Sample size calculation using absolute precision.

[anna]

I want to be clear about a basic query with reference that  how much absolute precision can be taken for sample size calculation for prevalence studies.

It is as follows "I want to determine the prevalence of depression in geriatric population using Geriatric Depression Scale (GDS) in a rural community of India"

Assuming a prevalence of 12%, significance of 5% and abs. precision of 3%.

[BXC (Bendix Carstensen)]

A rough approximation to the c.i. of a proportion (prevalence) is
p \pm 2*sqrt(p*(1-p)/n)

So if you want an absolute precision of 3% (assuming that this means a confidence interval of the form p \pm 3% you are facing to solve:

sqrt(p*(1-p)/n) = 0.015

for p=0.12.

But the most illustrative would be to do this calculation for a range of prevalences:

n <- 100:1000
p <- seq(6,18,3)/100
prc <- function(n,p) 2*sqrt(p*(1-p)/n)
aa <- outer( n, p, FUN=prc )
matplot( n, aa, type="l", lty=1, col="black", lwd=2,
xlim=c(100,1100), xlan="N", ylab="Precision (2*sd)" )
text(1050,aa[length(n),],p*100,adj=1)
abline(h=0.03,col="red")

______________________________________________

Bendix Carstensen
Senior Statistician
Epidemiology

Steno Diabetes Center A/S
Niels Steensens Vej 2-4
DK-2820 Gentofte
Denmark
+45 44 43 87 38 (direct)
+45 30 75 87 38 (mobile)
b...@steno.dk http://BendixCarstensen.com
www.steno.dk



________________________________

--
To post a new thread to MedStats, send email to MedS...@googlegroups.com .
MedStats' home page is http://groups.google.com/group/MedStats .
Rules: http://groups.google.com/group/MedStats/web/medstats-rules

[Kornbrot, Diana]

Try g*
Best
diana

---

Emeritus Professor Diana Kornbrot
email:  d.e.ko...@herts.ac.uk    
web:    http://dianakornbrot.wordpress.com/
Work
School of Psychology
 University of Hertfordshire
 College Lane, Hatfield, Hertfordshire AL10 9AB, UK
 voice:   +44 (0) 170 728 4626
   fax:     +44 (0) 170 728 5073
Home
 19 Elmhurst Avenue
 London N2 0LT, UK
    voice:   +44 (0) 208  444 2081
    mobile: +44 (0) 740 318 1612

[anna]

Thanks Diana and Bendix, i got the ans using OpenEpi.

[Sreenivas.V]

Generally I recommend in my work, a 10% relative error margin, which in your case translates into 1.2% or 0.012 in absolute terms.  Incidentally this is close to what Bendix suggested.

 

Sreenivas

 

--

To post a new thread to MedStats, send email to MedS...@googlegroups.com .
MedStats' home page is http://groups.google.com/group/MedStats .
Rules: http://groups.google.com/group/MedStats/web/medstats-rules

--
V. Sreenivas
Additional Professor
Department of Biostatistics
All India Institute of Medical Sciences
New Delhi 110029

[Anna]

Thanks Sreeniwas, In first instance i find your answer absolutely right. But I studied lots of epidemiological studies using which used precision 1-5%, even 10% (i think 10% in those with proportions are high e.g. 40%, 50% etc.).

We want to use 3-4% absolute precision for feasibility purpose, can our idea be maintained?

Please help us.