We need to improve post-operative pain management for patients, and have collected data for 2 years from 2 different samples. Data for year 1 (52 patients) was before we put interventions to reduce pain. Data for year2 (55 patients) was collected after we put new interventions. Pain levels were measured at 4 points (0, 12, 24 and 48hrs post-op) on the Visual Analogue Scale (VAS), taking the values 0-10 (pain level). What test can i use to find if the pain levels were statistically significantly lower @ 48hrs than it was at 24hrs using R. Data is arranged as PATIENT VAS0hrs VAS12hr VAS24hr VAS48hr.
1
$\begingroup$
$\endgroup$
From what I understand your question to be, you could use one of the following techniques.
1) A Cox proportional hazard model. https://stat.ethz.ch/R-manual/R-devel/library/survival/html/coxph.html
2) Wilcoxon Signed Rank test. https://stat.ethz.ch/R-manual/R-devel/library/stats/html/wilcox.test.html
-
$\begingroup$ Thank you. Was also thinking of using Wilcoxon Test but can i also use it in a situation when the subject gives 2 responses after a single treatment Or its best used say in a crossover design (the same subject receiving 2 diff treatments and give responses at each of the treatments) $\endgroup$ – mapiye Aug 11 '15 at 6:07
-
$\begingroup$ When I worked in healthcare, I used WSR to analyze the effect of treatment on a patient pre - and post on ER visits, IP admissions, etc. You might want to also look into a Wilcoxon Rank Sum Test. $\endgroup$ – Hidden Markov Model Aug 11 '15 at 6:23