QI Tools

Enter inputs

To make either a Run chart or Control chart, either:

• Replace the example data below with your data.
• Upload a csv file formatted like this example.
  
• 1st row is column names? TRUE FALSE

Columns: 1) time period, 2) events or means & STD, 3) total, 4) trial.
For trial, 0 = pre-intervention or no trial and 1 = after intervention.
Separate columns with commas. Click here to add commas.

1, 4, 30, 0, 2, 6, 30, 0, 3, 5, 30, 0, 4, 15, 30, 0, 5, 20, 30, 0, 6, 25, 30, 0

Plot type: **Select chart type Control chart (proportions) Control chart (counts) Segmented regression (proportions) Segmented regression (means) Segmented regression (counts) Run chart (avoid) Box chart for before-after (avoid)

Plot: Events Total

Outcome: nonconforming conforming means    Time period: Day Week Month Year
Goal (lower): Goal (upper):
Topic (for title - optional):

Plot logo: None KU

Update Plot (below)

Example scenarios

QI project 1: implementing screening for depression

We now have added 3 months of data after an intervention to increase screening for depression.

First analysis (box chart) Did this intervention help?

Second analysis (run chart) Did this intervention help?

Third analysis (p-chart) Did this intervention help?

• What shape would the line be if the intervention were perfect?

P-chart with longer trial Did this intervention help?

• What is different now?

Fourth analysis (segmented and multivariable regression) Huh?

• What does this chart add?
• What are the significant findings?

Stronger trial wih regression analysis (note the events dropped from 15 to 9 in the 4th row) What is significant if the drop is more abrupt?

Longer trial with regression analysis What is significant now?

Why control for coincident interventions or trends over time? Answer.

To control for these, use of these analyses:

• Regression analysis
• Segmented time series regression

1. Fan E, Laupacis A, Pronovost PJ, Guyatt GH, Needham DM. How to use an article about quality improvement. JAMA. 2010. PMID: 21098772

QI project 2: speeding antibiotics in sepsis

We have 3 months of of baseline data than 3 months of data after an intervention.

First analysis (interrupted time series) Did this intervention help?

 

Hint: if bad spacing between title/footer text and the plot, click 'png' file type in upper right of plot. Then after the image opens in a new browser window, alter the image's dimensions in the url of the window. Click here to see screen capture of an example. 

Interpreting control charts

14 data points should be present to best use these rules

1. Are all points within the 3-sigma control limits? If so, the process is in control.
2. Is there a signal that an effective change has occurred (if a prior QI effort has been done)?
   • Astronomical Point - A dramatically different value (red points)
   • Shift - 6 or more consecutive points above or below the median (orange points)
   • Trend - 5 or more consecutively increasing or decreasing points (blue line)
   • Runs - Are there too many or too few for just common cause variation (not implemented)?
   • From: Run Chart Tool. Institute for Health Care Innovation (IHI)
3. More detail from Wikipedia: Control Charts: Western Electric rules and Nelson rules