Welcome to the NEW CHUG— your ultimate destination for cutting-edge insights and discussions at the intersection of healthcare and technology.

Our platform is dedicated to exploring the dynamic landscape of healthcare technology, empowering professionals, enthusiasts, and innovators alike with the knowledge and resources to navigate the rapidly evolving field.

From telemedicine revolutionizing patient care to the transformative potential of artificial intelligence in diagnostics, we delve into a myriad of topics shaping the future of healthcare delivery. Whether you’re a seasoned healthcare IT professional, an aspiring entrepreneur in the digital health space, or simply curious about the latest innovations, our curated content, expert interviews, and thought-provoking articles provide a comprehensive exploration of healthcare technology’s most pressing issues.

Join us as we navigate the forefront of innovation, driving forward the transformation of healthcare for a healthier, more connected world.

Have questions or suggestions? Please use the contact button below.

Notifications

Clear all

Pooled Risk Equations- a little higher math help?

Trading Post

Last Post by Anonymous 1 second ago

11 Posts

4 Users

0 Reactions

87 Views

RSS

czelnick

(@czelnick)

Posts: 31

Trusted Member

Topic starter

The ACC/AHA recently revised lipid/cholesterol guidelines, and have thrown out the Framingham risk calculators we all know and love in Centricity, recommending we replace them with the Pooled Cohort Risk Calculations.

They have helpfully published their methodology on-line here:

http://content.onlinejacc.org/.....id=1770220

page 37-40 or so

I could deal with the exponent of "e" part of these equatoins, and even the need to build large arrays to set up the calculator...

BUT the data to be input requires LOGARITHMs to be extracted from the patient data of age, sex, bp, cholesterol values, etc.

Does anyone know how to calculate a Log in MEL? Its only been 45 years since I had high school math, and I'm not sure I remember how to crunch it out, and I can't find a LOG functoin in my MEL documentation..

Thanks.

Charles Zelnick ME

Stonington ME

Posted : December 7, 2013 5:45 am

czelnick

(@czelnick)

Posts: 31

Trusted Member

Topic starter

Alright, it bugs me that there are apps on the iPhone for this but nothing in Centricity.

so because I knew there was a way to do this, (thank you Mr Potter, R.I.P.), I rooted around online and found a polynomial expansion that will calculate the log. Here is the code below. because you need more interations to get accuracy for larger numbers, the loop goes one order of magnitude beyond the value of z, which comes out fine in when i check this in the spreadsheet.

Next for the arrays...

Anyone want to help me tackle this?

{fn ln(z) {

double a b c d f z

a=(z-1)/z

d=a

b=a

f=z*10

for t=2, t<f, t=t+1

b=b*a

c=(1/t)*b

d=d+c

endfor

return d}

}

Paste this in a letter and you can test some numbers as below:

Natural Ln of 50: {ln(50)}

Natural Ln of 5: {ln(5)}

Ln of 200: {ln(200)}

Ln of 2000: {ln(2000)}

Posted : December 8, 2013 6:50 am

StephenAdamsMD

(@stephenadamsmd)

Posts: 23

Eminent Member

Did you find any takers willing to help with the coding? I wonder if another option is to build a risk calculator outside of the EMR and use {RUNSHELLOPEN()} to feed it the values (age, cholesterol, etc) and have it return the calculated risk. (Feel free to poke holes in the idea - I don't know if if would work or not.)

Stephen Adams

Posted : December 18, 2013 1:19 am

joeg1962

(@joeg1962)

Posts: 843

Noble Member

Runshellopen ??

Is there a good reference for the MEL commands? This command, as well as "Mel_send_flag" are but a couple of the commands that I have never read anywhere. I am always eager to try new approaches, and maybe even help with this.

However, I am not even aware of the runshellopen command to know what I can or cannot do (parameters, etc..)

Thanks.

Posted : December 18, 2013 3:09 am

StephenAdamsMD

(@stephenadamsmd)

Posts: 23

Eminent Member

In the help files under "using data symbols" look for "MEL utility functions." RUNSHELLOPEN passes arguments to an outside program and looks for a returned value. The only place I've used it is within quicktexts to open a web browser to a specific page (e.g. a drug formulary page, etc.)

Place this text in a quicktext and give it a try.... (if you like pizza)

{RUNSHELLOPEN(" http://order.papajohns.com/index.html ")}

Posted : December 18, 2013 4:43 am

czelnick

(@czelnick)

Posts: 31

Trusted Member

Topic starter

I think I sorted this, by converting the equation to MEL and outputting it in a text component. It seems to work.

I need to further release it and test it within my organization before I can share it here.

The ln function was the key above; otherwise, just sorting out if the patient is male/female, caucasian/African American; then the rest of it is just crunching the coefficients in 4 equations.

Cheers.

Charles Zelnick MD

Posted : January 19, 2014 6:29 am

StephenAdamsMD

(@stephenadamsmd)

Posts: 23

Eminent Member

Excellent! Please post a solution when you get it worked out.

Steve

Posted : January 22, 2014 11:58 am

glenntuckermd

(@glenntuckermd)

Posts: 9

Active Member

czelnick said:

Alright, it bugs me that there are apps on the iPhone for this but nothing in Centricity.

so because I knew there was a way to do this, (thank you Mr Potter, R.I.P.), I rooted around online and found a polynomial expansion that will calculate the log. Here is the code below. because you need more interations to get accuracy for larger numbers, the loop goes one order of magnitude beyond the value of z, which comes out fine in when i check this in the spreadsheet.

Next for the arrays...

Anyone want to help me tackle this?

{fn ln(z) {

double a b c d f z

a=(z-1)/z

d=a

b=a

f=z*10

for t=2, t<f, t=t+1

do

b=b*a

c=(1/t)*b

d=d+c

endfor

return d}

}

?

Paste this in a letter and you can test some numbers as below:

Natural Ln of 50: {ln(50)}

Natural Ln of 5: {ln(5)}

Ln of 200: {ln(200)}

Ln of 2000: {ln(2000)}

Thank you so much for posting this! I'm just starting on integrating the ACA/AHA guidelines. The lack of support for logarithms in MEL is frustrating.

Posted : April 17, 2014 5:04 am

StephenAdamsMD

(@stephenadamsmd)

Posts: 23

Eminent Member

We are in the process of rolling out a risk calculator to our users. We addressed the lack of higher order mathematical functions in MEL by passing the relevant parameters to a program written in C++; the results are written to a temporary file and retrieved by Centricity via the {RUNTEXTPROCESS()} function. The MEL to launch it looks like this:

{arg=DURATIONDAYS(str(PATIENT.DATEOFBIRTH),str(._TODAYSDATE))+"|"+PATIENT.SEX+"|"+LASTOBSVALUE("BP SYSTOLIC")+"|"+LASTOBSVALUE("HDL")+"|"+LASTOBSVALUE("CHOLESTEROL")+"|"+USERYESNO("Current smoker?")+"|"+USERYESNO("Diagnosed with diabetes?")+"|"+USERYESNO("On BP med?")+"|"+get(PATIENT.RACE,1);RUNTEXTPROCESS(DIR_EMR()+"\scripts\pooled_risk_calculator.exe",arg)}

Posted : December 31, 2014 8:11 am

czelnick

(@czelnick)

Posts: 31

Trusted Member

Topic starter

Ok, I built this last year and it has been working in our system.

You must be a subscriber to the CCC forms and functions, as the code calls those functions.

Pasted below is a text component which we call with a quicktext (.pccvr) -- it will insert information about the patients CV risk and current treatment into the form.

This text component contains 2 important functions - one is the function [ln(z)] which computes a natural log of a number. Had to go back to my high school math over 45 years ago to figure that out.

The Pooled Cohort Cardiovascular Risk calculation [PCCVRisk()] then uses that ln function to compute the patients risk.

I have attached 3 tools that use this code: 1st a Handout that explains the pros and cons of statins; 2nd, a HISTORY page, for easy lookup of cardiovascular information including risk, and finally our current lab letter which fires the code if there are lipid test results.

Statin Decision Balance_4 code fix BHMH Cardiovascular 20141023 Lab Letter 2014_3

Good luck with this, your mileage may vary.

Cheers and Happy New Year.

Charles Zelnick

Island Family Medicine, Stonington, ME

==============code for text component=======================

{fn CVRsk() {
local cvr=""
If match(Toupper(OBSANY("SMOK STATUS")),"CURRENT")>0 then cvr=cvr + "Smoking/Tobacco Use" else "" endif
If OBSANY("HX OF HTN")=="yes" then cvr=cvr + "
Hypertension" else
If ccc_sys_hasdx("ICD-401") then cvr=cvr + "
Hypertension" else "" endif
endif
If ccc_sys_hasdx("ICD-585^ICD-403^ICD-404") then cvr=cvr + "
Chronic Kidney Disease" else "" endif
If ccc_sys_hasdx("ICD-250^ICD-249") then cvr=cvr + "
Diabetes Mellitus" else "" endif
If OBSANY("FH<55MALE MI")=="yes" then cvr=cvr + "
Family History of Heart Disease in a Male before Age 55" else "" endif
If OBSANY("FH<65FEMALMI")=="yes" then cvr= cvr + "
Family History of Heart Disease in a Female before Age 65" else "" endif
If OBSANY("HDL") <40 then cvr=cvr + "
Low HDL, below 40" else "" endif
If OBSANY("CARDCASCORE") <> "" then cvr=cvr + "
Cardiac Calcium Score= " + OBSANY("CARDCASCORE") + " [Normal <300 or <75%]" else "" endif
If OBSANY("CTCARTCASCO") <> "" then cvr=cvr + "
Cardiac Calcium Score= " + OBSANY("CTCARTCASCO") + " [Normal <300 or <75%]" else "" endif
If OBSANY("CRP") <> "" then cvr=cvr + "
C-Reactive Protein(CRP)= " + OBSANY("CRP") + " [Normal <2 mg/ml]" else "" endif
If OBSANY("CRP HI SENS") <> "" then cvr=cvr + "
C-Reactive Protein(High Sensitivity CRP)= " + OBSANY("CRP HI SENS") + " [Normal <2 mg/ml]" else "" endif
If OBSANY("CRP QUANTI") <> "" then cvr=cvr + "
C-Reactive Protein(Quantitative CRP)= " + OBSANY("CRP QUANTI") + " [Normal <2 mg/ml]" else "" endif
If PATIENT.SEX == "M" then
If PATIENT_AGE() >45 then cvr=cvr + "
Male, Age >45" else "" endif
else
If PATIENT_AGE ()>55 then cvr =cvr + "
Female, Age >55" else "" endif
endif
If cvr=="" then cvr="No Major Cardiac Risk Factors Recorded" endif
return cvr
}
}{fn ln(z) {
double a b c d f z
if z<1/2 then return "Cannot find ln of number below 0.5" endif
a=(z-1)/z
d=a
b=a
f=z*10
for t=2, t<f, t=t+1
do
b=b*a
c=(1/t)*b
d=d+c
endfor
return d}
}{fn PCCVRisk() {
local insum,e,w,r,q,s,t,MCV,PCRisk,SMOKSTATUS,DIABSTATUS

cond
case (PATIENT_AGE())<21
return "Patient too young to calculate risk" break
case (PATIENT_AGE())>79
return "Patient too old to calculate risk" break
case (OBSANY("CHOLESTEROL"))==""
return "NO Total Cholesterol value in flowsheet" break
case (OBSANY("HDL"))==""
return "No HDL value in Flowsheet" break
case (OBSANY("BP SYSTOLIC"))==""
return "No Systolic BP value in Flowsheet" break
case (PATIENT_AGE())>20

lAGE=ln(PATIENT_AGE())
lCHOL=ln(OBSANY("CHOLESTEROL"))
lHDL=ln(OBSANY("HDL"))
lBPS=ln(OBSANY("BP SYSTOLIC"))

If match(Toupper(OBSANY("SMOK STATUS")),"CURRENT")>0 THEN SMOKSTATUS=1 ELSE SMOKSTATUS=0 ENDIF
If ccc_sys_hasdx("ICD-250.^ICD-249.") then DIABSTATUS=1 else DIABSTATUS=0 ENDIF

If PATIENT.SEX=="M" then
If MATCH(PATIENT.RACE,"Black")> 0 then MCV=19.54 s=0.8954 //BLACK MALE

If (ccc_sys_hasdx("ICD-401"))=="FALSE" then
insum=2.469*lAGE+0.302*lCHOL+(-0.307)*lHDL+1.809*lBPS+0.549*SMOKSTATUS+0.645*DIABSTATUS

Else
insum=2.469*lAGE+0.302*lCHOL+(-0.307)*lHDL+1.916*lBPS+0.549*SMOKSTATUS+0.645*DIABSTATUS

Endif
Else
MCV=61.18 s=0.9144 //WHITE MALE

If (ccc_sys_hasdx("ICD-401"))=="FALSE" then
insum=12.344*lAGE+11.853*lCHOL+(-2.664)*(lAGE*lCHOL)+(-7.99)*lHDL+1.769*(lAGE*lHDL)+1.764*lBPS+7.837*SMOKSTATUS+(-1.795)*lAGE*SMOKSTATUS+0.658*DIABSTATUS
Else
insum=12.344*lAGE+11.853*lCHOL+(-2.664)*(lAGE*lCHOL)+(-7.99)*lHDL+1.769*(lAGE*lHDL)+1.797*lBPS+7.837*SMOKSTATUS+(-1.795)*lAGE*SMOKSTATUS+0.658*DIABSTATUS

Endif
Endif
Else //FEMALE

If MATCH(PATIENT.RACE,"Black")> 0 then MCV= 86.61 s=0.9533 //BLACK FEMALE

If (ccc_sys_hasdx("ICD-401"))=="FALSE" then
insum=17.114*lAGE+0.94*lCHOL+(-18.92)*lHDL+4.475*(lAGE*lHDL)+27.820*lBPS+(-6.087)*(lAGE*lBPS)+0.691*SMOKSTATUS+0.874*DIABSTATUS
Else
insum=17.114*lAGE+0.94*lCHOL+(-18.92)*lHDL+4.475*(lAGE*lHDL)+29.291*lBPS+(-6.432)*(lAGE*lBPS)+0.691*SMOKSTATUS+0.874*DIABSTATUS

Endif
Else MCV=(-29.18) s=0.9665 //WHITE FEMALE

If (ccc_sys_hasdx("ICD-401"))=="FALSE" then
insum=(-29.799)*lAGE+4.884*(lAGE*lAGE)+13.540*lCHOL+(-3.114)*(lAGE*lCHOL)+(-13.578)*lHDL+3.149*(lAGE*lHDL)+1.957*lBPS+7.574*SMOKSTATUS+(-1.665)*lAGE*SMOKSTATUS+0.661*DIABSTATUS

Else
insum=(-29.799)*lAGE+4.884*(lAGE*lAGE)+13.540*lCHOL+(-3.114)*(lAGE*lCHOL)+(-13.578)*lHDL+3.149*(lAGE*lHDL)+2.019*lBPS+7.574*SMOKSTATUS+(-1.665)*lAGE*SMOKSTATUS+0.661*DIABSTATUS

Endif
Endif
Endif

e=2.71828
w=insum-MCV
r=e^w
q=s^r
PCRisk=(1-q)*100
OBSNOW("PCCVR10YRRSK",str(PCRisk))
t= "Based on: Total Cholesterol = " + (OBSANY("CHOLESTEROL")) + " and HDL= " + (OBSANY("HDL")) + " and Systolic BP = " + (OBSANY("BP SYSTOLIC")) + fmt( "
10-Year Pooled Cohort Cardiovascular Risk = ","B") + PCRisk +"%"
return t
endcond}
}{fn sys_filter_all(GCIlist) {

// Returns list of Name and date of current meds in allergy list
// that match GPI fragments in list.
// Returns null otherwise.

local meds=getfield(ALL_AFTER("DELIMITED"),"|","")
local types=getfield(GCIlist,"^","")
local r="", i=0, j=0
for j=1, j<=size(meds), j=j+1 do
for i=1, i<=size(types), i=i+1 do
if match(CCC_sys_dp(meds[j],6),types[i])==1
then if r<>"" then r=r+hret endif
r=r+CCC_sys_dp(meds[j],1)+" ("+CCC_sys_dp(meds[j],2)+")" break
else continue
endif
endfor
endfor
return r }
}
Cardiovascular Risk Analysis
{ if ccc_sys_hasdx("ICD-410^ICD-411^ICD-413^ICD-414^ICD-V45.8^ICD-429.2^ICD-440^ICD-441^ICD-442^ICD-43^ICD-443.89^ICD-443.9^ICD-443.2^ICD-V15.1") then "A Cardiovascular Disease diagnosis is on the Problem List" else "
Major Cardiac Risk Factors:
" + CVRsk() endif + "

" + cond
case CCC_sys_filter_meds("3940")<>""
"Current Statin on Medlist:
" + CCC_sys_filter_meds("3940")+ "
"
case OBSANY("NOTONSTATIN")<>""
"Past history Statin declined or contraindicated
"
case sys_filter_all("3940")<> ""
"Allergy list includes a Statin
"
else
"No Statin on Med or Allergy List and No documented contraindication
"
endcond + (If OBSANY("CHD 10YR RSK")<>"","
Framingham 10 Year Cardiac Risk was " + OBSANY("CHD 10YR RSK") + " on " + LASTOBSDATE("CHD 10YR RSK") + "
", "") + "POOLED COHORT 10 Year Cardiac Risk is
" + PCCVRisk()
}

Posted : January 3, 2015 8:54 am

StephenAdamsMD

(@stephenadamsmd)

Posts: 23

Eminent Member

You did a brilliant job coding this. Thanks for sharing! One quick suggestion: The current version doesn't detect Vytorin as a statin but adding an additional case statement detects that combination: (the case CCC_sys_filter_meds("3999") is what I added.

You have the following Major Cardiac Risk Factors:
" + CVRsk() endif + "
" + cond
case CCC_sys_filter_meds("3940")""
"-Current Statin on Medlist:
" + CCC_sys_filter_meds("3940")+ "
"
case CCC_sys_filter_meds("3999")""
"-Current Statin on Medlist:
" + CCC_sys_filter_meds("3999")+ "
"
case OBSANY("NOTONSTATIN")""
"-Past history Statin declined or contraindicated"
case sys_filter_all("3940") ""
"-Allergy list includes a Statin"

Thank you,

Steve

Posted : January 5, 2015 10:23 am

24 Forums
9,940 Topics
39.9 K Posts
0 Online
6,470 Members

Forum Icons: Forum contains no unread posts Forum contains unread posts

Topic Icons: Not Replied Replied Active Hot Sticky Unapproved Solved Private Closed