Maximising Hospital Capacity with Predictive Analytics
While everyone had to adapt to the constraints of the pandemic over the past year, healthcare institutions had the additional task of dealing directly with the impact of the virus. We didn’t have the luxury of simply shutting down for a while to reorganise into a new operating model. The need to continue providing routine medical services while addressing the virus was incredibly challenging for many of us.
Fortunately for us and the residents we serve, (WWL) had already started our data journey before the pandemic. By early 2020, we’d hired our first data scientist and were well on our way to using data-based decisions to enhance our healthcare. Because of this preparation, we were ready to leverage the power of data during the pandemic.
Establishing a Data Foundation
WWL is a teaching hospital responsible for the medical needs across Wigan Borough in the North West of England. We have an annual budget of about £397 million and serve an area with a population of 326,000. Our Trust operates several facilities, including the Royal Albert Edward Infirmary, Leigh Infirmary and the Hanover Diagnostic Treatment Centre, and Wrightington Hospital, as well as services in 16 community clinic settings within the borough. Each year, we receive more than 70,000 inpatient admissions, 676,000 outpatients, and 2,400 baby deliveries.
Even with a large population to serve, we’re still a small trust, with the lowest number of beds for our population in the region. That reality is part of the reason for harnessing data. Even before the pandemic, our margin for error was tiny compared to other institutions.
Having been with the NHS for 15 years, and with WWL for eight years, I was part of the drive toward data. Years before the pandemic, we were already using to collect data to help with our operational decisions. I also became a in 2017. As part of that group, I was exposed to the passion for data intelligence shared by some of the best and brightest Qlik users. I learned some of the best practices for converting mountains of data into real insights.
At WWL, we were tracking the number of people visiting our facilities and what services they requested. Whereas previously this information was monitored via various spreadsheets or multiple paper reports, Qlik allowed us to create a single source of truth, so everyone got data from the same source. With Qlik, we also had the benefit of near-instantaneous updating and access.
This access to one pool of data also enabled us to ensure that clinical services, hospital executives, and external regulators all used the same information. Then, in 2019 we hired Jamie-Leigh Chapman to serve as our resident data scientist. She used the information we were already collecting and added some complex algorithms to look at “what-if” scenarios. She would often manipulate data using tools like R and Python, but present the information to our organisation using Qlik. That gave WWL staff that were not data proficient the ability to digest the information in a relatable format.
Our Data Analytics & Assurance Team had also already implemented a writeback capability into our "Hospital Live" dashboard. That gave us an even better ability to collect, manage, and access our data at the same time. Going into 2020, we already believed our data operation was in a good place.
Little did we know just how useful these tools would be in the coming year.
Capturing New Insights
Like most of the world at the time, we watched the news about the pandemic with naivety in the beginning. It almost took us by surprise when the first few patients with COVID-19 entered our doors. Once we began to realise the severity of the situation, our Data Analytics & Assurance Team needed data that was not available within the previous system. Suddenly, our data initiative went from a long-term improvement effort to our best survival tool for the moment. We began actively manage hospital capacity, including which wards were being used to isolate infected patients. Near real-time tracking of demand meant we could anticipate the need to flip wards to serve larger numbers of COVID-19 patients. We were also able to track equipment use like oxygen tanks to know when we may need inventory changes.
Managing wards in real time was critical. Prior to the pandemic, there was no reason we could not place a cancer patient in the same ward as one suffering from heart disease. We made our decisions based on patient needs like equipment. With the pandemic, however, we had the very real fear of cross-contamination. By using Qlik to manage patient services data, the system would flag patients with the virus before they were moved to another area or a different site.
The power of Qlik transformed our response to the virus in ways we could not have anticipated. Not only could we capture new data using Qlik, but it could be aggregated, visualised, and shared with the organisation in real time. This meant our leaders never lost track of the level of infection within a hospital or the overall demand. We were able to make data-driven decisions about service and space allocation and minimise the risk of cross-infections.
Predicting the Future
Our previous “what-if” scenarios used complicated predictive models. These algorithms were designed to let us make smart decisions about the future using the data available at the time. Since the entire world was managing the virus, many of these models were able to take advantage of external information. We could easily send reports to other institutions using Qlik and incorporate outside data into our analysis.
The biggest strength of our data operation was visualisation. It may have been a brilliant move to hire a data scientist in advance, but that expertise is meaningless without the ability to easily convey the information. It is often an unfortunate reality that powerful insights can be hidden within systems with little to no visualisation tools. Qlik visualisations remove the complexity from the equation so that everyone from ward to board can make better decisions. This ability was especially critical during the pandemic, with every single department making critical decisions under shifting circumstances.
Aside from visualisation, real-time data recording and dashboards were another critical strength of Qlik. Prior to the pandemic, much of our predictive modelling was based on eight-week averages. This was a standard created by several academic experts and other data partners. However, the onslaught of COVID-19 made long-term averages practically irrelevant. The demands on acute care often depended on the results of virus testing in the past few days. The two-month average was inadequate as a predictive tool.
By automating the modelling and data recording, we could make predictions based on current trends. Even in the ever-changing pandemic environment, we found these models incredibly accurate and insightful. With WWL having the fewest number of critical care and general acute beds in Greater Manchester, we needed that level of precision. Our hospital could not afford to underestimate demand. Using Qlik ensured that our entire organisation had access to the data we needed to make department-specific decisions.
Data Sharing for the Greater Good
Another impact of the coronavirus was the need to share data with external partners. The entire world was grasping something entirely new. Every medical institution was experimenting with different treatment combinations and doing their best to allocate equipment and supplies. Even government agencies that were not directly involved in medical care needed information to make better policy decisions. Our data solution had to be designed to facilitate data sharing with external partners.
We decided that Qlik Cloud was our best option. It made financial sense to invest in the cloud environment versus continually paying for licenses. In fact, WWL was one of their first customers using this service. After a few months of configuration, we were able to use the cloud to securely share real-time data with our external partners. This tool even worked with groups not using Qlik, removing any barriers that may have existed otherwise.
Being one of the first customers for this service brought us closer to Qlik. They introduced features like alerting and chatbots in perfect timing. On other occasions, we challenged their staff to adjust features to meet our specific requirements, such as providing external access to our directory. Fortunately, Qlik was always willing to work with our organisation to make sure we had the right tools.
A Look Back on Our Success
More than a year into the pandemic, we are now in a better position to reflect on our transformation. We implemented a new and innovative way to use Qlik to capture data and were able to maintain that single source of truth. Our data scientist has successfully created “what-if” predictive algorithms and uses visualisation tools to effectively convey the information. Qlik Cloud features created a powerful channel for sharing real-time information with our external partners, allowing better coordination of our virus response.
In approximately a year, we have progressed from initial data analysis to a full cloud transition. This included expediting our migration to a data warehouse. In the future, we expect to continue to explore the power of predictive analysis. As we begin to deal with the backlog of elective surgeries, we expect these data tools to provide powerful management insights.
Qlik enabled WWL to connect, utilise, and share data more efficiently. The insights we gleaned from these tools helped us to better serve our patients during a worldwide pandemic. But don't take my feelings as evidence. We ranked third best in North West England in meeting the NHS' standard of emergency room patients being seen within four hours. We also had the fourth-lowest risk for prolonged length of stay and one of the lowest numbers of delayed discharges.
These are real accomplishments made possible by Qlik’s ability to visualise predictive insights and share information from a single source of truth. I shudder to think how we would have responded to the pandemic without it.