adjusted Confidence intervals for age, gender, region standardized prevalence rates

[Philmsk]

Hi,

I am trying to obtain the adjusted confidence intervals for prevalence rates of a disease standardized by Region, Gender and Age.

I was able to calculate the crude prevalence rates and obtained the weights from the 2016 National Census. 

I am wondering if anybody can help me with the confidence interval calculation for the adjusted prevalence rates.

This is how the sample data looks like with thousands of observations.

SAS code:

Sas code;

PROC MEANS DATA=T1 FW=8 MAXDEC=2 CLM MEAN SUM noprint;
CLass REGION GENDER AGECAT;
VAR EVENT;
OUTPUT OUT=Dat_SUM SUM= EVS;
RUN;

DATA CALC1 ;
SET dat_SUM (rename=(_FREQ_=TOTN));
length CIS $19.;
If gender=. then Gender=3; If Agect=. then Agect=9;
/************ Prev rate and conf intervals for crude prevalence rates**********/
If EVS>0 then do; p=EVS/TOTN;
PREV=ROUND(100000*EVS/TOTN,.01);
F = log(p)-log(1-p);
SE = 1/SQRT(p*(1-p)*TOTN);
Lf = F - TINV(1-(0.05/2), TOTN-1)*SE;
Uf = F + TINV(1-(0.05/2), TOTN-1)*SE;
Lp = ROUND((exp(Lf)/(1+exp(Lf)))*100000,.01);
Up = ROUND((exp(Uf)/(1+exp(Uf)))*100000,.01);
CIS='['||STRIP(PUT(Lp,6.2))||', '||STRIP(PUT(Up,6.2))||']';
end;
/**** adjusted rates using the national weights******/
PREV_Ad=ROUND(PREV*weight,.01);
/**** how to calculate the CI's for adjusted rate ???????******/
run;

Thanks in Advance.

Phil

[Reeza]

There’s a PROC for that. PROC STDRATE. 

---

@Philmsk wrote:

Hi,

I am trying to obtain the adjusted confidence intervals for prevalence rates of a disease standardized by Region, Gender and Age.

I was able to calculate the crude prevalence rates and obtained the weights from the 2016 National Census. 

I am wondering if anybody can help me with the confidence interval calculation for the adjusted prevalence rates.

This is how the sample data looks like with thousands of observations.

SAS code:

Sas code;

PROC MEANS DATA=T1 FW=8 MAXDEC=2 CLM MEAN SUM noprint;
CLass REGION GENDER AGECAT;
VAR EVENT;
OUTPUT OUT=Dat_SUM SUM= EVS;
RUN;

DATA CALC1 ;
SET dat_SUM (rename=(_FREQ_=TOTN));
length CIS $19.;
If gender=. then Gender=3; If Agect=. then Agect=9;
/************ Prev rate and conf intervals for crude prevalence rates**********/
If EVS>0 then do; p=EVS/TOTN;
PREV=ROUND(100000*EVS/TOTN,.01);
F = log(p)-log(1-p);
SE = 1/SQRT(p*(1-p)*TOTN);
Lf = F - TINV(1-(0.05/2), TOTN-1)*SE;
Uf = F + TINV(1-(0.05/2), TOTN-1)*SE;
Lp = ROUND((exp(Lf)/(1+exp(Lf)))*100000,.01);
Up = ROUND((exp(Uf)/(1+exp(Uf)))*100000,.01);
CIS='['||STRIP(PUT(Lp,6.2))||', '||STRIP(PUT(Up,6.2))||']';
end;
/**** adjusted rates using the national weights******/
PREV_Ad=ROUND(PREV*weight,.01);
/**** how to calculate the CI's for adjusted rate ???????******/
run;

Thanks in Advance.

Phil

---

[StatDave]

If those weights are survey weights, then what you might need is the SURVEYFREQ or SURVEYLOGISTIC procedure.