Table 7 Problems detected in research procedures and solutions generated for the main trial

1

For the control group, as a standard treatment, we gave the participants PA and sitting reduction health education brochures after the baseline measurements ended. The control participants did not find this practice pleasurable or sensible.

Change the control group to be a “no-treatment control”, i.e. refrain from giving any additional brochures to participants.

2

Questionnaire burden. The questionnaires were perceived as too lengthy and boring by the participants.

Decrease length and number of questionnaires:

1) No T2 questionnaire for controls.

2) No questionnaires after each session for intervention participants.

3) Decrease number of questions in all questionnaires.

3

Accelerometers were not returned quickly or at all in pre-paid envelopes.

Vacation periods interfered with return.

Research assistants aim to collect accelerometers directly from schools in person.

Measurements not scheduled too close to vacation period.

4

Several students’ non-participation was due to being ill or other reason for not attending school on the day(s) when research team were in school to collect data.

Reserve enough days to return to schools for bioimpedance and accelerometer measurements, schedule several days for same class in order to maximize participation.

5

The strategies in which students were motivated and instructed to wear the accelerometer were too scarce (in research assistant – participant face-to-face session). Even slight changes from T3 to T4 instructions were related to increase in the days worn.

Improved instructions for RCT: For example, we instructed the participant to immediately put on the accelerometer (instead of just giving it to them), and simplified the self-report log associated with the accelerometer. We also added motivational content to the accelerometer instruction script.

6

Several students cited “not remembering” as the reason for not wearing the accelerometer.

SMS reminders to help participants remember to put on accelerometer in the mornings in the RCT.

7

In recruitment of one class, initial reception of research was positive, but in the session where consent was to be signed and questionnaire filled, negative group norms arose perhaps due to 1) different researchers were present for recruiting and subsequently hosting survey measurement or 2) lag between recruitment and survey.

1) Avoid changes in personnel per class.

Script even more carefully the recruitment session to avoid recruitment bias (to ensure similar recruitment between classes/recruiters).

2) Ask consent and provide survey in the same session as recruitment occurs. Filling in baseline survey right after recruitment and study info.

8

Difficulty in reaching third year students at follow-up (graduation).

1) Include only first or second year students in RCT.

2) Shorten the originally intended 24-month-follow-up into 14 months, to avoid the realistic risk of not locating participants for the last follow-up.

9

M (SD) of the outcome variables.

Power calculations accordingly (we used these data to inform a power calculation for sample size of the RCT).

10

Recruitment rate was 67.2% and drop-out after intervention (T3) in intervention arm and control arm 23.1% and 23.5%. Accelerometer drop-out due to human error by research assistant in preparing the accelerometer devices.

1) Adjust target recruitment rates accordingly.

2) Improve procedures to prevent study drop-out: E.g.: training of interpersonal and communication skills among data collection staff, improve motivation for participating in accelerometer study, add SMS reminders for accelerometer wear time, improve staff skill in preparing accelerometers.

11

Intervention timing was suboptimal, starting at the middle of the fourth (last) period of the school year.

1) As intervention effects need to be investigated in different seasons (PA seasonal effects), intervention start cannot be timed at the beginning of the school year for all of the RCT batches. Instead, ensure that intervention activities start promptly after each period starts.

2) Teacher activities to be placed in the beginning of school years.