Background and aim. Intensive care unit (ICU) demand is rising while long-term outcomes remain heterogeneous. This project aims to (1) identify risk factors for ICU admission and subsequent outcomes during and after critical care and (2) develop and evaluate risk-assessment tools to support triage and post-ICU follow-up. Two overarching questions guide the work: (i) To what extent do respiratory, circulatory, endocrine, kidney, infectious, and cancer-related diseases and socioeconomic circumstances influence ICU admission and mortality, and can long-stay and other outcomes be predicted using new models? (ii) Do ICU survivors experience elevated risks of organ-specific complications, and are school performance, socioeconomic status, and quality of life affected after intensive care? Design and data. We will conduct a series of registry-based longitudinal cohort studies using nationwide Swedish data. The ICU cohort comprises all admissions recorded since 2005, with Statistics Sweden generating a pseudonymized key and selecting ten population controls per ICU patient matched at the patient’s admission date. Linked registries provide longitudinal information on healthcare encounters and diagnoses, dispensed medications, mortality and causes of death, malignancies, social insurance outcomes (e.g., sickness absence, disability pension), and disease-specific quality registries relevant to diabetes, renal disease, and long-term respiratory support. Variable specifications follow the respective registry holders. Individual-level data are pseudonymized before delivery; the key remains with Statistics Sweden. Exposures and outcomes. Exposures include pre-existing and incident morbidities across organ systems, pharmacologic treatments, demographics, migration/refugee status, and socioeconomic indicators. Definitions draw on ICD-10 diagnoses, ATC codes, and registry metadata. Primary outcomes are ICU admission (risk and characteristics), short- and long-term mortality, prolonged ICU stay, and post-ICU complications in respiratory, circulatory, renal, infectious, and cancer-related domains. Secondary outcomes capture functional and social consequences, including sickness absence and other social insurance indicators; pediatric analyses will consider educational outcomes where available. Adults and children will generally be analyzed separately. Analytical approach. Directed acyclic graphs (DAGs) will inform identification of confounding structures and adjustment sets. Depending on the question, we will use generalized linear models, time-to-event methods (Cox, flexible parametric survival, competing risks), multinomial/ordinal models for cause-specific endpoints, and models for recurrent events where relevant. Prediction models for long stay and post-ICU outcomes will be developed with internal validation (bootstrap or cross-validation), and, where feasible, temporal or cohort-split validation. Performance will be summarized using discrimination (e.g., AUC/C-index) and calibration (plots, calibration slope, calibration-in-the-large). Missing data will be handled using multiple imputation under plausible mechanisms; residual confounding will be probed via quantitative bias analysis where applicable. Heterogeneity by age, sex, calendar period, and clinical strata (e.g., diagnostic groups, severity) will be assessed through interaction terms and subgroup analyses. Robust or clustered standard errors will account for repeated admissions within individuals. Governance, ethics, and transparency. Data handling complies with Swedish law and GDPR under approvals from the Swedish Ethical Review Authority and registry holders. Individual-level data cannot be shared publicly; transparency will be ensured through openly available analysis code, metadata, and aggregated outputs. The Uppsala University is data controller; all processing occurs within secure university environments with strict access control.