Interpret continuous variables in binary regression model


New Member
I have produced three binary regression models, each showing the odds ratios (ORs) for an outcome (attendance versus non-attendance)

Each regression model looks at age in 10 year intervals and its effects on non-attendance of a particular health service.

The OR for patients between 0-10 was 0.52; (3 year intervals used here)

The OR for patients between 11-49 was 0.66;

The OR for patients over 50 was 1.15.

How does one interpret these values?

Is it correct to say that each average increase in 3 years, there is a 48% reduced risk for non-attendance? Does this also suggest that younger patients have an increased risk for non-attendance?

I would appreciate any advice, I'm quite confused by this!

Anyone? :(
Last edited:


Omega Contributor
So there seems to be quite a bit a vagueness here. Why do have to run three models. Are there other variables in these models? What is with the age groupings (e.g., 0-10,...,etc.). Did you select 3-year intervals? Is it feasible to make generalizations from three different models to the the overall population? I would guess not. What does the confidence intervals look like?


New Member

I have left out other details to simplify the post.

There are three age groups (and three models) because the condition I'm looking at is managed differently at a clinical level.

There are other covariates (i.e. gender, location, diabetes, etc.) which I have left out, again for simplicity.

I have used three years for the youngest group as this was recommended by lecturer and statistician.

The confidence intervals are nothing unusual. I am working with a dataset of over 100,000 cases and most relationships are significant.

I was more confused about the interpretation.



Omega Contributor
The OR for patients between 0-10 was 0.52; (3 year intervals used here):

For every unit increase in age (3 years), the odds are 48% less for attendance. I think for ever unit decrease (3 years) odds for attendance is 72% greater for attendance.

The OR for patients between 11-49 was 0.66; Same interpretation here.

The OR for patients over 50 was 1.15. Same as 72% interpretation above.

Sidenote, not "risk" but odds.


Omega Contributor
It feels like you could run a model with all data, just control for a the age groups and look for potential interactions between groups and covariates.


New Member
Thank you very much hlsmith!

I will have a word with my friend about interactions between groups.

Could you also let me know where the 72% figure comes from, is this an inverse value?