Online Dissemination of Clinical Practice Guidelines as Narrative - - PowerPoint PPT Presentation

Online Dissemination of Clinical Practice Guidelines as Narrative Texts and Structured Pathways

A Case Study with the Treatment of Type 2 Diabetes

• B. S´

eroussi1,2, H. Falcoff3, V. Ertel-Pau4, V. Raimond4,

• J. Bouaud1,5

1Sorbonne Universit´

es, UPMC Univ Paris 06, UMR S 1142, LIMICS, Paris, France ; INSERM, U1142, LIMICS, Paris, France ; Universit´ e Paris 13, Sorbonne Paris Cit´ e, LIMICS, (UMR S 1142), Bobigny, France

2AP-HP, Hˆ

• p. Tenon, D´

epartement de sant´ e publique, Paris, France; APREC, Paris, France

3Universit´

e Paris Descartes, Sorbonne Paris Cit´ e, D´ epartement de m´ edecine g´ en´ erale, Paris, France; SFTG, Paris, France

4Haute Autorit´

e de Sant´ e, Saint-Denis-La-Plaine, France, France

5AP-HP, DRCD, Paris, France

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 1/16

Tekst

Context & Objectives Method Results Discussion & Conclusion

Clinical practice guidelines and CDSSs

Clinical practice guidelines (CPGs)

Promotion of Evidence-based medicine Recommendations for patients with specific conditions Support practitioners’ decisions

Publication as textual documents

Low impact on clinical practices

Guideline-based clinical decision support systems (CDSSs)

Deliver patient-specific recommendations Efficient tool to promote CPG implementation Potential impact on clinical pratices

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 2/16

Context & Objectives Method Results Discussion & Conclusion

Problem and objectives

CPG content modelling from text - formalization

Ambiguities, incompleteness Delay in CDSS delivery Static CPG textual knowledge

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 3/16

Context & Objectives Method Results Discussion & Conclusion

Problem and objectives

CPG content modelling from text - formalization

Ambiguities, incompleteness Delay in CDSS delivery Static CPG textual knowledge

Aim of the work Simultaneous CPG modelling and authoring of textual (CPGs) and structured formats (knowledge base)

Beneficial for both productions

Synchronous dissemination of textual and structured formats

Audience analysis User attitudes - online questionnaire

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 3/16

Context & Objectives Method Results Discussion & Conclusion

Problem and objectives

CPG content modelling from text - formalization

Ambiguities, incompleteness Delay in CDSS delivery Static CPG textual knowledge

Aim of the work Simultaneous CPG modelling and authoring of textual (CPGs) and structured formats (knowledge base)

Beneficial for both productions

Synchronous dissemination of textual and structured formats

Audience analysis User attitudes - online questionnaire

Management of type 2 diabetes (T2D-CPGs)

Collaboration with French National Health Agency (HAS) Update of 2006 CPGs Structured representation: RecosDoc-diab` ete

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 3/16

Context & Objectives Method Results Discussion & Conclusion

CPG authoring method

A HAS-controlled standard methodology Working group - 1 HAS project leader and 2 medical project managers First draft based on literature review 5 to 6 WG meetings Each proposed recommendation is voted Grading: EBM (A, B, C) or PA

Stratégie médicamenteuse du contrôle glycémique du diabète de type 2

RECOMMANDATIONS Janvier 2013 RECOMMANDATION DE BONNE PRATIQUE

Stratégie médicamenteuse du contrôle glycémique du diabète de type 2

Méthode « Recommandations pour la pratique clinique »

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 4/16

Context & Objectives Method Results Discussion & Conclusion

CPG formalization

WG1: project presentation Work with 1 medical project manager (H. Falcoff) Formalization of each intermediate version of CPGs

Structuration as a decision tree Grading: EBM, PA, or ExpB Abstractions, ambiguities, incompleteness reported to WGs

A browsable knowledge base Last WG: presentation of RecosDoc-diab` ete Synchronous online publication

HbA1c
>
therapeu/c
target

 and
pa/ent
under
Me4ormin


• Me4ormin
+
Sulfonylurea


Sulfonylurea
contraindicated


• Me4ormin
+
Alpha
glucosidase
inhibitors

• Me4ormin
+
DPP4‐inhibitors


No
 Yes


• Me4ormin
+
Insulin

• Me4ormin
+
GLP1‐receptors
agonists



 Yes

 No

 No

 Yes

 HbA1c
near
target
 Overweight


S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 5/16

Context & Objectives Method Results Discussion & Conclusion

Online questionnaire on RecosDoc-diab` etes

Short, open, anonymous questionnaire 5 questions related to the delivery

• f structured CPGs

Context of use of the system Ease of use Usefulness for practice Intention to re-use Usefulness for understanding CPGs

Free comments or suggestions Respondent profiles

Profession Years of practice Type of practice

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 6/16

Context & Objectives Method Results Discussion & Conclusion

Double format publication

Textual CPGs production

31 recommendations Some improvements

Structured version - RecosDoc-diab` ete

First version of the knowledge base

89 criteria 805 clinical profiles: 2 EBM, 37 PA, 766 ExpB

Expert-based profiles not endorsed by HAS Second version

30 criteria 88 clinical profiles: 2 EBM, 77 PA, 0 ExpB

Press conference by HAS (Feb. 13th, 2013)

T2D-CPGs dedicated page on HAS website Link to RecosDoc-diab` ete

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 7/16

Context & Objectives Method Results Discussion & Conclusion S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 8/16

Context & Objectives Method Results Discussion & Conclusion S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 9/16

Context & Objectives Method Results Discussion & Conclusion

Consultation statistics

Analysis ‡ Dec. 31st, 2013 (awstats v0.7)

55,203 visits of the HAS CPG page 10,565 visits of RecosDoc-Diab` ete

Mostly (68%) from the HAS website

13272% 6794% 4592% 5717% 4065% 2987% 2354% 3507% 4111% 4475% 3329% 4850% 1494% 709% 664% 548% 370% 298% 463% 419% 420% 330% 0% 2000% 4000% 6000% 8000% 10000% 12000% 14000% 16000% Feb.% March% April% May% June% July% Aug.% Sept.% Oct.% Nov.% Dec.% HAS%web%site%visits% RecosDoc%visits%

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 10/16

Context & Objectives Method Results Discussion & Conclusion

Consultation statistics

Analysis ‡ Dec. 31st, 2013 (awstats v0.7)

55,203 visits of the HAS CPG page 10,565 visits of RecosDoc-Diab` ete

Mostly (68%) from the HAS website

13272% 6794% 4592% 5717% 4065% 2987% 2354% 3507% 4111% 4475% 3329% 4850% 1494% 709% 664% 548% 370% 298% 463% 419% 420% 330% 0% 2000% 4000% 6000% 8000% 10000% 12000% 14000% 16000% Feb.% March% April% May% June% July% Aug.% Sept.% Oct.% Nov.% Dec.% HAS%web%site%visits% RecosDoc%visits%

[Update] ‡ July 31st, 2014

+1,935 RecosDoc-Diab` etes visits – ≈ 276/month, decreasing

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 10/16

Context & Objectives Method Results Discussion & Conclusion

Online questionnaire – Respondent profiles

186 responses 67 free comments

GPs$ 64%$ Diabetologis ts$ 6%$ Other$ specialist$ 11%$ Other$HC$ prof.$ 12%$ Diabe<c$ pa<ent$ 2%$ Other$ 5%$

Professions)(n=184))

<"5"y" 21%" 5(10"y" 13%" 10(20"y" 18%" >"20"y" 48%"

Years&of&prac+ce&(n=179)&

Healthcare) facili,es) 21%) Individual) prac,ce) 30%) Group) prac,ce) 26%) Health) centers) 5%) Coordinated) care) structures) 3%) Other) 15%)

Type%of%prac+ce%(n=178)%

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 11/16

Context & Objectives Method Results Discussion & Conclusion

Online questionnaire – Answers

Physician subgroup analysis (n=147)

Study of answers wrt:

Profession (GP vs specialist) Years of practice (< 10 vs > 10)

No significant statistical association

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 12/16

Context & Objectives Method Results Discussion & Conclusion

Online questionnaire – free comments

67 free text comments (36% of respondents)

10 non interpretable 3 reporting technical problems 54 interpretable

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 13/16

Context & Objectives Method Results Discussion & Conclusion

Online questionnaire – free comments

67 free text comments (36% of respondents)

10 non interpretable 3 reporting technical problems 54 interpretable

Comment categorization

17 about CPG content (mostly criticisms) 37 about RecosDoc-Diab` ete

25 positive – 16 suggestions for development 12 negative “The decision tree is too complex” “Medicine is an art and cannot be computerized”

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 13/16

Context & Objectives Method Results Discussion & Conclusion

Discussion

Fewer visits of the structured version

20% of HAS CPG webpage visits Cultural effect Poor visibility But visits continue

Criticisms of HAS CPGs (30% of comments)

Poorly evidence-based, professional agreements Drug-centered, incomplete, abstract

Remarks on the system

Contradictory – “one size does not fil all” too complex and too simple Add documentation (patients, students, physicians) EMR integration, smarphone application

Limitations

> 2% of system visits Voluntary polling

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 14/16

Context & Objectives Method Results Discussion & Conclusion

Conclusion

First experiment of dual development for HAS

Successful dual textual-structured CPG publication

Perspectives

Renew dual developpement of CPGs and publication Deeper integration of content formalization Explicit acknowledgment of knowledge/evidence gaps Provide room for additional decision support (expert-based) More integrated CPG-based tools to promote CPG implementation

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 15/16

Context & Objectives Method Results Discussion & Conclusion

Thank you!

S´ eroussi et al. MIE 2014, Sept. 1st, Istanbul, Turkey 16/16