Introduction: Inflammatory diseases (e.g. eczema, psoriasis, asthma, rheumatoid arthritis) and their treatments (e.g. glucocorticoids, targeted immune-modifying therapies) are associated with adverse health outcomes. Population-based studies in electronic health records data are suitable to study associations with many of these adverse health outcomes, and when well conducted can provide actionable evidence for decision makers, e.g. for which outcomes screening or preventive measures should be put in place. Objectives: Using UK routinely collected health data, I aimed to 1. generate evidence on multiple outcomes to inform clinical care for people with eczema and other inflammatory diseases, 2. assess the validity of disease definitions by using linkage between data sources, and 3. efficiently conduct studies on multiple adverse outcomes. Methods: 1. I investigated associations between inflammatory diseases or their treatments and various outcomes, making use of cohort study and cross-sectional designs and implementing different exposure definitions in the OpenSAFELY, CPRD GOLD and Aurum, and UK Biobank databases. 2. I assessed agreement concerning disease definitions between population cohorts (ALSPAC, UK Biobank) and linked electronic health records data and attempted to predict eczema subtypes in ALSPAC using linked EHR data. 3. I developed an approach to conduct studies on multiple outcomes and applied this to investigate adverse health outcomes for people with eczema. Results: 1. I found evidence for associations between immune-mediated inflammatory diseases (IMIDs), but not most targeted immune modifying drugs, and severe COVID-19 outcomes. 2. I found that in older adults who receive large cumulative doses of oral glucocorticoids, those who receive them in low-intensity patterns (intermittently, over a longer period of time or a larger number of prescriptions or with more gaps between prescriptions) were less likely to be prescribed recommended fracture preventive care (e.g., bisphosphonates) 3. I found evidence for an association between eczema/psoriasis and anxiety/depression across multiple study designs and data sources. 4. I found that there was considerable disagreement between eczema diagnoses derived from EHRs and questionnaires, and that agreement was better for other conditions including psoriasis and asthma. The poor agreement precluded using EHRs to predict eczema subtypes derived from questionnaires. 5. I showed that having eczema was associated with the subsequent development of several different adverse health outcomes, including a strongly increased risk of atopic and allergic conditions, skin infections and some immune-mediated skin conditions, a moderately increased risk of some liver and gastrointestinal conditions, a weakly increased risk of some cardiovascular, neurological and other outcomes and no increased risk of cancers, except lymphomas. Conclusions: I approached electronic health records research on adverse health outcomes of inflammatory conditions from multiple angles. I generated evidence to inform clinical care, highlighted a need for further research into eczema definitions in observational studies, and demonstrated and put to use an efficient approach to investigate multiple outcomes.