PakMediNet Discussion Forum : Biostatistics : sample size calculation
sample size calculationi am starting a study to find out the frequency of structural changes in lumosacral spine in infants with spina bifida who present with neurogenic bladder. its incidence is 1 in 100000. how to calculate the sample size for this study? can you please help me
Posted by: BVHOSPITALPosts: 131 :: 17-11-2009 :: | Reply to this Message
Re: sample size calculation
Dear sir
to calculate sample size we need population size, confidence limits,then we will be able. for further plz contact for free consultation on biostats 0300-4668681
Posted by: ibrahim_apPosts: 138 :: 17-11-2009 :: | Reply to this Message
Re: Re: sample size calculationpopulationof 2000
Posted by: BVHOSPITALPosts: 131 :: 22-11-2009 :: | Reply to this Message
Re: Re: Re: sample size calculation
when the prevalence is i out of 100000, how can you take sample from 2000. for such type of study you take cluster and then enumerate the whole population, at the end of study you fix your sample size
sorry for late reply
Posted by: ibrahim_apPosts: 138 :: 02-12-2009 :: | Reply to this Message
Re: Re: Re: Re: sample size calculationi have to submit my synopsis to RMC. ow i then calculate the sample size and what i write for this? kindly guide me.
Posted by: BVHOSPITALPosts: 131 :: 05-12-2009 :: | Reply to this Message
Re: Re: Re: Re: Re: sample size calculation
Sir please send your information again, i will send you the size.
level of significance, power of test, prevalence, population size
Posted by: ibrahim_apPosts: 138 :: 08-12-2009 :: | Reply to this Message
Re: Re: Re: Re: Re: Re: sample size calculation
i have to conduct a study to find the radiological features in MRI lumbosacral spine among paediatric patients with neurogenic bladder. the most common cause of neurogenic bladder in this population is spina bifida (90%. the incidence of spina bifida is 1 to 1.5 per 10000. we have 3000 patients each year and i have to find the sample size among these 3000 patients. last year we had 28 patients with neurogenic bladder. my study duration will be 6 to 9 months. kindly help me.
[Edited by BVHOSPITAL on 09-12-2009 at 03:21 PM GMT]
[Edited by BVHOSPITAL on 15-12-2009 at 05:26 PM GMT]
Posted by: BVHOSPITALPosts: 131 :: 09-12-2009 :: | Reply to this Message
Re: Re: Re: Re: Re: Re: Re: sample size calculationSir This a rare disease, you please all the patients. you fix the time then take all the patients. 0300-4668681
Posted by: ibrahim_apPosts: 138 :: 09-12-2009 :: | Reply to this Message
Re: sample size calculation
First, with the information provided it is not possible to calculate sample size. But even a bigger question is what is the study question? From the whole thread, I am not clear what exactly you want to study. May be presenting your question slightly more clearly may help in figuring out sample size (I think once your question is clear, you will find that sample size calculation will become easy).
Here is so far what I understand, you want to study the frequency of lumboscaral spine features as evident on MRI in infants presenting with neurogenic bladder. Right? But what does it tells... nothing... nothing at all unless we can compare it with something. Let me explain you; lets say that you find that 20% of infants with neurogenic bladder has some abnormality on MRI. It does not tell us whether this is more than what is found in other hospitalized infants or less. You don't have a control group. Now if you have a control group, and a neurogenic bladder group, and you have lumbosacral abnormalities in each group, you can easily do a chi-square test (or Exact test). For sample size calculations, you can find formula in any book or software. You will still have to come up with some numbers, but you can make an educated guess or use values given in literature.
Now lets say that you only want to study neurogenic bladder infants, the only thing you can control here is that how much you want to be sure that your calculated frequency is the true population frequency, in other words, how narrow confidence interval you want to have. The whole game is then of resources, the larger the sample, the narrower will be confidence interval.
You may still get by with having data on neurogenic bladder infants only if you assume that the prevalence of lumbosacral abnormalities in otherwise normal population is zero. Then you can test the null hypothesis that your discovered frequency is greater than zero. This is also a relatively straightforward calculation.
Bottom line, a clearer study question is important, define your exposure, define your outcome, and define your population. Have a rough estimate of the prevalence of your outcome in the exposed population and unexposed population, rest of the calculation is relatively straightforward.
Posted by: rqayyumPosts: 199 :: 18-12-2009 :: | Reply to this Message
Re: Re: Re: Re: Re: Re: Re: sample size calculation
http://www.nss.gov.au/nss/home.nsf/pages/Sample+size+calculator.
can this link be used to calculate sample size?
If i take confidence level 95, proportion is 0.8, confidence interval is 0.141 and standard error is 0.072, then the sample size is 31. is it accurate or there is some error in this calculation? kindly guide me.i shall be highly obliged if you rectify any error in calculation
Posted by: BVHOSPITALPosts: 131 :: 27-12-2009 :: | Reply to this Message
Re: sample size calculationWhat is your research question because that is going to determine sample size. Please be very clear with your research question. If the only thing you want to get is the frequency, you don't need to have sample size. In fact, you can't have a sample size until you want to do something more than calculating frequencies.
Posted by: rqayyumPosts: 199 :: 27-12-2009 :: | Reply to this Message
Re: Re: sample size calculationwe want to find the MRI changes in spinal cord and lumbosacral vertebra. it will also give us the causes of neurogenic bladder in this population. kindly guide
Posted by: BVHOSPITALPosts: 131 :: 29-12-2009 :: | Reply to this Message
Re: sample size calculationI really don't think that you understood what I am asking. Anyway, enlist all abnormalities that you will be looking at and their expected proportions.... if that is too much, write one primary abnormality on MRI that you will be looking at and the expected proportion in the population under study. Give me your sample size. write down your hypothesis and null hypothesis.
Posted by: rqayyumPosts: 199 :: 30-12-2009 :: | Reply to this Message
Re: Re: Re: sample size calculation
Here is an article that discusses calculating sample size for a proportion. It may answer some of your questions.
http://www.stat.auckland.ac.nz/~wild/ChanceEnc/Ch08.psampsize.pdf
Posted by: rqayyumPosts: 199 :: 30-12-2009 :: | Reply to this Message
Re: Re: sample size calculationthe commonest abnormality on MRI is spina bifida (90%) and its incidence is 7/1000 live births. my hypothesis is that the spina bifida is the commonest abnormality on MRI LS Spine in patients with neurogenic bladder. i have to calculate the sample size and help of the friends who know statistics, is needed.
Posted by: BVHOSPITALPosts: 131 :: 31-12-2009 :: | Reply to this Message
Re: sample size calculation
Dears
I am intending to conduct a study to find out which one is faster , freehand vs. laser guide device using ultrasound on a model. only information got from previous study, freehand (median=10.5; IQR=6.5–14.5), laser device 43 (33; 21–55).
please advice on sample size, and which test i should use for comparison?
Thanks
Mahajan
[Edited by msxhj on 15-01-2010 at 09:19 PM GMT]
Posted by: msxhjPosts: 5 :: 15-01-2010 :: | Reply to this Message