Background: Exact numbers of breast cancer recurrences are currently unknown at the population level, because they are challenging to actively collect. Previously, real-world data such as administrative claims have been used within expert- or data-driven (machine learning) algorithms for estimating cancer recurrence. We present the first systematic review and meta-analysis, to our knowledge, of publications estimating breast cancer recurrence at the population level using algorithms based on administrative data.Methods: The systematic literature search followed Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines. We evaluated and compared sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy of algorithms. A random-effects meta-analysis was performed using a generalized linear mixed model to obtain a pooled estimate of accuracy.Results: Seventeen articles met the inclusion criteria. Most articles used information from medical files as the gold standard, defined as any recurrence. Two studies included bone metastases only in the definition of recurrence. Fewer studies used a model-based approach (decision trees or logistic regression) (41.2%) compared with studies using detection rules without specified model (58.8%). The generalized linear mixed model for all recurrence types reported an accuracy of 92.2% (95% confidence interval = 88.4% to 94.8%).Conclusions: Publications reporting algorithms for detecting breast cancer recurrence are limited in number and heterogeneous. A thorough analysis of the existing algorithms demonstrated the need for more standardization and validation. The meta-analysis reported a high accuracy overall, which indicates algorithms as promising tools to identify breast cancer recurrence at the population level. The rule-based approach combined with emerging machine learning algorithms could be interesting to explore in the future.

This study aims to describe end-of-life (EOL) care in older patients with cancer and investigate the association between geriatric assessment (GA) results and specialized palliative care (SPC) use. Older patients with a new cancer diagnosis (2009–2015) originally included in a previous multicentric study were selected if they died before the end of follow-up (2019). At the time of cancer diagnosis, patients underwent geriatric screening with Geriatric 8 (G8) followed by GA in case of a G8 score ≤14/17. These data were linked to the cancer registry and healthcare reimbursement data for follow-up. EOL care was assessed in the last three months before death, and associations were analyzed using logistic regression. A total of 3546 deceased older patients with cancer with a median age of 79 years at diagnosis were included. Breast, colon, and lung cancer were the most common diagnoses. In the last three months of life, 76.3% were hospitalized, 49.1% had an emergency department visit, and 43.5% received SPC. In total, 55.0% died in the hospital (38.5% in a non-palliative care unit and 16.4% in a palliative care unit). In multivariable analyses, functional and cognitive impairment at cancer diagnosis was associated with less SPC. Further research on optimizing EOL healthcare utilization and broadening access to SPC is needed.

12041 Background: Long-term outcomes after cancer diagnosis in older persons are largely unexplored because of limited follow-up in clinical studies. By linking clinical data with population-based data, studying long-term outcomes in large cohorts becomes feasible. The current study aims to explore long-term outcomes in the care trajectory of older patients with cancer and to assess their association with baseline geriatric screening and assessment (GS/GA) results. Methods: A large cohort study of older patients with a new cancer diagnosis was set up by linking clinical, cancer registry and administrative health data based on a unique patient identifier. Clinical data were derived from a previously performed prospective multicentric Belgian study (2009-2015). Patients aged ≥ 70 years were screened with G8 followed by GA in case of an abnormal result (≤14/17). Tumor characteristics and vital status were derived from cancer registry data and long-term outcomes (general practitioner (GP) contacts, hospitalizations and nursing home transfers) from administrative health data. In patients that survived at least 3 months since inclusion, outcomes were assessed from the day after inclusion until 3 years after. Event rates were calculated using person-time at risk to allow for varying follow-up time. Patients were censored 3 months before death to exclude influence of end-of-life care. Results: After data linkage, 6,391 older patients with a new cancer diagnosis were available for this analysis. The median age was 77 (range: 70–100) and 59.8% was female. Diagnoses included solid (92.8%) and hematologic malignancies (7.2%). In the patients with a solid tumor, breast, colorectal and lung cancer were the most common and 20.1% of patients had stage IV. 64.3% of patients had an abnormal baseline G8 score. During the 3 year follow-up, 2,602 (40.7%) of the included patients died. In these 3 years, 5,985 (95.2%) patients had at least one contact with a GP and 4,634 (72.5%) had at least one new hospital admission (event rates in Table 1). Of the 3,724 patients living independently at inclusion and still alive after 3 years, 281 (7.5%) had been transferred to a nursing home and of those, 240 (85.4%) patients had an abnormal baseline G8 score. Conclusions: Older patients with an abnormal baseline G8 score have more GP contacts, hospital admissions and nursing home transfers in the 3 years following a new cancer diagnosis compared to patients with normal baseline G8 score. Baseline G8 could help identify patients at risk for higher long-term healthcare utilization.