Patterns of health service use in community living older adults with dementia and comorbid conditions: a population-based retrospective cohort study in Ontario, Canada

This study makes an important contribution to our understanding of health-system burden of community living older adults with dementia. Like many studies, we found that health service use increased substantially with the number of comorbidities, with those with three or more comorbid conditions seeing a GP and specialist more than once a month on average. In a systematic review by Lehnert et al. [38] the authors reported that almost all studies observed a positive association of multiple chronic conditions and health service use and costs, and that several studies found a near exponential relationship between the number of chronic conditions and costs.

We also wanted to examine how the level of comorbidity impacts both dementia-related and non-dementia-related health service utilization, as one could make a case that either one or both could increase as comorbidity increases. We found that overall health service use and costs increased with number of comorbidities, but this was driven mostly by non-dementia-related use. In fact, even in people with no other comorbidities, we found that non-dementia-related use was greater than dementia-related use. The dramatic increase in non-dementia related services may indicate that dementia is negatively impacting self-care and management for other chronic conditions. This could be exacerbated by the under-diagnosis of chronic conditions in older adults [39] especially in those with dementia [40]. However, it may also be partially explained by our data source. For example, for hospitalizations we used the “most responsible diagnosis”, which is based on the diagnosis considered to have most contributed to the overall length of stay, to determine if a hospital episode was dementia-related or non-dementia-related. This could happen if dementia is accelerating other chronic conditions (i.e., impacting non-dementia-related health service use) or increases the likelihood of other acute events, such as falls or pneumonia, the hospitalization may not be directly attributed to dementia or captured in the comorbid conditions that we counted. The most common reasons for hospitalizations for people with AD are syncope, falls and trauma, which would not likely be counted as “dementia-related use” in these analyses [3].

Hypertension and arthritis were the most common comorbidities in the dementia cohort. In a recent systematic review of disease clusters in older adults, Sinnige et al [41] found that hypertension and stroke were the most common diseases to cluster with dementia, although they reported that dementia was considered in only one-fifth of the studies included in the review. Our results are similar to the most common combinations of chronic conditions found in the general population [42]. In a systematic review of patterns of comorbidity in primary care, Violen et al. [43] reported that osteoarthritis and cardiovascular and/or metabolic conditions was the most common combination of chronic conditions. When comparing dementia patients and non-dementia patients in primary care, Schubert et al. [24] found that comorbidity profiles were similar in the two groups, however Bynum et al. [26] did not.

We found that the drivers of costs differed in people with no comorbidity compared to those with three or more chronic conditions. In those with no comorbidity, the biggest driver was homecare and in those with three or more chronic conditions, the biggest driver was acute care episodes. In a 3-year follow-up study, Rudolph et al. [44] found that 66 % of persons with Alzheimer’s disease were hospitalized at least once and 47 % were hospitalized two or more times. Furthermore, even after adjusting for age, gender and other potential confounders, Phelan et al. [6] found that dementia patients were more likely to be hospitalized than non-dementia patients (rate ratio 1.41; 95 % CI 1.23–1.61) especially for ambulatory care-sensitive conditions for which proactive outpatient care may prevent the need for hospital stay (rate ratio 1.78; 95 % CI 1.38–2.31). Similar results were reported by Feng et al. [45] who found that community-living people with dementia were more likely to have potentially preventable hospitalization, an emergency room visit that was potentially avoidable, and an emergency department visit that resulted in a hospitalization, than those without dementia. Our results suggest that this phenomenon occurs increasingly with the level of comorbidity.

The impact of comorbidity on utilization is a particularly important issue for ALC days. The estimates of the percent of ALC patients with dementia range between 25 % [46] to over 60 % [37]. The use of hospital beds for ALC patients can contribute to a decrease in acute care capacity, emergency department overcrowding, and patient flow inefficiencies throughout the entire healthcare system [47]. We found that overall in older adults with dementia that the ALC LOS increased with level of comorbidity, but this again was mostly driven by non-dementia-related hospital stays. This suggests that the impact of dementia on ALC days may be under-estimated in administrative-data based studies and underscores the complexity of these patients. Most ALC patients are waiting for placement in long-term care facilities [37, 48]. A study exploring ALC patients waiting for nursing home admission, however, found that some of these patients could be discharged to a community setting with the support of transitional programs and increased community care [49]. As most people with dementia want to stay at home for as long as possible [50], this would require a greater integration of care between the acute and post-acute care providers [51].

Using comprehensive administrative data, this study represents all community-living older adults in Ontario with dementia. However, this study has limitations. We follow a cohort over time, but have restricted the assessment of comorbidity to the baseline. As we would not be capturing new comorbidities over the 5 years, this would tend to underestimate the impact of comorbidity on utilization. For example, those with 0 comorbidities at baseline could have three or more comorbidities by year 5. As well, we included a restricted list of comorbidities to 14 conditions. This list does, however, include the most common combinations of conditions identified in Canada [42] as well the most common conditions included in multi-morbidity research [52]. Although we were able to present a snapshot of community living older adults with dementia in our cohort, we lacked information on their degree of cognitive impairment. The degree of cognitive impairment is related to healthcare utilization and costs [25] and impacts the ability to self-manage other chronic conditions. We excluded medications from our estimates of utilization and costs. Although medications are an important aspect of healthcare expenditures in older adults, research suggests they contribute a relatively lower proportion of overall healthcare costs in people with dementia. In a study of Medicare beneficiaries, those with Alzheimer’s disease or dementia had higher average annual per-person payments for all healthcare services (hospital, physician and other medical providers, nursing home, and home healthcare) except for prescription medications [3]. Because of our health-service perspective, we did not include utilization or costs associated with informal care which are significant in older adults with dementia [3, 53]. Finally, our classification of health service use into dementia-related and non-dementia-related is based on the most responsible condition. In cases where utilization was for the treatment of sequelae of dementia (e.g., falls or delirium), it may not be counted as dementia-related and thus underestimate dementia-related use. We did, however, find a consistent trend over all types of health service use and different administrative data sources.