analysis

Published: 22/04/2004, Volume II4, No. 5902 Page 19

There is increasing interest in the use of statistical process control charts in the management of healthcare.

SPC was originally developed as an approach to quality control and safety in manufacturing and engineering, but there is now widespread recognition of its value as a way of monitoring healthcare systems.

SPCs can be used to monitor systems to provide the earliest warning that something is moving out of acceptable control limits. They can also be used to measure the impact of any changes to a process or system.

In 2001 problems were identified with the stroke service at Worcestershire Acute Hospitals trust, with input from the Royal College of Physicians. A series of service improvements followed, including the creation of a stroke unit and the introduction of specialist nursing staff.

Chart 1 shows the in-hospital standardised mortality ratio (SMR) for each year from 1995-96 to 2002-03. There is a downward trend in the mortality ratio over the period. However, it is not possible to say with 95 per cent certainty - the standard level of statistical certainty - that the SMRs differ from the national average SMR of 100.

This is shown on chart 1 by the two parallel horizontal lines above and below the annual recorded points.

Chart 2 is an SPC designed to identify a doubling in the odds of death. The red line moves up every time a patient dies, moving more strongly upwards if, given the age, sex and deprivation of the patient, they were less likely to die.

The chart alerts when the line passes through a threshold - the level at which the threshold is set will determine how sensitive the system is and how often it alerts purely as a result of random fluctuation. Using a value of five, as in this chart, about 1 per cent of alerts after 500 patients have been monitored are likely to be statistical flukes.

In the period from 1996-2000 there were 27 alerts.

In the period from 2001-02 there was none. Of course there are a number of ways in which this could be achieved - for example, by discharging patients before they died. However, the average length of stay of stroke patients does not fall in the last two years but instead rises slightly (chart 3).

SPC charts can be used to measure positive outcomes. At St Mary's trust, London, the SMRs for stroke were better than the national average even before the introduction of the stroke unit in September 2000. However, it is still possible to track the impact of this change using SPC.

The final chart measures survival rather than death.

The chart moves up every time a patient survives and is designed to identify a doubling in the chances of survival compared to the national average. There is a clear difference between the period before the creation of the stroke unit, in which there were no positive alerts, and the period after the introduction of the unit, in which there were eight.

These examples illustrate the way in which SPCs can be used to monitor whether the introduction of a change to the way a patient group or procedure is managed is having the desired impact on key patient outcomes. l