Healthcare Reporting: Centralized vs. Decentralized
Click to Enlarge
One of the questions that often comes up as healthcare organizations look to get more deeply into data analytics is whether their reporting functions should be centralized or decentralized. The answer I will usually give is “yes.”
I’m not trying to be flippant. It’s that we at Health Catalyst believe in a concept that Reed Hastings, co-founder and CEO of Netflix, refers to as “highly aligned, loosely coupled.” Highly aligned means everyone in the organization shares the same specific strategic goals, and that team interactions are focused on strategies rather than tactics. Loosely coupled means different groups still have the flexibility to take the approach they need to regarding the tactics they use – so long as those tactics remained aligned with the organization’s goals.
The approach is much like the offense on an NFL team. The coaching staff develops a game plan and the overall strategic approach the team will take to win the game. However, during the game, the offensive and defensive coordinators call the plays according to the game plan. That is highly aligned.
At the line of scrimmage, however, if the quarterback sees a certain defensive formation and realizes there is a better opportunity available, he has the freedom to call an audible and make a tactical decision that is still highly aligned with the team’s overall strategy.
The purpose of analytics in a healthcare organization is to create insights that help achieve strategic goals such as improving clinical quality and patient safety, reducing the cost of care, lowering readmissions, or improving the health of populations. Analysts are the conduits that bring together data from different areas to tell a story in a meaningful way so non-analysts can take the information and act on it. The tactics they use depend on the organizational structure.
Basics of Centralized vs. Decentralized Model
Organizations generally take one of two approaches to analytics and reporting. One is a centralized model, where the analytics group is its own entity, independent of any particular group. If Cardiology, or Women’s and Children’s, or another group wants an analysis of its data to gain insight on how to improve a process or address an issue, they make a request of the analytics group, which goes into a queue, and the analytics team prioritizes it into their workflow.
The request is not randomly assigned. Normally, the best practice is for one or more analysts to work with a specific group, as they intimately understand the clinical context of the requests. However, the analytics group manages the analysts, not the groups making the request.
In a decentralized model, the analysts work directly for the different groups or departments. As such, the domain does not have to compete with others for the attention of the analysts. The analysts can focus on serving those customers well, and departments may purchase their own analytic tools.
Each of these models has its pros and cons; the pros of one are usually the con of the other. Let’s look a little more deeply at each to see where their strengths and weaknesses lie–and why a hybrid of the two tends to work best.
Centralized Reporting Pros
The centralized model has many advantages. Overall, if you had to choose a single approach, this would be the one I would recommend based on my experience with both. Its strengths are:
- Standards and best practices vs. a maverick approach – If the goal is for the organization to be highly aligned, standards and best practices for analytics are required. That begins with having standard tools. If everyone is sharing the same tools, the analysts are able to share their knowledge about those tools as they learn. The organization as a whole gets much smarter, rather than having individual pockets of expertise. It also means there is always someone available to run a particular report, even if the person who usually works with a particular department is sick, on vacation or otherwise unavailable. Standardizing on the tools means analysts or data architects can move between domains much more easily. When I was an analyst at one provider with a decentralized approach, it could take six to nine months for a data architect to get up to speed in a new area because he or she didn’t know the technology being used.Finally, while individual groups may only be concerned with their own areas, management is interested in the bigger picture. Rolling up reports that were created using different tools so they can be viewed in the aggregate is extremely difficult, time-consuming, and costly. Having a standard set of reporting tools ensures it can be done almost instantly.
- Flexibility – Although the ideal is to keep analysts tied to specific internal clients, the reality is there are often peaks and valleys in workloads. Organizations that use a centralized approach can shift resources where and when they’re needed; those with a decentralized approach cannot. For example, a major presentation requires a lot of analytical detail. If analysts from another department have the bandwidth, they can temporarily be shifted to take on the increased workload. Another example is a backlog of requests yet there is no one specifically assigned to a particular department because the volume doesn’t justify it. During these times, resources can be adjusted to address those requests.
- Ability to support individuals with different skillsets – If a domain in a decentralized model can only hire one analyst to be its sole resource for reporting, it cannot be a junior or mid-level person. It must be someone with a high level of expertise who can work independently. In a centralized model, the organization can have a mix of skill levels since the junior or mid-level people can always go upstream to get questions answered or learn new skills.
- Spot analytics trends vs. analytics islands – In a decentralized model; what is learned in the domain tends to stay in the domain. When reporting is centralized the organization can aggregate reports from multiple areas. These reports can be used to build customized dashboards that show executives the direction the business is heading overall, enabling them get ahead of trends.
- Better management of resources – In a decentralized model, analysts generally report up to a medical or operations director, i.e., someone who doesn’t have the technical background to understand what they do. If the analyst says it will take three weeks to build a report, the manager has no idea whether that is too long or not long enough. In a centralized model, management is provided by experts in analytics who know how long a particular task should take–and when a heroic effort is needed to deliver a particular report.
Centralized Reporting Cons
Despite all those positives, there is one big negative to the centralized model – the inability for a specific department to own their own destiny. They are subservient to a broader prioritization process. I have seen this happen over and over where the specific needs of a department or higher-level initiative usurp the needs of another department. This is generally when the dissatisfied department says… fine… if IT can’t help me, I will go hire my own people. This is a real problem and is the crux of why we see decentralized models.
Typically, a particular department or domain with deep pockets will provide the bulk of the funding for an analyst, who essentially becomes a dedicated resource for them. The normal ratio is 80 percent from the department and 20 percent from the central organization. When that department has an urgent need, its request is given priority in the queue due to the financial stake it has in the analyst (even if it it’s not the highest priority for the organization). It is an effective way to ensure requests are properly prioritized and things get done.
To Create Hospital Reporting Efficiencies, Play to Organizational Strengths
Centralized vs. decentralized reporting doesn’t have to be an either/or choice. In my experience, the organizations that are most successful are the ones that combine the two to keep their reporting processes highly aligned yet loosely coupled – giving them the best of each while overcoming each model’s negatives.
The primary benefit is that analytic efforts are focused on a specific need. There is a lot of low-hanging fruit out there, and this approach allows those who are down in the details to take advantage of immediate opportunities. As long as their work aligns with the overall organization’s strategy, they don’t need to follow the formal processes that are part of the centralized model.
How does this compare to your experience? Does your organization follow the highly aligned, loosely coupled model? Any additional pros or cons in either model I’ve missed?
Would you like to use or share these concepts? Download this presentation highlighting the key main points.