What is the sound of one hand clapping? Can analytics drive better business outcomes when used in isolation, and what is the value of insight that is not acted upon? These questions served as notional underpinning to the key challenge posed at Information Builders’ second annual Healthcare Symposium held in Ontario’s Hockley Valley last month – exploring techniques that can help create the data-driven culture and processes needed to transform healthcare delivery. This agenda offered stark contrast to last year’s event, which featured a recounting of multiple analytics projects developed by healthcare institutions across southern Ontario: the point of this year’s symposium was to extend the utility of technology implementations by outlining methods to encourage broader adoption and use, a complex proposition involving change at the management, operational and technology levels.
Well attended by healthcare reps from Ontario and beyond, IB’s Hockley Valley Symposium launched with a presentation by Dr. A. Thomas McGill, infectious disease practitioner with Pennsylvania-based Butler Health System. Focused on behavioural change within his institution, Dr. McGill pointed to some of the issues facing modernizers looking to enable transformation through analytics: while volumes of raw data produced by healthcare systems speak to the increasing complexity of problems to solve, clinicians, who are typically non-mathematically inclined, often have a hard time articulating what information they need. Citing Accenture, McGill noted that approximately 22 percent of employees within an organization typically use analytics; at Butler, the corresponding figure was a less impressive 2-3 percent.
Analytics projects in action
Subject to the “The Analytics Imperative” due to a funding model that can penalize Butler up to 20 percent of its Medicaid incentive for underperformance on specific healthcare KPIs, McGill’s team evolved a number of projects to support performance improvements based on use of the Information Builders (IB) WebFocus InfoAssist business intelligence solution and the InfoAssist reporting to tool. These include a simple report for Butler Nursing Homes that allows them to better manage doctor scheduling; a Realtime Antibiogram Report Generator that replaced sporadic reporting over the year with real time, location-based reports; the Emergency Department LOS Project analytics cube which provided visibility into process change to enable rapid cycle improvements, resulting in improved ratios of patients per bed; and an E Checkbook for monitoring work attendance and the number of patients managed per shift that was internalized to encourage staff self-improvement. According to McGill, the E Checkbook offers the best example of staff “thinking their way to a new kind of thinking” on job performance.
McGill’s list of analytics achievements is longer still – an Infection Surveillance App, a C. Difficile room location graphic, a predictive model aimed at freeing up beds through Local Influenza testing, and a Post Acute Care graphic that tracks and supports rapid post discharge visits with outside practitioners to reduce readmission rates. But as impressive as the creative uses analytics were put to in these BI examples, the source of Butler’s success in operationalizing them is even more interesting. For McGill, this lies in collaboration between clinicians, healthcare practitioners and BI specialists – not only in high level discussions of process improvements and outcomes for specific healthcare issues, but also in report/app creation itself. In Butler’s case, collaboration involved having clinicians pre-identify information needs, and actually sit with the developer in front of the screen through several iterations so that clinician and developer could learn each other’s language and logic. The result was 2-3 iterations of an app as opposed to the 6-7 that were typical when staff worked in isolation. Another tactic employed by Butler was analytics-enabled information feedback. As McGill explained, though doctors were initially reluctant to enter data on admission performance (even though this might impact their funding), by providing non-judgemental email feedback on individual performance, comparing this with average performance and institutional goals, Butler has “inculcated this system of value” among many practitioners, helping them to recognize the importance of performance measurement. Summarizing Butler’s experience with BI, McGill noted that ready access to information has enabled new ways of acting which in turn have changed how staff think, has supported deeper collaboration across departments in the organization, a very rapid cycle (for health care) of improvement, and more curiousity and a new degrees of involvement from staff at nearly all levels – in addition to improvement on cost containment: labor and benefits account for less than 47 percent of revenue at Butler due to WebFocus, McGill argued.
Fusing top down and bottoms up analytics innovation
In her presentation, Laure Pitfield, project manager BI, decision support at Queensway Carlton Hospital, described a similar focus on collaboration as the institution adopted “Lean Healthcare” via a Continuous Performance Improvement (CPI) program powered by IB’s business intelligence. As Pitfield explained, within the hospital are discrete “staff specializations – people who understand the process, and others who understand the data – but collaboration across both organizational performance and decision support departments is critical to addressing gaps in understanding.” Pitfield’s own area of specialization is organizational psychology, and her research topic is change management: at Queensway Carlton, her goal was to stimulate cultural transformation, retraining the institutional hierarchy so that managers better understand how to remove barriers to performance improvement in hospital operations.
In terms of method, this involved development of a problem solving culture – PDSA (Plan, Do, Study, Act), which entails finding the root cause of a problem, building an improvement team that will test various solutions, using data to evaluate results of various test cases, and deciding on action based on this assessment. According to Pitfield, before implementation of BI, project leaders would call decision support for information and then wait weeks for a recommendation; post CPI/BI deployment, leaders ask for record-level data and conduct their own analysis. Today, staff members identify a problem and receive coaching on how to itemize the data necessary to solve the problem; CPI analysts use a WebFocus Reporting Object created by decision support to collect insight, and then use InfoAssist to decide on appropriate action.
In Pitfield’s view, BI tools have made collaboration between CPI and Decision Support a lot easier, and business outcomes more certain. The hospital’s staged roll out CPI/BI has now reached 9 of 35 units, which rely on CPI in their daily work. As example, she described the ‘monthly scorecard’ which tracks KPIs for individual units against targets. Since each of these had 15-20 indicators and multiple sources, Queensway decided to automate the scorecard, providing results through a portal which serves as the main access point for BI. By offering staff a quick snapshot of their KPIs and the ability to drill down on specific metrics, and by disseminating information via clear paper graphs of the report, Pitfield’s team has encouraged a sense of responsibility for performance improvement that extends to everyone in the business unit – not just the manager. For example, physical ‘Improvement Boards’ located within the medical units help individuals see the relationships between a nurse’s performance and how well the department is tracking as a unit on initiatives such as the LINEN project which managed to save the hospital a thousand dollars a week. Pitfield’s team began work at the frontlines, with “some pretty strange projects” to go move beyond the stereotype that change is top down. Now they are working on ways to use True North metrics – the top ones identified by the hospital administration – based on the notion that “if you do these well, the others fall into place” as improvement projects fall from each of the KPIs that are tracked. The goal is find a positive response to the question “is organizational hierarchy and structure an enabler or a barrier to CPI?”
The human psyche component
Jim Hornell, CEO Brant Community Healthcare System, tackled the concept of change head on, in an engaging presentation featuring film, memory exercises and finger fights to highlight the importance of new ways to look at old issues. While the need for transformational change in healthcare – an aging population, new funding models that reward efficiency, user demand for more innovation and the inability to take due advantage of automation and ehealth systems, and the fact that medicine is now “complex, effective and potentially dangerous” – was likely familiar to audience members immersed in the field, how to effect adoption of new approaches was less so. In his presentation, Hornell focused on the how, with valuable advice on using human psychology and organizational behaviours to advantage in change management. For example, the animated video A Bug’s Life served to remind the audience that the status quo is powerful, change is scary, managers are typically too busy to think differently, next gen workers are more open to innovation and frustrated with lack of change (and may be lost to competitors), and that people in power are challenged by innovation. From the Volkswagen ‘Beetle’, Howell was able to draw advice on how to overcome these problems. People either make, help, watch or OMDB (over my dead body) change, and you need to send the right people to speak to the resisters: sending a “make” to talk with the “OMDB cohort” is setting up the change agent for failure, he argued. Hornell also shared some “Universal Laws of Teamwork” – he started with Plurality (the number of people on a team is usually greater than one), on Decadactyly (this number is often greater than the number of toes), and on Inevitability (you will step on people’s toes as you suggest change). He next offered some observations on relationships (real power and energy is generated through relationships and the capacities to form them is more important than tasks), some lessons on communication (drawing on a paper-folding exercise to illustrate the fact that everyone interprets isolated instructions differently, so communication needs to incorporate an understanding of context), and some advice on effecting transformation in the risk adverse healthcare environment (start with why, identify need, expect pushback, use available supports, face the music when things go wrong, interact and be prepared for anything). Ultimately, Howell placed most value on continuous education since “Learners will inherit the earth,” while “the learned will be splendidly prepared for a world that no longer exists.”
Technology techniques to drive analytics adoption and ROI
As befits an information session hosted by an analytics firm, Healthcare Symposium 2015 also offered new perspectives on how technology can be harnessed to support culture change, and with it improved business performance metrics. In many ways, this is a process of enablement through use of new tools – the HealthcareAppExchange, for example, a portal created by regional technical manager at Information Builders, Joe Walsh, that is designed to facilitate the sharing of information apps and reports among users of IB business intelligence solutions among the healthcare community. The Symposium also featured presentations by Dan Grady from IB corporate sales on “Five Hot Trends That are Shaping the Future of Business Analytics” (SaaS/cloud usage, social media, Big Data, mobile BI and predictive analytics) and on techniques for achieving ROI in analytics acquisition. Starting with the premise that “user adoption is the holy grail of BI” (borrowed from Wayne Eckerson), Grady advised implementers to “know your target audience: what they do, what information they need, what incents them, and how they make decisions.” Typically, he argued, this knowledge leads to the separation of staff into casual users who see analytics as a means to an end and power users who stretch the limits of analytics systems in search of insights. Each group has unique needs and characteristics, which Grady outlined in five keys for raising adoption rates within each segment:
- Since these are “answer people,” fast performance is critical, and it should take less than three seconds per click for a user to get the answer they need to improve performance. Information delivery that follows the “3 F rule” – that is familiar, flexible and flows – will drive adoption. Grady suggests implementers think about how people get other information, and package accordingly: tools alone don’t drive utility, a point that he illustrated by noting that the average business user only knows how to use 1.6 percent of Excel functionality.
- Data must be “trustworthy.” Data lineage embedded in the metadata or in the watermark, and a description of the data can help ensure that people trust the information.
- Actionable dashboards that allow users to drill down to the root cause of an issue makes information contagious. The gateway to all information assets, the dashboard should “start high and get low” in the way it communications information.
- Modifiable reports encourage use across a broader base. While it is relatively easy to roll out content, everyone wants a slightly different view of slice of information. To serve the broadest numbers and encourage ongoing usage, it’s important to embed the ability to modify the content as you go.
- Mobile delivery will help ensure information is actionable. Pushing information out to people, rather than expecting them to stop to pull data, makes it more likely casual users will make use of KPIs, and leverage the value of data in working life. The best approach is avoid static reports – to embed access to more information in the document, as is the case with IB’s Active Technologies, an easy to use template that allows people to ask for more detailed data via drop down menus and radio buttons.
- Data mashups must be enabled to allow the power user to explore relationships between data sets across the organization. The ability to ask questions across several different data sources is a huge factor in raising adoption, as is the ability to use a single dashboard, rather than open several applications or connect with another person to access information that is needed.
- High end visuals are needed to discover trends, relationships and patterns that might exist. Power users need the ability to click on an item and get a “360 degree view,” and to drill down to see the outliers which may be the source of problems in answers provided by the data.
- Analytics functions in the BI solution must be buildable, not static. Power users may want to import modules, created with statistical tools such as R, so the solution must support open source languages.
- Content sharing must be an easy proposition. If the goal is to turn insight into impact, it must be possible for the power user to share this insight.
- A system has “got to service everyone” with a solution that is easy to use but also offers flexibility as “no one size fits all” to get maximal adoption in information management.
To wrest maximum ROI from an analytics solution, the typical organization will need to simultaneously serve the (few) power users and the (many) casual users. To satisfy both audiences, the solution should offer a high degree of customization in information access and reporting, and provide both tools (for the power user) and apps for the front line casual user. This will enable self-service, a critical functionality necessary to creating the data-driven culture that will effect change in healthcare – or indeed the productivity outcomes that are required across industries.
For a quick view of these Symposium presentations and more, including The Hospital for Sick Children’s Kathleen Lavoie (formerly of Grand River Hospital) and BlackBerry’s Sarah Jost on securing doctors’ mobile communications, Guelph Hospital’s Doug Mitchell and Conestoga College student Joe Tran on financial benchmarking with WebFocus, and the intro to a Symposium panel on the determinants and outcomes of health system funding reform delivered by Anthony Reddick of McKenzie Health, see the video presentation which follows.