Evaluating a treatment in sigle arm trial

[ran20]

Hello all,

I am going to Evaluation of combined surgical and antibiotic treatment for Diabetic foot ulcers, 30 patients with Diabetic foot ulcers were enrolled in this study, and the date of first and last visit was recorded, I considered this study as single-arm treatment as there I had no control group. I recorded the CRP before and after the treatment, the patients with an absolute difference in CRP less than 10 were considered as healing otherwise no healing will be recorded. How I can start with R cran evaluating my treatment. Statistics approach, and methodology

Thanks in advance.

## My data

    crp_before = c(96.1,90.4,114.4,88.3,76.1,191.2,69.8,122.3,188.6,77.3,126.8,189.3,165.2,116.8,72.3,120.9,122.3,115.2,90,142.3,87.2,195.5,184.3,110.2,113.6,147.4,96.8,116.4,55.3,209)
    crp_after = c(5.3,7,6.2,3.5,4.2,9.6,5.2,5.3,9.6,8,7.6,11,10.3,4.6,3.2,8.6,7.5,8.4,6.3,7.6,6.8,112,6.3,8.5,9.2,5.3,4.1,7.6,3,100)
    time_week = c(9,8,12,8,4,24,4,8,24,4,12,24,20,12,5,12,13,12,8,16,8,24,24,8,8,16,8,12,3,4)

[TJH-research]

If you have before and after data, you have two groups. Any of the statistical tests for differences between groups will work (t test, regression, anova). Even descriptive statistics will work (I.e. what was the mean of before and the mean of after?)

Before/after research designs are generally pretty poor. They do not provide a strong estimate of the counterfactual. If you have measures of relevant antecedent variables, then using these as control variables will help improve the estimation of the counterfactual.

If that did not make sense to you, let me put it this way: some people’s ulcers might have gone away with time no matter what they did. Treatment, no treatment, does not matter because the ulcer was going to disappear anyways. If you also measured things that are related to ulcers disappearing (maybe BMI), then you could control for these.

Hopefully this was not a real experiment! I say that because you would want to have a plan for data analysis before starting. Otherwise, we do things like not collect a control group or not measure relevant antecedent and intervening variables.