1 Using an Informatics Tool to Improve Implementation of Recommendations by Consultants Martin C. Were, MD, MS. Regenstrief Institute, Inc. Indiana University School of Medicine BHI Lecture Series May 13, 2008
2 Outline Regenstrief Institute, Inc Background on project Creation of Consultant Recommended Orders (CROs) tool Evaluation and results Discussion
3 Regenstrief Institute, Inc.
4 Regenstrief Institute International Work Mamlin B, et. al. Cooking Up An Open Source EMR For Developing Countries: OpenMRS A Recipe For Successful Collaboration. AMIA Annu Symp Proc. 2006
6 OpenMRS Implementations
7 Regenstrief Institute Regional Work Indiana Network for Patient Care (INPC) Health Department Practice/Clinic Hospital
8 Health Information Exchange Services Health Information Exchange Services Hospital Payers Hospitals Clinical Messaging Medication Reconciliation Shared EMR Credentialing Eligibility checking Labs Health Information Exchange Physicians Results delivery Secure document transfer Shared EMR Clinical Decision Support Credentialing Eligibility checking Data repository Network applications Labs Clinical Messaging Orders Outpatient RX & PBMs Public health Needs Assessment Surveillance Reportable conditions ADE detection Physician office Ambulatory centers (e.g. imaging) Public health Payer Researchers Clinical Quality Measurement Claims Ajudication Secure document transfer De-identified, longitudinal clinical data
9 Regenstrief Institute Standards HL7 DICOM NCDP IP JPEG, MPEG 1 Code Standards: CPT ICD LOINC
10 Regenstrief Institute Institutional Work The Medical Gopher CPOE in use since 1986 Over 20 millions orders entered by 2003 (1) McDonald CJ, Overhage JM, Tierney WM, et al. The Regenstrief Medical Record System: a quarter century experience. Int J Med Inform. 1999; 54(3):225-53
11 Computerized Physician Order Entry (CPOE) CPOE is a computer application that accepts physician orders and may provide clinical decision support Medications Diagnostic studies including laboratory tests Ancillary services Nursing orders Consultation request
12 Providers Using Gopher
13 Gopher and Clinical Decision Support (G CARE) 2 Allergy and drug interactions check Restrictions of orderables to formulary Display of costs Reminders Active logic to redirect from one option to another (2) Overhage JM, Mamlin B, Warvel J, Warvel J, Tierney W, McDonald CJ. A tool for provider interaction during patient care: G-CARE. Proc Annu Symp Comput Appl Med Care. 1995;
14 Cost Savings
16 Preventive POE inpatient reminders 2001 (Dexter et al.) Dexter PR, Perkins S, Overhage JM et al. A Computerized Reminder System to Increase the Use of Preventive Care for Hospitalized Patients. N Engl J Med 2001;345:
18 Background Question at hand: How do we improve implementation of recommendations made by inpatient consultants? Only about half of recommendations by hospital consultants are implemented 1 Some reasons for low implementation rates: Lost paperwork Faulty communication Insufficient clinical information Lack of timeliness Differences in medical knowledge Differences of opinion about best practice Administrative or systems based barriers (1) Cefalu CA. Adhering to inpatient geriatric consultation recommendations. J Fam Pract. 1996; 42(3):
19 The hand writing of a doctor
20 The Typed Note
21 Role of the primary physician Autonomy for writing orders on their patients 2 Knowledge and familiarity with their patient Preferences and opinions (2) Salerno SM, Hurst FP, Halvorson S, Mercado DL. Principles of effective consultation: an update for the 21st-century consultant. Arch Intern Med. 2007;167(3):271-5.
22 Background: Preliminary work at Wishard Only 51% of recommendations made by an inpatient geriatrics consultation team were implemented 3 Implementation more likely when directly facilitated in some way (odds ratios greater than four, p < 0.001) Computerized Provider Order Entry (CPOE) can help: Improve legibility, timeliness, and specificity in making recommendations Leverage CPOE based Clinical Decision Support (CDS) (3) Weiner M, Lamar V, Munger S, et al. Variation in implementation of recommendations of inpatient geriatrics consultation. J Am Geriatr Soc 2006; 54 (S4):S94-95 [abstract].
23 Intervention Consultant Recommended Orders (CROs)
24 Consultant Recommended Orders
25 Consultant recommending a new medication
26 Consultant recommending a new medication
27 Consultant recommending change to active med
28 Consultant recommending D/C of active order
29 Consultant recommending change to active order
30 Consultant recommending change to active order
31 Consultant recommending change to active order
32 Consultant Recommended Orders
33 Consultant modifying previously entered CRO
34 Consultant Recommended Orders
35 Consultant Note
36 Methods: Consultant note with appended CROs
37 Primary team acting on CROs
38 Primary team acting on CROs
39 Primary team acting on CROs
40 Primary team acting on CROs
41 Primary team acting on CROs
42 Technical considerations Avoid confusion with active orders in user interface, notes, and in data repository Avoid confusion with student orders Re-use existing CPOE functions for CROs Take advantage of existing CDS system Figure a way to recommend discontinuation or revision of items implemented without active orders in the CPOE CDS = clinical decision support CRO = consultant-recommended orders CPOE = computerized provider order entry
43 Administrative and Workflow considerations Privileges to enter, view, or act on CROs Fate of CROs at discharge or during transfer to ICU or another team Prioritization of CROs Should CROs be purged after a pre-established number of days? When to display CROs to providers Whether display should be automatic or by user s choice CDS = clinical decision support CRO = consultant-recommended orders CPOE = computerized provider order entry
44 Our Implementation Only physicians could enter, view, or act on CROs CROs were stored separately in the data repository CROs were removed when a patient was transferred or discharged CROs were not prioritized CROs were displayed at the beginning of an order-entry session CROs = consultant-recommended orders
45 You have a tool to lets consultants enter recommendations as suggested orders and allows primary team to implement these but does it make any difference? i.e. so what?
46 Methods: Study design System piloted with Geriatrics Consulting Team and Medicine Hospitalist Teams Before and after study with respect to implementation of CROs Inclusion Criteria: Patients admitted to hospitalist service who underwent geriatrics consultation Exclusion Criteria: Patients not admitted to hospitalist service Discharged less than 24 hours following initial geriatrics consultation Died in the hospital Transferred to the ICU or to a non medicine team We provided instructions on how to use the tool and user support
47 Methods: Impact of CROs on implementation We compared implementation of recommendations before (20 patients) and after (20 patients) implementation of CROs Before Intervention Recommendations conveyed as part of a free text note After Intervention Recommendations conveyed as part of a free text note + recommendations entered as CROs using our tool One investigator (MCW) reviewed medical records for all study patients and documented: recommendations made by Geriatrics team recommendations that were implemented To assess reliability of identifying recommendations and their implementation, a second investigator (CK) independently rereviewed 40% of the study records
48 Survey: A questionnaire administered to a convenience sample of 29 (81%) of the 36 medicine residents who used the CRO tool Survey questions re: CROs 5-Point Likert Scale - Usefulness - Clarity of recommendations - Overall consultation quality - Quality of communication - Ease of implementation of recs Yes / No / Not Sure - Do CROs save time - Should Geriatrics continue to use - Should all consultants use CROs - Other Feedback
49 Methods: Data analysis Inter rater reliability estimated with the kappa statistic Total number of recommendations per patient compared between groups using an extra Poisson model Implementation rates of recommendations between intervention and control patients compared using a nonlinear mixed model with binomial outcome predictors included intervention as a fixed effect and patients as random effects to account for multiple recommendations per patient We computed descriptive statistics for the survey
50 Results: Patient Characteristics
51 Results: Inter rater reliability Seventeen records were reviewed by two investigators Of 202 recommendations identified by either reviewer, 169 (84%) were identified by both Among the 169, reviewers agreed on implementation in 146 (86%, kappa = ), regardless of study group (control = 82%, intervention = 89%, p = by chi square) Established confidence about the primary reviewer s assessment for the study Note: Recommendations for control patients were not made using CROs. For intervention patients, use of CRO tool occurred at the consultant s discretion
52 Results: Implementation of recommendations More (247 vs. 192, p < 0.05) were made for intervention patients mean recommendations (12.4±4.2 vs. 9.6±3.7) Among intervention patients, consultants entered 190 (77%) of the recommendations electronically as CROs; 164 (86%) of these were implemented Overall, implementation of recommendations was significantly higher in the intervention group (78% vs. 59%, p = 0.01)
53 Results: Survey The response rate was 83% (24/29) On a 5 point Likert scale, respondents agreed that CROs were: More useful (4.29 ± 0.69) Clearer (4.46 ± 0.59) Improved quality of consultation (4.25 ± 0.68) Improved communication between primary and ACE team (4.08 ± 0.88) Were easier to implement (4.79 ± 0.41) All respondents stated that CROs saved time and that all consultants should use the CRO tool
54 Discussion The deeply held idea that more recommendations per patient decrease implementation rates does not necessarily hold 4 Implementation of recommendations improved by > 30% when recommendations were entered and processed via CPOE Key to using informatics tools to improve co ordination of care is to simplify tasks which require complex decision making (4) Goldman L, Lee T, Rudd P. The Ten Commandments of Effective Consultations. Arch Intern Med. 1983;
55 Discussion The CRO tool is scalable and requires no fundamental changes to workflow for hospital wide adoption Work submitted to the AMIA Annual Symposium Proceedings 2008 RO1 application to conduct a larger randomized trial to evaluate impact of CRO tool on clinical outcomes and patient safety NLM Express Research Grant application to extend the use of CROs for recommendations from inpatient providers to outpatient providers
56 Discussion: Limitations Before and after design may accommodate bias Reviewers were not blinded Small sample size did not allow for adequate control for clustering by provider, length of stay, or patient characteristics
57 Discussion: Other possible uses of recommended orders Outpatient consultations Inpatient to outpatient recommendations Pre admission orders Recommendations between departments (e.g. medication reconciliation during an admission) Recommendations from public health to individual providers
58 Acknowledgements Michael Weiner, MD MPH (PI) Greg Abernathy, MD (Programmer) Siu Hui, PhD (Statistician) Carol Kempf, RN (Reviewer) Geriatrics & Medicine Teams Regenstrief Institute, Inc Grant Support 5K23AG from the National Institute on Aging LM from the National Library of Medicine.
59 Congrats! Akhila Balasubramanian, Ph.D.
60 Thank you.
61 Feedback, questions, and suggestions welcome. Thank you.