Text this: Impact of time-varying exposure on estimated effects in observational studies using routinely collected data: protocol for a cross-sectional study