Data collection

The medical notes were reviewed following guidelines suggested in a study by Gilbert et al. [13]. These guidelines included the use of standardised data extraction forms, uniform handling of ambiguous data and regular meetings between the research team to ensure standardisation of data extraction. All patients aged 16 years and under presenting to the ED of Morriston Hospital in 2008 to 2014 were examined and data recorded on a pre-designed database. A validation check was completed in which an additional researcher checked the accuracy of data input for 10 % of all patients, in order to reduce information bias. If a patient’s notes had missing or incomplete data for the variables under investigation, they were still included in the database.

The dataset included demographic variables such as age, sex, type of injury, percentage total burn surface area (TBSA) estimation, injury mechanism, injury agent, whether the injury was recorded as non-accidental, where the injury occurred (home or other) and primary anatomical area injured (according to the medical records). Outcomes recorded included mortality, hospital admission and hospital length of stay. Mechanism of injury for scald was further categorised into hot beverages (all hot drinks), domestic water (kettle, saucepan, bath, shower, tap, hot water bottle, flask, facial steamer and bowl of water) and food item (soup, cooking oil, pot noodle, gravy, curry and sauce) [1]. Contact burn mechanisms were further categorised as portable household agents (iron, hair straighteners/tongs, light bulb, saucepan, oven tray, cigarette), fixed household items (oven, hob, grill, radiator, heated pipe, fireplace/wood burner and steam press), outdoor items (exhausts, firework, BBQ, bonfire, sand) and miscellaneous (not fitting other categories) [1]. Chemical burns were further categorised as spills, ingestion, and eye splash. Radiation burns refer to thermal injury caused directly by the sun or sunbeds.

To ensure confidentiality, patients’ names were not recorded during the data collection period. The dataset was also stored on a hospital encrypted computer to ensure data security (safe-end protector encryption).