Empirical Grounding of Guideline Implementation in Cooperative Clinical Care Situations

Transcription

1 1 AI techniques in healthcare: evidence-based guidelines and protocols Workshop 29 August 2006 at ECAI 06, Riva del Garda, Italy Empirical Grounding of Guideline Implementation in Cooperative Clinical Care Situations Inger Dybdahl Sørby, Øystein Nytrø, Thomas Brox Røst Norwegian University of Science and Technology (NTNU) Norwegian EHR Research Centre (NSEP) Trondheim, Norway

2 2 Research focus: The relationship between a guideline and reality

3 3 Objectives Defining concepts - Effects - Goals - Indicators Achieved by: - describing - defining - formalizing Guidelines Defining concepts: - Actions - States -Measures Implemented by: - allocating - assigning - timing Care plans Defining concepts: - Actors -Resources - Interactions Realized by: -Performing -Writing -Talking... Clinical reality Defining concepts: - Acts - Events Observation, surveys and measurements Grounded guideline implementation EBM-based guideline development Soc/org learning

4 4 Empirical Grounding of Guideline Implementation in Cooperative Clinical Care Situations Clinical care situation: A time-limited process or sequence of actions/tasks (for an individual patient) in which the cast (persons filling roles) does not change, and which has identifiable start, preconditions, end, and result Cooperative: Several different roles and persons are involved in each patient care process

5 5 Empirical Grounding of Guideline Implementation in Cooperative Clinical Care Situations A plan (for clinical work) is the implementation or instantiation of one or more guidelines that is relevant for a patient and which eventually is acted out, or realized, by real people on real patients

6 6 Empirical Grounding of Guideline Implementation in Cooperative Clinical Care Situations Structured observation of clinical care situations in order to study and analyse guideline implementation

7 7 Methods: Observational studies Structured observation of complex care situations Information source: (Medium: E/P/Other) Supplementary Inform.: Information (what, why, difficulties, time used etc.): Observer s comments/questions: Patient Record Electronic Patient Record ICD-10 X-rays pictures/reports Pat. administrative system PDR Colleagues Patient Control Verification Consistency check Spends appr. 1 minute to collect relevant Sit. papers Part for the ID Source Information I/O Purpose Result Type Trigger Location Participants Physical Result no no. of preliminary discharge summary 1 1 Res Patient list Patient I Divide patient Resident X Date of admission name resp. responsible 2 Res F1a Regular I Overview X Regular medications at discharge med. X 3 Res Pre-record All I Familiarization Admission record to find previous 4 Head Admission All I Familiarization illnesses phys. note 5 Head Out-patient Hist I Familiarization X GP phys note (?) X Searches for the address of the GP Does not find 6 the address Res Test result Blood I Find High heam. anomalous percentage X Calls the secretary and asks her to find values High INR the address of 7 the GP Res Realizes ECG that he Heart I Former heart No heart 1 resident, rhythm illness? illness Room for 1 head Prerounds S1.7 tried to look up the patient s middle Regular group physician, 8 Res R-med Medications O Sep. activities 1 nurse name in the EPR 9 system Res Marevan- O INR and date S1.6 X Form Enters the correct name of the patient 10 Head Out-patient Former and finds the address phys note med. I Dosage No findings The patient states the correct name of 11 Res Requisition Blood O Test order INR X the GP; what is noted in the admission record is wrong Res Res Reg. med. Test result O I Sign. Summary P Diagnosis code for dictating 14 Res Marevan I Summary Spends some time (appr. 3 form minutes) to 15 Res Admission I Summary record 16 Res Requisition Blood I Summary Leads to S1.9 S1.8

8 8 Interpretations Frameworks for representing observable attributes of situations, actors, and action trajectories Aim: Being able to trace (potential) correspondence between guidelines and observed/documented reality

9 9 Example framework for categorizing clinical situations Attribute group Attribute Example values Situation Type Planned Location Trigger Patient information Actors Environment History Reason for admission Category Roles Abilities Systems Medium Events Information source Misc. Pre-rounds, Ward rounds, Discharge conversation Yes, No Office n, Patient room n, Hallway, Meeting room After pre-rounds, Ad-hoc Diabetes, hypertension, cerebral infarction Confusion, Chest pains, dyspnea New patient, Well-known patient, Ready for discharge Head physician, resident, nurse Interventions, delegation, decisions, medication EHR, Patient Administrative System, PACS, LIS Electronic, Paper-based * Patient list, Preliminary discharge report, Prescription Start time, end time

10 10 Example: Classification of observed events

11 11 Example Analysis Actors/information systems: Patientlist, Patient Chart, EHR, PAS, etc. Name Patientlist Actor: SeniorResident7 Name Patientlist Nav. into comm. underst. Actor: Nurse8 New Inform All EHR All EHR Evaluate Med Med Assess Legend: Info.type Actor A Info.source Actor B Synchronous event Time Event1 Event2 Event/force Event/force Findex PAS Info.type Actor A Actor B Asynchronous event Info.source Findex PAS Evaluate Inform Explan Blood Blood Remind Note book

12 12 Example Analysis Actors/information systems: Patientlist, Patient Chart, EHR, PAS, etc. Name Patientlist Actor: SeniorResident7 Name Patientlist Nav. into comm. underst. Actor: Nurse8 New Inform All EHR All EHR Med Med Findex PAS Findex PAS Evaluate Assess Evaluate Inform Explan Blood Blood Remind Note book

13 13 Discussion Structured observations possibly useful for -guideline grounding, -implementation, -validation -design Not useful for compliance analysis

14 14 Current progress More observations by means of the framework Synthesis of communicative act trajectories Mapping between observable situation attributes and various dimensions of existing guideline models (based on the eight dimensions from Mor Peleg et al. (2003): Comparing computer-interpretable guideline models: A case-study approach, JAMIA 10(1), 52-68)

15 15 Thank you! Questions?