Hi,

I have some questions for the calculation of prevalence rate for pre-1996 data.

I understand that I need to use different weight for different times. I also learned that there will be a universe for each questions. For example, to calculate the prevalence for diabetes, I should use DIABETICYRC, and that:

For 1973 and 1975, DIABETICYRC should be weighted with PERWEIGHT.

For 1978 to 1981, DIABETICYRC should be weighted with DIABWT.

For 1982 to 1996, DIABETICYRC should be weighted with CONDWT4.

Question:

How should I calculate the prevalence rate specifically?

What I have done is like this:

```
if year == 1996:
numerator = ONE(answering "20/21/22" for DIABETICYRC in 1996)*CONDWT4
denominator = ONE(answering "10/20/21/22" for DIABETICYRC in 1996)*CONDWT4
prevalence rate = numerator/denominator*
where ONE(`) returns 1 if "`" is satisfied and 0 if not*
DIABETICYRC:
00: not in universe
10: no
20: yes
21: Yes, indicated by response to direct survey question
22: Yes, indicated by other source
```

Is my method correct? Since I calculate for diabetes and the result seems unreasonable.

Thank you!

Dongyue