Table 1.

Multivariate linear regression with UpToDate usage as the dependent variable

Variable	Coefficients	P value
Cohort	−0.04	0.21
Year at enrolment	0.12	<0.001
Own any device	0.07	0.29
Own smartphone	−0.03	0.49
Hours devoted to school	0.00	0.91
Google use frequency	−0.02	0.30
UpToDate use frequency	0.00	0.70

The table shows a multivariable linear regression with average daily topic viewing frequency (natural logarithm transform) as the dependent variable. The dependent variable was calculated as ‘number of UpToDate topics viewed’ / ‘days with an active subscription’ for each user. It was set to zero for users who did not log on to UpToDate. The dependent variable was transformed with the equation Y’=ln(Y+1) to approximate normality. The independent variables were set as follows: ‘Cohort’ was set to one for students enrolling in 2015–2016 and 2 for student enrolling in 2016–2017. ‘Year at enrolment’: PCL1 was set to 1, PLC2 was set to 2, Doc1 was set to 3, Doc3 was set to 5 and Doc4 was set to 6. ‘Own any device’ and ‘Own smartphone’ were set to 0 if the student did not report ownership and to 1 if they did. ‘Hours devoted to school’ is a sum of student reported hours spent in the classroom, in clinical activities, and on studying. ‘Google use frequency’ and ‘UpToDate use frequency’ were set based on student responses at the time of enrolment (before UpToDate subscriptions were given to them). They were set to 4 if student replied ‘almost every day’, 3 if ‘a few times per week’, 2 if ‘a few times per month’, 1 if ‘a few times per year’ and 0 if ‘never’ or ‘i don’t know this resource'. P value bolded if <0.05.