i believe that age group for adults may not differ much in age as long as weight do not vary so much,because the pathological process does not differ much ,rather than age ,instead the weight matters more.secondly your exclusion criteria is far more important,removal of expose persons,old and  elderly,immuno-compromised  and those and medications of any form or special diet ,these may affect your rate on immune response of the studies cell series.  age group between 20-40 is not out of line but weight between 40-60 is appreciable.

I see a couple of issues in your matching and selection of controls.  One thing to be aware of is that when you match on a variable, e.g., age or gender, ideally you already know that those factors make a difference in relation to what you are studying so you want your results of comparing the cases and controls to be able to put those factors aside and look at other things.  In short, once you have matched on a characteristic, you no longer can study that characteristic in relation to the outcome (being a case).  One risk in neighborhood controls is that you may be introducing misclassification.  If a lot of people in that neighborhood have, like the cases, been exposed to tubercle bacilli - if the exposure is most likely to have occurred in the neighborhood, they could have the same characteristics as cases, indeed they could have subclinical cases of TB, and could have the same immune responses.  Therefore, misclassification diminishes the likelihood that you will show a difference between cases and controls for your factor of interest even though there actually is a difference.  Sometimes that risk is worth taking.  I once did a case control study using as controls siblings of cases who were within the age range of the cases.  In that instance there was not a 1-to-1 match, but the overall group of cases and controls were within the same age group.  There was a real risk that the controls could have been subclinical cases, but my reasoning was it was convenient to use that control group, and IF I showed a difference (I did), to the degree there was a bias in the selection of controls it would decrease the estimated magnitude of the effect but be hard to criticize the notion that there was a real effect.

Thank you very much Prof Schoenbaum for your helpful reply. I agree that once we match for age and sex we won't be able to study those variables in the analysis. However as my study is looking at immune responses in TB-helminth co-infected individuals, and both variables can influence immunity, I'm pretty sure I want to control for them as confounders. The rate of TST-positivity in my study population is high (reportedly up to 50%) so we will be taking account of TST results when analysing the results, and it will be interesting to see the difference in immune responses in the two control groups.  . 

You've stated exactly the reason for matching on age and sex, i.e., you already know that there is a relationship between those and immune response and it isn't the objective of your study.  That is the principle that should be applied to any characteristic one chooses to match.  

Great, many thanks for the advice, much appreciated.

Interesting topic.  As is typical, the answer will depend on your research question.  It isn't clear to me whether the cases in your study are of TB disease or latent TB, and if they are co-infected with helminths.  The strongest risk factor for all of these is probably foreign birth, so if you select controls from the general population you will likely find this to be the strongest association in your study.  So you may want to match on that if it is not a variable of interest.  Age is less strong for TB but still important so matching might be important.    

If you are interested in studying whether factors, such as co-infection, might lead to progression of latent TB to TB disease, then you might want to select controls who have LTBI.  This would probably obviate the need for matching on foreign birth or age.

If you simply want to see the differences in immune responses between co-infected vs mono-infected, you might want to compare 4 groups: uninfected for both, infected for both, and mono-infected TB and helminth.  In that case, I would think that matching would be related to those variables which are known to affect your immunological outcomes of interest, which will depend on which outcome you select.  I would think about age and foreign birth as potentially important sources of confounding, but also comorbidities and other factors.

It is important to really understand the reason you are trying to do a "case-control" study - the reason is that you want to control (e.g. minimize) the variation between the control and the experiment.  So, if you believe that gender affects and/or confounds the experiment then you need to control for it (e.g. match an experimental-control subjects between the genders)

It is also very important to realize that if you control on an attribute that has nothing to do with your study (e.g. eye color) then you can actually lower the power/effectiveness of your study!

James Mancuso is on the right track.  Think about the true risk factors that affect your results and control for those.  So, we know that there are several factors that affect risk of TB - HIV/AID status, diabetes, etc...  

Many thanks to you both for your advice.

Maria

Defintely you are  now better informed by our colleagues who so far commented on your question. Just a few addition for your sake. 

It is logical to match for age and sex. Matching for age is done in two ways:

1. Individual matching. Case1  and 1-10 controls  probably 2 controls are sufficient if you have good number of cases. The age of the controls is recommended to be with one year above and one year below the age of the case. 

2. Group matching. Stratify the cases in age groups (e.g., 5-9, 10-14...) and identify controls in the same age range.

Regarding who are the controls? if the cases are identified in the community, the controls may be their relatives, neibours or people in a population registry covering the area.

The controls need to be carefully ascertained that they are free from TB at the time of study or previously. 

Also you need to identify, list and define all possible risk factors for tuberculosis to study their distribution among cases and controls. Do not forget smoking.

interesting question,interesting interpretation