Brain HealthResearch PaperOpen Access

Pre-existing Dementia Raises Cancer Death Risk by 28% in Large Danish Cohort

A study of 100,000+ Danish cancer patients finds dementia sharply raises both cancer-specific and all-cause mortality, demanding updated clinical guidelines.

Thursday, October 1, 2026 1 view
Published in Lancet Healthy Longev
An elderly woman in a hospital gown sitting across from a physician at a clinical consultation desk, medical charts visible in the background

Summary

Researchers used Denmark's national health registries to compare 17,425 people who had dementia before a cancer diagnosis with 86,823 matched cancer patients without dementia. Within five years, 44.0% of those with dementia died of their cancer, versus 39.4% of those without. Deaths from any cause were far higher: 79.7% versus 57.6%. The gap was widest for colon, rectal, and breast cancers. It was smallest for cancers caught at a localized stage. Younger patients with dementia were hit hardest in relative terms. Survival improved over 25 years in both groups. The authors say cancer screening and treatment guidelines for people living with dementia are needed, since these patients are underrepresented in cancer research.

Detailed Summary

As populations age, cancer and dementia increasingly occur in the same person, yet cancer guidelines rarely address that overlap. Dementia affects an estimated 40 to 50 million people worldwide, and lifetime dementia risk after age 55 has been estimated at 42%. Earlier work shows people with dementia are less likely to be diagnosed with cancer, but little was known about how they fare once cancer is found. This study addresses that gap, which matters for both patients and clinicians weighing treatment intensity.

Researchers used linked Danish national registries covering all residents, with free universal health care. They identified 996,689 people diagnosed with a first cancer between 1995 and 2022 and found 17,425 who had a hospital-based dementia diagnosis beforehand. Each was matched at a 5:1 ratio to cancer patients without dementia (86,823) by cancer site, age, sex, and calendar period. The median age at cancer diagnosis was 80, 52.1% were women, and 44.6% had Alzheimer's disease. Follow-up ran through 2023.

Five-year cancer-specific mortality was 44.0% with dementia versus 39.4% without, a 4.5 percentage point difference. All-cause mortality showed a much larger gap: 79.7% versus 57.6%, a 22.2 point difference. Overall, 85.7% of patients with dementia died during follow-up versus 71.4% of comparators. The excess cancer-specific mortality was greatest in colon cancer (13.8 points), rectal cancer (14.8 points), and breast cancer (8.9 points). Non-melanoma skin cancer showed no excess, and the authors suggest incomplete registry capture of these cancers.

After adjustment for cancer stage and comorbidity, dementia was linked to a 28% higher hazard of cancer-specific death (HR 1.28, 95% CI 1.24–1.31) and roughly double the hazard of death from any cause (HR 2.03, 1.98–2.07). The cancer-specific hazard was 1.36 for all-cause dementia and 1.15 for Alzheimer's disease. The association was strongest in patients younger than 50 (cancer-specific HR 1.65; all-cause HR 3.01) and weakened with age (HR 1.17 at 80 or older). It was smaller for localized disease (HR 1.10) than for regional disease (HR 1.43).

Five-year cancer-specific mortality fell substantially over the study period, from 54.3% in 1995–99 to 38.2% in 2015–19 among patients with dementia, and from about 50% to 32.7% in comparators. This suggests treatment advances have benefited both groups. The authors interpret the smaller gap in localized cancers as support for early detection where appropriate. They propose that decisions about screening and treatment in people with dementia should weigh survival, quality of life, and caregiver input, and that formal guidelines are needed.

Several caveats apply. Vascular and other non-Alzheimer dementia subtypes could not be separated reliably and were grouped as all-cause dementia. The study also had no data on dementia severity and lacked information on screening, treatment received, and social support. Median follow-up was short (1.2 years with dementia), and lifestyle confounders were unmeasured. As an observational registry study, it shows association rather than causation. Less aggressive treatment, frailty, and delayed diagnosis may all contribute, and the data cannot distinguish among them.

Key Findings

  • Five-year cancer-specific mortality was 44.0% in patients with pre-existing dementia vs 39.4% in matched patients without dementia (risk difference 4.5 points, 95% CI 3.7–5.4)
  • Five-year all-cause mortality was 79.7% vs 57.6%, an excess of 22.2 percentage points; 85.7% vs 71.4% died during follow-up
  • Adjusted hazard of cancer-specific death was 28% higher with dementia (HR 1.28, 95% CI 1.24–1.31); all-cause mortality hazard was about doubled (HR 2.03, 1.98–2.07)
  • Excess cancer-specific mortality was largest in rectal (14.8 points), colon (13.8 points), and breast cancer (8.9 points)
  • Relative risk was highest in patients under 50 (cancer-specific HR 1.65; all-cause HR 3.01) and fell with age (HR 1.17 at 80 or older)
  • Cancer-specific HR was 1.10 for localized disease vs 1.43 for regional disease, supporting the value of early detection
  • Five-year cancer-specific mortality in patients with dementia fell from 54.3% (1995–99) to 38.2% (2015–19), a trend mirrored in patients without dementia

Methodology

This was a nationwide, registry-based matched cohort study in Denmark covering cancers diagnosed 1995–2022 (N=996,689 screened). It compared 17,425 patients with pre-existing dementia with 86,823 patients without dementia, matched 5:1 by cancer site, age, sex, and calendar period, with follow-up through 2023. Cumulative incidences used Aalen–Johansen (cancer-specific, with competing risks) and Kaplan–Meier (all-cause) estimators. Hazard ratios came from Fine-Gray and Cox models adjusted for cancer stage and a modified Charlson Comorbidity Index, with subgroup analyses by dementia type, stage, cancer type, sex, age, comorbidity, and period.

Study Limitations

Vascular and other non-Alzheimer dementia subtypes could not be assessed separately because ICD codes validate poorly for them; they were grouped within all-cause dementia, while Alzheimer's disease was analyzed separately. Dementia severity was not recorded. Data on cancer screening, treatment received and its timing, and caregiver and support systems were unavailable. Lifestyle confounders were unmeasured, which could bias the estimates. Median follow-up was short (1.2 years with dementia), non-melanoma skin cancer is imperfectly captured in registries, and the observational design precludes causal conclusions. The funders had no role in the study.

Enjoyed this summary?

Get the latest longevity research delivered to your inbox every week.

Enter your email to subscribe: