Stress, Transition-to-Practice, Burnout, and Turnover Intention in Early-Career Nursing (1981–2025): A Bibliometric and Science-Mapping Analysis

Stress during training, transition-to-practice strain, burnout, and turnover intention are recurring early-career nursing concerns and can be viewed in occupational safety and health (OSH) terms as work-related psychosocial hazards. Web of Science records (1970–31 Aug 2025) were screened, yielding 664 eligible articles (earliest eligible year 1981). Using VOSviewer (v1.6.20) and Biblioshiny (v5.0), we mapped the intellectual base via co-citation (minimum co-citation = 12), themes via co-word analysis (minimum occurrence = 15; thesaurus-cleaned keywords), and change over time via thematic evolution (1981–1999, 2000–2009, 2010–2019, 2020–2025). Five co-citation clusters centered on transition theory, stress/coping, burnout measurement, resilience, and organizational determinants of retention. Co-word mapping yielded four thematic clusters spanning transition supports, student stress/mental health, early-career burnout/turnover intention, and well-being/self-efficacy interventions. The literature frequently links stress, transition-to-practice strain, burnout, and turnover intention; when these constructs are framed in a sequential narrative, we interpret this as thematic convergence rather than evidence of a causal or developmental mechanism. Retention was often operationalized via intention rather than observed turnover. The mapped literature depicts an interconnected psychosocial hazard domain linking education and early practice to retention-risk signals with implications for staffing stability, patient safety, and safety culture. We propose a three-pillar heuristic synthesis: structured residency/preceptorship with protected preceptor time; longitudinal screening using validated stress/burnout tools; and psychologically safe leadership. This framework is an interpretive synthesis informed by mapped themes and central empirical literature, not a validated intervention model and not a direct algorithmic output of the clustering procedure.