Internet dependence as a predictor of mental health among medical students before and during the COVID-19 pandemic: a longitudinal cohort study in Brazil (2019–2021)

Purpose This study aims to compare internet dependence (ID) and mental health outcomes among Brazilian medical students before the COVID-19 pandemic (2019) and during peak distancing (2021), and to assess whether baseline ID predicted mental health outcomes at follow-up. Design/methodology/approach This prospective longitudinal cohort study was conducted at a private medical school in Brazil with two waves (2019, in-person; 2021, online). Instruments included the Internet Addiction Test (IAT), PHQ-9, GAD-7 and the Life Orientation Test-Revised (LOT-R). Paired t-tests and McNemar tests compared within-person changes. Correlations and multivariable linear regression models adjusted for age, sex, family income and mental health service use assessed associations, including change scores (? = T1–T0). Statistical significance was set at p < 0.05. Findings Ninety-four students completed both waves. Mean PHQ-9 scores increased from 10.56 to 14.80 (? = +4.24; p < 0.001), and mean GAD-7 scores increased from 9.74 to 13.08 (? = +3.34; p < 0.001). Optimism declined from 8.54 to 5.42 (p < 0.001), pessimism increased from 6.97 to 7.73 (p = 0.028) and ID increased from 40.59 to 47.43 on the IAT (p < 0.001). At baseline, ID correlated with higher depressive symptoms (r = 0.282; p = 0.038) and anxiety (r = 0.260; p = 0.042). In adjusted regression models, baseline ID predicted higher anxiety at follow-up (ß = 0.699; p = 0.042), but not depressive symptoms at follow-up (ß = 0.087; p = 0.148). Higher baseline ID also predicted greater increases in PHQ-9 (ß = 0.201; p = 0.022) and GAD-7 (ß = 0.398; p = 0.042) scores from 2019–2021. Originality/value This study provides longitudinal evidence from a middle-income country tracking ID and mental health among medical students from a pre-pandemic baseline to the period of peak distancing. By examining both negative psychological symptoms and positive psychological markers such as optimism, the findings underscore digital hyperconnectivity as a salient, modifiable correlate of distress in medical training. The results support routine screening and the integration of digital literacy and self-regulation strategies into medical curricula to mitigate risk and promote well-being.