This algorithm is for community associated MRSA (Methicillin-resistant Staphylococcus aureus, read more at Note this algorithm will use lab results and not ICD-9 codes, as ICD-9 codes are not specific enough for this algorithm and/or are not used consistently for this phenotype.  Thus, we expect the actual number of cases to be higher than what the eMERGE RC (Record Counter) estimated, and, as we will be studying patients aged 0 to 89, we would like for all sites to participate, please... Due to initial low numbers found, there are now both gold and silver standard definitions in the algorithm in an attempt to find all the cases.

Phenotype ID: 
Do Not List on the Collaboration Phenotypes List
Type of Phenotype: 
Jennifer Pacheco and Abel Kho
Contact Author: 
Date Created: 
Wednesday, March 13, 2013
Network Associations: 
Owner Phenotyping Groups: 
View Phenotyping Groups: 

Suggested Citation

Jennifer Pacheco and Abel Kho. Northwestern University Feinberg School of Medicine. caMRSA. PheKB; 2013 Available from:


- does the algorithm begin on p. 2?  i.e. do we ignore the last paragraph of p. 1 which describes the CDC def of CA-MRSA?

- at GH we can't identify cases that occur during inpatient admit - we have outpatient lab results only.

- are there other drug terms we should consider when identifying staph aureus infections that are resistant to methicillins? (i see that cephalosporins are also considered amongst the resistant drugs in the wiki reference)



- for cases, i've made the assumption that the correct algorithm is the one

on p. 2 of the attached algorithm document.

- for cases, i've assumed that age is calculated at date of MRSA infection.

- for controls, at what point in time should age be calculated

(e.g. either now/at death)?

- for cases, i've assumed that the exclusion on prior stay in a long term care

facility is for year prior to MRSA infection only.

- for cases, i've looked only at methicillin and oxacillin resistant cultures

- for cases, assume that hx of dialysis and hx of diabetes is prior to

MRSA infection

- for controls, should that be "ever" had dialysis or diabetes?




i think male should be C46109, not C46119.


thanks for data dictionary!  i'm now clear on date to use for controls.


Thank you Jane for pointing out our mistake.  All of your assumptions are correct,

except do not use cephalosporins, only methicillin & oxacillin, which I believe you did.

the data dictionary is updated.


Should the nursing home exclusion be if they ever have mention of a nursing home, or in some temporal relationship to the qualifying MRSA lab?

Using the ICD-9 codes that you have suppiled in a literal fashion does not result in many hits.

Did you mean for the 3 digit codes to be wildcarded?

For instance:

 Carbuncle and furuncle 680

should be

 Carbuncle and furuncle 680.*

At our site, MRSA screening is handled by a PCR usually given on inpatient admission. If positive, isolation measures

may be taken. It has nothing to do with any specific would or infection, and usually there isn't one that has manifested yet.

In fact, using the PCR test as a sign of MRSA plus all of the other case criteria yields no cases at our site. On the other hand, we have

culture tests for various organisms. Based on what I've seen, the culture tests seem to fit the needs of the algorithm better than

the PCR screens. Which should I be using? Do you have LOINC codes that you can provide?

In the review  of the control definition, we are a little concerned we will miss people who had MRSA but didn't get the test here.  One thought is to remove anyone with mention of MRSA in their PL.  Do you have other suggestions?  Is this modification ok?

yes, thank you, that is a good suggestion.  If it's possible to search the problem list using keywords &/or codes,

please exclude those as controls who have MRSA in the problem list, I'll add this to the document now.



I am not sure if I am interpreting the alg correctly.  To be a case the person has to have  MRSA site be an SSTI.  But then in the dictionary you ask us to provide the site for the cases.  Won't this be 1 for all?

  All questions should now be answered or addressed by updating the algorithm documentation, 

and the data dictionary.


  The only question/comment we have not addressed yet is the request for LOINC codes

for the labs, we will get that soon.


  Regarding Geisinger's question as to culture vs. PCR screening, we want cultures,

not PCR-based assays which instead are screening for colonization.