Análisis de la calidad metodológica en los ensayos clínicos aleatorizados publicados sobre ventilación no invasiva

Abstract

Justification. Noninvasive ventilation is a widely used respiratory support technique. In the last four decades, numerous studies have been performed analyzing this technique within different medical fields and pathologies. Despite this, the evidence on which its use is based consists of very heterogeneous randomized clinical trials whose methodological characteristics have not been studied. The aim of the present study is to analyze and compare the formal and methodological characteristics of the randomized clinical trials published on noninvasive ventilation, based on the type of clinical trial, the number of centers in which it was performed, and the year and impact factor of the journal in which it was published. Methods. An analytical cross-sectional study was performed, in which randomized clinical trials on noninvasive ventilation in the acute patient published in Medline, Embase and Cochrane from 1980 to 2019 were reviewed. The extraction of variables was carried out independently by two members of the research team, previously trained in research methodology. In the event of a discrepancy in a result, a third researcher was consulted. The variables are expressed as absolute and relative frequencies. Comparisons were made using Pearson&apos;s Chi-squared test. Results. We analyzed 294 randomized clinical trials published from 1980 to 2019. Of the 294 articles analyzed, adequate specification of study design was present in 20.1% (CI-95% 15.7 to 25.1), performance of sample size calculation was specified in 51.7% (CI-95% 45.8 to 57.5), method of randomization sequence generation in 48.3% (CI-95% 42.5 to 54.2), and allocation concealment in 47.6% (CI-95% 41.8 to 53.5). Registration number was published in 30.6% (CI-95% 25.4 to 36.2) studies. Parallel studies more frequently showed design specification (17.6% vs 30.9%, p = 0.026), sample size calculation (56.1% vs 32.7%, p = 0.002), randomization sequence generation (53.1% vs 27.3%, p < 0.001), allocation concealment method (56.1% vs 10.9%, p < 0.001) and registration number (31.8% vs 25.5%, p = 0.357). Multicenter studies more often adequately specified the design (37.3% vs. 10.4%, p < 0.001). More multicenter studies had a sample size estimate (85.1% vs 46.3%, p < 0.001), randomization sequence generation (70.1% vs 47.6%, p = 0.002), allocation concealment (80.6% vs 47.6%, p < 0.001) and registration number (49.3% vs 26.2%, p < 0.001). Articles published in first quartile journals more frequently presented sample size calculation (65.4% vs 59.4% vs 59.4% vs 46% vs 47.1%, p = 0.012) and allocation concealment (60.3% vs 50% vs 42% vs 47.1%, p = 0.03), with no differences in design specification, randomization sequence generation, and registration number. Articles published from 2010 to 2019 more frequently showed design specification (31.1% vs 9.3%, 6.7%, 0%, p <0.001), sample size calculation (64.2% vs 45.5% vs 20% vs 0%, p < 0.001), generation of the randomization sequence (58.3% vs 45.4% vs 13.3% vs 20%, p < 0.001), method of allocation concealment (53% vs 48.1% vs 23.3% vs 20%, p = 0.004) and record number (54.4% vs 7.4% vs 0% vs 0%, p<0.001) compared to those published in 2000-2009, 1990-1999 and 1980-1989. Conclusions. The randomized clinical trials published on noninvasive ventilation frequently present methodological and formal deficiencies that influence the quality of the scientific evidence supporting its use, which can be considered suboptimal. Key words: noninvasive ventilation, respiratory failure, research methodology, randomized controlled trial.