Table 1

Examples of SPC chart performance

Chart construction parametersPercentage of signals detected by clinical severity (1=low, 4=high)
Chart typeBaseline dataBaseline windowBaseline lagEWMA weight λ
or MA span
Control limit widths (σ)1234
MAHospital303200.20.10
EWMAHospital630.20.583.984.385.986.2
EWMAHospital18FW0.8421.816.719.137.9
S+WEDICON1824N/A185.983.594.896.6
S−WEHospital1824N/A243.245.454.669.0
MAHospital1231230.43.312.027.6
MAHospital924331.99.924.937.9
EWMADICON24120.6414.38.420.165.5
MADICON9241233.23.521.848.3
S+WEDICON12FWN/A248.840.167.289.7
EWMADICON2400.4319.514.042.175.9
MADICON18FW60.535.139.077.486.2
  • Charts clustered into 12 groups based on the proportions of signals detected for each clinical severity. For illustration, chart types, construction parameters and percentage of signals detected for each clinical severity category are shown for one representative chart per group in increasing performance (in terms of percentage of type 3 and 4 signals detected, and type 1 and 2 signals not detected).

  • λ, EWMA weight parameter; σ, estimated SD; DICON, Duke Infection Control and Outreach Network; EWMA, exponentially weighted moving average; FW, fixed window;MA, moving average; N/A, not applicable; SPC, statistical process control; S+WE and S−WE, Shewhart p-chart with and without Western Electric rules, respectively.