[EpiData-list] Charting SPC (statistical process control) data with EpiData Analysis

epidata-list at lists.umanitoba.ca epidata-list at lists.umanitoba.ca
Sun Nov 9 16:09:29 CST 2008

Shakeel asked on the list: "Is there any
way I can create SPC Charts that show actual counts (e.g of MRSA)
against targets?"

There are three SPC strategies here, depending on the frequency of MRSA infections:
1. If rare: Record date of incidence (e.g. dateinci) and save in a rec file, then in analysis
    (latest test release):
    read mrsadata ;  Gchart dateinci  /t

2. If a certain proportion of patients. Record dates (e.g. hospdate) number of infections (count) and total volume of patients (total):
   read mrsadata ; pchart hospdate count total /t

3. If you only have the counts, but not the totals, you can either do a runchart or ichart:
   read mrsadata ; runchart hospdate count /t ; Ichart hospdate count /t;

In other words it depends on the nature of the contact to the patients, 
the similarity of the underlying data in subgroups etc. 

For adding the target an option with the graphs is required, e.g.:
Say the target was 1 % of all patients with a certain infection:

pchart hospdate count total /t /yline=0.1 /sub="Target indicated at Y=10%"

A good advice is always regardless of the above, that a lot can often be learned by just plotting the raw data on a time scale. 
A "manual" for SPC is underway, but meanwhile read some of the textbooks available.


Jens Lauritsen
EpiData Association

More information about the EpiData-list mailing list