topic Re: How to calculate relative risk, and 95%CI in Statistical Procedures

How to calculate relative risk, and 95%CI

Bal23 — Wed, 13 Jan 2016 16:18:52 GMT

I would like to know how to use SAS macro/array to calculate relative risk. It has 5 IDs, I need to calculate lower bound and upper bound.

Would any of you present me a code or a simiar example?

Thanks.

Re: How to calculate relative risk, and 95%CI

Reeza — Wed, 13 Jan 2016 16:23:27 GMT

Why do you want a macro or array?

Proc freq can calculate relative risk.

https://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_freq_sect029.htm

proc format;
      value ExpFmt 1='High Cholesterol Diet'
                   0='Low Cholesterol Diet';
      value RspFmt 1='Yes'
                   0='No';
   run;
   data FatComp;
      input Exposure Response Count;
      label Response='Heart Disease';
      datalines;
   0 0  6
   0 1  2
   1 0  4
   1 1 11
   ;
   proc sort data=FatComp;
      by descending Exposure descending Response;
   run;

  proc freq data=FatComp order=data;
      format Exposure ExpFmt. Response RspFmt.;
      tables Exposure*Response / chisq relrisk;
      exact pchi or;
      weight Count;
      title 'Case-Control Study of High Fat/Cholesterol Diet';
   run;

Re: How to calculate relative risk, and 95%CI

Bal23 — Wed, 13 Jan 2016 16:44:35 GMT

Thanks. Those are varibles that have more than two levels of catogory. (>2)

So I am looking for a code that can do this.

Re: How to calculate relative risk, and 95%CI

Reeza — Wed, 13 Jan 2016 16:50:04 GMT

Please provide sample data, expected output and some details surrounding your question.

Re: How to calculate relative risk, and 95%CI

Bal23 — Wed, 13 Jan 2016 17:00:57 GMT

agegroup <10 10-20 20-30 >30

the risk of surgery 2.37 2.4 2.8 3.2 (%)

now I need to caculate the crude relatve risk, using <10 agegroup as a reference (1.00)

I also want to caculate same things among different communities (up to five communities). This is why I want to use array and/or macro

Thanks

Re: How to calculate relative risk, and 95%CI

ballardw — Wed, 13 Jan 2016 18:24:06 GMT

You may need to provide additional information about how your current SAS data set is structured. For example, you give a percent which seem to be your current value. But what was the actual size of the sample that the percent was calculated from? Since you are asking for "upper and lower bounds" I am assuming that is a confidence interval you are requesting and the calculation for that will require the original sample size.

Can your print a few rows of your data to show us what it looks like? You may need restructuring to calculate the risks and confidence intervals.

Re: How to calculate relative risk, and 95%CI

Bal23 — Wed, 13 Jan 2016 18:29:53 GMT

It is about 200,000 obs. For some age groups, it has 50,000. Others have more.

sorry, the data cannot be printed out. I just list an example that is very similar to my original data.

I would like to see whether there is any reference. I read some books and articles showing examples on variables with two catogories. But mine is over two. So it will be beneficial if there is code for three catogories or over three.

Thanks

Re: How to calculate relative risk, and 95%CI

ballardw — Wed, 13 Jan 2016 19:13:41 GMT

You need to show us what your data looks like. Dummy values are fine but the structure of current data set is important.

Does each record relate to a single person? Is that person in the data more than once? If a person is in the data more than once which record is to be used for the relative risk calculation.

You can get ODDS ratios between a base group and other groups using Proc Logistic (and others).

If you are looking at age as a contributor to specific outcome you might want to look at a regression model which would yield a change in likelihood per year of age.

Re: How to calculate relative risk, and 95%CI

Bal23 — Wed, 13 Jan 2016 19:30:32 GMT

I do not quite understand what "structure" you mean here.

In my dataset, some people have multiple records. But I have already disingushi this by using sort by their ID

relative risk is simply the rate I put and calculated by the total number of people. Then I need to calculate its related 95%CI.

I want to have a macro or array code to calculate since it has multiple subgroups.

RR= count of patients who had surgery/total people

lower CI=exp(ln(RR-1.96???? (formula should be available available, but I cannot find code)

so it really does not matter how much my data looks like, I am looking for a sample code to deal with data with >2 catogories

Re: How to calculate relative risk, and 95%CI

Reeza — Wed, 13 Jan 2016 20:40:49 GMT

Of course it matters how your data is. For starters, is your data already in counts, so can you divide two variables for RR? Or do you have individual level information and need to calculate the # of patients who had surgery and the total number?
How are your groups defined as well, is there a variable that defines your groups or multiple variables that have information for each group?
Typically subgroup analysis in SAS is via BY group processing, not macro or array. And sometimes transposing your data makes your analysis simpler.

If you know exactly what you want and need help converting that to a macro or array please post what you have and someone can help from there.
Otherwise vague problem descriptions get vague answers.

Re: How to calculate relative risk, and 95%CI

Bal23 — Wed, 13 Jan 2016 20:58:21 GMT

my data is not already in counts. First, I used SQL to put those different datasets together (different communities). Then, I use proc freq to calculte counts. Then, I calculate RR.

I have to use excel to calculate, one by one, which is very slow. So I am thinking whether I can speed up with macro code

Re: How to calculate relative risk, and 95%CI

ballardw — Wed, 13 Jan 2016 21:06:28 GMT

If a person is in your data more than once do you count them as a person (once for the entire data set) or each time they appear? How your result is interpreted depends on knowing that.

How is a person treated in the counts if a person is in the data once in one age group without the surgery but then has the surgery when in another age group and then appears later in yet another age group?

I think you may be at a point where you need to go back to what the research question that is being asked is defined to get clarification on how to treat your data, probably such that a person only occurs once in the analysis. At this point, I would not provide code on just how to do multiple chi-sq tests as I would not want to be associated with any result.

Re: How to calculate relative risk, and 95%CI

Bal23 — Wed, 13 Jan 2016 21:47:43 GMT

count only once

the agegroup is classfied at the beginning of research, no more agegroup variable later

we only analyze the patient once;

Just as the first one who provided code for me---unfortunately, he provided a sample code for variables with two categories----that is same as what I have read and learned online. I need, in stead, a sample code for variables with more than two catogories, like age groups.

Re: How to calculate relative risk, and 95%CI

Reeza — Thu, 14 Jan 2016 01:03:24 GMT

Are you sure you need relative risk? With multiple levels I'm more used to seeing odds ratio.

You usually need case/control scenarios for relative risk.

Re: How to calculate relative risk, and 95%CI

Reeza — Thu, 14 Jan 2016 01:05:33 GMT

@Bal23 wrote:

unfortunately, he

She

Re: How to calculate relative risk, and 95%CI

Bal23 — Thu, 14 Jan 2016 15:00:22 GMT

I am requried to calulate relative risks. Certainly, I am interested the codes that can do both works, calculating RR and OR.

Thanks.

Re: How to calculate relative risk, and 95%CI

Reeza — Thu, 14 Jan 2016 15:41:52 GMT

I'm too lazy to make sample data.

If you post sample data and sample calculations someone can help with the code.

Re: How to calculate relative risk, and 95%CI

Bal23 — Thu, 14 Jan 2016 16:15:57 GMT

Re: How to calculate relative risk, and 95%CI

Bal23 — Thu, 14 Jan 2016 16:19:34 GMT

relative risk or risk ratio (RR) is the ratio of the probability of an event occurring (for example, developing a disease, being injured) in an exposed group to the probability of the event occurring in a comparison, non-exposed group. Relative risk includes two important features: (i) a comparison of risk between two "exposures" puts risks in context, and (ii) "exposure" is ensured by having proper denominators for each group representing the exposure [1][2]

Risk Disease status Present Absent Smoker Non-smoker

Consider an example where the probability of developing lung cancer among smokers was 20% and among non-smokers 1%. This situation is expressed in the 2 × 2 table to the right.

Here, a = 20, b = 80, c = 1, and d = 99. Then the relative risk of cancer associated with smoking would be

Smokers would be twenty times as likely as non-smok

please note the first group's RR=1

Re: How to calculate relative risk, and 95%CI

ballardw — Thu, 14 Jan 2016 23:05:23 GMT

Your example data only shows the numbers of the event surgery, 2126 (implies nonsurger of 217486-2126. You do not have an exposure. The rate you calculated was a prevalence rate of surgery within the age group.

agegroup1

217486

2126

2126/21746*100

Note that the definition you post shows that the exposure group, which I have to assume, you have been attempting to use agegroup, can only have two levels. You can do RR for agegroup1 vs agegroup2, agegroup1 vs agegroup3 and agegroup2 vs agegroup3 but not all 3 at once. What you are attempting really is much more in the way of odds ratios where compare the ratio to a base group, likely the one with the lowest prevalence.

The value you show 2126/21746 * 100 = 9.776. I could not tell from your data how you get a relative risk of 1, there is not another exposure group. Please show the values for a, b, c and d you used to get a 1.