The KPI Cure: How Healthcare Data Analytics Can Improve Medical Center Finances

US healthcare providers could collectively cut more than $1.5 trillion in costs over the next six years by implementing productivity improvements, according to a 2022 report from McKinsey & Co. When it comes to increasing efficiency, healthcare is a particularly complicated industry. There are natural limitations on reengineering healthcare organizations because lives are at stake, not just profits. Yet providers are under immense pressure to optimize productivity, especially since the COVID-19 pandemic, which disrupted service delivery and left US hospitals collectively facing billions of dollars in losses.

The rapid adoption of electronic health records (EHRs) over the last decade has greatly enhanced patient care and operating efficiency, but there is still much more to be done.

As the founder and CEO of management consulting firm Marion Street Capital, I have worked with numerous healthcare providers over the last few years to improve productivity and bolster the bottom line—without sacrificing the quality of patient care. My team and I have deep experience in developing data analysis tools, and we use the insights they provide to develop key performance indicators (KPIs) and a robust plan to manage these metrics. Using this playbook, we’ve helped clients optimize care for thousands of patients and save over $50 million.

Using Healthcare Data Analytics to Drive Strategy

KPIs are different from other business metrics, in that they are specifically engineered to drive strategy by providing performance targets, milestones, and decision-making insights. For instance, sales volume is a common metric, but in and of itself, it represents only a result. In contrast, KPIs meant to help increase sales volume might include measurements like the number of new inbound leads and salesperson response time.

Given the unrelenting financial pressure on the healthcare industry, KPIs are critical and can help providers increase the value of their services, reduce costs, streamline operations, optimize resource allocation, and deliver higher-quality care.

Finding the Right KPIs

The key is to pick the most productive KPIs and design a framework that aligns stakeholders, such as heads of surgery and financial executives, across the organization. We use two broad categories to organize a provider’s data: operational and financial.

Operational KPIs

These metrics are aimed at improving resource use and patient outcomes and can include various measurements, such as the following examples we helped develop for one of our clients, an innovative opioid use disorder clinic in the US:

• Provider utilization, i.e., how much of each healthcare professional’s time (in number of hours or percentage) is spent on billable services with patients
• Patient no-shows
• Unique patient encounters
• Patient churn (admissions and discharges)
• Unique patients seen
• Services per patient, per visit

News headlines tend to focus on medical procedures and drug prices when it comes to healthcare costs, but labor is typically the biggest factor driving spending growth. That’s why providers that want to reduce costs would do well to focus on better using their workforce—for instance, collecting data so administrators can optimize patient flows and reduce wait times, deploying computer and mobile applications to help physicians manage administrative tasks and increase time spent with patients, and sending automated appointment reminders to patients to reduce no-shows and save staff time.

Financial KPIs

The other category of KPIs is financial, and many of these are similar to the metrics other industries use, like operating cash flow or net profit margin.

However, some finance-related KPIs are unique to healthcare, such as patient drug cost per stay and the entire subcategory focusing on the relationship with payers, most commonly insurance companies. These are crucial metrics for managing and enhancing a provider’s revenue stream, and include data like:

• Claims denial rate
• Insurance claim processing time
• Reimbursement rate: the dollar amount paid by insurers

Setting goals for improving operational KPIs can translate into better care and operating results, while doing the same for the financial indicators bolsters the bottom line directly. The overall objective of selecting KPIs is to figure out which targets have the biggest impact while keeping the staff and management aligned on shared goals. It’s also crucial to adjust KPIs over time, revising them as conditions change.

Before providers get to that point, though, they need to have an effective dashboard for visualizing and interpreting these metrics.

Creating Dashboards

The power of KPIs is that they turn data into insights. But to manage these insights and translate them into efficiencies, providers need an effective way to present that data. Typically, I find the most efficient and appealing way is through a visual dashboard.

The first step we usually take with a client is to make sure we understand the organization’s specific goals—for instance, reducing an operating deficit while mapping out the business’s technology infrastructure and where it’s generating data. We also survey staff members across different functions to understand how they can leverage data to improve decision-making—for example, by studying electronic medical records to glean high-level trends like average patient stays—without breaching patient confidentiality.

Once we determine which data to track, we begin to build a data stack, an information resource that we will help manage and refine along the way. The first step is organizing the information into a database, data warehouse (a collection of databases), or data lake (a repository that stores more raw information for modeling and analysis). Each of these is a useful means of storing content, but there are different use cases in which one is more advantageous than the others.

Then we create the dashboard using a business intelligence (BI) platform like Tableau, Power BI, or Qlik Sense and use that to pull information from the organization’s database or data lake.

Segmenting the Information

Deploying a successful KPI strategy requires determining the most useful metrics and features for each dashboard so the people who interact with them can do so productively. Different stakeholders need different data points and will have different KPIs, so it’s important to segment data for the right audiences. For example, you might want to design an operational dashboard for physicians and medical staff while moving financial KPIs into a different dashboard for administrators and executives. Data literacy will most likely vary widely across an organization, so it’s important to present information and provide guidance, view options, and filters in a way that doesn’t overwhelm people.

The goal is to maximize the return on investment, developing a stable and repeatable process. But we also want to maximize automation and minimize the need for human interaction with the data procurement and updating.

A Dashboard Case Study

For our opioid use disorder clinic client, we developed three dashboards: one for medical staff, another for the billing department, and a third for clinic executives. This system and the benchmarks it measures helped the organization generate $19.5 million in free cash flow in its second year of operations—money that was reinvested in social programs for the community.

Clinic Data Dashboard

This board pulls data directly from the clinic’s electronic medical record system and gives a detailed look into daily performance and operations across multiple areas of the business. We further divided the dashboard into the following pages for even more-specific audiences:

• Summary: This page is relevant to everyone at the clinic and shows some of the main clinical KPIs: billable services provided, admissions, discharges, and census (the number of unique patients who visited the clinic that day).
• Utilization: The chief medical officer, COO, and operations leads (such as directors or senior managers) look at this page, which shows the number of services and time spent on services by each clinician per day or week.
• Retention: The COO and operations leads use this page to do a deep dive into admissions and discharges by date, medical staff member, referral source, program patients are assigned to when admitted, and more.
• No-shows: This page allows the COO and operations leads to note specific trends, such as appointment no-shows. For example, our client saw that patients who have been in the program for fewer than 30 days miss 20% of appointments, while those who have been in the program for more than 30 days miss 10% of them.
• Chief medical officer report: The chief medical officer needed a specific report made especially for him, which quantified and gave information on positive drug tests and quantified medical errors that might require quick resolution or reporting.
• Staffing: The COO and operations leads use this page to understand how to staff the clinic on any given shift. Both retrospective and predictive, it tells providers about any past staff shortages or surpluses and then uses that data to project the number of staff members that will be needed to meet demand at specific days and times in the future. Created as part of an ad hoc request using Six Sigma and Lean processes, this page remains as a live report in the dashboard for ongoing monitoring as the clinic grows.

Billing Data Dashboard

This dashboard pulls information from the medical billing system and displays an estimate of medical claims we expect the clinic to collect as revenue. It also displays trends among specific insurance payers, such as how quickly they pay and how much they pay relative to our expectations.

The finance team uses the estimate of outstanding claims for Accounts Receivable, and the billing team employs this dashboard to understand claim trends and identify issues they need to address. For example, if a specific payer’s average DSO—days it takes to process and pay a claim—goes from 15 days to 25, the team can use filters in the dashboard to identify and examine specific types of claims that may be taking longer and brainstorm solutions.

C-suite Dashboard

This distills data from all of the other dashboards into a summarized view of the KPIs that are most critical to operational and financial performance. This dashboard also ties in the weekly targets we helped establish for the clinic—including patient encounters, services per patient, admissions and discharges, and revenue and accounts receivable—so executives can compare actual results to the forecasts.

Since we track KPIs across different areas of the healthcare organization, each one may have a different owner or responsible party. We use what we call a “dynamic feedback loop” to support the flow of information, regularly reviewing, interpreting, and discussing KPI values relative to targets with stakeholders. This allows us to constantly reassess the effectiveness of the metrics and the effect they’re having on the organization’s operations and profitability.

Crowdsourcing Solutions

As any organization evolves, there are challenges that emerge, requiring solutions beyond just meeting and beating benchmarks. Sometimes the data also signals the existence of problems for which there are no obvious solutions. Surveying all of the parties involved can help in this situation.

For example, one dashboard we designed for a healthcare client last year highlighted a substantial decrease in group counseling sessions at their clinic. By crowdsourcing information from several areas of the organization, we obtained a holistic view of the group counseling initiative and put in temporary correctives—such as more granular reporting on group counseling and increased promotion of the service by schedulers—to mitigate the downtrend and increase sessions. Here are some examples of what we discussed with different staff groups:

• Medical billers: How much revenue does the clinic receive from a group session versus an equivalent service?
• Other medical staff: How will an increased focus on group sessions affect their schedules and interactions with patients? Was the medical staff pushing patients to attend individual sessions instead of group sessions? And if so, why?
• Other operational staff: Was there negative feedback regarding group sessions in patient surveys that may have led to patients avoiding them? Are there other operational changes we can use to make the group sessions more effective and enjoyable?

As a result of the data we collected via these dashboards and discussions, clinic executives launched several initiatives to schedule more group sessions and promote them through patient communication channels. The number of group sessions increased by 476% from October 2021 to June 2022, while the no-show rate among patients fell by 13.2%. This increase contributed to 58% growth in average weekly revenue over that same time period.

Other Healthcare KPI Do’s and Don’ts

In a healthcare setting, positive patient outcomes should supersede profitability—so the fact that KPIs rely so much on data to help produce efficiencies may raise concerns about potential conflicts. With this in mind, one indicator we specifically avoid is attaching staff incentive compensation to the volume of services provided to patients, as that could encourage shorter interactions and poor provider-patient relationships.

Another way we’ve tried to address potential conflicts while still looking at volume is by developing a KPI called Active Census, defined as the number of active patients enrolled in a treatment program at a clinic. This encourages clinicians and other key staff members to develop practices that lead to patients staying with the program. This kind of quality interaction results in greater patient retention—and patient outcomes.

Taking multiple KPIs into account is also crucial to understand and relate a complete narrative. Looking at just one metric often provides an inaccurate picture. If the chief medical officer were to examine provider utilization alone and see that one provider worked 30 to 35 billable hours out of 40 per week over multiple weeks, while another worked 20 to 25 hours in each of the same weeks, the leader might conclude that the first provider was more productive and deserved a larger raise or bonus. However, the second clinician may have had a higher no-show rate, meaning his hours were unrelated to job performance or work ethic.

Every organization is different, and there’s no single answer for charting the path forward for the healthcare industry. Still, I’ve seen how seemingly minor changes like reducing the number of no-show appointments can have cascading effects throughout an organization. Used properly, healthcare data analytics offer tremendous opportunities to improve both the bottom line and the quality of care as staff and leadership work toward common goals.