# Thread: Medical stats - which statistical test to use? Stuck!

1. ## Medical stats - which statistical test to use? Stuck!

Dear all,

I need some help with determining what statistical test to use for the following data.

I have a column for age of patient and I want to compare this statistically with another column where it is stated YES if a infection occured or NO if no infection occured. I have grouped the ages into under 25, 26-35, 36-45......etc and I want to see if there is a statisically significant variation beween the different age groups and the frequency of infections (yes/no) in each age group.

For example

under 25: 23 no infection, 3 yes infection
26-35: 18 no infection, 2 yes infection
36-45: 22 no infection, 0 yes infection
46-55: 28 no infection, 4 yes infection
56-65: 3 no infection, 15 yes infection
66-75: 23 no infectin, 2 yes infection

Similarly, I have a column of surgical procedures and I want to also compare this statisically with a column for infection (yes/no) to see if there is a statistical variation between the 30 surgical procedures and the number of infections.

Also , I have a column of 23 countries of residence and I would like to compare this statically with the number of infections.

Could someone kindly assist me with this? Which stats test should I use?

Many thanks!

2. ## Re: Medical stats - which statistical test to use? Stuck!

For the age/infection as you have them organized - into age groups - chi-square goodness of fit would work, with age groups as the categories and "proportion having an infection" as the proportions.

Or if you used age/infection and did not group the ages, you could do logistic regression with age as the predictor and infection (yes/no) as the DV.

You didn't give a lot of detail about the surgical procedures and countries of residence variables, but they both sound like nominal variables in which case chi-quare GOF test works for those analyses as well.

3. ## Re: Medical stats - which statistical test to use? Stuck!

Dear bruin,

With regards to Chi square goodness of fit, would it still be appropriate if more than 20% of variables had a frequency of less than 5?

With regards to procedure and country of origin the data is like.

Procedures

Upper GI surgery: 123 no infection 5 yes infection
Lower GI surgery: 130 no infection 3 yes infection
Urology surgery: 46 no infection 1 yes infection
Orthopaedic surgery: 8 no infection 0 yes infection

Countries
Argentina 1 no infection 0 yes infection
Burma 2 no infection 0 yes infection
Cyprus 6 no infection 1 yes infection
UK 128 no infection 6 yes infection
America 60 no infection 3 yes infection

There are several frequencies that may be 0 or small numbers.

Would a Chi squared GOF test be okay?

Thank you!

4. ## Re: Medical stats - which statistical test to use? Stuck!

That's a good question; no, it may not be appropriate to do a Chi-Square GOF when the expected frequencies are very low.

Check these out, they may help:

http://www.theanalysisfactor.com/obs...est-no-biggie/

http://www.biostathandbook.com/small.html

 Tweet

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts