Pareto Charts

A Pareto chart provides facts needed for setting priorities. It organizes and displays information to show the relative importance of various problems or causes of problems. It is essentially a special form of a vertical bar chart that puts items in order (from the highest to the lowest) relative to some measurable effect of interest: frequency, cost, time. The chart is based on the Pareto principle, which states that when several factors affect a situation, a few factors will account for most of the impact. The Pareto principle describes a phenomenon in which 80 percent of variation observed in everyday processes can be explained by a mere 20 percent of the causes of that variation.  Pareto charts help improvement teams identify the most important causes of variation.

Placing the items in descending order of frequency makes it easy to discern those problems that are of greatest importance or those causes that appear to account for most of the variation. Thus, a Pareto chart helps teams to focus their efforts where they can have the greatest potential impact.

When to Use a Pareto Chart

Pareto charts help teams focus on the small number of really important problems or causes of problems. Pareto charts are useful in establishing priorities by showing which are the most critical problems to be tackled or causes to be addressed. Comparing Pareto charts of a given situation over time can also determine whether an implemented solution reduced the relative frequency or cost of that problem or cause.

Pareto Chart example

Here is an example of a Pareto chart:

How to Use a Pareto Chart

Step 1. Develop a list of problems, items, or causes to be compared.

Step 2. Develop a standard measure for comparing the items.

  • How often it occurs: frequency (e.g., utilization, complications, errors)
  • How long it takes: time
  • How many resources it uses: cost

Step 3. Choose a time frame for collecting the data.

Step 4. Tally, for each item, how often it occurred. Then add these amounts to determine the grand total for all items. Find the percent of each item in the grand total by taking the sum of the item, dividing it by the grand total, and multiplying by 100.

Tallying Items in a Compilation Table

Causes for sample not processed

Number of Occasions

Percentage

No pipettes

9

7

No reagents

18

13

Sample clotted

16

12

Insufficient sample

2

1

Sample haemolyzed

7

5

Machine broken

82

61

 

 

100

 

Step 5. List the items being compared in decreasing order of the measure of comparison: e.g., the most frequent to the least frequent. The cumulative percent for an item is the sum of that item’s percent of the total and that of all the other items that come before it in the ordering by rank.

Arranging Items in a Compilation Table

Causes for sample not processed

Number of Occasions

Percentage

Cumulative Percentage

Machine broken

82

61

61

No reagents

18

13

75

Sample clotted

16

12

87

No pipettes

9

7

93

Sample haemolyzed

7

5

99

Insufficient sample

2

1

100

 

 

100

100

 

Step 6. The horizontal axis of a graph should have columns for all the items listed from highest to lowest. Label the left vertical axis with the items, then label the right vertical axis with the cumulative percentages (the cumulative total should equal 100 percent).

Step 7. The cumulative percentages are illustrated by a line graph. The first point on the line graph should line up with the top of the first bar.

Step 8. Analyze the diagram by identifying those items that appear to account for most of the difficulty. Do this by looking for a clear breakpoint in the line graph, where it starts to level off quickly. If there is not a breakpoint, identify those items that account for 50 percent or more of the effect. If there appears to be no pattern (the bars are essentially all of the same height), think of some factors that may affect the outcome, such as day of week, shift, age group of patients, home village. Then, subdivide the data and draw separate Pareto charts for each subgroup to see if a pattern emerges.

Points to Remember

Try to use objective data instead of opinions and votes.

ASSIST publication: 
no
Facebook icon
Twitter icon
LinkedIn icon
e-mail icon