# Thread: Significance of change after treatment application

1. ## Significance of change after treatment application

Hi all,
I'm facing a statistical problem much greater than my basic statistics knowledge.

I am trying to determine if a change in the percentage of patients considered at risk in a group is statistically significant after applying a treatment.

My population is divided in groups (or treatment centers) and each group is initially assessed for behavioral health issues with a test. The test assigns a score between 0 and 3 to a client and a client is considered at risk if his score is 2 or 3.
Each patient is then placed in one of two "buckets" (at risk, not at risk) based on the score on their assessment.

The admission data looks something like this:
Group | Population | Not at risk | At risk
A | 150 | 70 (47%) | 80 (53%)
B | 50 | 10 (20%) | 40 (80%)
C | 230 | 180 (78%) | 50 (22%)

A treatment is then applied to to all the clients in the groups and the clients that have not left the program are re-assessed before discharge. This means the population for the second measure is not the same as the one for the first measure, but it's in most cases a subset.

The discharge data looks something like this:
Group | Population | Not at risk | At risk
A | 110 | 59 (54%) | 51 (46%)
B | 34 | 10 (29%) | 34 (71%)
C | 145 | 100 (69%) | 45 (31%)

If statistically possible, I need to determine the significance of the change in percentage of clients at risk between admission and discharge for each group. A group is doing well if the percentage of clients at risk gets lower.

Can I say that the 9 point decrease in the percentage of at risk clients for group B is as or more significant than the 6 point decrease for group A?
Or can I say that a change is significant at all and is not simply a random event?
Is there a way to find the minimum number of patients in a group for percentage change to be considered significant?

We don't know the characteristics of the population of each group. For example, we don't know if a particular group has higher probability to have patients with high risk.

D.

2. ## Re: Significance of change after treatment application

You can perform 3 McNemar's tests (one for each group), using the data of those with 2 assessments.
In additions, you can check whether those who were re-assessed were different at baseline from those
who could not be re-assessed.

With kind regards

K.

3. ## Re: Significance of change after treatment application

Thanks Karabiner! I'll look at McNemar's test.

D.

 Tweet

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts