By submitting your email address, you agree to receive emails regarding relevant topic offers from TechTarget and its partners. You can withdraw your consent at any time. Contact TechTarget at 275 Grove Street, Newton, MA.
This article originally appeared on the BeyeNETWORK
Walk down the halls of any hospital, clinic, long-term care facility or home-health agency and you are likely to see a large number of run charts decorating the walls. These charts, as I am sure you are aware, are used to measure an almost endless array of clinical, service, administrative and financial indicators. The data feeding these charts may come from many sources, but the goal is always the same: to focus management attention in order to improve performance across time.
Run charting and business intelligence seem to have been made for each other. Not only can you use business intelligence data to power up your run charts, but you can also use the insights gained using run charts to power up your business intelligence.
Throughout this article, I will be using a simple efficiency service variable to illustrate the power of merging run charts with business intelligence. The variable is called time to first clinical entry. Picture a visit to the doctor. You register at the front desk and sit down in the waiting room. You get the call to come into examining room #3, where you wait a bit longer. The nurse comes in and takes your vitals. He or she signs into the system and records your blood pressure, pulse, temperature and so forth.
From the moment your record was brought up at the front desk until the moment your vital information was entered is called time to first clinical entry. As you can see, a lot of steps were involved, including giving your information at the front desk, the time you spent in the waiting room, the time it took to get to the examining room, the exam room wait time plus, of course, the time it took to get your vitals.
Why is this time measure important? It is important because you came in for clinical service, not for administrative recordkeeping or to wait around. Until the nurse entered your vital information, everything else is overhead. And there are a huge number of potential improvements in each of these steps that can affect your health, your satisfaction and your compliance with the advice you eventually get from the doctor.
This is a simple measure that can be done easily with an integrated patient care system (i.e., time of first clinical input minus time of first administrative input). Simple yes, but for a number of reasons, it is a highly effective measure.
Run Chart Fundamentals
First, a quick description of run charts. Run charts have been in use for nearly a century, so there is a great deal of information available on the Web about how and why to use them. In the references listed at the end of this article, I have included some sources that are pretty good. Figure 1 is a picture of a basic run chart.
Figure 1: Basic Run Chart
For our purposes, we need to focus on eight run chart fundamentals:
- Variable (y-axis). This is the measure being tracked. It needs to be controllable to some degree and relevant to the reader of the chart. Using our example variable (time to first clinical entry), it would be relevant to the clinicians, the administrative staff and to the patients who see it. In addition, these same people (even the patients) can see how their actions contribute to the performance being produced. There is no sense in posting any run charts where people wonder why they are looking at it.
- Time (x-axis). This is the timeline for taking the measurements. It should align with the pace of change of the variable. For our variable, an appropriate measurement frequency might be weekly average time to first clinical entry. This measure is pretty dynamic, so improvement efforts should be made often.
- Data Points. These are the intersections of the variable across time (for instance, average weekly time to first clinical entry).
- Average. The standard average used is the mean of all of the data points shown on the run chart. Other measures of central tendency such as the median or mode are really more appropriate for other types of analyses.
- Upper and Lower Control Limits. These lines mark the boundaries that are three standard deviations above and below the average. They are used to separate data points that are in control (normal within a stable system) versus those that are out of control (not normal within a stable system).
- Points inside the control limits. Any points inside the control limits represent normal variation within a stable process. Fluctuations up or down represent statistical noise. Chasing these fluctuations and attempting to act on them is a waste of time. In other words, you will never be able to flatten out the line completely.
- Points Outside the Control Limits. Any points outside the control limits represent the existence of a special cause. They are not noise, but the bang of a gong that something has happened that was not a normal part of the process. These points need to be investigated and acted upon. This is where the value of run charts is greatest in terms of focusing management attention and action.
- A New Average. A second chart (see Figure 2) shows the effect of a disruption in the previously stable process. If we caused the disruption, then this is a good thing, and we want to measure the impact and test this disruption in other parts of the organization. Policy changes, procedural changes and automation are examples of deliberate disruptions. To paraphrase W. Edwards Deming, 94% of all improvement is in the hands of top management (through policy changes, procedural changes, investments in automation, etc.).
Figure 2: The Effect of Disruption
Run charts are usually borne of quality improvement efforts such as lean events, resulting in a high degree of ownership by those tracking and posting the measurements. But they can also be high maintenance tools to keep up to date using ad hoc reports and manual tracking. Plus, they are often not integrated with other data sources or with other areas of the organization, which limits their effect on the enterprise. Traditional run charting efforts might be best described as pilot projects.
Using Business Intelligence to Feed Run Charts
The first applications of run charts in most organizations cover small, defined data points in a single subject area. In addition, every clinic is likely to be measuring different variables, or measuring similar variables using different criteria and business rules. For instance, one clinic might measure the time to first clinical entry from the beginning of the registration process while another might measure it from the end of the registration process. This is fine for early efforts, but limits the value of improvement efforts across the organization.
In order to get this value, tapping into your business intelligence data repositories would be a wise move. Doing so would give your run charts a boost by allowing you to drill into formal hierarchy. For instance, begin by evaluating performance at the corporate level, then drill down by clinic group (region, district) and then to the individual clinic. Plus, if you have data at the department level, you can drill to it. Furthermore, with data at the specific person level (PCP, PA, NP, RN), you can measure performance and standardize practice.
With data at the individual visit level, you can drill across various subsets that don't follow strict hierarchical lines. Some examples include looking at:
- Days of the week (Are we better on Tuesdays than Saturdays?)
- Seasons of the year (Are we better at back to school time?)
- Patient groups (Are we better with diabetes patients than the general public?)
- Demographics (Are we better with kids, adults, elderly, etc.?)
- Geographic areas (Are we better in rural areas, upscale urban neighborhoods, downtown?)
Slicing and dicing our data this way leads to earlier identification and intervention of special causes that are not part of the normal functioning of a stable process. In addition, they provide us with ideas to focus our improvement efforts (i.e., deliberate disruptions).
The real power from using business intelligence data as a feeder for run charts comes from measuring variables that are tracked by different sources. One example is combining revenue data with service metrics, which can improve performance on revenue per hour. With run charts, the entire organization can see improvement over time (or not) and can see how their improvement efforts contribute to the overall good of the organization. In short, the effect of new programs and different methods becomes powerfully visible.
Using Run Charts to Feed Business Intelligence
Business intelligence data is very useful in supporting statistical analysis, giving you the ability to slice, dice, sort and sum your analysis. But the real power of merging run charts with business intelligence comes from using the insights gained from the charts in your other business performance analysis applications.
Let's say you analyze your run charts and find these situations:
- Variation across Organization. While the corporate average is fairly stable (12 minutes), there is wide variation of averages across your organization (ranging from 6 minutes to 18 minutes). In this situation, the slow group needs to be sent to learn the secrets of the fast group. Then, they need to document what they find so that all of the other clinics can change to the faster process. This documentation then becomes part of your organization’s business intelligence.
- Special Cause Tracking. Another type of documentation you want to add to your intelligence base are the special causes for points that fall outside of the control limits when the process is stable. It could be that a group of new nurses comes in each June after graduation; and as they learn processes and procedures, overall performance is negatively impacted. Or it could be that the systems get overloaded right before quarter end that slows people down. That's on the negative end. On the positive side, a data point could fall outside the control limit because you recently hired a new PA who is very efficient. The added person and increase in skill level is a special cause, and this needs to be documented. This could be happening across the organization, and publishing this special cause could lead others to find explanations faster.
- Spreading Deliberate Disruptions. Still another situation you need to document is a major disruption and its impact. Adding this documentation and its measurement to your improvement effort evidence base is useful for promoting similar efforts across the organization for marketing purposes, for supporting funding requests and for justifying investments that lead to increased revenue and profitability.
If business intelligence is using the collective knowledge of your organization and of your environment to find and exploit opportunities, then it is imperative that you record your history of special causes and disruptions that are highlighted by your run charts.
Merge your business intelligence with your statistical analysis. Both will be significantly improved by this merger. Then, create deliberate disruptions. Using the data you already own gives you the ability to find, test, judge and even predict the outcomes of the disruptions you introduce into the system. Your organization, your stakeholders, your staff and, of course, your patients will be better for it.
Thanks for reading!
Schmidt S, Kiemele M, Berdine R, Knowledge Based Management: Unleashing the Power of Quality Improvement, Colorado Springs: Air Academy Press, 2005.
Deming, W E, Out of the Crisis, The MIT Press, 2000.
NIST/SEMATECH e-Handbook of Statistical Methods, September 30, 2007.