Outcomes and instruments
The primary outcome variable of this study was the wearing of sun-safe hats during break periods of the school day (i.e. recess and lunch) by students. Whilst sun-safe behaviours have been observed in a beach setting [
16,
17] these were deemed inappropriate for school settings due to the variables being observed (i.e. type of swimwear). There have been a few studies to date that used a mix of surveys and on-site evaluations of SunSmart programs in Hawaii (Glanz et al, [
18]) and observations of public recreation venues (Dobbinson et al, [
19], Nikles & Harrision, [
20]). Very few however record sun-safe behaviours of children in school settings.. In this study, we adapted the System for Observing Play and Recreation in Communities (SOPARC) [
21] to capture sun-safe behaviour data in children. The SOPARC [
21] is based on momentary time sampling techniques in which systematic and periodic scans of individuals and contextual factors within predetermined target areas are made. Computer tablets (Apple Inc, USA) installed with the iSOPARC Application Version 1.75 (CIAFEL, Portugal:
https://ciafel.fade.up.pt/isoparc/) were used to provide an objective measure of hat wearing by children during recess and lunch.
In the traditional application of the iSOPARC tool, a scan of each subject is electronically coded and identified by: sex (male or female), intensity of activity (Sedentary, Walking, or Very Active), and whether they are a Child, Teen, Adult or Senior. For this study, given all the subjects were children, the third battery of coding (Child, Teen, Adult or Senior) was changed to detect whether the student was Unprotected (i.e. wore no hat), Partially-Protected (i.e. was wearing a baseball cap), Fully-Protected (i.e. was wearing a sun-safe hat: broad-brimmed, bucket hat or legionnaire).
Separate scans were made for females and males, and entries are also made for time of day, temperature, UV radiation level, area accessibility, area usability and presence of supervision. Each observation was conducted twice during the recess and lunch breaks for both females and males. A single scan involved a researcher observing each student, of one sex, within a pre-defined Target Area sequentially from left to right without pausing.
Direct observations were made in the designated Target Areas that had been identified by school principals (or their proxy) as areas that were likely to provide opportunities for students to have sun exposure during their recess and lunch periods (i.e one shaded and one unshaded play area). It is important to note here that shaded areas had to be man-made structures and were all metal structures with solid roofing. No areas were covered with ‘shade cloth’ or other permeable materials.
Additional data recorded prior to the direct observation scans included; temperature and UV level at the start and end of the observation period; whether the observation was made at recess or lunch; start and finish times of recess and lunch; and whether the area was shaded or not.
Researchers also recorded whether the teacher supervising the children’s behaviour in the Target Area was role-modelling sun-safe behaviour. An observation note was added to the final iSOPARC data on whether the teacher was wearing a) a sun-safe hat; b) sunglasses or other appropriate eye protection (i.e. transition lensed optical glasses; and c) a sleeved shirt and collar. As there was no more than 4 teachers present during any observation period, research assistants were able to record all of the teacher’s sun protective behaviours. Student clothing was not recorded as State requirements are that all students wear a school uniform that requires a collared shirt that covers the shoulders and upper arm as a minimum.
Thirty two field-based inter-rater reliability checks were conducted during the 10-week observation period. During reliability checks, two observers independently coded the same students in the same lunch or recess period. A high degree of reliability was found between measurements. The average measure Intraclass Corelation Coefficient (ICC) was .912 with a 95% confidence interval from .885 to .932 (F = 11.324, p < .001).
In an effort to minimise bias, the inter-rater reliability checks on 4% of the iSOPARC observations were randomised in order to prevent possible collusion. Recess and lunch break observations were randomly selected and observers and schools were given limited notice of when a reliability check was going to occur (usually less than 24 h).
Sunscreen consumption was also recorded during the observation period. All Stage 3 (i.e. Grades 5–6 or children aged 10–12 years) classes within the 20 schools were given accurately weighed and filled one-litre sunscreen containers (1 l = 1.07kg) at the start of the period. These containers were placed near the door of the students’ classroom. The containers were removed at the end of the observation period and weighed again. This consumed weight of sunscreen was divided by the number of students within the class to produce a baseline consumption rate per student.
Demographic data including socioeconomic status based on postcode of the school (Socio-Economic Indexes for Areas—SEIFA) [
22] and national school enrolment data collected by the federal government (Index of Community Socio Educational Advantage—ICSEA) [
23] was recorded for each school.
The quantitative variables reported in this paper include: student sex, student hat wearing behaviour, teacher sun safety behaviour, environmental factors, school SEIFA and ICSEA status, and student activity levels. Student sex was reported as male or female. Student hat wearing behaviour was reported as a) no hat; b) baseball cap; c) legionnaire hat; d) 360° broad brimmed or bucket hat; or e) sun-safe hat which was the sum of (c) and (d). Teacher sun safety behaviours were reported as a) hat wearing; b) sunglass wearing; c) sun protective clothing (i.e. covered shoulders and collar); and d) shade seeking.
The remaining quantitative variables were reported as correlation alphas to student wearing, or not wearing, a sun-safe hat during recess or lunch break. Environmental factors were shaded area (dichotomous), UV levels and temperature (continuous). The school factor was the school ICSEA value (continuous) and student enrolment numbers. Student activity levels were the proportion of time spent in each of the activity levels (Sedentary, Walking, or Vigorous).
Percentages were for the entire sample and then stratified by sex. Pearson and bi-serial correlations were then calculated for the continuous and dichotomous variables accordingly. This was done to ascertain the relationship between the primary outcome variables and the covariates at the smallest discernible level. All data were analysed using Statistical Package for Social Science (SPSS) version 22.