Healthcare analytics best practice scrum

Healthcare in Ontario received a shot in the arm this month when data specialists from hospitals across the southern region of the province met to discuss the use, abuse and potential of analytics to resolve issues in institutional management and patient care delivery. Hosted by BI and analytics provider, Information Builders, the Healthcare Analytics Symposium was held in Niagara-on-the-Lake, a bucolic location that set a convivial tone for the open sharing of information and best practices in information management. Information Builders has identified the establishment of software user groups as an important means to drive awareness of analytics business value; this event demonstrated the opportunity for building knowledge around the use of data management technologies afforded by the coming together of peer-to-peer networks representing a specific — and in this case, complex — industry. Interestingly, impetus for the event came not from the host, but rather from the community. According to Canadian marketing manager and event organizer Caterina Didio-Duggan, the Symposium represented Information Builders’ response to repeated requests from healthcare analytics users for a forum for the exchange of ideas and experience.

Identifying issues

Wayne Samuels, partner, health industries, PWC
Wayne Samuels, partner, health industries, PWC

To provide a framework for discussion, Wayne Samuels, partner, health industries at PwC and Dipak Pandya, director, health industries at PwC, opened the Symposium with a review of four key trends that are shaping a new competitive healthcare landscape. While “frequent flyers” — the top 5% of patients who consume 66% of total health expenditures — are an increasing focus of attention in policy circles, Samuels also described reform in financing models, epitomized by Accountable Care Organizations in the US which tie provider reimbursement to metrics showing achievement of specific patient outcomes (and potentially, reductions in the total cost of care for patients). And while growth rate increases for spending (as opposed to total spending) in Ontario healthcare have been in remission since 2009, a shift towards funding programs aimed at improving care for specific patient groupings is on the rise: citing an early Ministry of Health study, Samuels noted that funding for Quality Based Procedures (such as hip replacements or COPD) has grown from 5 percent in 2012 to 30 percent of the total in 2014. To improve care delivery, many providers are turning to digitization and mobility in particular — by 2020, virtual, remote healthcare is expected to account for 25 percent of care delivery. And a final phenomenon is “Health everywhere,” or, Samuels explained, the rise of new healthcare players which hail from telecom, transport, hospitality, financial services, retail and media sectors that are looking to leverage their existing relationships with consumers as they develop health products and services. According to Samuels, US consumers spent $267 billion on ancillary health services in 2012 as they looked to “DIY.”

Dipak Pandya, director, health industries, PWC
Dipak Pandya, director, health industries, PWC

So if the future of healthcare is more patient-centric, more coordinated, more transparent on cost and quality, digital, and disrupted by non-traditional players, what is the impact on the hospital CFO? According to Pandya, he/she has a “new agenda” marked by the need for efficient and effective resource allocation to meet reform goals; performance management based on new funding methods that mandate tracking and meeting goals; and better understanding of organizational needs/performance to inform adoption of new tools for service delivery. To support this agenda, Pandya advocated for the establishment of “integrated EPM” (Enterprise Performance Management), in which information, processes, data and systems are integrated and aligned to accelerate the achievement of strategic objectives — and for quality analytics that can help the institution mark progress towards targets and better track the utilization of resources and services, which in Ontario have formed the basis for Health Based Allocation Model (HBAM) funding since 2012.

Samuels and Pandya’s deep insight into performance, competitive and funding model change in the Ontario healthcare environment were not presented as ‘challenges’ but rather as “innovation trends” that may form the basis for transformation. This positive theme was picked up in other presentations at the Symposium by individual data specialists who detailed specific initiatives at their hospitals that leverage analytics technologies as an innovation engine. In each case, presenters highlighted a unique component of the implementation process which when combined offer comprehensive guidance that Symposium attendees and other adopters may use in developing their own analytics strategies.

Continuous improvement at Grand River Hospital

Grand River Hospital slideStrategic planning and implementation of a data warehouse that could offer up one version of the patient truth, and which would anticipate future service needs (15 years out was the expectation) was a complex process at the Grand River Hospital (GRH), involving input from 75 different stakeholders and five years’ time. According to presenters Kathleen Lavoie, chief privacy officer & director, HIM at GRH, and clinical data analyst Ben Windling, “culture” — or collaboration with different groups — was a huge focus in Grand River’s implementation as the data warehouse would eventually have to support the needs of 68 different departmental systems. Working in a “Fragmented System Environment,” where each business unit relied on best-of-breed BI and reporting tools, Lavoie’s team relied on Information Builders iWay data migration products, and WebFOCUS InfoAssist and Developer Studio reporting technologies as these “are agnostic — IB was able to work with all [departmental] tools and so could accommodate people with different tool preferences.” Care taken to build consensus, and institute 16 analytics best practices  has generated its own measure of success; so far, the project team has managed to onboard data from nine administrative and patient care systems (including Admissions, Discharge and Transfer), has reduced the amount of time needed for information delivery from months to three days (based in part on the separation of reporting from the systems that were in use every day), has introduced consistency in language and terminology and has other units queuing up to participate.

But Grand River is not done. Lavoie is developing a profile-based “triangle” of users as input to a new front end that will offer easier, more intuitive access to the data — relieving IT and decision support of the many questions that are coming with increased interest and use of the data warehouse, but also enable new levels of data driven decision support for hospital staff and practitioners.

Cambridge Memorial Hospital integrates for real time patient tracking

Long wait times, and long length of stay for patients in Cambridge Memorial’s Emergency Department (which did not meet provincial targets) and testing delays prompted a small team led by Mike Meyette, director of informatics and corporate services at Cambridge Memorial, to look at improvements. Though the hospital had an Emergency Department (ED) tracker system that had been implemented by Information Builders using WebFOCUS, this did not connect to the hospital’s text-based Meditech Patient Care system and there was no data warehouse to store data for reporting. The solution was integration — achieved through deployment of Blue Elm’s OpenGate software, which accessed Meditech directly in real time, and deployment of IB’s iWay Service Manager to pull data from OpenGate and make it available to WebFOCUS through the IB Web Service Adapter.  With integration and a data warehouse in place, Meyette’s team was able to evolve the tracker app according to requests from users for additional information (patient status, room status, etc.) and formatting, and ultimately in version two, to use OpenGate to refresh a table every minute while the original tracker continued running — for near real time information updates. And through integration with the hospital’s new WebFOCUS portal, have developed further enhancements in v3, such as secure access to information by authorized users located outside the emergency department, access to data for different interested groups — nursing, for example, which can now “see” ED patients that are ready for admittance to their unit — as well as tracking for the Clinical Decision Unit to ensure patients are held no longer than the provincially mandated 24 hour maximum, and a medicine unit tracker. A next phase involved application of tracking to the operating room to improve patient experience and offer status information to families. Based on experience building the ED tracker, Meyette’s group has developed an application that delivers information via a screen in the waiting room or web portal in the form of visual codes (patients are identified by number only).

In addition to real time Tracker info, the system also generates reports that will show various unit managers what their performance was on a daily basis, a key input to building greater accountability in processes and procedures. But according to Meyette, an important advantage to implementation of an integrated data warehouse and reporting system is that the “technology has allowed us to build applications rather than reports.” Additionally, the hospital has saved money through use of the Trackers, savings that can be applied to new technology acquisitions and projects, though as Meyette concluded, “we don’t measure ROI on the application per se — it’s more about quality performance improvements.”

Brant Community Healthcare System transitions from “hunch-driven to data-driven” Lean decisioning

For Tim Armstrong, Horatio Stewart and Casey Scholl of the Brant Community Healthcare System, implementation of BI was a strategic initiative designed to support the institution’s transition to Lean management principles. With backgrounds in accounting and decision support (as opposed to IT), Scholl and Stewart were tasked with building an information system that would move Brant from “experience-based to evidence-based decision making,” and address growing internal and external (LHIN) pressures for more data at a time when “budget crunch means no new human resources.” An initial approach to the institution’s data service front page, based on Microsoft Excel, VBA programming and integration with Meditech using NPR Reporting proved inadequate: the solution involved a labour-intensive process to populate datasets, long development cycles, reliance on a narrow internal knowledge base (neither Stewart nor Scholl are DBAs) and offered no ad hoc end user capabilities. As a result, the group moved to Information Builders BI technology to enable restructuring of information management according to (Lean) value streams. According to Stewart and Scholl, some immediate benefits of the new Brant BI portal included support for restructuring, identification of system issues, new views of the data (day of the week, weekly trends, yearly comparisons, consistency in ER admissions and discharge rates), in rapidly built and accurately deployed dashboards.

But Brant’s relationship with IB’s BI is relatively new, and for the team, the future looks even brighter. Going forward, the team looks forward to embracing:

User driven technology — develop a process to engage leadership, find their business requirements and address them.

Culture change — out with the paper; in with the new by showing users the benefits of BI.

Planning for the forever build: the world does change — senior leadership was informed they would need two years for system implementation, but the team will also purchase software for data modelling.

Self-Service — since Brant doesn’t have the staff resources to deliver all information queries, the team is focused on helping different units to run their own reports — even half of the time.

Integration of data, getting answers to tough questions — for example, OR is looking for access to survey information on surgeries, or for data estimating costs of new patient or processes to drive shift towards a “patient-based hospital.”

Proactive, not reactive — IB technology has freed up Brant’s information professionals from manual processes so they can engage in forecasting, etc. They are redefining decision support, building capacity through automation of manual tasks in Decision Support, training the team on using the software and providing them with access to report writing.

Going mobile — IB technology can be run on mobile devices (tablets, smartphones) to facilitate information access.

Real time, or as close as you can get — reports are currently run over night, but in ED, for example, delivering reports in minutes would allow better decision making.

Bridging the data gap to drive aboriginal cancer strategy

Cancer care protocol signing with Nishnawbe Aski Nation, 2014
Cancer care protocol signing with Nishnawbe Aski Nation, 2014

The second Symposium keynote, focused on issues around access to cancer data for First Nations people, was delivered by Joseph Imre from Cancer Care Ontario and Hash Qureshi and Sarah Lyons from consulting and SI firm A Hundred Answers. According to Imre, cancer is rising more quickly among FN people, who are diagnosed with later stage cancers and for whom survival rates for major cancers is significantly worse than the population as a whole. But though the priorities of Aboriginal Cancer Strategy, including surveillance, prevention and screening, are dependent on high-quality, relevant and reliable data, Cancer Care Ontario (CCO) had no ability to identify First Nations Inuit and Métis individuals in its existing data repositories. To build relationships with FN people, CCO signed “relationship protocols” with a number of groups, which formed the basis for creating access to people data, while working with Regional Cancer Programs and FN networks to identify core FN health tables.

CCO also worked with A Hundred Answers, which was tasked with sourcing new and existing data holdings that could be used to link COO data and capture FN identifiers. A Hundred Answers has delivered its first report on the creation of identifiers that will be used for program measurement, community-level reporting, targeted interventions and culturally appropriate correspondence — with the ultimate goal of building regional capacity and addressing barriers for under screened FN populations in delivery of surveillance, diagnosis and symptom management. To underscore the importance of data collection in this process, Sarah Lyons noted: “the work CCO had done to establish relationships with the communities was instrumental in creating access to people data — and the way they approached issues of privacy helped to build trust” with aboriginal communities.

Quinte Health Care separates the users from the users

In his presentation, Peter Papadakos, director, decision support & analytics, health records, FOI & privacy, at Quinte Health Care, addressed familiar challenges in analysis (formatting, repeatability, automation, customization and data discovery), as well issues in data integration (standards for data types, naming, documenting data models and the process or gating for change). To improve access and usability, Quinte Health identified five key objectives: to improve turnaround times on data and analysis, enhance accessibility, automate where possible, shift towards more value add work, grow skill sets in Decision Support, engage in agile iterative information systems development with end user input and move the organization as a whole towards more self-service.

On this last goal, Papadakos offered a fascinating account of the thinking behind the creation of self-service at Quinte. Papadakos’s team asked “who accesses the data” and “how is data accessed” to identify five different user types, including  push users, dashboard users, analytical users, advanced users and power users, and their unique needs. “User types are various,” he explained, “some just want one specific dashboard, some want straight data, some want to be able to slice and dice data, some want performance metrics, while some users want all different types of reports. Because users exhibit varying levels of skill in working with data and with technology, and have varying needs, Quinte Health leveraged IB’s business intelligence software to create a series of dashboards, push access alerts, predictive apps that correlate symptoms or hospital encounters that align with HBAM Groups, such as COPD or pneumonia, analytics to align QBP strategy, multi-dimensional reporting for acute inpatients, and access for power users who can create their own reports. Automation and customization capabilities in Quinte’s BI software have enabled achievement of many of the original goals outlined by Papadakos: access to a small set of data in static reports has been transformed into access to numerous interactive data sets in many forms; recurring data operations have been reduced by 25%, automation has been extended from data flows to reporting, virtually all internal (and many external) stakeholders now have service, and most importantly, users that had a one-dimensional cut of data now have more flexibility and opportunity to gain insight from hospital data.

Guelph General Hospital develops “Daisy”

In the last presentation of the Symposium, Doug Mitchell spoke at length about the importance of mapping the implementation of BI — a process that at Guelph entailed establishment of a “program framework” where the Steering Committee defined business objectives, and leverage of Decision Support teams in BI, Process Improvement, Financial Management and Patient Flow. To avoid “silos,” the BI team was made up of Decision Support and staff from IT, health records and the data warehouse; to measure progress, a “maturity update” model created, and to balance the need for user/executive buy-in with high value analytics, the team identified specific projects in three categories: “quick wins, flashy success and long-lasting, sustainable value.”

Guelph General Hospital data warehouse information systems
Guelph General Hospital data warehouse information components

In Guelph’s “Daisy” system, the information source systems represent the petals of the daisy, and the centre is where these comes together as business intelligence. As Mitchel explained, the data warehouse was designed by Florida-based healthcare data warehouse architect Rick Biehl to house information in as simple a manner as possible, and features 19 dimensions and 3 facts that accommodate the information needs of all units outlined in the slide presented here. But in access and reporting, Daisy is anything but simple. With the help of IB’s Data Migrator (for data warehouse admins), and DevStudio and InfoAssist products, the Guelph BI team has managed to create a panoply of apps and dashboards, ranging from a real time GGH patient tracker, to a financials spending tool, to a “metric mart” library of metrics that have been captured in other reports to track and trend KPIs, such as wait times or numbers of procedures completed.

Information managers representing hospitals across southern Ontario who attended the Healthcare Symposium face similar challenges — fragmented systems, issues with data integration, data gaps, and the need to accommodate different user groups — while working to support provincial mandates to monitor, speed, and report on improvements in patient care delivery. Speaking the same language, presenters and the attending audience were able to absorb lessons learned from each implementation, and consider their application in the home setting. The problems in data management that the presenters identified will be familiar to organizations outside the healthcare vertical as well, and project principles and processes they discussed an aid in a variety of implementation environments.


Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.