In 2004, nine hospitals in Michigan began implementing a cleanliness process in their intensive care units. Three months after it began, the process had cut infection rates of ICU patients by 60%. Within 18 months of adoption, these hospitals had saved $200M according to a WHO report. Best of all, this single intervention saved the lives of more than 1,500 people in just a year and half. After so much success, the check list was published as the Keystone ICU project.
The Keystone ICU project was simple, yet powerful. It aimed to identify a simple 5-step process to help lower and mitigate infection when hospital staff had to put a line into the patient. To gain awareness, the 5-step process was typed onto a piece of paper and displayed in visible locations so every medical professional could see and read it. The check list included the following:
It’s worth noting that the above steps were never new. They were basic steps that every medical professional were taught in their first few days of medical school. Yet they were critical in the project’s success, and instrumental in saving 1500 lives.
Today, there’s nearly a 0% error rating when administering lines in patients. Think about that– there were no new technical innovations, no new pharmaceutical discoveries, and no new cutting edge procedures involved. It was just the time to identify the existing 5-step process, put pen to paper and display it in an area as a reminder to doctors not to skip any steps. Talk about impact!
The above technique is called Poka-Yoke – a Japanese term that means “mistake proofing” or “inadvertent error prevention.” By using a checklist, errors can be detected and prevented before reaching the customer, or patient, in this example. This approach was used successfully in the automobile manufacturing industry when the first auto-transmission vehicles were introduced in 1940. There were many deaths and injuries due to unintentional gear selection by drivers.
At the time, General Motors used a poke-yoke to fix the issue by adding a switch that required the car to be in park or neutral before it could be started. It added an additional “mistake proof approach” when it required drives to press the brake pedals before shifting the gear from park/neutral to drive, eliminating the possibility of the mistakes it had witnessed just years earlier. The curved gear shift in today’s car is one additional poke-yoke that prevents accidental gear shifts.
Poka-Yoke’s Application to Data Quality
The above poka-yoke approach to mitigate mistakes and ensure success can be leveraged in data management to prevent data quality issues. Creating the right data quality checks and balances across the entire data flow can help ensure the right data is processed, as well as eradicate propagation of error data further downstream. Automated data quality business rules (a.k.a. data controls) institute a repeatable and standardized process that is application independent to monitor, measure, and validate 100% of your data by strategically placing data controls at key junctures within your data process. Data controls that systematically reconcile data across multiple process touch points prevent data error propagation from occurring and deliver trustworthy data required to make confident business decisions.
It’s critical to identify a benchmark to compare data quality checks to a known good data element to ensure conformance to expected values, and to automate validation of the data points against the benchmark through templated data controls within the process. It’s this due diligence that applies the poka-yoke technique to data quality. The ability to introduce inter system and intra system balancing of data points in a complex data flow with multiple aggregation points is a crucial prevention check list that yields a “mistake proof” data flow. This is required to safeguard that only accurate data moves through the process and any irregularities are promptly identified as data quality issues.
Tracking end-to-end data flows is an instrumental element that is performed by monitoring multiple data control points in order to prepare an end-to-end exceptions report which aids as a tool to prevent data error propagation across the process while also building trust and data assurance to promote active data use within the end user community.
Applying Poka-Yoke Starts with a Shift in Thinking
The problem experienced by data management professional today is, “Everybody in data management already knows about the data quality issue.” However; does there exist a poka-yoke opportunity which changes the conversation to, “Everyone already does something about preventing the data quality issues.” This change begins when we institute a solution that is no longer implemented occasionally, but rather is implemented consistently. The above example of the Keystone ICU project clearly demonstrates the importance of implementing a checklist that is used daily by all stakeholders. Every physician knew about the steps, but they needed a check list to remind them about doing the right thing.
With data quality, the issue remains that no one person owns the problem. That’s similar to the Keystone project. One person took on the issue and created a solution. While finding an enterprise-wide data quality solution is difficult, it might be even more difficult to identify a stakeholder willing to take on the responsibility. But it doesn’t have to be a single person. It can be a group that applies poka-yoke principles that become a culture that changes the norms to consistently apply repeatable data quality processes.
Once that person or group is found, identifying the poka-yoke solution comes next. What can be implemented to help stop the data inconsistencies and quality issues plaguing the organization? The first step begins by articulating the root of the data quality problem(s). Are systems seeing issues with data movement between data hops? If yes, it’s critical to implement an enterprise data analysis platform that can monitor data flow as it moves across the enterprise. In this instance, it’s critical to find a platform that will automate the monitoring of data with business rules that span the end-to-end business process so 100% of transactions are monitored. Implementing this step will ensure that data is complete, checking to make sure that data type conformance, value conformance and data consistency at its origination are diligently checked and completed. Then data missing values can be immediately flagged and an exception management process can promptly fix data errors. It’s only at this point that you can confidently say that you’ve put in an “error prevention” system which ensures maximum data quality.
Unfortunately most organizations don’t have just one data quality problem. That’s why it’s necessary to scope the breath of challenges faced in order to find a comprehensive solution that can span across an array of applications which comprise a business process.
Understanding the missing link in data governance to transform your data quality initiatives which follow the spirit of the poka-yoke technique are outlined in this white paper
For a deeper dive into this topic, visit our resource center, where you will find a broad selection of content that represents the compiled wisdom, experience, and advice of our seasoned data experts and thought leaders.