Main findings
Dementia, DM, and COPD were all independently associated with increased mortality, however, highest for dementia. Mortality from combinations of either two or all three diagnoses was higher than mortality from the single diagnoses, and highest when dementia was included. Nevertheless, the interaction effect on mortality was lower for all combinations of two, and for the three, diagnoses, and lowest when dementia was included in the combination.
Comparison with existing literature
In line with our results, previous studies have shown that older adults with dementia have higher mortality compared to those without dementia [3, 27], and even though we found DM and COPD to be individually related to mortality, individuals with dementia were worse off. Likewise, we found an increased mortality for all combinations of two diagnoses with a particularly high mortality for the combination of dementia and COPD. The combination of lung and mental health diagnoses (including dementia) has earlier been found to be hazardous [11]. In our study, having all three diagnoses concurrently (i.e. dementia, DM, and COPD) was related to the highest mortality, and it is well known that mortality increases with the number of conditions among individuals with multimorbidity, i.e. individuals with two or more chronic conditions [11, 28, 29]. The high mortality related to dementia could be explained by the fact that dementia is more prevalent among the oldest age groups, who often contend with concurring conditions and frailty [30]. Furthermore, the concomitant lower ability to maintain and respond to health advices may play a role [6], since neurological and cognitive functions are required to understand and handle symptoms, including navigating in a healthcare system. Finally, another possible explanation for the high mortality in the dementia group could be that during annual control of a chronic condition in general practice (the patients in our cohort can have other chronic conditions apart from DM, COPD, and dementia), there might be an increased risk of finding a new disease, simply because the patient is examined by a doctor. In that light DM and COPD are typically easier and faster to diagnose in general practice compared to dementia, and are likely identified earlier in the disease trajectory, making them less lethal [31]. Furthermore, it is likely that DM and COPD more often occur among younger persons.
On the other hand, regular check-ups for DM or COPD might facilitate earlier detection of dementia, however, hypothetically most likely when DM and COPD are less advanced. Additionally, in research on multimorbidity, dementia is only included in about half of the studies [32], which is why knowledge about the interplay of dementia with other chronic diseases remains relatively scarce. Our study quantifies the effects of the different combinations of chronic conditions, which adds to the knowledge about mortality and dementia.
On the contrary, the interaction effect on mortality was below one for any combination of two comorbidities, with the lowest effect observed when dementia was one of the two. Similar results were found in the aforementioned study on multimorbidity combinations exploring interaction effects, which found lower mortality for mental health diagnoses (where dementia was one of them) in combinations with either lung diagnoses (including COPD) or endocrinological diagnoses (including DM) [11]. However, interestingly the interaction changed direction if the mental health condition appeared before the endocrinological condition, with a resulting slightly higher mortality [33]. In contrast, a British study found no interaction effect among individuals with newly diagnosed stroke in combination with DM and dementia [34]. To the best of our knowledge, no other studies have reported on this interaction effect on mortality for patients with dementia.
The interaction effect could partly be explained by the higher risk of being institutionalized among persons with dementia [35], a risk that seems independent of other comorbidities [36], and maybe partly explained by the strain dementia puts on family and caregivers [37]. Yet, when individuals are finally institutionalized, they are more likely to receive adequate overall health care for their comorbidities, possibly leading to increased general health benefits, compared to those who are not institutionalized. Finally, the effect could also partially be explained by the care management of chronic diseases mentioned above, also relevant for dementia, with yearly or half-yearly controls in general practice. These controls induce a risk of finding new diseases earlier in the disease trajectories where the clinical signs are few, why the coexistence falsely seems protective [31].
Strengths and limitations
One major study strength is that this study was based on a well-defined, large nationwide cohort of all Danish residents. Furthermore, the Danish central registries are overall reliable and well-validated [38], and bias due to non-participation or loss to follow-up can be considered negligible [39]. However, estimates based on hospital diagnoses may underestimate the incidence and prevalence of dementia, DM, and COPD in the general population. To appear in the register patients must be referred to, and visit, a hospital. For example, DM is often well treated and managed in general practice for years or decades and does not necessarily end up in secondary care. To address this underestimation, we have also based our diagnoses on redeemed medications specifically targeting dementia, DM, and COPD in the present study. Despite this, some patients with mild dementia may not have been referred to secondary care, hence not yet received anti dementia medication, and consequently underrepresented in the cohort. Therefore, it is likely that our study is biased towards more severe diseases. However, we believe a higher disease severity would result in excess mortality for disease combinations, which suggest that such bias is unlikely to explain the results of our study.
We used multivariable Cox regression analysis to adjust for possible confounding. Important lifestyle factors, such as high body mass index and smoking, both of which have been associated with DM and COPD, cannot be obtained from the registers, and it is a limitation that residual confounding cannot be ruled out. However, by adjusting for several socioeconomic factors we believe some lifestyle-related confounding effects are accounted for.
Our use of the CCI could result in overcompensation, since adjusting for all other comorbidity effects may reduce the real effects on mortality from dementia, DM, and COPD. However, excluding CCI from the analyses did not change the overall conclusions.
We only analyzed the effect of two comorbidities, DM and COPD, which acted as “model comorbidities” on the mortality of patients with dementia. The comorbidities were chosen because they are common and demanding for both patients and the healthcare system, and therefore likely to be less well-managed by patients with impaired cognition. Still, we cannot rule out that other comorbidities would interact differently with dementia on mortality.
Lastly, it may be a limitation that we only had a five-year follow-up period. Mortality may be different with longer follow-times. That said, a systematic review and a meta-analysis have shown average survival time after dementia diagnosis to be between approximately 4-5.5 years, depending on type of dementia [40], indicating that our five-year follow-up period is reasonable.
Implications
It is well known that patients with dementia have increased mortality. This study shows that dementia, both as single and in combination with other chronic diseases, is related to comparably high mortality. On the other hand, mortality seems lower than expected when dementia co-occurs with DM and COPD, compared to the mortality related to the individual diseases. Although this reduced mortality could be partly explained by diseases found earlier in the disease trajectory it points at something “protecting” in the co-occurrence. This may be due to intensified care for institutionalized patients, the general health benefits of different treatments e.g. smoking cessation, better nutrition, and lipid lowering treatment, as well as increased social contact for these patients. This underlines that there could be a preventable part of the high mortality among patients having dementia in combination with other chronic diseases. Therefore, it is important to improve health factors related both to the development of dementia, but also to chronic diseases [2], e.g. antihypertensive treatment, physical activity, smoking, etc. Consequently, this study adds clinically relevant information valuable to consider when organizing healthcare for patients with dementia.