Background. The Northern Sweden Health and Disease Study (NSHDS) is a population-based, prospective, longitudinal cohort including approximately 1.5 million person-years of follow-up and more than 240?000 biobanked blood samples. It was initiated in the mid-1980s and continues to include participants. The NSHDS has three subcohorts. The largest is the Västerbotten Intervention Program (VIP), an open cohort based on a cardiometabolic prevention programme integrated within primary healthcare and targeting 40-, 50-, and 60-year-olds (currently 127 069 of NSHDS participants). The other two are the Northern Sweden Monitoring of Trends and Determinants of Cardiovascular Disease (MONICA) Study and the Mammography Screening Project (MSP). In addition to research blood samples, participants provide health measurements, such as anthropometry, blood pressure, blood lipids and glucose tolerance, and extensive health and lifestyle questionnaire data.  

Objectives. The aim of this study is to assess selection bias in the NSHDS. Given the full-population intent in VIP (annual participation rates 50-80%), a substantial proportion of the population of Västerbotten is included in the cohort, renderingofficial full-population data insufficient for comparison. Therefore, we will compare NSHDS participants to eligible non-participants with respect to demographic and health-related factors. 

Methods. The study population was defined for each calender year from 1986 to present and included residents of Västerbotten and Norrbotten who were eligible for at least one of the NSHDS cohorts, based on age, geographical region and, for the MSP, female sex. We obtained data on education, socioeconomic status, birth country and other demographic variables from Statistics Sweden and extensive diagnosis and prescription drug data from registries held at the Swedish Board of Health and Welfare. Clinical data for prospective disease cases will be requested from national quality registries and have already been acquired for breast cancer, from the Swedish Breast Cancer Registry. A complete list of NSHDS participants was sent to Statistics Sweden for creation of indicator variables, allowing for delivery of fully pseudonymized registry. Ethical approval was obtained from the Swedish Ethical Review Authority.  

Results. Data were delivered very recently and comprise, for example, around 143?000 observations from the National Cancer Registry, 6?200 000 observations from the National Patient Registry and 130?000?000 observations from the Prescribed Drug Registry. Data cleaning is ongoing. Data analysis will involve basic descriptive statistics with extensive subgrouping, as well as graphs to visualize time trends in selection bias. 

Conclusions/Implications. These large data resources will allow rigorous characterizations of differences between NSHDS participants and eligible non-participants. We are planning one major paper presenting results for the full NSHDS andrelevant subgroups, followed by a series of articles focusing on specific types of cancer and other diseases. The results will support interpretation of results and planning of future studies utilizing the valuable NSHDS resources.