Portal de Periódicos da CAPES

Use of Shewhart's technique

2001; Elsevier BV; Volume: 358; Issue: 9280 Linguagem: Inglês

10.1016/s0140-6736(01)05661-6

1474-547X

Mohammed A. Mohammed, Kar Keung Cheng, Andrew Rouse, Tom Marshall,

Medical Coding and Health Information

R Henderson (May 12, p 1528)1Henderson AR Use of Shewhart's technique.Lancet. 2001; 357: 1528Summary Full Text Full Text PDF PubMed Scopus (1) Google Scholar and Tim Wilson and colleagues (May 12, p 1529)2Wilson T Holt T Greenhalgh T Use of Shewhart's technique.Lancet. 2001; 357: 1529Summary Full Text Full Text PDF PubMed Scopus (1) Google Scholar highlight the importance of previous work in the use of Shewhart's technique, in response to our report.3Mohammed MA Cheng KK Rouse A Marshall T Bristol, Shipman, and clinical governance: Shewhart's forgotten lessons.Lancet. 2001; 357: 463-467Summary Full Text Full Text PDF PubMed Scopus (184) Google Scholar We unequivocally acknowledge the work of others in this area (especially workers they mention) and feel there is a need for a systematic survey of the application of Shewhart's approach to health care. We are undertaking such a survey, to see who did what, and to learn more about the shortcomings of control charts in health care.We see the work of Shewhart (and Deming4Deming WE Out of the crisis. Massachusetts Institute of Technology, Boston1986Google Scholar) not so much as a means of refining widget production2Wilson T Holt T Greenhalgh T Use of Shewhart's technique.Lancet. 2001; 357: 1529Summary Full Text Full Text PDF PubMed Scopus (1) Google Scholar but primarily as a means of delivering clinical governance, and we attempted to relate the two.We agree with Wilson and colleagues that processes need to be changed when there is only common-cause variation. We tried, perhaps insufficiently, to emphasise this in our report. Under the headings “common-cause and special-cause variation” and variation “cannot be eliminated”, we underscore the need for action on the system to reduce common-cause variation. This message is repeated in case study 4: neonatal deaths for which only common-cause variation was seen.Wilson and colleagues allege that we had to find the right epoch groupings before we found the right results for Bristol Royal Infirmary. This is incorrect. The epochs were originally devised by Spiegelhalter and colleagues (reference 9 in our report) as part of their statistical evidence to the Bristol Inquiry. We aimed not to find the right results for Bristol, rather to guide action to improve paediatric cardiac surgery for all centres by acting on signals indicating special-cause variation.We agree with Wilson and colleagues that statistics alone could not detect a murderer such as Dr Harold Shipman. Shewhart's approach enables us to identify when we should act to find and eliminate special-cause variation.Henderson makes the point that additional rules have been shown to improve the sensitivity of Shewhart control charts. We did not emphasise this. However, we note that Shewhart himself was aware of the need for additional rules,5Shewhart WA Statistical method from the viewpoint of quality control. Dover Press, New York1987Google Scholar and that our case studies show there is much mileage in the use of Shewhart's primary rule for detecting special causes. R Henderson (May 12, p 1528)1Henderson AR Use of Shewhart's technique.Lancet. 2001; 357: 1528Summary Full Text Full Text PDF PubMed Scopus (1) Google Scholar and Tim Wilson and colleagues (May 12, p 1529)2Wilson T Holt T Greenhalgh T Use of Shewhart's technique.Lancet. 2001; 357: 1529Summary Full Text Full Text PDF PubMed Scopus (1) Google Scholar highlight the importance of previous work in the use of Shewhart's technique, in response to our report.3Mohammed MA Cheng KK Rouse A Marshall T Bristol, Shipman, and clinical governance: Shewhart's forgotten lessons.Lancet. 2001; 357: 463-467Summary Full Text Full Text PDF PubMed Scopus (184) Google Scholar We unequivocally acknowledge the work of others in this area (especially workers they mention) and feel there is a need for a systematic survey of the application of Shewhart's approach to health care. We are undertaking such a survey, to see who did what, and to learn more about the shortcomings of control charts in health care. We see the work of Shewhart (and Deming4Deming WE Out of the crisis. Massachusetts Institute of Technology, Boston1986Google Scholar) not so much as a means of refining widget production2Wilson T Holt T Greenhalgh T Use of Shewhart's technique.Lancet. 2001; 357: 1529Summary Full Text Full Text PDF PubMed Scopus (1) Google Scholar but primarily as a means of delivering clinical governance, and we attempted to relate the two. We agree with Wilson and colleagues that processes need to be changed when there is only common-cause variation. We tried, perhaps insufficiently, to emphasise this in our report. Under the headings “common-cause and special-cause variation” and variation “cannot be eliminated”, we underscore the need for action on the system to reduce common-cause variation. This message is repeated in case study 4: neonatal deaths for which only common-cause variation was seen. Wilson and colleagues allege that we had to find the right epoch groupings before we found the right results for Bristol Royal Infirmary. This is incorrect. The epochs were originally devised by Spiegelhalter and colleagues (reference 9 in our report) as part of their statistical evidence to the Bristol Inquiry. We aimed not to find the right results for Bristol, rather to guide action to improve paediatric cardiac surgery for all centres by acting on signals indicating special-cause variation. We agree with Wilson and colleagues that statistics alone could not detect a murderer such as Dr Harold Shipman. Shewhart's approach enables us to identify when we should act to find and eliminate special-cause variation. Henderson makes the point that additional rules have been shown to improve the sensitivity of Shewhart control charts. We did not emphasise this. However, we note that Shewhart himself was aware of the need for additional rules,5Shewhart WA Statistical method from the viewpoint of quality control. Dover Press, New York1987Google Scholar and that our case studies show there is much mileage in the use of Shewhart's primary rule for detecting special causes.