Electronic Thesis and Dissertation Repository

Thesis Format



Doctor of Philosophy


Epidemiology and Biostatistics


Hachinski, Vladimir

2nd Supervisor

Stranges, Saverio


Background: Scarce information exists about whether the relationship between multimorbidity and cognitive decline has changed over the past decades.

Objectives: 1) To summarize knowledge about the association of multimorbidity and cognitive decline by performing a systematic review. 2) To estimate the association between multimorbidity in two cohorts selected three decades apart. 3)To identify which multimorbidity combinations have the strongest associations with cognitive decline 4) To identify protective factors that reduce the risk of cognitive decline in the presence of multimorbidity.

Methods: We performed a systematic review following the PRISMA statement. We then addressed objectives 2-4 by analyzing data from two longitudinal studies. The Canadian Study of Health and Aging (CSHA, n = 497) collected a baseline in 1991. The Canadian Longitudinal Study of Health and Aging (CLSA, n = 23654) had a baseline in 2015. Both studies collected information on several chronic conditions and used validated measures of different cognitive domains. Incident dementia was available in the CSHA only. Statistical models included multilevel linear and logistic regression and classification and regression trees (CART).

Results: We identified 19 publications evaluating the relationship between multimorbidity and cognitive decline, of which 17 studies reported statistically significant results for this association. We found no association between multimorbidity and cognitive scores in the CSHA, while in the CLSA, we found associations with frontal function (ß:–0.049) and RAVLT (ß:–0.05). We did not find an association between multimorbidity and 5-year dementia incidence in the CSHA. In the CLSA, CART identified cardiopathies and stroke as part of the multimorbidity combinations associated with the lowest cognitive test scores. Finally, in the CSHA, physical activity decreased dementia risk (OR = 0.45, 95% CI: 0.20–0.98).

Conclusion: Our systematic review supports the association between multimorbidity and cognitive decline without evidence of change over time. We found an association of multimorbidity on cognitive scores only in the ongoing CLSA; future research should clarify this apparent increasing effect of multimorbidity on cognition. We noticed that multimorbidity combinations containing cardiopathies and stroke were associated with low cognitive scores. Finally, we found that regular exercise reduced the risk of dementia in multimorbidity patients.

Summary for Lay Audience

Background: An increasing number of older adults live with more than one chronic disease at the same time, a concept known as multimorbidity. In the past few decades, we have learnt that multimorbidity may increase the chances of loss of mental capacity and developing dementia. Our objectives were to evaluate if this relationship has changed in the past three decades. We began by searching for all the published papers on the topic using a systematic review. Then we analyzed data collected about thirty years apart in Canada to see if certain combinations of diseases increase the possibility of mental decline. We also looked for characteristics that could protect people with multimorbidity from developing dementia.

Methods: We searched published research to see how many previous researchers have looked at this topic. We analyzed data from two community studies in Canada. First, the Canadian Study of Health and Aging (CSHA, 1991–1996), and second, the Canadian Longitudinal Study on Aging (CLSA, 2015–2018). We determined whether people with multimorbidity have a higher risk of cognitive decline or lower scores on mental tests using appropriate statistical analyses that let us control for factors other than multimorbidity that may be responsible for cognitive decline. We used different statistical methods to see if we got similar or different results.

Results: We found 19 articles on the topic, of which 17 suggest that multimorbidity increases cognitive deterioration. We found that in the CSHA, there was no association, whereas in the CLSA, there was an association between multimorbidity and lower cognitive scores. We found that one of the statistical methods, called Classification and Regression Trees (CART) found multiple combinations of diseases associated with worse cognitive scores. We noticed that many disease combinations linked with low cognitive scores included heart diseases and stroke. Finally, for the protective factors, we found that exercise reduced the risk of dementia almost by half in the CSHA. In the CLSA, we did not have enough information to corroborate this association.

Supplementary 1 Full for render in knit - thesis - subgroups2 - after defense.Rmd (250 kB)
R studio script for the analysis

Supplementary 2 cshammcombs2022-03-30T172529.txt (8 kB)
Multimorbidity Combinations CSHA - Regression

Supplementary 3 clsammcombs2022-03-29T002243.txt (50 kB)
Multimorbidity combinations CLSA - Regression

Supplementary 4 CART Models summary R.docx (156 kB)
CART summary output from R

Supplementary 5 Thesis changes at jury request - oct 2022 copy.xlsx (19 kB)
Changes requested by evaluators

2 Form used to extract information for the SR - Multimorbidity association in cognitive impairment.pdf (47 kB)
Extraction form used in the SR