ResoPrim Experimental network in primary health care Primary - - PowerPoint PPT Presentation

UCL

Gefinancierd door de Programmatorische Federale Overheidsdienst Wetenschapsbeleid Financé par le Service Public fédéral de Programmation Politique Scientifique

ResoPrim Experimental network in primary health care

Primary Healthcare Research Network: the Belgian ResoPrim Recommendations

Etienne DE CLERCQ, Viviane VAN CASTEREN, Pascale JONCKHEER, Peter BURGGRAEVE MIE2009 congress Sarajevo, 30/08 -02/09/2009

UCL – IRSS, Clos Chapelle aux Champs 30.41, 1200 Brussels | Belgium email: etienne.declercq@uclouvian.be

2

Overall aim of the project

Development of a stable experimental network/ framework for collection (from the EMR), analysis and dissemination of data from primary health care To become a reference network/framework, enabling to improve the quality of other health information networks in primary care in Belgium

3

Objectives

• To issue recommendations for sustaining

development of other networks (early 2009)

• On a longer term : to implement (some of) these

recommendations to set up a stable EPR-based national research network (2009 - /)

Setting-up the ACHIL research laboratory (Ambulatory Care Health Information Lab)  Part of the Belgian health information strategy

4

ResoPrim

• Financed by Federal Public Planning Service –

Science Policy (framework of multi-annual Programme for the Development of the Information Society)

• October 2003 – December 2008 (2 pilot phases)
• 4 partners, 64 GPs, 6 software systems, an

international expert committee

5

Specific context

In Belgium:

10,000 GPs; 6,000 GPs using an EPR; > 18 software systems, …

For the ResoPrim project

Volunteer participation of the GPs, No highly specific IT skill required , use of current running systems, No GP patient lists

6

Description of the Resoprim network

7

Objectives (1)

Analyse conditions regarding organisation and use of a

pilot network

• possible routine EPR-based data exploitations in

various domains: epidemiology, quality of care, socio- economy, health research information system

• organisation and management of network
• validity of collected data in various domains
• benefits/drawbacks for GPs in participating in the

network

8

Methods (1)

• Extended literature review
• Definition of research questions
• Methods used

 Quantitative research  Qualitative research  Questionnaires to GPs

• Recruitment
• Satisfaction survey after data collection
• Satisfaction survey after feedback

 Analysis of data previously collected

9

Methods (2): Research domains

Epidemiology Quality of care HRIS assessment Denominator and sampling Socioeconomy GPs’ education and benefits

10

Defining required data

Could the network be a tool for GPs to manage their proper quality of care? Can we appraise patients’ health risks and their management? R016: How many hypertensive patients have an undefined CVR because of missing data? + RO17, RO18

11

RETROSPECTIVE

JAN 2002 DEC 2004 COMPLETELY AUTOMATIC EXTRACTION

PROSPECTIVE

(+/- 6 weeks)

14 FEB 2005 27 MAR 2005 A:AUTOMATIC EXTRACTION B:INPUT FROM THE GP

Quantitative research: Data collection (contact based)

JAN 2004 DEC 2007 28 May 2007 Dec 2007 PHASE 1 2

12

Quantitative research : Data types

Demographic and contact related data

• age, sex, contact date, …)

Clinical data

• Diagnoses (ICD10 – ICPC 2)
• Some drugs prescribed
• The fact that a referral to a specialist has been

made during the contact

Some parameters: height, weight, blood pressure, tobacco use

13

Results

The 18 Recommendations

• Some key features

Much more information on the ResoPrim website: http://www.iph.fgov.be/epidemio/epien/index38.htm

14

Privacy protection

15

Privacy protection

Do we need explicit (written) informed consent from the patients? (i.e.: should the GP actively inform the patient?)

• Not for retrospective data
• Yes for prospective data …

…if it does not interfere with the care process Otherwise …

16

Privacy: Patient’s informed consent

• The patient should at least be informed « passively » by means
• f a poster and a folder (information that could be taken back

home). The GP should also be able to provide additional information.

• At least, the patient’s refusal has to be recorded explicitly. A

period to express this refusal has to be foreseen after the consultation/information of the patient (for instance 15 days). GP should also be able to express his/her own refusal.

17

Privacy: Communication to third parties

• The GP should sign a contract in order to restrict the

potential usage of the anonymous data (that eventually could be done by third parties).

18

Quality control & assessment

19

A wide & well known underestimation

Sensitivity of automatic extraction from EPR

• Diagnostic (HTA): 54%
• Antihypertensive medication: 60%
• Referral (hypertensive patients): <11%
• Parameter availability (height): 52%

20

Quality control & assessment

HRIS = Tool  needs to be calibrated

GP EPR DB Quality of the DB content Quality of the extraction modules

(Dummy patient technique)

21

Source validation

GPs’ thoughts EPR Research Data Base Data Base

“gold standard” Source validation

22

Automatic extracted Diagnoses, drugs, referrals

Sensitivity and Positive Predictive Value (PPV) of automatic extraction vs questionnaire Sensitivity (min – max) PPV (min – max) Drugs 34.1% - 59.34% 91.5% - 94.1% Diagnoses 53.6% - 67.1% 42.1% - 94.90% Referrals < 17% < 37% % = min. and max. per question

23

HRIS content improvement

We can improve it! How?

• Participation to (thematic) data collection
• networks. This effect could last.
• Global measures (e.g.: software homologation)

24

Participation impact ICPC

25

Persistence of some effects

2004 2005

26

Additional strategies

To improve the HRIS content we can also act on

• For the drugs sensitivity:

 The extraction procedure design  The ‘active drugs’ management

• For the diagnosis sensitivity:

 The education?

• For the Diagnosis PPV

 The extraction module (family history, derived codes)  The questionnaire? (list of questions?)

• For the parameters sensitivity & completeness

 The software interfaces

27

Can we do something with the data currently available in the EPRs?

28

HRIS and Quality of care

• If no quality control assessment is foreseen, we presently

do not recommend to use routine EPR data for identification of a target population.

• It is presently not recommended to compare the quality of

care between GPs (too much uncertainty about the “EPR use” of GPs).

• If, however, we assume that incompleteness of data and

changes in EHR use are homogeneously spread across various goups or types of patients within any one group of GPs, then it may be possible to compare various subgroups of patients or to monitor some aspects of the quality of care.

29

How to encourage (volunteer) GPs?

30

GPs’ involvement – key factors (1)

• GP invited by a personal contact
• Exhaustive information
• Trust in the organizers
• Use of “anonymous” data
• Protection (encryption) of data sent
• Scientific objective of the network
• Acceptable daily workload
• No supplementary request from patient

31

GPs’ involvement – key factors (2)

• An efficient help desk
• Learning opportunities or individual feedback

(more controversial)

• Financial incentive (also controversial)

…To manage all that at a national level is challenging!

32

Conclusions

A HRIS can be set up with a high acceptable privacy protection. Such EPR based network can be a useful tool to monitor some aspects of the quality of care …but … We need a quality control and assessment procedure to calibrate this tool (and monitor its improvement). To encourage GPs’ involvement at a broad national level is still challenging.

33

34

Quality of a research DB content

Patient GP EPR DB

Facts Thoughts Registered thoughts Extracted info

Information System Consultation