<p><b>Background: Knowledge translation literature shows a delay between publication and uptake of research findings into clinical practice. There is uncertainty about whether this delay exists in the intensive care unit (ICU) context and whether participation in research influences changes in clinical practice. Knowing whether ICU participation in research increases the likelihood of research findings being incorporated into clinical care is important to funders, researchers and patients.</b></p>
<p>Aim: To investigate the effects of participating in the Intensive Care Unit Randomised Trial Comparing Two Approaches to Oxygen therapy (ICU-ROX) randomised controlled trial on attitudes and practices in relation to ICU oxygen therapy.</p>
<p>Methods: The research question was examined using three different methods: a survey, an inception cohort study, and a retrospective cohort study. The practitioner attitudes survey and inception cohort study were conducted before ICU-ROX started, after ICU-ROX finished but before the results were known, and, finally, after publication of the ICU-ROX results. The retrospective cohort study (using the Australian and New Zealand ICU adult patient database) compared data from 48 weeks before and 48 weeks after publication of the ICU-ROX trial.</p>
<p>Results: Practitioner attitudes survey: Responses were received from 112 specialist doctors and 153 ICU nurses from 11 ICU-ROX and 11 non-ICU-ROX sites. Before the ICU-ROX trial was conducted, respondents from ICU-ROX trial sites and those from non-ICU-ROX trial sites were similarly likely to indicate they would use a fraction of inspired oxygen (FIO2) of 0.21, in a ventilated patient with peripheral oxygen saturations (SpO2) of 91-96%. However, after participation in the ICU-ROX trial, and before the trial results were known, ICU-ROX trial site respondents were significantly more likely than their non-ICU-ROX trial site counterparts to state they would use an FIO2 of 0.21 in a patient with an SpO2 of 91-96%. The differences in all patient admission categories, except the ‘other ICU patients’ category, persisted after the ICU-ROX trial was published. Nurses were generally less likely to report they would use an FIO2 of 0.21 in a ventilated patient than doctors. Following ICU-ROX publication, 41.1% of the doctors compared to 13.6% of the nurses knew the ICU-ROX results (P<0.001), and 53.1% and 5.4%, respectively, reported having read the publication (P<0.001).</p>
<p>Inception cohort study: Data were analysed from 650 patients from 11 ICU-ROX and 11 non-ICU-ROX sites. ICU-ROX site patients spent 8.2% of the time ventilated at FIO2 of 0.21, before ICU-ROX started, 7.7% after their participation in the ICU-ROX trial and 7.1% of the time after ICU-ROX publication, a change over time was not statistically significant. For the non-ICU-ROX site patients, time spent ventilated at FIO2 of 0.21 was 0.9% before ICU-ROX, 2.4% after ICU-ROX was conducted and 3.3% after publication. This change for non-ICU-ROX site patients after ICU-ROX publication was statistically significant (relative rate 3.63; 95% CI 1.39 to 9.47, P=0.01).</p>
<p>Retrospective cohort study: The FIO2 from the arterial blood gas associated with the worst Acute Physiology and Chronic Health Evaluation III-J score in the first 24 hours of admission to ICU was analysed for 51,498 pre-publication admissions and 47,523 post-publication admissions from 15 ICU-ROX sites and 188 non-ICU-ROX sites. Although the average FIO2 values varied considerably from week to week, there was a statistically significant reduction in the average FIO2 over time in ICU-ROX sites (P=0.01) but not in non-ICU-ROX sites (P=0.49).</p>
<p>Conclusion: Differences between attitudes of practitioners in relation to oxygen in ICU-ROX sites compared with non-ICU-ROX sites changed over time (before and after the ICU-ROX trial, and after publication of the results). Differences in oxygen therapy practice changes for ICU-ROX versus non-ICU-ROX sites shown in the inception cohort study and retrospective cohort study were small, and are unlikely to be clinically important. These studies did not demonstrate a consistent effect of participation in ICU-ROX on knowledge translation.</p>