Hospitals are not only places where childhood cancer is treated—they also generate records of earlier health encounters. A new population-based study now asks whether patterns of hospital contact before a cancer diagnosis can foreshadow the risk of dying early, years before the disease is formally recognized.
Researchers led by Kenborg and colleagues analyzed nationwide health data from Denmark and Sweden, taking advantage of the rich linkage between registry-based hospital records and long-term survival outcomes. Their central question was whether “pre-diagnosis” healthcare use reflects underlying vulnerability that later affects mortality after childhood cancer.
The team compared children who eventually received a childhood cancer diagnosis with those whose hospital contact histories differed in timing and frequency. Instead of focusing solely on the cancer itself, the study treated prior hospital visits as a proxy for earlier disease burden, comorbid conditions, diagnostic delay, or subtle early symptoms that may not have been recognized as cancer.
To estimate early mortality risk, the investigators used cohort methods appropriate for time-to-event outcomes, producing hazard-based measures that quantify how hospital-contact patterns relate to survival. They adjusted for key confounding factors that could distort interpretation, including demographic characteristics and other health-related differences captured in the registries.
Importantly, “hospital contact” was operationalized in a way that captures real-world healthcare behavior rather than symptom reports alone. This design strengthens external validity: it reflects how children actually interact with healthcare systems in the years preceding diagnosis.
From a technical standpoint, registry completeness and consistent coding practices across two countries improve statistical robustness. Cross-national comparison also helps test whether observed associations are driven by system-specific factors or represent broader biological and clinical signals.
The authors report that prior hospital contact is meaningfully associated with mortality soon after diagnosis. In other words, early healthcare utilization before cancer is not merely background information—it may carry prognostic information relevant to survival.
These findings have potential implications for clinical awareness and risk stratification. If future work confirms the mechanisms, clinicians could use pre-diagnosis contact patterns to identify children who may require closer follow-up, expedited diagnostic pathways, or tailored supportive care.
As survival outcomes for childhood cancer improve, understanding predictors of early death becomes increasingly important. This study highlights that the story begins earlier than the cancer diagnosis date—embedded in the hospital record long before treatment starts.
The research appears in Br J Cancer (2026) and is published under DOI: 10.1038/s41416-026-03560-5.
Subject of Research: Hospital contacts before childhood cancer and early mortality risk (Danish and Swedish population-based cohort study)
Article Title: Hospital contacts before childhood cancer and the risk of early mortality: a Danish and Swedish population-based cohort study
Article References: Kenborg, L., Mogensen, H., Hasle, H. et al. Hospital contacts before childhood cancer and the risk of early mortality: a Danish and Swedish population-based cohort study. Br J Cancer (2026). https://doi.org/10.1038/s41416-026-03560-5
Image Credits: AI Generated
DOI: 10.1038/s41416-026-03560-5
Keywords:
Tags: childhood cancer early diagnosischildhood cancer mortality risk factorscomorbid conditions childhood cancerDenmark and Sweden health registry studydiagnostic delay childhood cancerearly symptoms and childhood cancerhazard-based survival analysis childhood cancerhospital contact patterns childhood cancer prognosisimpact of healthcare contact on childhood cancer outcomespopulation-based study childhood cancer survivalpre-diagnosis healthcare utilization childhood cancerregistry-linked health data childhood cancer
