Process Control Charts
In the current healthcare environment, providers and healthcare organizations strive to do more with less and to make their current resources such as physicians, hospitals, systems and sites more effective. It is often difficult however, to determine whether or not a current process is running within acceptable parameters or is need of intervention. Process control charts are a tool that allows healthcare organizations to judge how effectively their current processes are functioning and determine whether or not minor tweaking or a major overhaul is required.
Process Control Charts, also known as Shewart Charts are a statistical process tool used to determine if a process is functioning within a state of control. Process Control Charts were developed by Walter A. Shewhart, an American physicist, engineer and statistician in the 1920s, as a tool to improve telephone transmission quality at Bell Labs. (Deming, 1967) Process Control Charts are used to determine if a process is currently under control. Under control means that the output of a process is stable and no adjustments to the parameters of the process are required. Process Control Charts can also be used to identify the source of variation. (Graban, 2017)
There are 2 types of Process Control Charts; univariate and multivariate. A univariate chart tracks one quality characteristic vs time or the sample. For example, a univariate control chart could be used to track Emergency Department wait times over time, with the average wait time in minutes on the Y axis and the day of the week on the X axis. Multivariate Process Control Charts on the other hand, track a statistic that summarizes 2 or more quality characteristics. For example, a radiology department may want to track the amount of radiation delivered to a patient over time. The quality characteristic, amount of radiation is actually the product of 3 different characteristics, namely exposure time, number of frames and other factors such as the patient’s weight. (Wheeler, 1997)
A Process Control Chart consists of 3 features, a center line, an upper control limit and a lower control limit. The center line is the mean value of the quality characteristic being tracked. The upper control limit and lower control limit represent 3 standard deviations above and below the center line. If a process fluctuates between the upper and lower control limit, it is considered to be in control. If the process wanders outside of the upper and lower control limits, then it is considered out of control and corrective measures should be taken. (Berardinelli, 2015)
A process mapped to a Process Control Chart will exist within one of 4 states:
- Ideal State-The process is operating within the expected parameters, at 100% conformance and is producing the expected product.
- Threshold state-The process is still stable and operating within statistical control, but still produces the occasional non-conforming result.
- Brink of Chaos-A process at the Brink of Chaos is not in statistical control, but still meets customer expectation.
- State of Chaos-State of Chaos means that the process is operating outside of statistical control and does not meet customer requirements.
To understand these 3 states, let’s return to our Emergency Room waiting time example. By measuring the wait time over an extended period, we determine that the mean wait time for a patient to enter the emergency room is 2 hours and a standard deviation of 30 minutes. The upper control limit can be calculated by multiplying the standard deviation of 30 minutes by 3 and adding it to the center line or mean, for a total wait time of 3.5 hours. The lower control limit can be computed by subtracting 3 standard deviations from the mean wait time, for a total of 30 minutes. Next, observe and chart the emergency room wait time for 7 days.
- Ideal State-All patients are seen by an Emergency Room Doctor after waiting no more than 3.5 hours or no less than 30 minutes.
- Threshold State-For the most part, patients are being seen in the ER after waiting between 30 minutes and 3.5 hours. Occasionally, patients may have to wait more than 3.5 hours, or less than 30 minutes, but the wait times are predictable. For example, on Fridays and Saturdays, when emergency department utilization is high, wait times may occasionally exceed 3.5 hours.
- Brink of Chaos-Frequently patients may have to wait more than 3.5 hours or less than 30 minutes, but the wait times are not predictable. This indicates that unknown factors are causing wait times to fluctuate.
- State of Chaos-Wait times very frequently veer outside of the upper and lower control limits and there is no way to predict patient wait times. As a result, emergency room patients may indicate a high level of dissatisfaction.
Process Control Charts are a vital tool to improve processes in healthcare. The correct application of process control charts can be used to detect variables and improve performance in a variety of processes and departments. St Andrews War Memorial Hospital in Queensland Australia used data mining and Process Control Charts to study how long patients had to wait in the Emergency Department, before being admitted to the hospital as inpatients. This wait time is also referred to as Length of Stay (LoS). Long LoS were causing a backup in the Emergency Department, as well as poor patient satisfaction and stress and dissatisfaction on the part of the hospital staff. St Andrews War Memorial Hospital discovered that the hospital admission process was causing a significant delay in LoS.
Quartz Clinical is an advanced healthcare data analytics platform developed by Quartz, a leader in healthcare analytics. Quartz Clinical combines an Electronic Health Record (EHR) platform, quality reporting system and quality and improvement tools. Most importantly, Quartz Clinical uses artificial intelligence and data analytics to produce Process Control Charts natively making it easy to identify outlier processes at a glance.
A major metropolitan medical center built a heart and vascular center. Using Quartz Clinical, they discovered that the rate of return to the ER and the readmission rate for the heart and vascular center were higher than expected and out of compliance. Armed with the data, the medical center was able to reduce the readmission rate by 20% and the rate of return to the ER by 33%.
Process Control Charts are a critical tool to help healthcare organizations identify areas for improvement and operate more effectively. Quartz Clinical collects data from a variety of sources and creates Process Control Charts natively, making it easier and more efficacious to create and study process control charts.
Wheeler, D. J. (n.d.). SPC Toolkit. Retrieved May 02, 2018, from https://www.qualitydigest.com/dec97/html/spctool.html
“Control Limits – the Key to Control Charts.” QI Macros for Excel, www.qimacros.com/free-excel-tips/control-chart-limits/.
Berardinelli, Carl. “Control Limits – the Key to Control Charts.” QI Macros for Excel, Six Sigma, www.qimacros.com/free-excel-tips/control-chart-limits/.
Lean Improvement Tools, www.hasc.org/sites/main/files/lpc_module_10_lean_improvement_tools_revised_march_15_2013.pdf.
Adams, B., Woodall, W. H., & Benneyan, J. C. (2011). The Use of Control Charts in Healthcare. Retrieved May 4, 2018, from http://www.coe.neu.edu/healthcare/pdfs/publications/C13-Use_of_Control_C.pdf
Graban, M. (2017, January 31). How to Create a Control Chart for Managing Performance Metrics. Retrieved May 5, 2018, from https://blog.kainexus.com/continuous-improvement/how-to-create-a-control-chart-for-managing-performance-metrics
W. Edwards Deming (1967) “Walter A. Shewhart, 1891–1967”, American Statistician 21: 39–40.