Census Records Reveal Which Women Face Worse Breast Cancer Outcomes in Northern Ireland

A decade-long study of nearly 4,000 women diagnosed with breast cancer in Northern Ireland has uncovered a stark and troubling pattern: where a woman sits on the social ladder, and whether she lives with a hearing disability, can measurably change her chances of surviving the disease. The research, published in BMC Cancer by a team from Queen’s University Belfast and the Northern Ireland Cancer Registry, is among the first to link individual census records directly to cancer registry data, allowing scientists to move beyond the blunt instruments of postcode-level deprivation and examine health inequality at the level of the person.

The study drew on women newly diagnosed with breast cancer between 2011 and 2021, identified through the Northern Ireland Cancer Registry. Crucially, the researchers linked a representative sample of these patients to their responses in the 2011 Northern Ireland census, a resource that captures detailed information about occupation, disability, household composition and other social characteristics recorded before or around the time of diagnosis. This linkage was made possible through the Northern Ireland Longitudinal Study, a large, anonymised data resource funded by the Public Health Agency and the Economic and Social Research Council that connects census records to health events over time.

The primary outcome was breast cancer-specific death, tracked up to March 2023, meaning the team could follow women for more than a decade after diagnosis. To quantify the effect of social factors on survival, the researchers employed Cox proportional hazards regression, a statistical technique that estimates hazard ratios: the relative rate at which an event, in this case death from breast cancer, occurs in one group compared with another over time. A hazard ratio above 1 indicates an elevated rate of death, while confidence intervals that exclude 1 signal a finding unlikely to be due to chance alone. The models adjusted for potential confounders, variables such as age that might independently influence both the social factor and the outcome, helping to isolate the specific contribution of each measure of inequality.

The results on disease stage at diagnosis produced one of the study’s more counterintuitive findings. Women living with one or more disabilities actually had lower odds of being diagnosed with stage 4 disease, the most advanced stage in which cancer has spread to distant organs, compared with women reporting no disability. The adjusted odds ratio was 0.64, with a 95 percent confidence interval of 0.46 to 0.91 and a P value of 0.012, indicating a statistically significant reduction of roughly a third in the odds of late-stage presentation. The researchers used logistic regression for this analysis, a method suited to binary outcomes such as whether or not a cancer is detected at stage 4. By contrast, the odds of stage 4 diagnosis were broadly similar across socio-economic classification categories, suggesting that the social gradient in breast cancer outcomes in this population is not primarily driven by differences in how advanced the disease is when first found.

That distinction matters because it shifts attention away from screening access and early detection alone and toward what happens after diagnosis. When the team turned to survival, the picture changed considerably. Women with any disability overall did not experience a significantly different rate of breast cancer-specific death compared with women without disability; the adjusted hazard ratio of 1.08, with a confidence interval spanning 0.89 to 1.32, included the possibility of no effect. But when disabilities were examined by type, a specific and sobering signal emerged: women with hearing disabilities had a 36 percent higher rate of breast cancer-specific death, with an adjusted hazard ratio of 1.36 and a confidence interval of 1.04 to 1.78 that excludes the null value.

The finding on hearing disability deserves particular scrutiny. Communication barriers in clinical settings are a plausible mechanism: patients who struggle to hear consultations may have more difficulty understanding treatment plans, asking questions, reporting side effects or navigating the complex sequence of surgery, radiotherapy, chemotherapy and endocrine therapy that modern breast cancer care demands. Hearing loss may also compound the cognitive load of medical decision-making at a moment of acute stress. While the observational design of the study cannot prove causation, the specificity of the association, absent in the broader disability category, suggests that sensory impairment deserves targeted attention from oncology services rather than being subsumed under a generic disability label.

The socio-economic gradient in survival was equally consequential. The researchers used the National Statistics Socio-economic Classification, the standard UK framework that assigns occupations to classes ranging from higher managerial and professional roles down to routine and semi-routine occupations. Women in routine or semi-routine occupations had a 34 percent higher rate of breast cancer-specific death than women in managerial occupations, with an adjusted hazard ratio of 1.34 and a confidence interval of 1.05 to 1.72. Because the odds of stage 4 diagnosis were similar across these classes, the survival gap likely reflects differences in treatment receipt, treatment adherence, comorbidities, or the biological and behavioural factors that influence prognosis once cancer is detected, rather than differences in how early the disease is caught.

Methodologically, the study’s strength lies in its individual-level design. Much of the health inequalities literature relies on area-based deprivation indices, which assign a single deprivation score to small geographic units and can obscure variation within neighbourhoods. By anchoring exposure measures in census responses, the Belfast team reduced this misclassification and captured dimensions of disadvantage, such as disability type and occupational class, that area measures cannot distinguish. The population-based cohort design, drawing on registry records covering an entire nation rather than a selected hospital population, further limits selection bias. The trade-off is that census data were collected in 2011, so some women’s circumstances may have changed between the census and their diagnosis, a form of exposure misclassification that would typically bias results toward finding no effect, making the observed associations more notable.

The study also had limitations inherent to its design. Breast cancer-specific death requires accurate attribution of cause of death, and competing causes of mortality can complicate interpretation, particularly among older women. The cohort of 3,892 women, while substantial, means that some subgroup analyses, such as those for individual disability types, rest on smaller numbers and wider confidence intervals. The authors note that the work uses anonymised data provided by patients and collected by Health and Social Care Northern Ireland, with ethical approval granted on the basis that only anonymised information was made available to researchers. The research was funded by Breast Cancer Now, which had no role in study design, data collection, analysis or the decision to publish.

The implications reach well beyond Northern Ireland. Universal health systems are often assumed to equalise cancer outcomes, yet this study demonstrates that even within a system with no financial barrier to care, social position and sensory ability leave visible fingerprints on survival. The findings argue for interventions at specific points in the cancer pathway: communication support and accessible information for patients with hearing loss, and closer attention to treatment completion and follow-up among women in routine and manual occupations. As linked administrative data resources of this kind expand across the United Kingdom and beyond, researchers will increasingly be able to test whether these patterns replicate in other populations, and whether targeted service changes can close a survival gap that, on current evidence, is neither small nor inevitable.

Subject of Research: Individual-level health inequalities in breast cancer stage at diagnosis and survival in Northern Ireland

Article Title: Individual-level health inequalities in breast cancer stage and survival in Northern Ireland: a population-based cohort study

Article References: McShane, C. M., Bennett, D., McIntosh, S. A., Lohfeld, L., Middleton, D. R., McBrien, A., McCallion, D., Savage, G., Fitzpatrick, D., Gavin, A., Baxter, S., Sharma, M., & Cardwell, C. R. (2026). Individual-level health inequalities in breast cancer stage and survival in Northern Ireland: a population-based cohort study. BMC Cancer. https://doi.org/10.1186/s12885-026-17082-z

Image Credits: AI Generated

DOI: 10.1186/s12885-026-17082-z

Keywords: breast cancer, health inequalities, survival, cancer stage, socio-economic status, disability, hearing loss, Cox regression, census linkage, cancer registry, Northern Ireland, cohort study