A Randomized Study of Codeine Plus Paracetamol Versus Placebo for PRK Post-operative Pain
A Randomized Study of Codeine Plus Paracetamol Versus Placebo for PRK Post-operative Pain
Compare the use of codeine/paracetamol against placebo for pain control after cornea photorefractive keratectomy (PRK).
The aim of this study was to measure pain during PRK postoperative period, comparing placebo with the association of codeine phosphate 30 mg and paracetamol 500 mg for 72h after surgery.
This is a double-blind, prospective, randomized study involving 80 eyes of 40 patients (paired design), who underwent photorefractive keratectomy at hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo, Brazil. The main outcome was the evaluation of pain by the Visual Analog Scale (VAS) 24h post-operatively.
Patients who had PRK as their chosen technique to correct their refractive errors were recruited for the study.
Patients had to be older than 20 years, with an spherical component between -1.00 and -5.00, with or without astigmatism, a cylindrical component up to 1.5D, spherical anisometropia of less than or equal to 0.75D, cylindrical anisometropia of less than or equal to 0.5D, stable refraction errors (maximal variation of 0.5D in the spherical or cylindrical component during 1 year) and amenable to follow-up for at least seven (7) days.
Exclusion criteria included the presence of active disease of allergic, inflammatory or infectious nature, in the ocular surface or its attachments; previous ocular history (pathology, surgery or ocular trauma); best corrected visual acuity less than 20/25); autoimmune disease, or immunosuppression or Diabetes Mellitus; pregnancy and lactation.
The same surgeon (VPBP) performed all procedures. Patients' eyes were operated at two-week intervals, and the treatment order (placebo or codein/paracetamol) was randomized for each patient.
The distribution of the continuous variables was assessed visually by constructing histograms. Due to moderately symmetrical feature of the data, parametric approaches were used, following the guidelines of Fagerland.18 Data were presented as mean ± standard deviation (SD) or mean (95% confidence interval), as appropriate. Counts and percentages were used to summarize binary variables.
The intervention effect was estimated by the difference between the average of the group treated eyes (μtreatment) and the mean of the control eyes group (μcontrol). In other words, the effect of treatment was estimated by the variable Δ where Δ = μtreatment - μcontrol. The statistical approach considered the paired nature of the study ("paired-eye design") following the recommendations of Armstrong et al. 19 Specifically, Δ was calculated taking into account the correlation between the pair of eyes. Covariates were included in the models of multiple linear regression, which were adjusted for age, gender (1 = male, 0 = female) and ancestry (0 = white 1 = brown and 2 = black). These models were built with a robust estimator of the variance, which incorporates the intra-patient correlation of the pair of eyes.20 Furthermore, due to the characteristic of repeated measurements (1h, 24h, 48h and 72h), for each variable, the p values were adjusted for multiple comparisons by the method of Holm- Šidák. 21,22 The relative magnitude of the variation of the treatment effect over time was examined by analysis of variance (ANOVA ) one-way repeated measures. This evaluation explicitly asks whether longitudinally, the effect of treatment, i.e., the magnitude of Δ is the same over time (H0 : Δ1h Δ24h = = = Δ48h Δ72h ). A Huynh-Feldt correction was applied. When statistically significant, ANOVA models were followed by post-hoc tests (paired student t test) to detect differences between specific times.
Potential predictors of the magnitude of the treatment effect were investigated having as the dependent variable Δ and the explanatory variables age, gender, ancestry, fractional error group (1 = -3 to -5, 0 = -1 to -3) and sequence of the treatment. No corrections were applied for multiple testing in the exploration of predictors of the magnitude of the treatment effect, since they were considered exploratory.
P values of the post-hoc tests were also adjusted for multiple comparisons using the Holm-Šidák method. The α significance level for all analyzes was 5%. The software used was Stata 13.0 (Stata Corporation, College Station, TX , USA).
Inclusion Criteria:
Exclusion Criteria: