Predictive value of cancer for dementia in cohorts with and without T2DM: a national observational study

Application Number
Lay Summary

Previous studies have reported that people with cancer are less likely to develop late-onset dementia (occurs from 65 years old), especially dementia attributable to Alzheimer’s disease. The reported protective effect shows a lower risk of cancer among those diagnosed with dementia and vice-versa. The underlying reasons for this relationship remain largely unclear. Some research suggests that factors which increase the risk of cancer, can decrease the risk of dementia. Type 2 diabetes (T2DM) is an important risk factor for both diseases, but it is to date unclear whether it affects the relationship observed between cancer and dementia

The purpose of this study is to use a large national database to explore the relationship between cancer and dementia in a primary care population ≥ 65 years in individuals with and without T2DM.

Findings from this study will provide a greater understanding of common and increasingly prevalent diseases which co-occur, and how they relate to each other in terms of risk. In particular we anticipate that the data produced could pave the way for new observational studies which focus on biomarkers and or lifestyle interventions. This will aid the understanding of the mechanisms underlying the relationships between these common diseases and potentially demonstrate opportunities for reducing their risks.

Technical Summary

This study will be a retrospective open cohort study. The overall aim is to examine the relationship between cancer and late onset dementia separately in a T2DM and non-T2DM cohort, using routine English primary care data. Our outcome of interest will include participants with a diagnosis of dementia as well as those with a diagnosis of dementia attributable to Alzheimer’s disease. We will first aim to estimate the incidence rates of cancer and dementia separately in subpopulations with and without T2DM. We will then estimate the incidence rates of diagnosed dementia among those with and without cancer. Cox proportional hazard models, with time-dependent covariates, will be used to determine the risk of overall dementia and dementia likely attributable to Alzheimer’s disease in those with and without a cancer diagnosis. The hazard ratios (HRs) will be adjusted to account for various possible confounders identified in the literature. We will use propensity score analysis to correct for confounding. Additionally, Fine and Gray competing risk models with sub-distribution Hazard Ratios (sdHRs) for overall dementia and dementia attributable to Alzheimer’s disease will be used; along with cumulative incidence risk plots of dementia, to account for death as a competing risk in our study.

Health Outcomes to be Measured

1. Any dementia: record of CPRD medcode or prodcode, HES ICD10 code or ONS ICD9/10 code indicative of any dementia diagnosis (see list in Annex 2). Through other ISAC approved dementia research projects e.g. 16_143, we will also be developing and testing a dementia diagnosis algorithm using codes other than CPRD or HES diagnoses. 2. Dementia likely attributable to Alzheimer’s disease: record of CPRD medcode ,HES ICD10 code or ONS ICD9/10 code indicative of a diagnosis of Alzheimer’s disease (see list in Annex 10).The most recent and currently adopted 2011 NIA-AA Criteria for Alzheimer’s Disease will be used for a more accurate diagnosis: Dementia possibly attributable to Alzheimer’s disease (medcodes indicating a diagnosis for Alzheimer’s disease but with the evidence of another record of a medcode indicating another type of late onset dementia ) and dementia probably attributable to Alzheimer’s disease ( medcodes indicating a diagnosis for Alzheimer’s diseases without a later record of a medcode indicative of another type of late onset dementia).


Michael Soljak - Chief Investigator - Imperial College London
Darina Bassil - Corresponding Applicant - Harvard School of Public Health ( HSPH )
Ailsa McKay - Collaborator - Imperial College London
Azeem Majeed - Collaborator - Imperial College London
David Muller - Collaborator - Imperial College London
Ioanna Tzoulaki - Collaborator - Imperial College London
Lefkos Middleton - Collaborator - Imperial College London
Mahsa Mazidi - Collaborator - Imperial College London
Roger Newson - Collaborator - Imperial College London


HES Admitted Patient Care;HES Outpatient;ONS Death Registration Data;Patient Level Index of Multiple Deprivation;Practice Level Index of Multiple Deprivation