Dear Kelsey,

Based on the surveys in the United States and as mentioned in american cancer society, only 5% of females younger than 40 years old were diagnosed with breast cancer, and from that age it has exponentially increasing manner until the age 80. So, even considering interval lengths by 20 years won't be strange if we talk about only female population.

My first suggestion is to categorize your sample by gender and race. For example, african-americans tend to have breast cancer in younger ages much more than white-americans. So, the age intervals can be changed accordingly. 

In any case, considering the interval length as 5 years seems to be wiser than 10 or 20 years. In the most of studies related to breast cancer, the first class is 0 - 19, not 0 - 18. So, you might want to change that in order to have better comparison with other similar studies. 

I hope it was helpful. Good luck. 

Thank you Mani! That was extremely helpful. I will update the first age class to include 19, and I will look more into categorizing our sample by gender and race.

We are going to circulate this survey on social media and in our network, so we may not be able to pick and choose respondents. So we may aim to choose the most broadly relevant age categories. Thank you for giving me some food for thought. I appreciate it!

I would use 

Less than 40

40-50 

50-60

60-70

70+

Gareth Evans 

If I correctly understand your aim " To describe or quantify survivorship among patients with breast cancer, the classification need to be logical, informative and fits the study aim. Regarding age, I agree with Mani to use five-year interval. You may start with age groups:

15-19

20-24

25-29.................... 80+

 If you do not have cases aged less than 20 years then you can skip the first group and start from 20-24. The pattern of survival may be examined for patients by age, gender, race, geographic areas and any other interesting variable. Once you have done age pattern of survival, then you may decide to keep it as 5-year or the pattern may become clearer by using 10-year interval:

20-29

30-39.............80+

Remember that the larger the sample you study the easier is to be liberal in using small class interval

I can see why you may want to examine the data in terms of age groups, but usually I set up surveys simply asking participants their age. Then you can group people into the age categories of interest in your datafile by adding an extra variable, but you also have the freedom to use age as a continuous variable in analyses where this would be of interest.