Introduction: Primary care for older adults is facing challenges: there is an urgent need for integration of care, health problems could be identified at a more timely stage, and older adults demand more autonomy in their care process. We aimed to address these challenges by introducing the geriatric care model (GCM), based on the chronic care model (Bodenheimer et al., 2002). Methods: Design & setting 2-year stepped-wedge cluster randomized clinical trial, carried out among 35 primary care practices in West-Friesland and Amsterdam. Geriatric Care Model: Practice nurses acted as linchpins in integrated care for frail older adults who live at home. Every six months older adults received a geriatric comprehensive assessment, followed by a tailored care plan. Participants with complex care needs were reviewed in interdisciplinary consultations. Rationale: GCM addresses challenges (concerning integration, timely identification of health problems, and client autonomy), and this in turn leads to improved patient outcomes. Participants 1147 frail older adults aged 65 years and over. Data collected (every 6 months): 1- Patient outcomes: independence in (i)ADL, quality of life, hospitalisation, (un)met care needs 2- Costs 3- Process outcomes: fidelity, barriers and facilitators to implementation 4- Challenges addressed: Timely identification of health problems and care needs Client autonomy Integration of care Results: Despite the fact that adherence to the main components of the GCM was good (Muntinga et al., 2015), Intention-to-treat analyses based on multilevel modelling adjusted for time and baseline confounders showed no relevant differences between usual care and the GCM on patient outcomes (Hoogendijk et al., 2016), client autonomy, and integration of care; GCM was not cost-effective compared to usual care (van Leeuwen et al., 2015 ); Analyses of care registries among the 781 older persons who received at least one comprehensive geriatric assessment indicated that the GCM addresses timely identification of health problems and care needs. For example: 315 (40 %) older people experienced any type of pain. Practice nurses identified 20 (11 %) new pain cases, and 188 (60 %) older people with pain formulated at least one pain action plan together with a practice nurse. More than half of the older people whose pain had already been identified by a general practitioner wanted a pain action plan. Most pain action plans consisted of actions or agreements related to continuity of care (Muntinga et al., 2016). Discussions: Elements of the model have been adopted and continued in both regions. The outcome measures may have not been able to capture the perceived usefulness of these elements. Conclusions: GCM was not cost-effective compared to usual care. Lessons learned: Design interventions and outcome measures in close collaboration with end users Monitor sites after the end of a study to investigate long-term effects of interventions Limitations: Small study sample hindered subgroup analyses Suggestions for future research: Aggregation of data of similar proactive initiatives may provide necessary insights into outcomes in (subgroups of) older adults. Research on outcome measures is necessary: how to capture the perceived usefulness of integrated care approaches?