Chapter 17 Pareto Charts for Ranking Problems
The Pareto chart, named after Vilfred Pareto, was invented by Joseph M. Juran as a tool to identify the most important causes of a problem.
For this example, we use the dataset on adverse events causing harm to patients, collected using the Global Trigger Tool method (Plessen, Kodal, and Anhøj 2012).
## 'data.frame': 131 obs. of 2 variables:
## $ severity: chr "E" "F" "E" "F" ...
## $ category: chr "Pressure ulcer" "Gastrointestinal" "Infection" "Infection" ...The paretochart() function (from qicharts2) takes a categorical vector as argument and plots a Pareto chart as demonstrated in Figure 17.1.
Figure 17.1: Pareto chart of patient harm.
The bars show the count in each category, and the curve shows the cumulated percentage over categories. Almost 80% of harms come from 3 categories: gastrointestinal, infection, and procedure.
Figure 17.2 is a Pareto chart of harm severity demonstrating that nearly all events resulted in temporary harm (E-F).
Figure 17.2: Pareto chart of harm severity: E-I, where E-F = temporary harm, G-H = permanent harm, and I = fatal harm.
The paretochart() function takes a character or factor vector, but often data have already been aggregated into tabular format:
category count
1 Fall 1
2 Gastrointestinal 40
3 Infection 34
4 Medication 18
5 Other 4
6 Pressure ulcer 5
7 Procedure 29To make a Pareto chart from tabular data, we first need to convert data back into a vector. This can be achieved with the rep() function repeating each category by its count:
# make vector from counts
ae.cat <- rep(ae.tbl$category, ae.tbl$count)
# show first six rows of vector
head(ae.cat)## [1] Fall Gastrointestinal Gastrointestinal Gastrointestinal
## [5] Gastrointestinal Gastrointestinal
## 7 Levels: Fall Gastrointestinal Infection Medication Other ... Procedure
Figure 17.3: Pareto chart constructed from tabular data.
In conclusion, the Pareto chart is useful for identifying the most common causes of a problem. Often most of the problems are caused by relatively few of the causes. In the example above eliminating gastrointestinal harm (most often obstipation) and hospital associated infections would more than half the rate of adverse events.