# How to calculate the age range of participants in a published study?

#### Boon

##### New Member
Hello there

I've have read some published papers which related to psychological interventions in health care settings. I want to exclude the papers that include participants aged over 65, but many of the papers only tell me that participants were aged "18+". Is there a way to calculate the actual age range of participants in the sample? The papers generally include sample size, mean age and standard deviation.

To try and solve this problem, I have asked a supervisor who told me to "Google it". I googled it but didn't find how how to solve this. I have also tried to contact our stats dept but it's closed for Easter and I really need to work this out as soon as possible. If anyone can advise me, I would be most grateful.

#### hlsmith

##### Not a robit
Most conclusive thing you can do is just to contact the corresponding authors for the articles!

#### j58

##### Active Member
If the mean age plus 2 standard deviations is less than 65, then there are likely to be at most a few subjects over 65. That, I'd say, is the best you can do from just the mean and SD.

#### Boon

##### New Member
Most conclusive thing you can do is just to contact the corresponding authors for the articles!
Yes, I will do this. I also need a plan for how to interpret the papers where authors don't respond.

#### Boon

##### New Member
If the mean age plus 2 standard deviations is less than 65, then there are likely to be at most a few subjects over 65. That, I'd say, is the best you can do from just the mean and SD.
Thank you. I'd just welcome a little more guidance on how to calculate this please. Examples of the information available to me is:

Paper 1: n=137, mean 31.47 years (SD 11.65)
Paper 2: n=24, mean age 43.07 (SD 14.07)

#### hlsmith

##### Not a robit
Step one contact authors.

Step two run j58 formula.

Step three use your own content knowledge and the study setting description to weigh in on the likelihood.

There is no way to know if a couple of 65 YO were in the study beyond contacting the others. Were the studies registered on clinical trials.gov? And what is the big deal if a couple older individuals were in the study, does it complete!y negate the generalizability to your intended use?

#### j58

##### Active Member
If the mean age plus 2 standard deviations is less than 65, then there are likely to be at most a few subjects over 65. That, I'd say, is the best you can do from just the mean and SD.
Thank you. I'd just welcome a little more guidance on how to calculate this please. Examples of the information available to me is:

Paper 1: n=137, mean 31.47 years (SD 11.65)
Paper 2: n=24, mean age 43.07 (SD 14.07)
You multiply the SD by 2 and add it to the mean.

#### Boon

##### New Member
You multiply the SD by 2 and add it to the mean.
A ha! Thank you so much. Now that level of maths I can definitely manage!

#### Boon

##### New Member
Step one contact authors.

Step two run j58 formula.

Step three use your own content knowledge and the study setting description to weigh in on the likelihood.

There is no way to know if a couple of 65 YO were in the study beyond contacting the others. Were the studies registered on clinical trials.gov? And what is the big deal if a couple older individuals were in the study, does it complete!y negate the generalizability to your intended use?
Thanks for your advice - I like your suggestions. In terms of whether it negates generalisability, it's just that I'm looking at these intervention papers for a systematic literature review so I'm trying to be as thorough and consistent as possible. Thanks again.

#### Boon

##### New Member
You multiply the SD by 2 and add it to the mean.
Hmmmm one paper has a mean = 20 with a SD of 54.1. If I follow your formula that makes 128.2 years old. Do you mean divide the SD by 2?

#### hlsmith

##### Not a robit
The calculation is a general generic formula that works for normally distributed data. If data is skewed you can get wonky estimates. Your values are bounded by 18 and human longevity. Ages can be skewed if you have some older individuals, just like mean salaries can be skewed if you have some millionaires in your sample. You could try simulating some lognormal distributions given parameters and constraints.

#### j58

##### Active Member
Hmmmm one paper has a mean = 20 with a SD of 54.1. If I follow your formula that makes 128.2 years old. Do you mean divide the SD by 2?
I meant the formula I gave you. A distribution of human ages with mean 20 years and SD 54 years is impossible.

#### Boon

##### New Member
I meant the formula I gave you. A distribution of human ages with mean 20 years and SD 54 years is impossible.
It turns out I was reading the stats from the gender line not the age line!! It makes much more sense when I look at the correct figures: mean age 38.9 (SD 16.1).