When we want to define the socioeconomic state of an enrollee, what variables (items or data) should we look at and collect?
The most commons are income (different from salary), place of residence (proxy), and values assets (not sure you can collect this).
Hope this help.
Education (length of completed education) immigrant background (1. gen, 2.gen) and also the place of residence can be useful as already mentioned (for instance if there exists other aggregated data on poverty, crime etc for different districts etc.) Finally even if not strictly speaking socioeconomic there might be correlation between socioeconomy and health/lifestyle variables such as smoking, alcohol consumption and overweight. It really all depends where you are in the world!
Information on income, occupation and urbanization level of residential district is available.
Insurance status? Income, education, race, employment status, location, and proxy such as diet, BMI, tobacco, and drug use,
Total living cost per month (proxy), value of assets and house space (in square meter)
I think you need to be careful!
What you are looking for is likely to be dependent on cultures and place where you want to apply what you are looking for.
I would recommend you to contact me should you have more interest on this as I have been validating different tools for patients' usage/hospital use in different places over time.
While above commentators may be right in general and ideal situations, our world is not that ideal and less likely to be adherent to assumptions put using variables like "place of residence" or "level of education"
(e.g. while in Africa, it is quite possible to find out students' dormitories available in posh areas, in some previous states of the former Soviet union it is quite possible to find out people with multiple PhDs to live without reliable health insurance cover.)
message: BE CAREFUL!
What do you wish to predict with SES -- patterns of utilization? What would you collect anyway for the administrative function of the registry? Do you have data now to estimate correlations?
SES is a latent construct that is correlated with the above (likely correlated) manifest variables (and gender) whose value, meaning, and predictive significance change over time and with place -- other than building a dynamic (temporal and spacial) measurement model specific to your setting (culture), perhaps education, age, and sex (combined as a proxy for gender) would be a good place to start?
Often times, the SES variables available in a claims database are limited; however, if you have address data which can be geocoded, you can link records to additional SES measures. For example, the Townsend deprivation index can be constructed using census data from the geographical region in which an individual resides. It creates a score based on the following percentages 1) Households without a car, 2) Overcrowded households, 3) Households not owner-occupied , 4) Persons unemployed.
number of computer
number of cars
number of rooms in your house
In administrative claims data, you can't. There are no such variables in the standard medical claim forms in use in the USA for the past 20 years.
Measures are poor, and I tend to agree with Frost. May try address linked to small-area income/poverty rate data (see Hall above); health insurance status/type of health insurance, depending on the circumstances.
- Badges
- Science topic
- Similar topics
- Journal Impact Factor