Analysis

Monitor data were collected in Microsoft Excel and uploaded to R program (R Foundation for Statistical Computing, Vienna, Austria) for analysis. Light levels were averaged in 30 min intervals, with maximum and minimum values taken every hour. Equivalent sound was calculated at 15 min intervals, with hourly minimum and maximum measurements. Sound peaks, defined as the number of times a room's sound level was above various thresholds, from 70 to 90 dB, were also collected. Patient demographics were extracted from the electronic medical record and entered in a Research Electronic Data Capture (REDCap) database, hosted at Virginia Commonwealth University.

Demographic characteristics between the new vs. pre-existing unit were compiled, and chi-square tests were performed for these categorical values. Light and sound data are presented in a descriptive manner, without tests to assess statistical significance. With more than a million data points, even a minor difference in light or sound would likely be statistically significant. Results are presented as number and percentage or median and interquartile range (IQR), as appropriate. Additionally, results are presented as “daytime” (07:00–21:00) or “nighttime” (22:00–06:00). These times were chosen with the concept of merging a typical bedtime with the observed natural bustle of the ICU at shift changes. Light and sound levels between the new construction and existing PICU rooms were compared. Light and sound level daily trends over the PICU stay were also evaluated.