I think you are quite right, that under these conditions the answers to a Western style self reporting questionnaire will almost certainly be invalid. Since the formulation of any sort of robust questionnaire is actually extremely difficult, especially one that works across cultures,  you will indeed need an alternative means of assessing health. And don't forget that the Suva concept of health might be quite different to our Western one!

You can calculate at the population level some statistical indicators related to some variables, such as the body mass index (BMI), number of persons suffering from chronic diseases (blood pressure, ulcer, and diabetes). The stistical indicators are: mean, standard deviation, standard error, confidence interval, variation coefficient.

As indicated by you, these people living in remote areas or interior villages are less educated or ill-literates to understand what you want and to give you proper answer. Live alone questionnaire, even oral answer they may not be able to provide you. Usually in the already identified places where the illness is expected due to consumption of polluted / unsafe water, we contact / visit the nearby health centres and get the data from the Doctor on the number of incidences of vector borne diseases reported and treated. This will later be used to know the intensity of the effect of drinking water in the region on the human health. 

What about height,body weight and age as indices?

One more important indicator is "Child Mortality Rate".