*2.3. Variable Definitions*

Variable definitions and data collection protocols included the following. Staff measured blood pressure using an automated device, the Omron HEM-907XL (Omron Healthcare, Lake Forest, IL, USA). This ensured measurement consistency across participants and across survey years. Staff applied the cuff to students' right arm with measurements taken on a single occasion. After five minutes of quiet rest, the automated device took three successive measurements, while the participant sat in a chair with back support, feet flat on the floor, and arm supported with the antecubital fossa at heart level [36]. Staff recorded the first measurement, but it was not included in analysis, since these measurements can be falsely elevated [36]. Consequently, all analyses included only the average of the second and third measurements [36]. The definition of elevated blood pressure was at or above the 90th percentile for age, height, and sex or ≥120/80 mm Hg [36]. Staff recorded height to the

nearest 0.1 cm using a Shorr Board stadiometer (Shorr Productions, Olney, MD). Staff measured weight using a portable Tanita BWB-800S digital scale that was certified to be accurate to 400 pounds each year by a professional scale service and maintenance company (Tanita Corporation, Arlington, IL, USA). Overweight or obese categories were based on a body mass index (BMI) at or above the 85th percentile for age and sex, which follows the BMI-for-age- weight status categories provided by the Centers for Disease Control and Prevention (CDC) [37]. BMI was calculated as weight in kilograms divided by the height in meters squared. Waist circumference was measured to the nearest 0.1 cm using a plastic tape measure that was stretch resistant. A waist circumference at or above the 75th percentile for age, sex, and ethnicity was defined as abdominal obesity [36]. AN is a dark discoloration and/or thickening of the skin the back of the neck that is used as an indicator of high insulin levels or resistance. Staff used a similar approach to assess AN as those implemented in a Texas state-wide school- screening program [36,38,39]. The basis of work status was a self-reported annual work history of job type, dates of employment, and number of hours worked with survey items modelled after prior studies with working youth [40,41]. The recall period began from January 1st of the year prior to data collection with data collection occurring between January through March each year. The definition of current work status was working for pay or not for pay. The current work status definition was a student who reported a having a job during the same week when they participated in the survey. The basis of items pertaining to health behaviors was the Youth Risk Behavior Surveillance System YRBSS [42].

#### *2.4. Data Management and Analysis*

Data were managed using a Microsoft SQL relational database (Microsoft, Redmond, WA, USA). All data were double entered into the database to minimize data entry errors, as well as computer edited for out-of-range and contradictory values. SAS 9.4 was used for all data analysis (SAS Institute Inc., Cary, NC, USA). Descriptive statistics included means and proportions at baseline and years of follow-up. Appropriate statistical tests including chi-square tests, t-tests, and Fisher's exact tests were used to compare the distribution of variables across the two schools and by elevated blood pressure status. Generalized linear mixed models with a logit link function and binomial distribution were used to estimate the probability of high or high normal blood pressure at each time period by current work status after adjusting for potential confounders, including age, gender, AN, BMI, number of days physically active in the past 7 days, and school enrolled. Random intercepts at the individual level were also included in these models to account for potential correlation of repeated measures from the same student. Statistical significance was set a priori at a level of alpha <0.05.

## **3. Results**
