Towards a Structured Electronic Patient Record for supporting Clinical Decision- Making

Transcription

1 Towards a Structured Electronic Patient Record for supporting Clinical Decision- Making Bente Christensen, Gunnar Ellingsen Telemedicine and e-health Research Group, University of Tromsø, Norway bente.christensen@telemed.no.,gunnar.ellingsen@uit.no Abstract. At present, Electronic Patient Record systems (EPRs) in Western hospitals are mainly stores of free-text patient information and lack utility for purposes other than accessing such information. The need to improve support for clinical work processes and decision-making has put pressure on vendors and decision makers to put forward a strategy for enhancing the structure of EPR content. This paper reports on the first attempt to implement a structured EPR using a two-level modeling approach in Norwegian hospitals. Taking a work practice perspective, we show that implementing this new EPR in patient care planning implicates several socio-technical challenges that need to be solved in the process. Keywords. EPRs, work practice, clinical decision support. 1 Introduction Over the last few decades, Western hospitals have increasingly implemented Electronic Patient Records systems (EPRs) systems to document the treatment and care of patients. As a result, the EPRs have grown to considerable sizes, due to the accumulation of large amounts of patient information. A key problem is that much of the information in EPRs is free text (of which physicians narrative notes are the most prominent example). This makes it hard to use EPRs for purposes other than registering and looking up patient information [1]. Increasingly, it has been pointed out that EPRs also should be capable of providing tailored Clinical Decision Support (CDS) [2] at crucial points in patients treatment trajectories. This has put pressure on vendors and decision-makers to put forward a strategy that facilitates such a process by: a) structuring the content in EPRs, b) identifying crucial tasks in particular need for decision support and c) automatically triggering the next step in patients treatment pathways. Simultaneously we recognize that achieving CDS [3 5], as well as fostering the inclusion of more structured information in EPRs [6, 7], has proven hard to achieve in practice. In this paper, we look into a promising new strategy from the international openehr organization [8].

2 openehr is a two-level modeling approach aimed at supporting a high degree of interoperability between different healthcare systems and the reuse of data [8]. The openehr approach seems promising, as the use of archetypes allows user communities to define structured data in a very dynamic way, thus potentially avoiding out-ofhand user rejection of the structuration efforts due to unsuitable degrees of standardization. However, despite being a technical innovation, the openehr approach may still face many socio-technical challenges when the new technology is put into real use. We pursue this through the following research question: What challenges are associated with the introduction of decision-supportive EPRs based on the openehr approach, and how can these be dealt with? Empirically we report from a large-scale EPR project initiated in 2011 in the North Norwegian Regional Health Authority. A key aim of this project was to replace an existing, largely free-text-based EPR with a new archetype-based (i.e., highly structured) EPR offering extensive decision support. We focus on the first attempt to implement decision support in the surgery planning process. This study is positioned within a constructive paradigm and hence makes use of interpretive methods [9, 10] where we particularly focus on the users work practice. A variety of data sources are used to bring out different perspectives. The first author has participated in testing sessions and workshops with developers at software vendor dubbed BigVendor and health care personnel, conducting semi-structured interviews and document studies. In addition, informal talks with people involved in the development process clarified themes that are identified in this paper. Our analysis of the data has been guided by the philosophical perspective of hermeneutics, according to which a complex whole is understood from preconceptions about the meanings of its parts and their interrelationships [9]. 2. Theory The openehr framework separates the technical designs of systems from clinical concerns. A standardized reference information model represents the first level, while the openehr archetypes, based on the reference model, represent the second level [11]. The archetypes specify constraints on the data structures in the reference model. Hence, they serve as a constraint mechanism, ensuring that stored information is valid in terms of clinical knowledge. [1]. The outcomes of the openehr approach are systems and tools for computing with health information at a semantic level, thus enabling analytical functions like CDS. Clinical decision support depends on good quality clinical data repository and hence reinforces the need for standardized data representation and storage. Lack of good clinical data warehouse will have significant impact on the quality of advices emanating from CDS systems. Data mining algorithms require good quality clinical data repositories to be able to extract knowledge to support clinical decision-making [5]. CDS systems also depend profoundly on large volumes of readily-accessible, existing clinical datasets usually extracted from the repository content of EPRs. Lack of standardized data in the repository may lead to datasets not representative of the patient population (ibid). It is therefore essential that standardized data representation are

3 used for leveraging the knowledge base repositories to facilitate the generation of patient-specific care recommendations for physicians (ibid). Archetypes allow flexible terminology mapping. However, to enable interoperability, they require consensus on maximum definition. Being structured data elements, archetypes can also be displayed in different presentations. Together with the ability to match individual patients to a computerized knowledge base, in order to generate patient-specific assessments or recommendations, the reuse and easy access to data is considered decisionsupportive abilities. While acknowledging the great promise with the openehr approach, new technical solutions to be used in complex health care settings may still face many sociotechnical challenges [12]. For example, the temporal aspect of health care work makes it a process, not an event [13]. In our case, this means that making a plan for surgery, the planning process is not a result of a sequence of decisions, it is the emergent effect of the interlocking of entities performing subtasks [14]. In this respect collaborative activities including informal communication and coordination are of uttermost importance [15]. Munkvold et al. [16] found that pressing too hard on a practice to achieve more formalized routines, resulted in unexpected changes of the correlated informal practice. They demonstrated how informal routines, which initially were considered a problem, became part of the solution in order to make the new formalized routines work. They even found that the formalizing of routines nurtured the need for even more informality (ibid). Viewing clinical work as collaborative productions dependent upon and constitutive of teamwork has implications for the computing that supports the primary work [13, 17]. 3 The BigInvestment Project In 2011, the North Norwegian Regional Health Authority issued a call for tender, asking for new clinical ICT systems for all 11 hospitals in North Norway. The overall goal of the NNHRAs new investment was development of an EPR supportive of physician decision-making, with the objective of facilitating execution of patient treatment pathways and clinical workflows within the region. The process of developing and implementing the new EPR system took the form of a project known as Big- Investment. The cost of BigInvestment amounted to 82 million EURO for the period , making it one of the most ambitious healthcare-related ICT projects in Norway at the time. The leading Vendor of EPR systems in Norway, BigVendor, won the bid for tender. BigVendor s systems are used by 70% of hospitals in Norway, encompassing 65,000 users, including hospitals in the North Norwegian Regional Health Authority. In winning the bid for tender, BigVendor promised to develop what it denoted as the next generation EPR, NewArena. The NewArena will replace the old EPR, which, to a large degree, contains unstructured text that offers little CDS. Together with BigVendor, the BigInvestment project has planned the progressive development and implementation of NewArena. While NewArena is being developed, a substantial amount of clinical work will have to be done on the old platform, but this is not sup-

4 posed to affect users daily work. All information stored on the old platform must also be retrievable from NewArena after migration to this system and must be displayable as part of patients histories. NewArena is built on a two-level modeling approach, using the openehr architecture [11]. 4 Assessing patients for surgery Critically, NewArena provides clinicians with a totally new CDS tool that offers easy access and overview over patient treatment history and facilitates the documentation of patient treatment and care. The first functionality afforded by NewArena s product family benefited outpatient clinic physicians: The ambitions for Arena are high; they must be when we are talking about an Electronic Patient Record for the future ( ) The first version ready for customers in spring 2012 will allow physicians to work more efficiently in outpatient clinics. New functionalities will follow rapidly until the end of 2016 [18] Particularly in surgery planning, the role of physicians at outpatient clinics assessing surgery patients is vastly important. Patient assessment is the first step in the costly and complicated surgery planning process, a process characterized by the prevalent waste of resources, presumably due to missing information. The collection and documentation of clinical data earlier in the planning process is anticipated to improve the situation. Today, clinical data are registered as patients are admitted for surgeries; this might be the day before or the day of surgery. Pushing information gathering forward to outpatient consultations, where surgery is decided, is expected to improve the quality of the whole planning process, including planning for bedcapacity, anesthetics, use of surgery theatres and tools. NewArena shall also provide the ability to communicate plans and patient needs in a way that allows the system to catch up if anything is forgotten. Thus, for patients planned for surgery, for instance, a reminder will show up for the physician if treatment is not initiated according to the plan made by the physician in the outpatient clinic. Accessible and updated templates of patient pathways will also inform personnel involved in patient treatment about patient status and treatment plans and progress, thus making it easier to carry out treatment according to plans. Also in NewArena, clinical information, notes and other types of documents related to, e.g., medication, are expected be retrieved from patient records and displayed in surgery planning module, connected to surgery cases. Thus, the planning process will be facilitated by the reuse of structured, clinical data already registered in the EPR. 5 The first pilot After a one-year delay of the project plan, a version of NewArena was presented to clinicians for piloting and testing in October According to BigVendor, delivery was delayed because the surgery planning module in the new platform had to

5 «wait» for basic functionality to be achieved on the Arena platform, like structuring of data, in order to work optimally. The interaction between EPR and NewArena is complex, and BigVendor reports challenges related to interaction between the platforms and to the performance and implementation of new functionality in NewArena. (Dossiers to the steering committee meeting on BigInvestment, 10 October 2013) The starting point for surgery planning is defined to be outpatient consultation. It is important that physicians be provided enough information in outpatient consultations to plan and document decisions regarding further treatment. Fig. 1. The information displayed for physician in outpatient clinic in NewArena The use of new technologies that rely on archetypes to structure data is expected to facilitate the generation of overviews of patient health information in ways that allow physicians to quickly get overviews of problems, the anamnesis and possibly test results before patients enter consultations. The information displayed in the NewArena system is adapted to each physician s work tasks, so that if a physician enters the system from an outpatient clinic, the information needed in this setting is displayed. A surgeon who decides that a patient is going for surgery has to create a decisionnote within the NewArena system. This note is regarded as the go for the surgery planning process and serves as the head-document for what BigVendor has named a case. Other documents in a case typically include an admission note; a surgeon s assessment note; an anesthesiologist s note; nurses pre-, per- and post-operative

6 notes and a discharge letter. Case collections of such documents make it easy to retrieve and view relevant documents in relation to one another. As the decision-note is the head-document, other documents are attached to it, and it cannot be removed from its associated case. Only when a head-document is created may other documents be added to a case. This makes an inscription of the workflow meaning that others cannot start their jobs before the surgeon has produced a decision-note that establishes assignments. This approach may seem straightforward but is not, as today s working routines are that secretaries prepare for the surgeons work by creating decision-notes using information retrieved from EPRs, like diagnoses and codes and indications for surgery. This way, surgeons need only fill in decision-notes with data derived from consultations and can focus on patients, instead of their computer screens, while patients are present. The inscribed workflow in NewArena does not allow for this; due to accesscontrol rules, decision-notes, as part of surgery order forms, can only be established by physicians. The decision-note, the anesthetic pre-operative assessment and the surgeon s assessment note together are to replace the mandatory fields in the old surgery order form for two reasons: First, in the old surgery order form, the fields for surgeonsand anesthesiologists information could be filled by anyone with access to the system. Hence, from a juridical point of view, it was not possible to tie the data in these fields to the physicians responsible for assessments. Replacing these fields with documents whose validity must be verified by their authors will create accountability. Second, the structuring of text will make the importation of data into surgery order forms much easier and, hence, hopefully make available more complete information needed to prepare for surgery. However, in many cases, outpatient consultations reveal the need for further information collection via such complementary assessments as radiology or other tests, before a final decision on surgery can be made. Thus, under NewArena, it may remain as difficult to author decision-notes or patient health information may be just as insufficient as it is today. In the outpatient consultation, you must consider if you have sufficient information to decide if the patient is in need of surgery or if additional tests like radiology examinations are required. I might approach this in two ways: Either I agree with the patient that he is going for surgery, but I need additional assessments that will be carried out as he is admitted for surgeries. Then, there will be a risk that the test might show surgery is not accurate treatment and hence result in cancellation. The other approach is to order the necessary tests and schedule another outpatient consultation when the results are ready, hence postponing the decision of surgery and spending more resources at the outpatient clinic. (Surgeon) The new documents in NewArena the decision-note, anesthetic pre-operative assessment and surgeon s assessment notes are thus a part of the surgery case and are incorporated into the old standard surgery order form on the old platform, since the planning tool with current functionality is still there. At present, secretaries processing the planning for surgery have to work on the old platform, as the functionality available in NewArena is currently only supporting

7 physicians. This means that a secretary has to pick up the decision note in the surgery case, and send it to the head surgeon, who must set deadlines for treatment, in accordance with national guidelines regarding prioritization of treatment. When the head surgeon has returned this note, the secretary starts booking resources and sends appointment letters to the patient. Put simply, for the patient to turn up, it is necessary that the surgery decision-note go into an electronic workflow managed by secretaries. So far, there is no support for this electronic workflow in the new planning module. From what users have been presented at this stage, it may seem that surgery plans are made as decisions to operate are made and as surgery order forms are created. However, surgeons decision-notes are, in fact, only the triggers of surgery planning, and much work by secretaries and coordinating nurses goes into the planning process. This multidisciplinary perspective on the planning process is absent for now from the NewArena format. Though NewArena focuses on physicians work settings, data have not been structured, and hence data reuse is not yet possible. BigVendor has defined only a small number of archetypes to exemplify how these will work in NewArena. These are archetypes of clinical observations, like blood pressure, weight and height, data that surgeons register in the surgery order form. Archetypes of instructions and actions necessary to set up workflows have not been defined, and hence workflow support is missing. The further work on defining archetypes will have to be done by the hospitals themselves. The side-by side, alternating use of the old EPR and the NewArena platforms creates additional challenges: Documents produced by physicians in NewArena have to go into the old EPR to be processed further in the surgeryplanning module by secretaries, and since the old EPR does not support structured text, the data structured by archetypes in NewArena cannot be employed in the old EPR. Thus, clinical data registered by archetypes in NewArena act like free-text documents during further processing in the old EPR. This represents no improvement in the existing planning tool: If I could decide, the surgery order form would not look like it does today at all. It is too much work to set it up; you cannot import data from the electronic record into the form. As an example, the patient may use 20 different medicines, and you have to write it all into the surgery order form. That is quite time consuming, besides being an error trap. The surgery planning tool should be connected to the electronic record, so that there would be only one click set up the surgery order form and then all the structured data in the record would be loaded directly and all I would have to do was to check it. No writing. (Surgeon) At present, surgeons are not quite content with the functionality of the NewArena surgery-planning tool, as they have not yet got what they want the most: structured data that can be reused independent of documents, as described in the above quote. Still, they are happy with the new interface and the display of data. They say that the screen displayed in Figure 3 is a major improvement, as they have to open six different windows to display the same amount of information in the old system. They see the improved display as a kind of decision support in itself.

8 The secretaries and nurses who still have to work on the old platform cannot see their contribution to the planning process and are rather confused about how to ensure that patients are enrolled for surgery. Their major concern is the establishment of routines for being aware of new decision notes. Without an electronic workflow that notifies them of new notes for patients, they must find workarounds to replace such functionality. 6 Concluding discussion The intent of NewArena Clinician is to support health care workers decisionmaking at crucial steps in their clinical work. One such step occurs at outpatient clinics, where physicians decide whether a patient is going to have surgery. The key strategy supporting the design of NewArena focuses solely on this situation and the physician s role in it. Presumably, this does not only delimit the scope for the developers, but also promises to greatly enhance physicians decision-making capabilities. However, our empirical evidence challenges this assumption, pointing to the sociotechnical implications of the new software, which extend beyond the software s support for physician decision-making process at outpatient clinics. In particular, we discuss the relationship between the formal and informal as these relate to information, work responsibility and professional roles. Formalization of responsibility: In order to realize a working decision-support system, some degree of formalization appears necessary. The obvious initial step is, of course, the structuring of EPR content, which lays the basis for automatic/suggestive decision-making. However, the provision of appropriate decisionmaking support to the right professional (the surgeon) also requires the redistribution of work tasks between personnel involved in the work flow. This second, largely hidden step is an implicit formalization of the involved work processes. Just consider how the surgeon must set up the decision-note for the secretaries to be assigned their tasks, while earlier the secretaries set up and prepared the decision-note for the surgeon. Moreover, information that surgeons used to dictate, they must now write themselves and approve immediately in the form of decision-notes, before the rest of their teams may be assigned their tasks. The responsibility for decision-making hence becomes undisputed. The formalization of physicians work was expected to improve resource utilization later in the planning process. However, the information necessary to make determinations regarding surgery e.g., the results of x-rays or heart- and lung-function tests may not have been processed at the time of outpatient consultations, or such consultations may reveal the need for supplementary information, meaning that patient health information may not be complete even if information gathering is pushed to the front of the planning process. This jeopardizes the decision-note s role as the trigger of the planning process and the assignment of tasks to physicians teams. While it is relatively clear that formalization of decisions is necessary to surgery planning, a more open question is at what point in the planning process formalization should happen. Even if pushing formalization up front, to outpatient consultations, is

9 anticipated to result in better utilization of resources, this may not be a good thing for physicians experiencing added workload (particularly when the additional workload comes as a surprise), as they will have to ensure the completeness of decision-notes before the next step in the planning process can be initiated. The effect on surgeons work was worsened by the fact that data was not structured yet and hence could not be presented automatically to physicians, make suggestions or be reused. The importance of informal collaboration: In an implicit way, the formalization of roles and responsibilities appears to be a consequence of NewArena. The physician is accountable for the decision at the outpatient consultation stage. However, NewArena fails to capture the informal collaboration between physicians and secretaries in the planning process. Care processes do not rely solely on clinical information to be seamless, and secretaries and nurses make important contributions to the logistical work involved in surgical planning, although these contributions are largely hidden to the untrained eye. Solely focusing on a single user-role in the development process ignores the interdisciplinary perspective, which is perhaps one of the main characteristics of health care that makes health care work. Secretaries facilitate physicians work by preparing documents, transcribing notes, looking up and checking the status of activities, booking, scheduling and coordinating. They are aware of different physicians preferences and make preparations accordingly, hence smoothing the system and contributing to the effective use of one of the most costly resources, surgeons` time. Compatibility with installed base: This case illustrates the importance of ensuring that new ICT systems are compatible with installed bases. Stepwise implementation of the new functionality afforded by NewArena was anticipated to facilitate the surgical planning process but turned out to complicate it as well. In so far as they continue to serve as repositories for retrievable information, old EPRs must remain operable, so that users can alternate between new and old platforms. Such alternation is technically demanding, as well as challenging for users, as attention to what platform they are working on is required to accomplish their tasks. For now, secretaries are mediators, and their mediating between platforms is actually process-supporting the system. Their concern for making routines that ensure patients are enrolled for treatment is compensating for the process-support missing in the EPR system. References 1. Bird, L., Goodchild, A., Tun, Z.Z.: Experiences with a two-level modelling approach to electronic health records. J. Res. Pr. Inf. Technol. 35, (2003). 2. Kawamoto, K., Houlihan, C.A., Balas, E.A., Lobach, D.F.: Improving clinical practice using clinical decision support systems: a systematic review of trials to identify features critical to success. BMJ. 330, 765 (2005). 3. Aarts, J., Ash, J., Berg, M.: Extending the understanding of computerized physician order entry: Implications for professional collaboration, workflow and quality of care. Int. J. Med. Inf. 76, Supplement 1, S4 S13 (2007).

10 4. Ash, J.S., Gorman, P.N., Lavelle, M., Payne, T.H., Massaro, T.A., Frantz, G.L., Lyman, J.A.: A Cross-site Qualitative Study of Physician Order Entry. J. Am. Med. Inform. Assoc. 10, (2003). 5. Bonney, W.: Impacts and Risks of Adopting Clinical Decision Support Systems. In: Jao, C. (ed.) Efficient Decision Support Systems - Practice and Challenges in Biomedical Related Domain. InTeh (2011). 6. Rector, A.L.: Clinical terminology: why is it so hard? Methods Inf. Med. 38, (1999). 7. Chase, M.D.: Engineering a Learning Healthcare System: A Look at the Future: Workshop Summary. National Academies Press (2011). 8. Garde, S., Knaup, P., Hovenga, E., Heard, S.: Towards semantic interoperability for electronic health records. Methods Inf. Med. 46, (2007). 9. Klein, H.K., Myers, M.D.: A set of principles for conducting and evaluating interpretive field studies in information systems. MIS Q (1999). 10. Walsham, G.: Interpretive case studies in IS research: nature and method. Eur. J. Inf. Syst. 4, (1995). 11. Beale, T and S Heard: Beale T, Heard S. Archetype Definitions and Principles in The openehr foundation release Architecture Overview in The openehr foundation release Presented at the (2007). 12. Berg, M.: Health information management: integrating information technology in health care work. Routledge, London; New York (2004). 13. Gasser, L.: The integration of computing and routine work. ACM Trans. Inf. Syst. TOIS. 4, (1986). 14. Berg, M.: Rationalizing Medical Work: Decision-support Techniques and Medical Practices. MIT Press (1997). 15. Kraut, R.E., Fish, R.S., Root, R.W., Chalfonte, B.L.: Informal communication in organizations: Form, function, and technology. Human reactions to technology: Claremont symposium on applied social psychology. pp (1990). 16. Munkvold, G., Ellingsen, G., Koksvik, H.: Formalizing work: reallocating redundancy. Proceedings of the th anniversary conference on Computer supported cooperative work. pp (2006). 17. Hardstone, G., Hartswood, M., Procter, R., Slack, R., Voss, A., Rees, G.: Supporting informality: team working and integrated care records. Proceedings of the 2004 ACM conference on Computer supported cooperative work. pp (2004). 18. D MAG 2011, sep 2013.