Experience in improving healthcare decision-making with health IT: integrating theory, research, and practice

1. Experience in improving healthcare decision-making with health IT: integrating theory, research, and practice Matthew Samore, MD VA Salt Lake City Health Care System Professor of Internal Medicine Adjunct Professor of Biomedical Informatics University of Utah

2. Acknowledgments Michael Rubin, MD, PhD Kim Bateman, MD Brian Sauer, PhD Lucy Savitz, PhD Tom Greene, PhD R. Scott Evans, PhD Randall Rupper, MD, MPH Salt Lake VA Informatics, Decision Enhancement, ,and Surveillance (IDEAS) Center Selected Investigators and Collaborators Jonathan Nebeker, MD Charlene Weir, PhD Frank Drews, PhD

3. Thesis of this talk Theory and models provide scientific underpinnings for generalization Which supports comparative effectiveness research For health services research and epidemiology: Use of models understood For clinical decision support: Not so much Health information technology ? informatics Sub-disciplines such as cognitive informatics crucial

4. More succinctly expressed: �The difference between theory and practice is that in theory there is no difference but in practice there is�

5. Statement of the problem: �A disproportionate amount of literature on the benefits [of health information technology] that have been realized comes from a small set of early-adopter institutions that implemented internally developed health information technology systems..�

6. Addressing generalizability In what contexts will effects generalize? What accounts for variability in results? Why are impacts lower in magnitude or narrower in scope in larger trials compared to single institution studies? How to incorporate information about implementation, adoption, formative evaluation?

7. Relevance to comparative effectiveness research (CER) CER priorities Directly focused on health information technology Compare the effectiveness of alternative redesign strategies�using decision support capabilities, electronic health records, personal health records Indirectly tied to health information technology Compare the effectiveness of various strategies To control MRSA To control healthcare associated infection To enhance patients� adherence to medication regimens

8. Addressing CER challenges Need to explicitly formulate causal question Determining identifiability Defining level of inference Validating methods to reduce bias

9. Conceptual frameworks (THEORY) Natural & engineered systems Co-evolution Cognitive processing Information overload ? fit-to-workflow Cyclical models of control Feedback and feed-forward

10. System co-evolution Fundamental theorem in informatics C. Friedman J Am Med Inform Assoc. 2009;16:169-170 Proposed modification: Computers plus humans create a distinct socio-technical system Characteristics are not equivalent to other industries

11. Relevance Level of inference needed to assess causal effect of health information technology: Socio-technical system Potential benefits of simulation

12. Cognitive processing Motivation, mental models, tasks, goals Influenced by social context Lack of fit-to-workflow experienced as: Information overload Interruptions

13. Relevance Cognitive informatics methods Task analysis Direct observation Match implementation strategy to task complexity

14. Second law of thermodynamics as applied to cognition: Humans seek states of reduced cognitive effort Workarounds As cognitive load increases, automatic processing systems kick-in

15. For those who believe that there is a Simpson�s quote for every situation �In this house, we obey the laws of thermodynamics! Homer Simpson�s response when his daughter builds a perpetual motion machine in which energy increases with time

16. Contextual Control Model Feed-back systems not sufficient Need to anticipate and predict Pure feedback systems subject to loss of stability Time horizon is long in strategic control modes Relevance Link between decision support and surveillance Surveillance contributes feedback and feed-forward capabilities

17. Feedback & feed-forward decision support

18. Illustrative experience with decision support for antimicrobial prescribing Two different technologies studied Clinical task: Management of patient with acute respiratory infection in outpatient setting Whether or not to prescribe an antibiotic Choosing the antibiotic Diagnostic label Impact of perceived or actual patient demand

19. Application of theory to practiceimplementation of electronic health records in rural settings Socio-technical system Hook was electronic prescribing Stepwise approach to adoption Accommodating variation Readiness to change Social context and clinic culture Encouraging play Avoiding information overload

20. Community intervention plus clinical decision support system  Standalone algorithms on handheld computers Community randomized trial

21. Clinical decision support system integrated with computerized clinic order entry   Algorithm usually triggered by ordering antibiotic Clinic randomized trial

22. Interpretation Deciding whether to prescribe an antibiotic and choosing the drug involve different cognitive processes Given that decision to prescribe an antibiotic is made Possible to embed correct choice in workflow Feed-forward decision support needed to impact the �is this a situation that warrants an antibiotic� decision Relevant to drug-drug interaction alerting

23. Recommendations and conclusions Models fundamental to translation of research into practice Incorporation of theory and models into comparative effectiveness research Role of simulation