HIMSS 2016 Davies Enterprise Award Submission: Reduced Preventable Inpatient Deaths

Transcription

1 Title of Case Study: Reduced preventable inpatient deaths Organization: Location: North York General Hospital 4001 Leslie St, Toronto, Ontario, Canada M2K 1E1 Primary Point of Contact: Dr. Jeremy Theal MD FRCPC Chief Medical Information Officer Secondary Point of Contact: Linna Yang, Manager, Clinical Informatics Executive Summary Since 1968, North York General Hospital (NYGH) has served the culturally diverse communities of North Toronto with the best possible care experience. One of Canada's leading community academic hospitals, we offer a wide range of acute, ambulatory and long-term care services across our three sites (inclusive of 618 inpatient beds) covering the south central region of Ontario, centered on Toronto (Figures 1 and 2). Our dedicated team of 5,000+ staff, physicians and volunteers are proud to serve a growing population of over 400,000 people. Our patients and their families are at the heart of everything we do. Figure 1 Figure 2 NYGH has a history of adopting innovative technology to promote the ideal patient experience. Beginning in 2007, we embarked on a multi-year clinical transformation project to bring our electronic health record system (EHR) into the future, from HIMSS Stage 2 to a goal of HIMSS Stage 7. This project, called ecare, has a primary focus of improving the quality and safety of care that we provide to our patients every day. One of ecare s most significant achievements to date is how NYGH established a culture of evidence-based care to measurably improve the number of lives saved. In fact, over the past 6 years, the NYGH ecare project has won seven national awards for improving quality and safety of patient care. The ecare project team felt strongly that the implementation and broad adoption of electronic evidence-based order sets would be foundational to improving quality and safety. For this reason, the team engaged NYGH physicians and other interprofessional groups of clinicians in a two-and-a-half year, carefully managed process to create a library of over 350 electronic order sets. Unique to this process was the way in which evidence was carefully integrated into all clinical content and workflows, and how 1

2 front-line clinicians were involved at every step. The result was that NYGH clinicians fully embraced the new Computerized Physician Order Entry (CPOE) system and order sets as their own. Broad and deep adoption was evidenced by a dramatic increase in order set use, from 36 percent of medical patient volume (paper-based, pre-ecare) to over 97 percent (ecare CPOE). What s more, clinician adherence to the evidence-based order sets has significantly lowered the probability of in-hospital death from pneumonia or exacerbation of chronic obstructive pulmonary disease (COPD). Local Problem In 2004, the landmark Canadian Adverse Events Study 1 found that 7.5% of acute care hospital admissions in Canada were affected by preventable adverse events, of which 20.8% were serious enough to result in death. The study went on to estimate that the lives of between 9,250 and 23,750 Canadians could be saved every year if hospital-based adverse events could be prevented. To improve on this important outcome, the Canadian Institute for Health Information (CIHI) developed a metric that determines and compares the number of preventable deaths occurring in hospitals across Canada. The data are adjusted for factors that affect in-hospital mortality, such as patient age, sex, diagnosis, length of stay, comorbidities, admission status and goals of care (active treatment vs palliative). It is mandatory for every Canadian hospital to report this data to CIHI, from which CIHI calculates a national average for preventable deaths, and in turn reports every hospital s performance compared to the national average. This measure, called the Hospital Standardized Mortality Ratio (HSMR) 2, is reported by facility, region and province and is made available to hospitals, government and the public nationwide. Mandatory measurement and reporting of HSMR is therefore a means to track, compare and improve quality and safety of acute patient care across Canada. While HSMR is not currently tied to hospital funding, recent movement in Canada toward performance-based funding indicates that HSMR could be associated with financial incentives or penalties in the future. The HSMR is indexed to a score of 100, such that a hospital performing the same as the national average will have a score of 100. Hospitals with fewer preventable deaths than the national average will have a score lower than 100, and hospitals faring worse than the national average will have a score greater than 100. Prior to implementing CPOE with ecare, the HSMR measured at NYGH was worse than the national average in several large clinical areas. In 2010, our overall HSMR was for inpatients admitted to the Medicine Program, and for patients with a primary diagnosis of pneumonia or COPD exacerbation, our HSMR scores were and 140.4, respectively (Figure 17.1, 17.2, 17.3). Our goal was to improve these scores to be better than the national average (under 100), with the assistance of the advanced health information system and evidence-based care we were implementing with ecare. NYGH launched an organizational study of our HSMR. We began by focusing on Pneumonia and COPD exacerbation because these were two of our higher-volume diagnoses for inpatient care, and both conditions had worse than average HSMR performance at our organization. For these conditions, there was also plenty of available evidence to guide treatment, and an accurate diagnosis could often be made 1 Baker GR, Norton PG et al. The Canadian Adverse Events Study: the incidence of adverse events among hospital patients in Canada. Can Med Assoc J 2004 May 25; 170(11):

3 upon patient admission to hospital, which gave clinicians the opportunity to actuate evidence-based management as early as possible in a patient s stay (e.g. via use of admission order sets). With the introduction of CPOE and evidence-based, standardized care, we hypothesized that there would be a reduction in adjusted preventable inpatient mortality (HSMR), as well as potentially 30-day readmission rates and length of stay (LOS) when compared to traditional paper processes. Additionally, we predicted that upon hospital admission, if a physician used an evidence-based order set that matched the patient s diagnosis (as opposed to a general order set with no specific evidence to manage the patient s illness) there would be an associated additional reduction in adjusted mortality, and potentially 30-day readmission and LOS. Figure 3 outlines these two hypotheses, along with details of our study population both before and after CPOE (for further information on the population studied, please see the Value Derived section of this case study). Figure 3 Our study revealed that the extent of improvement was greater than we initially predicted, enough to earn national recognition for our reduction of in-hospital deaths. The following two sections describe how we achieved this reduction with implementation of CPOE and evidence-based order sets as part of ecare. Design and Implementation Before we could design evidence-based order sets, NYGH needed the foundational capabilities enabled by an enterprise-wide EHR. Our multi-year ecare project kicked off in Implementation of Cerner Millennium EHR advanced functionality occurred in two phases: Phase 1 launched in 2008 with interprofessional clinical documentation and enterprise-wide patient scheduling; two years later, Phase 2 introduced CPOE, Closed Loop Medication Administration (CLMA), Medication Reconciliation, Prescription Writer and electronic discharge process across all Medical, Surgical and Critical Care units, as well as for admitted patients in the Emergency Department. 3

4 ecare would not have been possible without a robust change management methodology, dedicated project management, transparent organizational communication and mandatory training. During Phase II, all staff (209 physicians, 730 nurses, 60 allied Health and Pharmacy) completed mandatory training. Our Simulation Room allowed clinicians to work through potential care scenarios, and a team of Super Users was trained to provide supplementary support (one per floor during every shift in 10-day increments). Together with IT staff, clinical informatics and NYGH leadership, they provided 24x7 onsite support for 6 weeks during go-live. Thanks to the combined efforts of this community of stakeholders, both implementation phases were implemented on time and within our budget. Additionally, in 2011 NYGH became the first community teaching hospital in Canada (and one of only 3 hospitals in Canada overall at that time) to achieve Stage 6 on the HIMSS Analytics Electronic Medical Record Adoption Model (EMRAM). We realized that implementing CPOE alone was not enough to significantly improve patient care. Accordingly, we built a layer of electronic clinical decision support into ecare, to integrate evidence and best practice into clinical workflow. As part of the ecare project, NYGH formed a centralized, interdisciplinary build team that focused on the goal of building static and dynamic decision support by clinicians, for clinicians. The team included physician, pharmacy and clinical informatics subject matter experts. The team prototyped order sets and decision support, and then directly involved front-line clinicians in content review, with a goal to build and approve a custom library of 350 order sets in two-and-a-half years. The initial stakeholder group included more than 80 physicians and 150 clinical staff. By nature of directly involving these clinicians in the development process, the result was better quality of clinical content, redesigned clinical workflows that make it easy to do the right thing, and full ecare buy-in from end users (Figure 4). Figure 4 4

5 The team integrated evidence into system content such as order sets (Figure 5) to guide clinical behavior in accordance with the Ontario government s Excellent Care for All Act 3, which emphasizes the importance of using evidence-based medicine in daily practice to improve the quality and safety of patient care. The build team reviewed each order set with clinical stakeholders and encouraged them to share their feedback on the content, logic and process embedded within each order set (Figure 6). Figure 5 Figure 6 3 About the Excellent Care for All Act. Ontario Ministry of Health and Long-Term Care. Queen s Printer for Ontario, 29 Dec Web. 23 Feb

6 In two-and-a-half years, NYGH created and approved a custom library of more than 350 electronic order sets; today, it has expanded to over 800. Since creation, each order set is revisited on an annual or biannual basis by an interprofessional team of clinicians to verify order utilization, review formulary changes, address user requests and policy updates, and incorporate the latest evidence from the medical literature. Together with the central build team, clinicians are directly involved in the steps of our master order set development methodology in order to review, finalize and approve each order set (Figure 7). Figure 7 How Health IT Was Utilized The NYGH ecare team knew from countless examples in medical literature that the provision of standardized, evidence-based care can significantly improve quality and safety outcomes for patients. However, the team also knew that changing clinical culture and workflows to regularly leverage evidence at the point of care is challenging. Physicians often feel that they already know what they need to know to provide the best care. On this basis, the pull model, which depends on physicians to recognize a knowledge deficit, stop during regular clinical workflow and look up information, is rarely effective (Figure 8). The failure of the pull model is one of the reasons why it is traditionally said that it takes 17 years from the time new evidence is first published in medical literature, to the time that it is regularly practiced at the bedside. Clearly, our patients deserved better. By reviewing research on how physicians utilize evidence in the course of care 4, we discovered that electronic systems can be used to provide clinical decision support in a push model, which is proven to 4 Kawamoto K, Houlihan CA, Balas EA, Lobach DF. Improving clinical practice using clinical decision support systems: a systematic review of trials to identify features critical to success. BMJ Apr 2; 330(7494):

7 be much more effective at ensuring that physicians use evidence in their daily work. In fact, if decision support is provided by a computerized system which presents clear recommendations at the time and location of decision-making, as part of clinician workflow, it is 112 times more likely that a physician will use the information in the course of patient care (Figure 8). Figure 8 The ecare order set development and utilization workflow was purpose-built for clinicians from the ground up, to leverage the push model for evidence utilization and make it easy to do the right thing. The steps of this workflow are as follows (Figure 9): 1. Regular development and review of electronic order sets in partnership with front-line clinicians, with integration of the latest evidence-based information (see Figures 5, 6, and 7 for additional detail on this process). This step is crucial to ensuring clinicians trust and embrace order sets as their own, use them regularly as part of organizational culture, and have access to the most upto-date information every time content is used. Also, because order sets are continuously reviewed and improved, regulatory compliance is ensured, since it is built into standard process. 2. Order set content is imported for use in our hospital information system (Cerner). We build all order sets during step 1 using a dictionary of orderables identical to our Cerner order catalogue. This ensures that clinical meaning of all content is interpreted correctly during development, review and approval of order sets by clinicians, and minimizes work during import into Cerner. 3. Physicians access Evidence-Based Order Sets at time of CPOE order entry (such as when admitting a patient to hospital). This process is made fast and easy for physicians using intuitive naming and synonyms for all order set content, verified during clinician review/usability testing. As electronic documentation of coded diagnoses is not yet in broad use by physicians at our hospital, we cannot rely on system-automated suggestion of order sets based on patient diagnosis. However, we feel we have achieved something better: culture change. By engaging physicians in order set design and ingraining use of evidence into organizational culture, we have increased order set utilization on admission from approximately 36 percent of patient volume (when using paper) to over 97 percent of patient volume (using CPOE with ecare). 4. As physicians complete order sets, evidence-based information is integrated into their decisionmaking workflow. Utilizing key success principles defined by Kawamoto and others (Figure 8), the design of our order sets integrates relevant evidence-based information into the physician decision making process. At each decision point in an order set, a succinct summary of relevant 7

8 evidence-based information is presented right beside the associated orders. Examples of content from the pneumonia and COPD exacerbation order set are depicted in Figure 10 and More detailed evidence-based information is available to physicians within one click, during the ordering process. We have associated clickable evidence links with the orders in our order sets (Figure 5, Figure 10 and Figure 11). Clicking on an evidence link in an order set presents the physician with additional information, such as a summary of the medical literature that corroborates the evidence statements being provided, as well as clinical guidelines, scoring systems, and so on. All of the linked evidence-based content is regularly updated in a continuous improvement process, as part of order set review as outlined in Step 1. Figure 9 ecare order set development and utilization workflow 8

9 HIMSS 2016 Davies Enterprise Award Submission: Reduced Preventable Inpatient Deaths Figure 10 example of evidence-based content to aid empiric antibiotic selection in the Community-Acquired Pneumonia order set Figure 11 example of evidence-based content to guide medication selection in the COPD exacerbation order set 9

10 Value Derived ecare was focused on the use of advanced health information technology to enable clinicians to enhance patient care by building best practice and evidence into clinical workflow. Our hypothesis was that implementing this system, and engaging clinicians to be part of it, would result in improved patient outcomes. As outlined at the beginning of this case study, we decided to determine the effects of implementing ecare on quality and safety of patient care by evaluating in-hospital mortality, LOS and 30-day readmission rates for all patients with a primary discharge diagnosis of pneumonia or COPD exacerbation. We conducted a before-and-after retrospective chart analysis that evaluated in-hospital mortality, LOS and 30-day readmission rates for all patients with a primary discharge diagnosis of pneumonia or COPD exacerbation. We selected all patients discharged from our hospital with these diagnoses in the pre-implementation time period of January to September 2010 (520 patients), and compared it to a seasonally-matched post go-live time period of January to September 2011 (511 patients). These time windows were selected because phase 2 of the ecare system went live in late October 2010, and it was appropriate to allow two months of stabilization before measuring postimplementation outcomes. The sample groups were statistically similar in sample size, age, gender distribution, LOS, comorbidities, and frequency of critical care unit admission (Figure 12). There was a difference between the two groups in the distribution of patients that had a diagnosis of pneumonia or COPD, but this was corrected using a standardized formula called Probability of Death, which is used nationally in inpatient mortality studies by CIHI. The formula corrects for differences in patient age, gender, LOS, comorbidities (using Charlson Weight), admission type, transfer type (if applicable), and diagnosis group 5. Since Critical Care unit admission is not included in the Probability of Death formula, all results were adjusted to correct for Critical Care admission, in addition to Probability of Death. Figure 12 5 HSMR Technical Notes. Canadian Institute for Health Information. Feb

11 Statistical analysis was conducted using Stata 12 software. Baseline population characteristics were calculated using Wilcoxon rank-sum test for continuous variables, and Chi-squared test for other variables. Odds of death and readmission were calculated using logistic regression. Our analysis revealed that the adjusted odds of dying while admitted to NYGH due to pneumonia or COPD exacerbation were 45.3 percent lower when a patient was treated using ecare, versus the previous paper-based processes (significant to p=0.005, Figure 13). Furthermore, there was a 56.3 percent reduction in the adjusted odds of in-hospital death when the physician entered an electronic order set at patient admission that contained evidence-based guidance and treatment that matched the patient s diagnosis (significant to p=0.024, Figure 14). No such mortality benefit was seen if the physician selected an order set that was only close to the diagnosis (such as selecting Congestive Heart Failure instead of Pneumonia [both conditions cause dyspnea/hypoxia]), or a general order set with no specific evidence (such as General Medical Care ). There was no significant difference found in the rates of readmission at 30 days, or length of stay between the two populations (Figures 13, 14, 15). Of note, the focus of our quality intervention was to reduce preventable inpatient mortality; as such, no specific strategy was applied to improve 30-day readmissions or length of stay. In addition, confounding factors may have prevented improvement in these metrics during the measurement period. As depicted in Figure 15, 30-day readmissions may actually have increased due to factors such as: a competing corporate program to expedite early discharge in 2011 (arrow 1), a change in care model where generalists became attending physicians rather than specialists in 2012 (arrow 2), another corporate program focusing on reducing care costs in 2014 (arrow 3), and implementation of a government-mandated Quality-Based Procedure (QBP) program in , which did not include evidence-based order sets for discharge. Figure 13 11

12 Figure 14 Figure day readmissions to the Medicine program for a related diagnosis, before and after ecare 12

13 Further analysis of the data demonstrated why implementation of advanced health information technology improved inpatient mortality: Clinician adoption of the ecare system was broad and deep. From go-live, 100 percent adoption was achieved. In addition, over 93 percent of physician orders were directly entered by physicians using CPOE (the remainder were medical directive, verbal, and telephone orders). Use of the ecare system, compared to paper-based processes, resulted in decreased inpatient mortality (Figure 13). Furthermore, the importance of using the correct evidence at the point of patient care was demonstrated by the fact that inpatient mortality was reduced when the physician selected an electronic admission order set that matched the patient s primary diagnosis, whereas no effect on mortality was seen if a non-matching or general order set was used (Figure 14). Commensurate to the reductions in inpatient mortality seen in our study, our hospital s HSMR score dropped significantly not only for the diagnoses of COPD exacerbation and pneumonia, but also overall for inpatients of the Medicine program (Figure 17.1 to 17.3, Figure 18). We were recognized in the press for this significant achievement, when CIHI reported in 2012 that NYGH had achieved the lowest HSMR scores in the Greater Toronto Area, and the second best in all of Canada a significant improvement compared to our prior performance, and something to be proud of nationally (Figure 16). Ongoing use of the ecare system since 2010 has demonstrated sustained benefits in reducing preventable inpatient mortality. While NYGH s HSMR score for all Medicine inpatients prior to ecare in 2010 was 112 (worse than the national average), our score in all subsequent years has been dramatically better, and among the best in the nation (between 74.8 and 81, years 2011 to 2015, Figure 17.1 and 17.3). Similar sustained benefits were seen for patients with the specific diagnoses of pneumonia and COPD exacerbation (Figure 17.1 and ). Trended data measured twice yearly from 2008 to 2015 also illustrate the significant and sustained improvement in mortality for Medicine inpatients, as well as specific patients with pneumonia or COPD exacerbation (Figure 18). Review of the trended data shows a temporary spike in COPD HSMR in 2013 and early 2014, which we hypothesize was related to a hospital policy transition to COPD patients being managed primarily by generalists rather than specialist respirologists. Subsequent incorporation of government-mandated Quality Based Procedure requirements into electronic order sets in further reinforced use of evidence in the treatment of specific conditions such as COPD exacerbation and pneumonia. This update appears to have resulted in further small improvements in mortality (Figure 18). Lastly, trended data show a recurring pattern of seasonal increase in COPD-related mortality in winter months, which we hypothesize is related to increased frequency of upper respiratory tract infections in winter, known to be associated with increased severity of COPD exacerbation (Figure 18). Organizational culture underwent a strong positive transition as a result of ecare. Prior to ecare, physicians at NYGH were quite individualistic and mostly wrote free-hand paper orders. They didn t openly discuss evidence-based practice with their peers, and only a few dozen paper order sets were in use. After ecare, physicians embraced evidence-based care. Order set utilization on medical hospital admissions increased from 36.5 percent (pre-ecare) to more than 97 percent with ecare (even though utilization was not mandatory). Asking physicians to review electronic order sets was initially quite a challenge during ecare s initial design phase, requiring many reminders and repeat requests. Fast-forward to today at NYGH: physicians readily and regularly participate in review of clinical content for order sets and decision support. In the past 6 Results based on 2015 HSMR Methodology. Data Source: NYGH Business Intelligence Data Warehouse. 13

14 12 months, in partnership with clinicians, over half of the order sets in the NYGH library (379 of 700 order sets) have been reviewed, updated with new evidence and/or created new. Quality monitoring and reporting has become an ingrained and automated part of NYGH s daily business. HSMR is a Key Performance Indicator (KPI) at NYGH, and is reported (among other KPIs) automatically through the NYGH Business Intelligence system, which receives live data from ecare (Figure 19). HSMR performance and other KPI s are reviewed regularly by our Hospital Quality of Care Committee, and are reported monthly to the Senior Leadership Team, Program Directors, the Medical Affairs Committee and Quality Committee of the Board (Figure 20). Interventions and change management intended to iteratively improve KPI s such as HSMR are devised within the committees of the Quality Governance Structure, enacted through the corporate Quality Improvement Office (QIO) using the DMAIC structure (Figure 20), and shared with front-line staff via interprofessional education programs, nurse educators, weekly unitbased quality huddles, and electronic Quality Boards on each nursing unit. We have calculated that NYGH clinicians saved 31 lives from pneumonia and COPD exacerbation during the timeframe of our HSMR study. Given that NYGH has been able to maintain exemplary HSMR scores for these diagnoses over the subsequent 5 years (Figure 17.1 and 17.2, Figure 18), we are confident that we have saved at least another 120 more lives just from these two disease groups alone, not to mention many more lives from other common conditions treated at NYGH every day (Figure 17.3, Figure 18). Figure 16 14

15 Figure 17.1 Figure 17.2 NYGH Medicine Program Overall Calendar Probability Actual Year of Death Death HSMR Figure

16 Figure 18 trended HSMR from 2008 to 2015 for Medicine program overall, Pneumonia, and COPD exacerbation 16

17 Figure 19 Key Performance Indicator (KPI) reporting from NYGH Business Intelligence System Figure 20 Quality governance, Quality Improvement Office core functions, and DMAIC structure for implementing quality improvement projects at NYGH 17

18 ecare s widely adopted standardized clinical content has enabled NYGH to be one of the first organizations in the province to fully comply with newly introduced government requirements for evidence-informed care of specific conditions (Quality Based Procedures 7 ). Even more significant, ecare enabled NYGH to become the provincial lead site for implementing recommendations of Choosing Wisely Canada, resulting in a 33 percent reduction in un-necessary pre-operative clinic and emergency department laboratory testing, among other benefits 8. All of the above successes have resulted in NYGH achieving recurrent national recognition, in the form of seven national awards for excellence in improving quality and safety of patient care with health information technology. Compelled to share our best practices to assist others, we approached Canada Health Infoway with an idea to create a national Toolkit to share content and lessons learned, at no cost, with other hospitals. We won a competitive bid to receive funding for this project, which is called the Canadian CPOE Toolkit. More information about this Toolkit is detailed in the Lessons Learned section. Lessons Learned The Way We ve Always Done It is not a valid argument: Prior to ecare, our HSMR score was, in several large clinical areas, worse than the national average. As we embarked on the implementation of advanced health information technology with ecare, at first many clinicians felt that we should simply automate the status quo in other words, simply make the existing clinical processes electronic. However, as we reviewed current state processes, policies and workflows, we quickly realized that some of the existing clinical processes at NYGH were out of date, inefficient and/or not representative of best practice. This may have been one of the reasons why our HSMR score was previously worse than the national average. We worked with clinicians to help them understand that The Way We ve Always Done It is not a valid argument to maintain the status quo. In partnership with clinicians, we redesigned hospital processes, policies and workflows to align with best practice and take advantage of the new capabilities of the ecare system. While this process did require more work and time, we feel it was a key success factor in ensuring that ecare wasn t just a technology project, but a clinical transformation project. Regular use of workflow-integrated evidence is important to drive improved outcomes: We learned that CPOE alone is not enough to fully optimize patient outcomes. We found it necessary to include clinical decision support in our ecare implementation as a catalyst to transform clinical practice. We integrated decision support into our electronic order sets by embedding the latest evidence into the physician ordering workflow. This Push Model greatly increased the use of best practices at the bedside. This improvement was evidenced in this case study, where we found a 56.3 percent reduction in the odds of in-hospital death when a physician selected and used electronic an admission order set that contained evidence-based guidance and treatment specifically matching the patient s diagnosis. No such mortality reduction benefit was seen when a physician used an order set that did not contain evidence specific to the patient s diagnosis (see Value Derived section above). 7 Clinical Handbooks for Quality-Based Procedures. Health Quality Ontario. Queen's Printer for Ontario, Web. 24 May Picard, André. When it comes to tests, more is not always better. The Globe and Mail. The Globe and Mail Inc., 05 Apr Web. 24 May

19 Development and maintenance of clinical content takes more time and effort than you think: It took two-and-a-half years to create and approve a custom library of over 350 evidence-based electronic order sets, and a few additional years to expand it to over 800. This process was time-consuming because it wasn t about quickly implementing electronic orders; it was about transforming clinical practice. Our master order set development methodology (Figure 7) involves engagement of all clinical professions that care for patients with a particular clinical condition. The clinicians work with the central build team to comprehensively review clinical content, evidence, workflows, policies and procedures that are affected by the order set. It is not uncommon to identify changes that need to occur in order to optimize not only the order set, but also the clinical processes that are connected to the orders within an order set. This process improves quality and safety in many facets of the organization, beyond the electronic system. Also, comprehensive cross-departmental engagement in this process encourages clinicians to take ownership of the ecare system as their own not only adopting its regular use, but becoming stewards of constant system improvement. This is important because the work does not end once Version 1 order sets are approved each is revisited on an annual or bi-annual basis by an interprofessional team of clinicians, to revise and optimize content in accordance with corporate policy. In fact, over the past 12 months, 379 of the more than 800 order sets in the NYGH library were revised or created new, demonstrating how actively our organization works to keep content aligned to the latest evidence and best practice (Figure 21). Figure 21 Order Set Clinical Content Review and Maintenance Activity Achieving measurably improved outcomes requires culture change: Before ecare, our physician culture was quite individualistic. Physicians preferred to write their own freehand admission orders, even when paper order sets were available. Data indicated that only 36.5 percent of medical inpatients were admitted using standardized paper order sets. Also, it was rare for physicians to discuss with their colleagues the evidence that might merit one investigation or treatment over another. During the 2.5 year ecare design process for evidence-based order sets, physicians were regularly educated about the benefits of standardization and use of evidence at the point of care. As they reviewed order sets, they 19

20 regularly engaged with their colleagues in discussing what the evidence says is the best management for each diagnosis. While at first the ecare team needed to constantly remind and encourage physicians to participate in order set development and review, over time our culture fundamentally changed. Now, physicians regularly use order sets, even though this is not mandatory. Order set usage on hospital admission has increased from 36.5 percent of medical inpatients (on paper) to over 97 percent of medical inpatients (using CPOE with ecare). Physicians have not only fully embraced the use of these standardized order sets, but have also become stewards of the content within them. Physicians now regularly reach out to the ecare team to make adjustments and improvements in order set content. Also, physicians now regularly review order sets when they are sent out for comment, without the need for constant reminders. Clearly, culture change was an essential driver in leveraging health IT to ensure the regular use of evidence-based care at NYGH. There was a clear cause and effect relationship between system adoption and improved outcomes. This included the clear mortality benefit shown in this case study, where adjusted odds of dying while admitted to NYGH from pneumonia or COPD exacerbation were 45.3 percent lower when a patient was treated using ecare, versus the previous paper-based processes. Furthermore, there was a 56.3 percent reduction in the adjusted odds of inpatient death when the physician entered an electronic order set at patient admission that contained embedded evidencebased guidance and treatment that matched the patient s diagnosis. Quality monitoring and reporting should become an ingrained and automated part of daily business: HSMR is a Key Performance Indicator (KPI) at NYGH, and is reported (among other KPIs) automatically through the NYGH Business Intelligence system, which receives live data from ecare (Figure 19). HSMR performance and other KPI s are reviewed regularly by our Hospital Quality of Care Committee, and are reported monthly to the Senior Leadership Team, Program Directors, the Medical Affairs Committee and Quality Committee of the Board (Figure 20). Interventions and change management intended to iteratively improve KPI s such as HSMR are devised within the committees of the Quality Governance Structure, enacted through the corporate Quality Improvement Office (QIO) using the DMAIC structure (Figure 20), and shared collaboratively with front-line staff via interprofessional education programs, nurse educators, weekly unit-based quality huddles, and electronic Quality Boards on each nursing unit. Quality programs improve the metrics they target, but don t necessarily improve other non-targeted metrics, even if they are conceptually related: The focus of our quality improvement intervention was to reduce preventable inpatient mortality. While this effort was successful, we had expected that other conceptually related metrics would also improve namely, 30-day readmission rates and length of stay. However, these other metrics did not improve. We hypothesize that the lack of improvement was because we did not specifically develop or enact a strategy to reduce readmissions and length of stay, and competing/confounding factors may have prevented these metrics from improving during the measurement period. Document and communicate best practices for success: The following key success factors were crucial to implementing ecare on time, within budget and with improved clinical outcomes. We continue to build upon these best practices and share them internally as well as externally via the Canadian CPOE Toolkit: Define your organizational vision, assess cultural readiness and consistently demonstrate your commitment to the project (from the leadership team, to the project team, to front-end users) 20

21 Directly engage respected, front-line clinicians to serve as champions for peer-driven change (this enables clinicians to take on true ownership of their EHR) Build an effective communication plan that clearly explains the why behind the change and recognizes the unique adoption characteristics of each clinician group Integrate clinical expertise by directly engaging clinicians in clinical content development for the EHR, including integration of evidence and optimizing workflows Utilize robust and reliable technical infrastructure that captures and efficiently shares real-time patient data across venues to support clinicians at the point of care Practice careful data governance to ensure clinical data recorded in the EHR are discrete, accurate and standardized Leverage the data in the EHR using business intelligence systems to automatically track outcomes, facilitate learning and enable iterative improvement on KPIs (Figure 19) Empowering the Canadian healthcare community: The Canadian CPOE Toolkit NYGH has come a long way since the launch of ecare in We have achieved exemplary outcomes thanks to a community of stakeholders who are embracing new IT-enabled processes to deliver care to our patients. Even with considerable changes to healthcare funding and care delivery in our province over the past several years, our efforts remain centered around one mission: Our Patients Come First in Everything We Do. We recognized that while CPOE systems can deliver substantial clinical benefits, success is certainly not guaranteed. CPOE implementation is difficult, expensive, and can result in failure to achieve clinician adoption and positive patient outcomes. In Canada, we have a non-competitive, publically-administered and funded healthcare system. At NYGH, we felt compelled to share our best practices and lessons learned, to assist other organizations in the implementation of CPOE. Through sharing, we hoped to reduce the cost and time required for CPOE implementations across the country. Equally important, we wanted to ensure that clinician engagement and clinical practice transformation became central goals of each CPOE project, to produce high rates of system adoption and positive patient outcomes. We approached Canada Health Infoway with an idea to create a national toolkit to share content and lessons learned, at no cost, with other hospitals. We successfully won a competitive bid to receive funding for this project, which is called the Canadian CPOE Toolkit (Figure 22). The Toolkit is a robust online resource that includes a 400+ page implementation guide and a searchable library of evidencebased order sets. It is based on a multi-publisher sharing model that allows contributing organizations to share their content at no cost (while retaining full ownership of their contributions). 21

22 Figure 22 CPOE Toolkit home page and Implementation Guide cover The Canadian CPOE Toolkit continues to expand as NYGH refines its own order sets, and as contributing organizations share their findings. Today, there are 57 member organizations, six contributing organizations and 473 active users from CPOE implementation teams across seven Canadian provinces. Each organization has access to more than 1,500 evidence-based electronic order sets that they can localize to their own clinical workflows in partnership with their clinicians, and then build into their own CPOE systems. The Toolkit is accessible via this web address: Financial Considerations The total cost of the multi-year ecare investment since 2010 is $36.9 million CDN; it consists of $13.9 million in capital and $4.6 million in annual operational costs. A more thorough breakdown of our total cost of ownership for ecare, inclusive of planned costs, is detailed in Figure 23. Initial Capital costs - ecare project costs: $12.8 million - Hardware/Infrastructure: $1.1 million - Total: $13.9 million Ongoing operational costs (per year) - Software/Hardware: $1.9 million - IT and CI Staff: $2.7 million - 5-Year Total: $23 million Total ecare investment from : $36.9 million Figure 23 As depicted in this case study (see Value Derived, above) we were able to realize a 45.3 percent reduction in odds of in-hospital death for patients with COPD exacerbation and pneumonia utilizing ecare vs. paper orders, and a 56.3 percent reduction in odds of in-hospital death when patients were admitted using CPOE with a correctly matched, evidence-based order set. These results quantify the 22

23 safety benefit of using an electronic system, and the quality benefit of integrating evidence into daily patient care. Specific to financial benefit, to date NYGH has not conducted case-specific costing for this HSMR study. However, we do know that the inpatients in this study benefitted from reduced risk of nosocomial adverse events as a result of implementing ecare, similar to all other NYGH patients. These adverse events can not only lead to increased morbidity and mortality, but also increased cost. The costs of common nosocomial adverse events are detailed in a 2012 study by the Canadian Patient Safety Institute (CPSI) 9, as follows: Cost per inpatient medication error: $402 to $632 CDN ($517 CDN median) Cost per nosocomial adverse drug event: $4,028 CDN Cost per case of deep-vein thrombosis (DVT): $24,411 CDN Cost per case of pulmonary embolism (PE): $26,426 CDN Cost per case of combined DVT and PE: $36,074 CDN Cost per case of nosocomial c.difficile: $10,809 CDN Since the launch of ecare, NYGH identified significant cost savings from prevention of all of the above adverse nosocomial events. Cost savings have been calculated and detailed in the other three submitted NYGH case studies, and are summarized in the chart below (Figure 24). One additional cost saving included in this chart that was not part of the submitted case studies was the prevention of recurrence of clostridium difficile infection. In partnership with the Infectious Disease and Antimicrobial Stewardship departments at NYGH, the ecare team developed a custom real-time decision support alert that activates if both of the following are true: a) the patient has had a positive c.difficile stool toxin result at NYGH in the past 180 days; b) a physician is attempting to prescribe an antibiotic that is known to increase risk of recurrence of c.difficile. Details of the saving calculation are listed below (Figure 24). Adverse Event Prevented Medication Error due to Drug Administered to Incorrect Patient (see separate case study on Closed Loop Medication Administration) Adverse Drug Event from Medication Discrepancy on Medical Admission to Hospital (see separate case study on Medication Reconciliation) Venous thromboembolism (VTE) ecare Cost Savings ($CDN) Nov 2010 to Dec 2015 $5,730,428 $31,062,140 $1,029,169 (see separate case study on VTE prophlylaxis) Recurrence of clostridium difficile infection * $293,376 TOTAL COST AVERTED $38,115,113 * C.difficile recurrence prevention calculation (alert first put into production on March 27, 2012): - Prevalence of c.difficile at NYGH is 4.6 cases per 1,000 discharges - There have been 59,004 medical and surgical discharges since the alert was introduced - The risk of recurrent c.difficile after repeat antibiotic prescription is 10 percent - The cost of nosocomial c.difficile is $10,809 per case (recurrent cases likely even higher, since they are more difficult to treat) Total savings = 59,004 * * 0.10 * 10,809 = $293,376 Figure 24 9 Etchells, E et al. Economics of Patient Safety in Acute Care. CPSI, July Print. 23

24 The overall return on investment for NYGH s ecare project from 2010 to 2015 is calculated as follows: $38.1 million in savings minus $36.9 million in capital and operating costs = $1.2 million in net savings. This calculation takes into account measurements for only four types of averted adverse events. There are many other types of savings that are not quantified here. Not included in this NYGH ROI calculation are other system benefits outside the scope of our case studies, such as efficiency savings from decreased medication turnaround time (reduced by 83% since inception of ecare), improvements in appropriate antimicrobial use, savings from reductions in unnecessary testing 10, and most importantly, the immeasurable benefit of saving lives. This case study on reduction of preventable inpatient deaths shows that we saved 31 lives from just the specific diagnoses of pneumonia and COPD exacerbation in less than one year after ecare was first introduced. Having sustained our improved HSMR score since 2011 (refer to Figures 17.1, 17.2, 17.3 and 18), we are confident that we have saved at least another 120 lives just from these two disease groups alone, not to mention many more lives from other common conditions treated at NYGH every day. In summary, while we are pleased to see an overall cost saving from our investment in health IT, most important to our patients, families, clinicians and staff is the knowledge that more people now go home from the hospital with their loved ones every year having survived a serious illness thanks to safe, high quality care provided at our organization with the help of ecare. Moving forward, North York General Hospital will continue to use ecare to design and iteratively improve IT-enabled, patient centered solutions that achieve the best clinical and financial outcomes for our patients, families, community and health care system. We will also continue to share our lessons learned with other organizations, so that they can achieve similar benefits