Mapping the Multidimensional Stress Burden in Nursing Students: A Systematic Review and Instrument-Specific Meta-Analysis

Undergraduate nursing students experience stress as a multidimensional construct shaped by academic, clinical, psychosocial and contextual demands throughout training, placing them at increased risk of adverse mental health outcomes. Contemporary literature suggests that stress measures vary by instrument, reflecting distinct constructs (perceived vs. symptom-based). This systematic review and meta-analysis synthesised instrument-specific pooled estimates of moderate-to-high stress among undergraduate nursing students and examined variation by training stage, country income level, geographical region, pandemic period and stress measurement approach. Electronic databases were searched for quantitative studies published between January 2010 and December 2025. Random-effects meta-analyses were conducted separately for studies using the Perceived Stress Scale (PSS), reflecting perceived stress, and the Depression Anxiety Stress Scale (DASS) stress subscale, reflecting affective stress symptoms. Subgroup, sensitivity and meta-regression analyses were undertaken to explore heterogeneity. Forty-two studies involving 19,655 students from 21 countries met the inclusion criteria, with 27 studies contributing to prevalence synthesis and 18 to pooled mean scores. Given the very high heterogeneity and the non-equivalence of instruments, an overall pooled prevalence of 49.98% was calculated only as a descriptive summary. Instrument-specific analyses indicated a substantial prevalence of perceived stress in PSS studies (74.37%) and symptom-based stress in DASS studies (30.46%), while mean scores across both scales suggested moderate stress levels. However, substantial unexplained heterogeneity persisted, indicating the influence of unmeasured contextual, psychosocial and methodological factors across studies. These findings emphasise the wide range and variability of reported stress prevalence, suggesting that a pooled estimate may not adequately capture the underlying complexity. Overall, estimates are presented according to instrument-defined constructs, and cross-instrument comparisons should be interpreted with caution, highlighting the need for measurement-aware, context-sensitive research and interventions.