RE: control charts in health care

[Thomas Raithby <traithby@mhsqa.com>]

Mike,

I'm new to the list and I've missed some of this discussion - so if it's
already been covered, please excuse my reply.

Most North American Emergency Medical Services (EMS) systems measure and
report on their response times. May provide the service under contract to a
municipality or health region and the contract includes incentives and
penalties relating to response times.

Although some services still report on average response times, most report
on the 90th percentile. For example, the response time on emergency calls is
8 minutes or less, 90% of the time. It's not the perfect measure. It's
described as an indicator because we think that systems sophisticated enough
to meet this response time standard are demonstrating good operational
practice. And from a consumer's perspective, it's somewhat reassuring to
know that if I call for an emergency ambulance, I can expect the paramedics
to arrive at my door in 8 minutes or less 90% of the time. It's a response
time reliability issue.

Response time is just one of many performance indicators in EMS. In addition
to operational performance indicators like time in facility, time on-scene,
transport time, hours between critical vehicle failures, etc, there are a
host of clinical performance indicators. Many clinical indicators pertain to
intermediate outcomes, like improvements to vital signs, protocol compliance
etc. Unfortunately, the better outcomes (like length of hospital stay,
survival at 30 days etc are nearly impossible to link due to patient
confidentiality issues. It is a big challenge.

I'm using MiniTab 14 to monitor response times in an X-Bar S chart now. I'm
using it on a daily basis with 18 to 30 cases per day, and looking at 10
weeks data at a time. A confounder I have is that the X-Bar S chart plots
means and control limits. But it has become the defacto standard to report
the 90th percentiles.

Is there a way I can use a process control chart to monitor the 90th
percentile and have Minitab plot (I know this sounds bizarre) the mean 90th
percentile with UCL and LCL? Is that a mathematically valid approach? Has
this approach been taken elsewhere?

Regards,

Thomas Raithby, BSc EMT-2 Consultant and Field Coordinator 
tel: (506) 649-2597 
fax: (506) 649-2529 
Mobile Health Services Quality Agency  www.MHSQA.com 
133 Prince William Street, Suite 802 
Saint John, NB E2L 2B5
Canada