Trial Outcomes & Findings for Attention Control Training (ACT) and Very Preterm Infants (NCT NCT03896490)

The recruitment phase encompassed 13 months from April 2018 to April 2019. Eligible participants were identified by two categories of gatekeepers: (a) Collaborating neonatology practitioners in hospitals within the Belfast, South Eastern, and Northern Trust in Northern Ireland; (b) A local charity for families of premature children. The gatekeepers ensured that parents or caregivers of eligible infants received information about the study.

One participant did not attend the pre-test session (withdrew before starting the study)

Recruitment as a percentage of the eligible families approached who agreed to take part in the study and were randomised, and retention, defined as the percentage of randomised participants for whom data are available at baseline and post-test.

Sessions attended are defined as sessions to which parent and infant are present at the scheduled appointment

Sessions completed are defined as sessions whereby infants engaged without interruptions for at least 240 sec in at least two tasks.

Duration is defined as the cumulative duration in seconds of the tasks delivered to infants during the training or control sessions.

Completion is defined according to time criteria specific for each of the four attention tasks. Specifically, a sustained attention stipulates an infant completes two consecutive looks at less than 50% of the longest look displayed, or 120 s of accumulated looking time, or 12 trials without showing consecutive looks at less than 50% of the longest look displayed (habituation criterion). A visual-paired comparison task stipulates that stimuli are displayed for 16,000 ms. A disengagement task stipulates administration of at least 52 trials. An information density preference task stipulates administration of 6 blocks of the task in total.

A random sample of the baseline and post-test attention assessments will be selected at random (20% of all the completed assessments). The quality of the eye-tracker data collected will be assessed by determining the number of usable fragments recorded during the task and the proportion of usable fragments recorded by the duration of the task in seconds.

Performance indicators included reaction times or level of task difficulty achieved. These indicators were standardized (std.): higher scores indicated better performance and followed a normal distribution with ranges roughly between -3 and +3. Each infant contributed several performance scores during each visit. We used multilevel growth regression models to test if trained infants displayed linear changes in performance across visits: linear effects would indicate significant performance improvement across tasks. In these analyses, the units of observations were the std. scores registered across visits, which were considered nested within infants. The models allowed to estimate the initial status (average std. scores in the first visit) and the rate of change (the average change in performance std. scores from one visit to another), as well as residual variances around these parameters