The idea that you can get to zero defects causes people problems. It sounds impossible. Most people agree with the aim, and are happy with the idea that there are only two numbers to which you should aspire in quality – 0% and 100%, depending on how the target is worded. Aspirations are fine, but moving from aspiration to real intention often stops people short.
I visited Nissan’s Sunderland plant in July – several previous blog posts have discussed the experience. Nissan’s presentation told us that over 80% of cars passed through the plant with ‘no touch’ – i.e. they needed only the initial work, and no correction of errors. A small percentage – less than 1% – needed to be returned to a process for correction. The remainder needed some kind of remedial work, however minor, which did not delay them passing through the process.
This doesn’t sound like Zero Defects, but the issue is in the definition of ‘Defect’. Shigeo Shingo, a very big name in quality, separated mistakes, and defects. Mistakes, he felt, would always happen, although you could do your very best to minimise them. Only if mistakes were not identified and corrected before they affected the customer, did they become defects, This is very like the Nissan process above – they try not to make mistakes, but when they do, they try to correct them before they reach the customer, and without delaying the process.
Shingo is associated with Poka Yoke devices – devices that physically prevent an error. He went to great lengths in his books, however, to emphasise that Poka Yoke devices could not prevent all defects alone.
Shingo identified several activities that he believed had to happen to prevent defects, including identification of defects. He suggested identifying errors in three ways: self-inspection, sequential inspection and 100% inspection.
Self-inspection implies that there is a clear standard for the work, that the worker knows what it is, and that he or she checks their work against the standard. Sequential inspection means that the next person in the process checks the relevant aspects of the work of the previous person in the process, so that they can be assured that the work they receive has no mistakes. If a mistake is identified, it is immediately corrected.
All work requires inspection, Shingo felt, because no defect is acceptable. If a defect reaches a customer, then that person experiences 100% defects, because the only product they experience is defective. That is not acceptable, said Shingo, so all defects must be found – you cannot, he argued, rely on statistical quality control, or sampling inspection alone.
You combine this, he argued, with inspection of any error and identification and resolution of the root cause. When you find the cause, you correct the conditions that caused the problem, which may require new standard work, in order to prevent that specific problem occurring in the future.
All of these ideas were visible at the Nissan plant. In a production line that I was allowed to see, each work station had three pieces of paper: a job allocation, which referred to the process, a quality standard and required inspection which specified the standard, and explained the self-inspection they were to conduct, and a ‘neighbour inspection’,. which described the checks they were to make on the work they received. This is the standard work, self-inspection and sequential inspection describe by Shingo.
If problems arose on the production line, and the worker could not resolve them immediately, they sounded an alarm, known as an andon. I saw rooms close to the lines which has immediate response staff, and one of their jobs was to go to any andon in their section, and help to resolve the problem, and identify the cause, so that the line could start again, and problems could be prevented in the future.
In addition to this every single car was inspected. I saw manual inspections, a ‘rolling road’ test, a water tightness test deliberately aimed at breaching any vulnerable seal, and finally the cars had a test on a road circuit. This is Shingo writ large – standard work, self-inspection, sequential inspection, root cause investigation and 100% inspection of every car.
There were additional checks. I was told that the Quality Assurance department would take vehicles at random off the production line, and check the entire car for mistakes. At team stations in each zone, there were problems reported by staff on the wall, and a statement of what was being done to resolve them. The mantra here is to produce no mistakes, accept no mistakes from a previous process, and pass no mistakes on to the next process. This will not happen by accident, or by avoiding pointing out problems as they occur.
Knowing what we want to do, checking the quality of our own work, checking the quality of the work we receive, investigating and resolving root causes, and inspecting everything we produce, does offer the possibility of preventing errors turning in to defects in patient care. In this case, I believe we really can learn from industrial methods, and improve care to patients by thinking through and applying mistake proofing.