Linear mixed effect models: Ibuprofens infuence on fractured wrists compared to other painkillers
Author
Nielsen, Nikolaj Beck
Term
4. term
Education
Publication year
2018
Abstract
This thesis applies linear mixed effect models to evaluate postoperative painkiller regimens after surgery for wrist fractures. Eighty-three patients were randomized to three regimens: ibuprofen 600 mg three times daily for seven days; ibuprofen 600 mg three times daily for three days followed by placebo for four days; or placebo for seven days. Outcomes included wrist moveability measured at 6 weeks, 3 months, and 1 year (using the healthy wrist as baseline) and daily pain ratings during the first 14 days; rescue analgesic (Tradolan) use was also recorded. The report describes the datasets and handling of missing values, introduces random-effect and linear mixed-effect modeling with estimation (including restricted maximum likelihood), and uses visual residual analysis supported by a simulation study. Two models were fitted, one for moveability over time and one for postoperative pain. Across both analyses, the models found no differences between the currently used painkiller regimen and the two alternatives, suggesting that the alternatives can be used without affecting patient outcomes.
Dette speciale anvender lineære mixed effekt modeller til at vurdere smertebehandling efter operation for håndledsbrud. 83 patienter blev randomiseret til tre regimer: ibuprofen 600 mg tre gange dagligt i syv dage; ibuprofen 600 mg tre gange dagligt i tre dage efterfulgt af placebo i fire dage; eller placebo i syv dage. Udfald inkluderer bevægelighed målt efter 6 uger, 3 måneder og 1 år (med det raske håndled som baseline) samt daglige smerteangivelser i de første 14 dage; brug af supplerende smertestillende (Tradolan) blev også registreret. Rapporten beskriver datasættene og håndteringen af manglende værdier, introducerer teori om tilfældige effekter og lineære mixed modeller med estimering (herunder restricted maximum likelihood), og benytter visuel residualanalyse understøttet af et simulationsstudie. Der opstilles to modeller, henholdsvis for bevægelighed over tid og for postoperative smerter. På tværs af begge analyser findes der ingen forskel mellem den aktuelt anvendte smertebehandling og de to alternativer, hvilket antyder at alternativerne kan anvendes uden at påvirke patienternes resultater.
[This apstract has been generated with the help of AI directly from the project full text]