Demographic change is leading to a higher proportion of older adults in most parts of the world. A minority of older adults have poor health, but this group has high care needs due to frailty and/or multimorbidity. Guidelines for the management of frailty emphasise early detection of frailty and recommend comprehensive care approaches in primary care, but the evidence for these interventions is low. To provide effective and individualised care, the health system needs to identify these patients and develop proactive interventions to improve quality of life and avoid treatments that are of no benefit to the individual.  

The aim of this thesis was to study the effects of a proactive primary care working model in which vulnerable older adults were identified and received individually tailored care, using an adaptation of comprehensive geriatric assessment (CGA). 

Methods: A pragmatic controlled trial was conducted in 19 primary care practices in Sweden from 2017 to 2020. A predictive model, using electronic medical records to assess the risk of hospital admission, selected participants at high risk. Participants in the intervention practices were offered a comprehensive geriatric assessment in their primary care practice and subsequent follow-up by a team consisting of a nurse and the patient's doctor. A new CGA tool - PASTEL (Primary care ASsessment Tool for Elders) was used for assessment and care planning. The primary outcome for the intervention was hospital care days and secondary outcomes were hospital care episodes, mortality, outpatient visits, healthcare costs and cost-effectiveness. The outcomes were adjusted for age, sex and risk score and ana-lysed according to intention-to-treat. 

The predictive model was validated, and performance was assessed using the C-statistic. Focus group interviews were conducted to explore primary care nurses' and doctors' experiences with the new tool PASTEL. 

Results: 1304 older adults were included in the trial. The mean age was 82.2 years, 51% were female. During the follow-up period of 24 months, the relative risk reduction of hospital care days in the intervention group was - 22% (CI 95% = -35% to - 4%, p = 0.02) compared with usual care. There was no significant difference in mortality and outpatient visits. The reduction in healthcare costs was - € 4324 (- € 7962 to - € 686, p = 0.02). The intervention was cost-effective compared with usual care, mainly due to lower costs.

The predictive model had an AUC of 0.69 (CI 0.68- 0.70). Primary care staff considered PASTEL valuable and feasible in the primary care context.

In conclusion, the results of this thesis indicate that vulnerable older adults at risk of hospitalisation can be identified by a predictive model. Proactive intervention with a comprehensive geriatric assessment adapted to pri-mary care can reduce the need for hospital care. Future studies in similar contexts are needed to determine whether these results are generalisable.