0

3

29-J

ul-0

3

5-A

ug

-03

12-A

ug

-03

19-A

ug

-03

26-A

ug

-03

Pro

po

rtio

n o

f N

HS

Dir

ec

t c

alls

0

5

10

15

20

25

30

35

Ma

xim

um

Ce

ntr

al E

ng

lan

d T

em

pe

ratu

re (

C)

Proportion of heat/sun stroke calls Seasonal mean proportion of heat/sun stroke calls

Daily mean temperature - 2003

6 NHS Direct sentinel sites (population 19 million): North West Coast, East Midlands, West Country, Kent/Surrey/Sussex, East Anglia, South London

‘Heat-health watch’

Published by Department of Health

26 July 2004

http://www.dh.gov.uk/PublicationsAndStatistics/Publications/

In the event of a “major incident” being declared, all existing emergency policies and procedures will apply.

Heatwave is “very severe or prolonged”.

Level 4 - Emergency

In addition to above:

Regional directors of public health ensure no utility suspensions.

PCTs and Social Service departments commission additional care and support to ensure daily contact with vulnerable individuals.

Hospitals and trusts alerted in case there is increase in admissions.

Met Office confirms threshold temperatures exceeded in any one region.

Level 3 – Heatwave

In addition to above:

Department of Health issues specific advice to general public.

Targetted media strategy.

PCTs and Social Service departments distribute advice to at-risk individuals and managers of care homes.

Met Office forecast of threshold temperatures for at least three days ahead in any region, or 80% chance of temperatures exceeding threshold on 2 consecutive days.

Level 2 - Alert

•Minimum state of vigilance.

Department of Health issues general advice to public and health care professionals

Regional directors of public health review utility suspension policies.

PCTs and Social Service departments review the identification of individuals at risk.

NHS Trusts review resilience of infrastructure and equipment.

No warning required unless there is 50% probability of the situation reaching Level 2 somewhere in UK in next 5 days.

Level 1 - Awareness

ResponseTriggerLevel

HPA responsibilities

In the event of a ‘major incident’ being declared, all existing emergency policies and procedures will apply. All Level 3 responsibilities will also continue.

Level 4

Same as for Level 2Level 3

The Health Protection Agency will continue surveillance of increased heat related illness reflected in calls to NHS Direct and GP consultations to provide daily real-time reports to the Department of Health. These will provide a source of intelligence on (a) how severe the effects are, and (b) how well services are responding.

Level 2

The Health Protection Agency, in collaboration with NHS Direct,will refine mechanisms for the surveillance of increased heat-related illness with the aim of being able to provide daily real-time reports to the Department of Health. These will provide a source of intelligence on (a) how severe the effects are, and (b) how well services are responding.

Level 1

Heat Watch Plan – NHS Direct surveillance

Surveillance – summer 2005

NHS Direct heat/sun stroke calls as a proportion of total calls: Northern, Central and Southern England

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1-M

ay-0

58-

May

-05

15-M

ay-0

522

-May

-05

29-M

ay-0

55-

Jun-

0512

-Jun

-05

19-J

un-0

526

-Jun

-05

3-Ju

l-05

10-J

ul-0

517

-Jul

-05

24-J

ul-0

531

-Jul

-05

7-A

ug-0

514

-Aug

-05

21-A

ug-0

528

-Aug

-05

4-Se

p-05

11-S

ep-0

518

-Sep

-05

25-S

ep-0

52-

Oct

-05

9-O

ct-0

516

-Oct

-05

23-O

ct-0

530

-Oct

-05

Pro

po

rtio

n o

f ca

lls

(%)

Proportion of heat/sun stroke calls in Northern England - 2005Proportion of heat/sun stroke calls in Central England and Wales - 2005

Proportion of heat/sun stroke calls in Southern England - 2005 Source: NHS Direct

Surveillance – summer 2005

Age distribution of NHS Direct heat/sun stroke calls (England and Wales)

0

20

40

60

80

100

120

1-M

ay-0

58-

May

-05

15-M

ay-0

522

-May

-05

29-M

ay-0

55-

Jun-

0512

-Jun

-05

19-J

un-0

526

-Jun

-05

3-Ju

l-05

10-J

ul-0

517

-Jul

-05

24-J

ul-0

531

-Jul

-05

7-A

ug-0

514

-Aug

-05

21-A

ug-0

528

-Aug

-05

4-Se

p-05

11-S

ep-0

518

-Sep

-05

25-S

ep-0

5

Nu

mb

er o

f ca

lls

Heat/sun stroke calls <5 years Heat/sun stroke calls 5-74 years

Heat/sun stroke calls >74 years

Source: NHS Direct

Surveillance – summer 2005

Outcomes (dispositions) of NHS Direct heat/sun stroke calls (England and Wales)

0

20

40

60

80

100

120

1-M

ay-0

58-

May

-05

15-M

ay-0

522

-May

-05

29-M

ay-0

55-

Jun-

0512

-Jun

-05

19-J

un-0

526

-Jun

-05

3-Ju

l-05

10-J

ul-0

517

-Jul

-05

24-J

ul-0

531

-Jul

-05

7-A

ug-0

514

-Aug

-05

21-A

ug-0

528

-Aug

-05

4-Se

p-05

11-S

ep-0

518

-Sep

-05

25-S

ep-0

5

Nu

mb

er o

f ca

lls

999 A&E GP Home Care Other

Source: NHS Direct

So what about heat related work?

NHS DIRECT QRES

FEASIBILITY Ongoing routinely generated data

National/ daily

Ongoing routinely generated data

National (could be daily)

SUSTAINABILITY Joint funding (NHSD and HPA)

Funded to 2008 (HPA)

ACCURACY Validated against established systems

Validated against established systems

TRANSFERABILITY TO ‘HEAT ALERT’

Are using but limited algorithms – outcomes useful

Morbidity – need further work on prescribing

OTHER COUNTRIES ? Similar national help lines

Good national coverage of one GP system

Is there utility and if so how might we improve this?

Can primary care data provide any ‘added value’ in either early detection of health problems or assessing size of health problems?

What morbidity indicators would be useful? – now ability to look at more unusual endpoints (neurological)

Can we be ‘clever’ in our prescribing linked to morbidity indicators?

?? Select risk groups for ongoing surveillance (able to do so with new GP system)

When should we ‘switch on’ daily reporting?

The English Summer!?